Merge #4621

4621: Bring back changes from v1.8.0 into main r=curquiza a=curquiza Co-authored-by: ManyTheFish <many@meilisearch.com> Co-authored-by: Tamo <tamo@meilisearch.com> Co-authored-by: meili-bors[bot] <89034592+meili-bors[bot]@users.noreply.github.com> Co-authored-by: Clément Renault <clement@meilisearch.com>
2025-10-31 16:06:31 +00:00 · 2024-05-06 13:46:39 +00:00
parent ecb5c506b3 c668043c4f
commit 4d5971f343
56 changed files with 1450 additions and 802 deletions
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -354,9 +354,9 @@ dependencies = [
 [[package]]
 name = "anyhow"
-version = "1.0.80"
+version = "1.0.82"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5ad32ce52e4161730f7098c077cd2ed6229b5804ccf99e5366be1ab72a98b4e1"
+checksum = "f538837af36e6f6a9be0faa67f9a314f8119e4e4b5867c6ab40ed60360142519"
 dependencies = [
 "backtrace",
 ]
@@ -889,9 +889,9 @@ dependencies = [
 [[package]]
 name = "charabia"
-version = "0.8.8"
+version = "0.8.10"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "60dc1a562fc8cb53d552d371758a4ecd76d15cc7489d2b968529cd9cadcbd854"
+checksum = "933f20f2269b24d32fd5503e7b3c268af902190daf8d9d2b73ed2e75d77c00b4"
 dependencies = [
 "aho-corasick",
 "cow-utils",
@@ -901,9 +901,7 @@ dependencies = [
 "fst",
 "irg-kvariants",
 "jieba-rs",
- "lindera-core",
+ "lindera",
 "lindera-dictionary",
 "lindera-tokenizer",
 "litemap",
 "once_cell",
 "pinyin",
@@ -1715,9 +1713,9 @@ dependencies = [
 [[package]]
 name = "env_logger"
-version = "0.11.2"
+version = "0.11.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6c012a26a7f605efc424dd53697843a72be7dc86ad2d01f7814337794a12231d"
+checksum = "38b35839ba51819680ba087cd351788c9a3c476841207e0b8cee0b04722343b9"
 dependencies = [
 "anstream",
 "anstyle",
@@ -2661,6 +2659,15 @@ dependencies = [
 "simple_asn1",
 ]
 [[package]]
 name = "kanaria"
 version = "0.2.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "c0f9d9652540055ac4fded998a73aca97d965899077ab1212587437da44196ff"
 dependencies = [
 "bitflags 1.3.2",
 ]
 [[package]]
 name = "kstring"
 version = "2.0.0"
@@ -2766,10 +2773,67 @@ dependencies = [
 ]
 [[package]]
-name = "lindera-cc-cedict-builder"
+name = "lindera"
-version = "0.28.0"
+version = "0.30.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ca21f2ee3ca40e7f3ebbd568d041be1531c2c28dbf540e737aeba934ab53f330"
+checksum = "a1bbf252ea3490053dc397539ece0b510924f2f72605fa28d3e858d86f43ec88"
 dependencies = [
 "lindera-analyzer",
 "lindera-core",
 "lindera-dictionary",
 "lindera-filter",
 "lindera-tokenizer",
 ]
 [[package]]
 name = "lindera-analyzer"
 version = "0.30.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "87febfec0e2859ce2154fb90dd6f66b774ddb0b6e264b44f8e3d1303c9dcedd7"
 dependencies = [
 "anyhow",
 "bincode",
 "byteorder",
 "encoding",
 "kanaria",
 "lindera-cc-cedict-builder",
 "lindera-core",
 "lindera-dictionary",
 "lindera-filter",
 "lindera-ipadic-builder",
 "lindera-ko-dic-builder",
 "lindera-tokenizer",
 "lindera-unidic-builder",
 "once_cell",
 "regex",
 "serde",
 "serde_json",
 "thiserror",
 "unicode-blocks",
 "unicode-normalization",
 "unicode-segmentation",
 "yada",
 ]
 [[package]]
 name = "lindera-cc-cedict"
 version = "0.30.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "fcb91bb8a93ab0f95dbc3c43b5105354bb059134ef731154f75a64b5d919e71d"
 dependencies = [
 "bincode",
 "byteorder",
 "lindera-cc-cedict-builder",
 "lindera-core",
 "lindera-decompress",
 "once_cell",
 ]
 [[package]]
 name = "lindera-cc-cedict-builder"
 version = "0.30.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "f6022a8309a287dbef425fd09a61585351670c83001d74f6c089979e2330b683"
 dependencies = [
 "anyhow",
 "bincode",
@@ -2778,6 +2842,7 @@ dependencies = [
 "encoding",
 "env_logger",
 "glob",
 "lindera-compress",
 "lindera-core",
 "lindera-decompress",
 "log",
@@ -2786,9 +2851,9 @@ dependencies = [
 [[package]]
 name = "lindera-compress"
-version = "0.28.0"
+version = "0.30.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "34da125091f3b3a49351f418484a16cb2a23f6888cd53fe219edad19d263da5d"
+checksum = "32363cbcf433f915e7d77c2a0c410db2d6b23442e80715cf2cf6b9864078a500"
 dependencies = [
 "anyhow",
 "flate2",
@@ -2797,9 +2862,9 @@ dependencies = [
 [[package]]
 name = "lindera-core"
-version = "0.28.0"
+version = "0.30.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "09d4b717a8a31b73a3cbd3552e0abda14e0c85d97dc8b911035342533defdbad"
+checksum = "d9a0e858753a02b1a3524fae4fbb11ca4b3a947128fd7854b797386562678be8"
 dependencies = [
 "anyhow",
 "bincode",
@@ -2814,9 +2879,9 @@ dependencies = [
 [[package]]
 name = "lindera-decompress"
-version = "0.28.0"
+version = "0.30.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "98f4476c99cb4ffa54fbfc42953adf69ada7276cfbb594bce9829547de012058"
+checksum = "0e406345f6f8b665b9a129c67079c18ca9d97e9d171d102b4106a64a592c285e"
 dependencies = [
 "anyhow",
 "flate2",
@@ -2825,29 +2890,73 @@ dependencies = [
 [[package]]
 name = "lindera-dictionary"
-version = "0.28.0"
+version = "0.30.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a45b92f0ce331c2202c6cec3135e4bfce29525ab3bb97a613c27c8e0a29fa967"
+checksum = "3e2a3ec0e5fd6768a27c6ec1040e8470d3a5926418f7afe065859e98aabb3bfe"
 dependencies = [
 "anyhow",
 "bincode",
 "byteorder",
 "lindera-cc-cedict",
 "lindera-cc-cedict-builder",
 "lindera-core",
 "lindera-ipadic",
 "lindera-ipadic-builder",
 "lindera-ipadic-neologd",
 "lindera-ipadic-neologd-builder",
 "lindera-ko-dic",
 "lindera-ko-dic-builder",
 "lindera-unidic",
 "lindera-unidic-builder",
 "serde",
 "strum",
 "strum_macros",
 ]
 [[package]]
 name = "lindera-filter"
 version = "0.30.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "1badaf51bad051185ea4917ba91bbbf2d6f8167e155647e21e0eaaef0982a95d"
 dependencies = [
 "anyhow",
 "csv",
 "kanaria",
 "lindera-cc-cedict-builder",
 "lindera-core",
 "lindera-dictionary",
 "lindera-ipadic-builder",
 "lindera-ko-dic-builder",
 "lindera-unidic-builder",
 "once_cell",
 "regex",
 "serde",
 "serde_json",
 "unicode-blocks",
 "unicode-normalization",
 "unicode-segmentation",
 "yada",
 ]
 [[package]]
 name = "lindera-ipadic"
 version = "0.30.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "129ec16366354998f9791467ad38731539197747f649e573ead845358271ce25"
 dependencies = [
 "bincode",
 "byteorder",
 "lindera-core",
 "lindera-decompress",
 "lindera-ipadic-builder",
 "once_cell",
 ]
 [[package]]
 name = "lindera-ipadic-builder"
-version = "0.28.0"
+version = "0.30.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "642dee52201852df209cb43423ff1ca4d161a329f5cdba049a7b5820118345f2"
+checksum = "7f0979a56bc57e9c9be2996dff232c47aa146a2e7baebf5dd567e388eba3dd90"
 dependencies = [
 "anyhow",
 "bincode",
@@ -2857,6 +2966,7 @@ dependencies = [
 "encoding_rs_io",
 "env_logger",
 "glob",
 "lindera-compress",
 "lindera-core",
 "lindera-decompress",
 "log",
@@ -2865,10 +2975,24 @@ dependencies = [
 ]
 [[package]]
-name = "lindera-ipadic-neologd-builder"
+name = "lindera-ipadic-neologd"
-version = "0.28.0"
+version = "0.30.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "325144b154e68159373e944d1cd7f67c6ff9965a2af41240a8e41732b3fdb3af"
+checksum = "20076660c4e79ef0316735b44e18ec7644e54786acdee8946c972d5f97086d0f"
 dependencies = [
 "bincode",
 "byteorder",
 "lindera-core",
 "lindera-decompress",
 "lindera-ipadic-neologd-builder",
 "once_cell",
 ]
 [[package]]
 name = "lindera-ipadic-neologd-builder"
 version = "0.30.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "eccd18ed5f65d1d64ac0cbfa1d6827bfbbaf6530520ae6847e6a91ee38f47e20"
 dependencies = [
 "anyhow",
 "bincode",
@@ -2878,6 +3002,7 @@ dependencies = [
 "encoding_rs_io",
 "env_logger",
 "glob",
 "lindera-compress",
 "lindera-core",
 "lindera-decompress",
 "log",
@@ -2887,9 +3012,9 @@ dependencies = [
 [[package]]
 name = "lindera-ko-dic"
-version = "0.28.0"
+version = "0.30.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b484a2f9964e7424264fda304beb6ff6ad883c347accfe1115e777dedef3661d"
+checksum = "59073171566c3e498ca048e84c2d0a7e117a42f36c8eb7d7163e65ac38bd6d48"
 dependencies = [
 "bincode",
 "byteorder",
@@ -2900,13 +3025,14 @@ dependencies = [
 "lindera-ko-dic-builder",
 "once_cell",
 "tar",
 "ureq",
 ]
 [[package]]
 name = "lindera-ko-dic-builder"
-version = "0.28.0"
+version = "0.30.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b9413d4d9bf7af921f5ac64414a290c7ba81695e8ba08dd2f6c950b57c281a69"
+checksum = "ae176afa8535ca2a5ee9471873f85d531db0a6c32a3c42b41084506aac22b577"
 dependencies = [
 "anyhow",
 "bincode",
@@ -2924,9 +3050,9 @@ dependencies = [
 [[package]]
 name = "lindera-tokenizer"
-version = "0.28.0"
+version = "0.30.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9987c818462d51ca67e131e40f0386e25e8c557e195059b1257f95731561185d"
+checksum = "457285bdde84571aa510c9e05371904305a55e8a541fa1473d4393062f06932d"
 dependencies = [
 "bincode",
 "lindera-core",
@@ -2938,26 +3064,27 @@ dependencies = [
 [[package]]
 name = "lindera-unidic"
-version = "0.28.0"
+version = "0.30.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0c379cf436b2627cd7d3498642e491eadbff9b3e01231c516ce9f9b1893ab7c3"
+checksum = "5839980be552dfa639b70964c61914a9ad014148663679b0e148aa72e5e30f23"
 dependencies = [
 "bincode",
 "byteorder",
 "encoding",
 "flate2",
 "lindera-core",
 "lindera-decompress",
 "lindera-unidic-builder",
 "once_cell",
 "tar",
 "ureq",
 "zip",
 ]
 [[package]]
 name = "lindera-unidic-builder"
-version = "0.28.0"
+version = "0.30.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "601ec33b5174141396a7a4ca066278863840221fec32d0be19091e7fae91ed94"
+checksum = "dcaab8f061d5b944b1e424f49c7efbf8f276e8a72e4f4ff956d01e46d481f008"
 dependencies = [
 "anyhow",
 "bincode",
@@ -4214,9 +4341,9 @@ dependencies = [
 [[package]]
 name = "regex"
-version = "1.10.2"
+version = "1.10.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "380b951a9c5e80ddfd6136919eef32310721aa4aacd4889a8d39124b026ab343"
+checksum = "c117dbdfde9c8308975b6a18d71f3f385c89461f7b3fb054288ecf2a2058ba4c"
 dependencies = [
 "aho-corasick",
 "memchr",
@@ -4226,9 +4353,9 @@ dependencies = [
 [[package]]
 name = "regex-automata"
-version = "0.4.3"
+version = "0.4.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5f804c7828047e88b2d32e2d7fe5a105da8ee3264f01902f796c8e067dc2483f"
+checksum = "86b83b8b9847f9bf95ef68afb0b8e6cdb80f498442f5179a29fad448fcc1eaea"
 dependencies = [
 "aho-corasick",
 "memchr",
@@ -4795,6 +4922,28 @@ version = "0.10.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "73473c0e59e6d5812c5dfe2a064a6444949f089e20eec9a2e5506596494e4623"
 [[package]]
 name = "strum"
 version = "0.26.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "5d8cec3501a5194c432b2b7976db6b7d10ec95c253208b45f83f7136aa985e29"
 dependencies = [
 "strum_macros",
 ]
 [[package]]
 name = "strum_macros"
 version = "0.26.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "c6cf59daf282c0a494ba14fd21610a0325f9f90ec9d1231dea26bcb1d696c946"
 dependencies = [
 "heck",
 "proc-macro2",
 "quote",
 "rustversion",
 "syn 2.0.58",
 ]
 [[package]]
 name = "subtle"
 version = "2.5.0"
@@ -5324,6 +5473,12 @@ version = "0.3.13"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "92888ba5573ff080736b3648696b70cafad7d250551175acbaa4e0385b3e1460"
 [[package]]
 name = "unicode-blocks"
 version = "0.1.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "6b12e05d9e06373163a9bb6bb8c263c261b396643a99445fe6b9811fd376581b"
 [[package]]
 name = "unicode-ident"
 version = "1.0.12"
@@ -5332,9 +5487,9 @@ checksum = "3354b9ac3fae1ff6755cb6db53683adb661634f67557942dea4facebec0fee4b"
 [[package]]
 name = "unicode-normalization"
-version = "0.1.22"
+version = "0.1.23"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5c5713f0fc4b5db668a2ac63cdb7bb4469d8c9fed047b1d0292cc7b0ce2ba921"
+checksum = "a56d1686db2308d901306f92a263857ef59ea39678a5458e7cb17f01415101f5"
 dependencies = [
 "tinyvec",
 ]
@@ -5350,9 +5505,9 @@ dependencies = [
 [[package]]
 name = "unicode-segmentation"
-version = "1.10.1"
+version = "1.11.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1dd624098567895118886609431a7c3b8f516e41d30e0643f03d94592a147e36"
+checksum = "d4c87d22b6e3f4a18d4d40ef354e97c90fcb14dd91d7dc0aa9d8a1172ebf7202"
 [[package]]
 name = "unicode-width"
@@ -5942,9 +6097,9 @@ dependencies = [
 [[package]]
 name = "yada"
-version = "0.5.0"
+version = "0.5.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b6d12cb7a57bbf2ab670ed9545bae3648048547f9039279a89ce000208e585c1"
+checksum = "aed111bd9e48a802518765906cbdadf0b45afb72b9c81ab049a3b86252adffdd"
 [[package]]
 name = "yaml-rust"
--- a/dump/src/lib.rs
+++ b/dump/src/lib.rs
@@ -256,8 +256,8 @@ pub(crate) mod test {
    pub fn create_test_settings() -> Settings<Checked> {
        let settings = Settings {
-            displayed_attributes: Setting::Set(vec![S("race"), S("name")]),
+            displayed_attributes: Setting::Set(vec![S("race"), S("name")]).into(),
-            searchable_attributes: Setting::Set(vec![S("name"), S("race")]),
+            searchable_attributes: Setting::Set(vec![S("name"), S("race")]).into(),
            filterable_attributes: Setting::Set(btreeset! { S("race"), S("age") }),
            sortable_attributes: Setting::Set(btreeset! { S("age") }),
            ranking_rules: Setting::NotSet,
--- a/dump/src/reader/compat/v5_to_v6.rs
+++ b/dump/src/reader/compat/v5_to_v6.rs
@@ -315,8 +315,8 @@ impl From<v5::ResponseError> for v6::ResponseError {
 impl<T> From<v5::Settings<T>> for v6::Settings<v6::Unchecked> {
    fn from(settings: v5::Settings<T>) -> Self {
        v6::Settings {
-            displayed_attributes: settings.displayed_attributes.into(),
+            displayed_attributes: v6::Setting::from(settings.displayed_attributes).into(),
-            searchable_attributes: settings.searchable_attributes.into(),
+            searchable_attributes: v6::Setting::from(settings.searchable_attributes).into(),
            filterable_attributes: settings.filterable_attributes.into(),
            sortable_attributes: settings.sortable_attributes.into(),
            ranking_rules: {
--- a/index-scheduler/src/lib.rs
+++ b/index-scheduler/src/lib.rs
@@ -3041,6 +3041,7 @@ mod tests {
            source: Setting::Set(milli::vector::settings::EmbedderSource::Rest),
            api_key: Setting::Set(S("My super secret")),
            url: Setting::Set(S("http://localhost:7777")),
            dimensions: Setting::Set(4),
            ..Default::default()
        };
        embedders.insert(S("default"), Setting::Set(embedding_settings));
--- a/index-scheduler/src/snapshots/index_schedulertestssettings_update-2.snap
+++ b/index-scheduler/src/snapshots/index_schedulertestssettings_update-2.snap
@@ -7,6 +7,7 @@ expression: task.details
    "default": {
      "source": "rest",
      "apiKey": "MyXXXX...",
      "dimensions": 4,
      "url": "http://localhost:7777"
    }
  }
--- a/index-scheduler/src/snapshots/index_schedulertestssettings_update-3.snap
+++ b/index-scheduler/src/snapshots/index_schedulertestssettings_update-3.snap
@@ -6,7 +6,7 @@ expression: embedding_config.embedder_options
  "Rest": {
    "api_key": "My super secret",
    "distribution": null,
-    "dimensions": null,
+    "dimensions": 4,
    "url": "http://localhost:7777",
    "query": null,
    "input_field": [
--- a/index-scheduler/src/snapshots/index_schedulertestssettings_update.snap
+++ b/index-scheduler/src/snapshots/index_schedulertestssettings_update.snap
@@ -7,6 +7,7 @@ expression: task.details
    "default": {
      "source": "rest",
      "apiKey": "MyXXXX...",
      "dimensions": 4,
      "url": "http://localhost:7777"
    }
  }
--- a/index-scheduler/src/snapshots/lib.rs/test_settings_update/after_registering_settings_task.snap
+++ b/index-scheduler/src/snapshots/lib.rs/test_settings_update/after_registering_settings_task.snap
@@ -6,7 +6,7 @@ source: index-scheduler/src/lib.rs
 []
 ----------------------------------------------------------------------
 ### All Tasks:
-0 {uid: 0, status: enqueued, details: { settings: Settings { displayed_attributes: NotSet, searchable_attributes: NotSet, filterable_attributes: NotSet, sortable_attributes: NotSet, ranking_rules: NotSet, stop_words: NotSet, non_separator_tokens: NotSet, separator_tokens: NotSet, dictionary: NotSet, synonyms: NotSet, distinct_attribute: NotSet, proximity_precision: NotSet, typo_tolerance: NotSet, faceting: NotSet, pagination: NotSet, embedders: Set({"default": Set(EmbeddingSettings { source: Set(Rest), model: NotSet, revision: NotSet, api_key: Set("My super secret"), dimensions: NotSet, document_template: NotSet, url: Set("http://localhost:7777"), query: NotSet, input_field: NotSet, path_to_embeddings: NotSet, embedding_object: NotSet, input_type: NotSet, distribution: NotSet })}), search_cutoff_ms: NotSet, _kind: PhantomData<meilisearch_types::settings::Unchecked> } }, kind: SettingsUpdate { index_uid: "doggos", new_settings: Settings { displayed_attributes: NotSet, searchable_attributes: NotSet, filterable_attributes: NotSet, sortable_attributes: NotSet, ranking_rules: NotSet, stop_words: NotSet, non_separator_tokens: NotSet, separator_tokens: NotSet, dictionary: NotSet, synonyms: NotSet, distinct_attribute: NotSet, proximity_precision: NotSet, typo_tolerance: NotSet, faceting: NotSet, pagination: NotSet, embedders: Set({"default": Set(EmbeddingSettings { source: Set(Rest), model: NotSet, revision: NotSet, api_key: Set("My super secret"), dimensions: NotSet, document_template: NotSet, url: Set("http://localhost:7777"), query: NotSet, input_field: NotSet, path_to_embeddings: NotSet, embedding_object: NotSet, input_type: NotSet, distribution: NotSet })}), search_cutoff_ms: NotSet, _kind: PhantomData<meilisearch_types::settings::Unchecked> }, is_deletion: false, allow_index_creation: true }}
+0 {uid: 0, status: enqueued, details: { settings: Settings { displayed_attributes: WildcardSetting(NotSet), searchable_attributes: WildcardSetting(NotSet), filterable_attributes: NotSet, sortable_attributes: NotSet, ranking_rules: NotSet, stop_words: NotSet, non_separator_tokens: NotSet, separator_tokens: NotSet, dictionary: NotSet, synonyms: NotSet, distinct_attribute: NotSet, proximity_precision: NotSet, typo_tolerance: NotSet, faceting: NotSet, pagination: NotSet, embedders: Set({"default": Set(EmbeddingSettings { source: Set(Rest), model: NotSet, revision: NotSet, api_key: Set("My super secret"), dimensions: Set(4), document_template: NotSet, url: Set("http://localhost:7777"), query: NotSet, input_field: NotSet, path_to_embeddings: NotSet, embedding_object: NotSet, input_type: NotSet, distribution: NotSet })}), search_cutoff_ms: NotSet, _kind: PhantomData<meilisearch_types::settings::Unchecked> } }, kind: SettingsUpdate { index_uid: "doggos", new_settings: Settings { displayed_attributes: WildcardSetting(NotSet), searchable_attributes: WildcardSetting(NotSet), filterable_attributes: NotSet, sortable_attributes: NotSet, ranking_rules: NotSet, stop_words: NotSet, non_separator_tokens: NotSet, separator_tokens: NotSet, dictionary: NotSet, synonyms: NotSet, distinct_attribute: NotSet, proximity_precision: NotSet, typo_tolerance: NotSet, faceting: NotSet, pagination: NotSet, embedders: Set({"default": Set(EmbeddingSettings { source: Set(Rest), model: NotSet, revision: NotSet, api_key: Set("My super secret"), dimensions: Set(4), document_template: NotSet, url: Set("http://localhost:7777"), query: NotSet, input_field: NotSet, path_to_embeddings: NotSet, embedding_object: NotSet, input_type: NotSet, distribution: NotSet })}), search_cutoff_ms: NotSet, _kind: PhantomData<meilisearch_types::settings::Unchecked> }, is_deletion: false, allow_index_creation: true }}
 ----------------------------------------------------------------------
 ### Status:
 enqueued [0,]
--- a/index-scheduler/src/snapshots/lib.rs/test_settings_update/settings_update_processed.snap
+++ b/index-scheduler/src/snapshots/lib.rs/test_settings_update/settings_update_processed.snap
@@ -6,7 +6,7 @@ source: index-scheduler/src/lib.rs
 []
 ----------------------------------------------------------------------
 ### All Tasks:
-0 {uid: 0, status: succeeded, details: { settings: Settings { displayed_attributes: NotSet, searchable_attributes: NotSet, filterable_attributes: NotSet, sortable_attributes: NotSet, ranking_rules: NotSet, stop_words: NotSet, non_separator_tokens: NotSet, separator_tokens: NotSet, dictionary: NotSet, synonyms: NotSet, distinct_attribute: NotSet, proximity_precision: NotSet, typo_tolerance: NotSet, faceting: NotSet, pagination: NotSet, embedders: Set({"default": Set(EmbeddingSettings { source: Set(Rest), model: NotSet, revision: NotSet, api_key: Set("My super secret"), dimensions: NotSet, document_template: NotSet, url: Set("http://localhost:7777"), query: NotSet, input_field: NotSet, path_to_embeddings: NotSet, embedding_object: NotSet, input_type: NotSet, distribution: NotSet })}), search_cutoff_ms: NotSet, _kind: PhantomData<meilisearch_types::settings::Unchecked> } }, kind: SettingsUpdate { index_uid: "doggos", new_settings: Settings { displayed_attributes: NotSet, searchable_attributes: NotSet, filterable_attributes: NotSet, sortable_attributes: NotSet, ranking_rules: NotSet, stop_words: NotSet, non_separator_tokens: NotSet, separator_tokens: NotSet, dictionary: NotSet, synonyms: NotSet, distinct_attribute: NotSet, proximity_precision: NotSet, typo_tolerance: NotSet, faceting: NotSet, pagination: NotSet, embedders: Set({"default": Set(EmbeddingSettings { source: Set(Rest), model: NotSet, revision: NotSet, api_key: Set("My super secret"), dimensions: NotSet, document_template: NotSet, url: Set("http://localhost:7777"), query: NotSet, input_field: NotSet, path_to_embeddings: NotSet, embedding_object: NotSet, input_type: NotSet, distribution: NotSet })}), search_cutoff_ms: NotSet, _kind: PhantomData<meilisearch_types::settings::Unchecked> }, is_deletion: false, allow_index_creation: true }}
+0 {uid: 0, status: succeeded, details: { settings: Settings { displayed_attributes: WildcardSetting(NotSet), searchable_attributes: WildcardSetting(NotSet), filterable_attributes: NotSet, sortable_attributes: NotSet, ranking_rules: NotSet, stop_words: NotSet, non_separator_tokens: NotSet, separator_tokens: NotSet, dictionary: NotSet, synonyms: NotSet, distinct_attribute: NotSet, proximity_precision: NotSet, typo_tolerance: NotSet, faceting: NotSet, pagination: NotSet, embedders: Set({"default": Set(EmbeddingSettings { source: Set(Rest), model: NotSet, revision: NotSet, api_key: Set("My super secret"), dimensions: Set(4), document_template: NotSet, url: Set("http://localhost:7777"), query: NotSet, input_field: NotSet, path_to_embeddings: NotSet, embedding_object: NotSet, input_type: NotSet, distribution: NotSet })}), search_cutoff_ms: NotSet, _kind: PhantomData<meilisearch_types::settings::Unchecked> } }, kind: SettingsUpdate { index_uid: "doggos", new_settings: Settings { displayed_attributes: WildcardSetting(NotSet), searchable_attributes: WildcardSetting(NotSet), filterable_attributes: NotSet, sortable_attributes: NotSet, ranking_rules: NotSet, stop_words: NotSet, non_separator_tokens: NotSet, separator_tokens: NotSet, dictionary: NotSet, synonyms: NotSet, distinct_attribute: NotSet, proximity_precision: NotSet, typo_tolerance: NotSet, faceting: NotSet, pagination: NotSet, embedders: Set({"default": Set(EmbeddingSettings { source: Set(Rest), model: NotSet, revision: NotSet, api_key: Set("My super secret"), dimensions: Set(4), document_template: NotSet, url: Set("http://localhost:7777"), query: NotSet, input_field: NotSet, path_to_embeddings: NotSet, embedding_object: NotSet, input_type: NotSet, distribution: NotSet })}), search_cutoff_ms: NotSet, _kind: PhantomData<meilisearch_types::settings::Unchecked> }, is_deletion: false, allow_index_creation: true }}
 ----------------------------------------------------------------------
 ### Status:
 enqueued []
--- a/meilisearch-types/Cargo.toml
+++ b/meilisearch-types/Cargo.toml
@@ -44,6 +44,7 @@ all-tokenizations = ["milli/all-tokenizations"]
 # chinese specialized tokenization
 chinese = ["milli/chinese"]
 chinese-pinyin = ["milli/chinese-pinyin"]
 # hebrew specialized tokenization
 hebrew = ["milli/hebrew"]
 # japanese specialized tokenization
@@ -56,3 +57,5 @@ greek = ["milli/greek"]
 khmer = ["milli/khmer"]
 # allow vietnamese specialized tokenization
 vietnamese = ["milli/vietnamese"]
 # force swedish character recomposition
 swedish-recomposition = ["milli/swedish-recomposition"]
--- a/meilisearch-types/src/settings.rs
+++ b/meilisearch-types/src/settings.rs
@@ -3,7 +3,7 @@ use std::convert::Infallible;
 use std::fmt;
 use std::marker::PhantomData;
 use std::num::NonZeroUsize;
-use std::ops::ControlFlow;
+use std::ops::{ControlFlow, Deref};
 use std::str::FromStr;
 use deserr::{DeserializeError, Deserr, ErrorKind, MergeWithError, ValuePointerRef};
@@ -143,21 +143,13 @@ impl MergeWithError<milli::CriterionError> for DeserrJsonError<InvalidSettingsRa
 )]
 #[deserr(error = DeserrJsonError, rename_all = camelCase, deny_unknown_fields)]
 pub struct Settings<T> {
-    #[serde(
+    #[serde(default, skip_serializing_if = "Setting::is_not_set")]
        default,
        serialize_with = "serialize_with_wildcard",
        skip_serializing_if = "Setting::is_not_set"
    )]
    #[deserr(default, error = DeserrJsonError<InvalidSettingsDisplayedAttributes>)]
-    pub displayed_attributes: Setting<Vec<String>>,
+    pub displayed_attributes: WildcardSetting,
-    #[serde(
+    #[serde(default, skip_serializing_if = "Setting::is_not_set")]
        default,
        serialize_with = "serialize_with_wildcard",
        skip_serializing_if = "Setting::is_not_set"
    )]
    #[deserr(default, error = DeserrJsonError<InvalidSettingsSearchableAttributes>)]
-    pub searchable_attributes: Setting<Vec<String>>,
+    pub searchable_attributes: WildcardSetting,
    #[serde(default, skip_serializing_if = "Setting::is_not_set")]
    #[deserr(default, error = DeserrJsonError<InvalidSettingsFilterableAttributes>)]
@@ -251,8 +243,8 @@ impl<T> Settings<T> {
 impl Settings<Checked> {
    pub fn cleared() -> Settings<Checked> {
        Settings {
-            displayed_attributes: Setting::Reset,
+            displayed_attributes: Setting::Reset.into(),
-            searchable_attributes: Setting::Reset,
+            searchable_attributes: Setting::Reset.into(),
            filterable_attributes: Setting::Reset,
            sortable_attributes: Setting::Reset,
            ranking_rules: Setting::Reset,
@@ -319,7 +311,7 @@ impl Settings<Checked> {
 impl Settings<Unchecked> {
    pub fn check(self) -> Settings<Checked> {
-        let displayed_attributes = match self.displayed_attributes {
+        let displayed_attributes = match self.displayed_attributes.0 {
            Setting::Set(fields) => {
                if fields.iter().any(|f| f == "*") {
                    Setting::Reset
@@ -330,7 +322,7 @@ impl Settings<Unchecked> {
            otherwise => otherwise,
        };
-        let searchable_attributes = match self.searchable_attributes {
+        let searchable_attributes = match self.searchable_attributes.0 {
            Setting::Set(fields) => {
                if fields.iter().any(|f| f == "*") {
                    Setting::Reset
@@ -342,8 +334,8 @@ impl Settings<Unchecked> {
        };
        Settings {
-            displayed_attributes,
+            displayed_attributes: displayed_attributes.into(),
-            searchable_attributes,
+            searchable_attributes: searchable_attributes.into(),
            filterable_attributes: self.filterable_attributes,
            sortable_attributes: self.sortable_attributes,
            ranking_rules: self.ranking_rules,
@@ -412,13 +404,13 @@ pub fn apply_settings_to_builder(
        _kind,
    } = settings;
-    match searchable_attributes {
+    match searchable_attributes.deref() {
        Setting::Set(ref names) => builder.set_searchable_fields(names.clone()),
        Setting::Reset => builder.reset_searchable_fields(),
        Setting::NotSet => (),
    }
-    match displayed_attributes {
+    match displayed_attributes.deref() {
        Setting::Set(ref names) => builder.set_displayed_fields(names.clone()),
        Setting::Reset => builder.reset_displayed_fields(),
        Setting::NotSet => (),
@@ -690,11 +682,13 @@ pub fn settings(
        displayed_attributes: match displayed_attributes {
            Some(attrs) => Setting::Set(attrs),
            None => Setting::Reset,
-        },
+        }
        .into(),
        searchable_attributes: match searchable_attributes {
            Some(attrs) => Setting::Set(attrs),
            None => Setting::Reset,
-        },
+        }
        .into(),
        filterable_attributes: Setting::Set(filterable_attributes),
        sortable_attributes: Setting::Set(sortable_attributes),
        ranking_rules: Setting::Set(criteria.iter().map(|c| c.clone().into()).collect()),
@@ -848,6 +842,41 @@ impl From<ProximityPrecisionView> for ProximityPrecision {
    }
 }
 #[derive(Debug, Clone, Default, Deserialize, PartialEq, Eq)]
 pub struct WildcardSetting(Setting<Vec<String>>);
 impl From<Setting<Vec<String>>> for WildcardSetting {
    fn from(setting: Setting<Vec<String>>) -> Self {
        Self(setting)
    }
 }
 impl Serialize for WildcardSetting {
    fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
    where
        S: Serializer,
    {
        serialize_with_wildcard(&self.0, serializer)
    }
 }
 impl<E: deserr::DeserializeError> Deserr<E> for WildcardSetting {
    fn deserialize_from_value<V: deserr::IntoValue>(
        value: deserr::Value<V>,
        location: ValuePointerRef<'_>,
    ) -> Result<Self, E> {
        Ok(Self(Setting::deserialize_from_value(value, location)?))
    }
 }
 impl std::ops::Deref for WildcardSetting {
    type Target = Setting<Vec<String>>;
    fn deref(&self) -> &Self::Target {
        &self.0
    }
 }
 #[cfg(test)]
 pub(crate) mod test {
    use super::*;
@@ -856,8 +885,8 @@ pub(crate) mod test {
    fn test_setting_check() {
        // test no changes
        let settings = Settings {
-            displayed_attributes: Setting::Set(vec![String::from("hello")]),
+            displayed_attributes: Setting::Set(vec![String::from("hello")]).into(),
-            searchable_attributes: Setting::Set(vec![String::from("hello")]),
+            searchable_attributes: Setting::Set(vec![String::from("hello")]).into(),
            filterable_attributes: Setting::NotSet,
            sortable_attributes: Setting::NotSet,
            ranking_rules: Setting::NotSet,
@@ -883,8 +912,9 @@ pub(crate) mod test {
        // test wildcard
        // test no changes
        let settings = Settings {
-            displayed_attributes: Setting::Set(vec![String::from("*")]),
+            displayed_attributes: Setting::Set(vec![String::from("*")]).into(),
-            searchable_attributes: Setting::Set(vec![String::from("hello"), String::from("*")]),
+            searchable_attributes: Setting::Set(vec![String::from("hello"), String::from("*")])
                .into(),
            filterable_attributes: Setting::NotSet,
            sortable_attributes: Setting::NotSet,
            ranking_rules: Setting::NotSet,
@@ -904,7 +934,7 @@ pub(crate) mod test {
        };
        let checked = settings.check();
-        assert_eq!(checked.displayed_attributes, Setting::Reset);
+        assert_eq!(checked.displayed_attributes, Setting::Reset.into());
-        assert_eq!(checked.searchable_attributes, Setting::Reset);
+        assert_eq!(checked.searchable_attributes, Setting::Reset.into());
    }
 }
--- a/meilisearch/Cargo.toml
+++ b/meilisearch/Cargo.toml
@@ -149,12 +149,14 @@ mini-dashboard = [
    "zip",
 ]
 chinese = ["meilisearch-types/chinese"]
 chinese-pinyin = ["meilisearch-types/chinese-pinyin"]
 hebrew = ["meilisearch-types/hebrew"]
 japanese = ["meilisearch-types/japanese"]
 thai = ["meilisearch-types/thai"]
 greek = ["meilisearch-types/greek"]
 khmer = ["meilisearch-types/khmer"]
 vietnamese = ["meilisearch-types/vietnamese"]
 swedish-recomposition = ["meilisearch-types/swedish-recomposition"]
 [package.metadata.mini-dashboard]
 assets-url = "https://github.com/meilisearch/mini-dashboard/releases/download/v0.2.13/build.zip"
--- a/meilisearch/src/analytics/mock_analytics.rs
+++ b/meilisearch/src/analytics/mock_analytics.rs
@@ -7,7 +7,6 @@ use serde_json::Value;
 use super::{find_user_id, Analytics, DocumentDeletionKind, DocumentFetchKind};
 use crate::routes::indexes::documents::UpdateDocumentsQuery;
 use crate::routes::tasks::TasksFilterQuery;
 use crate::Opt;
 pub struct MockAnalytics {
@@ -86,6 +85,4 @@ impl Analytics for MockAnalytics {
    }
    fn get_fetch_documents(&self, _documents_query: &DocumentFetchKind, _request: &HttpRequest) {}
    fn post_fetch_documents(&self, _documents_query: &DocumentFetchKind, _request: &HttpRequest) {}
    fn get_tasks(&self, _query: &TasksFilterQuery, _request: &HttpRequest) {}
    fn health_seen(&self, _request: &HttpRequest) {}
 }
--- a/meilisearch/src/analytics/mod.rs
+++ b/meilisearch/src/analytics/mod.rs
@@ -14,7 +14,6 @@ use platform_dirs::AppDirs;
 use serde_json::Value;
 use crate::routes::indexes::documents::UpdateDocumentsQuery;
 use crate::routes::tasks::TasksFilterQuery;
 // if the analytics feature is disabled
 // the `SegmentAnalytics` point to the mock instead of the real analytics
@@ -117,10 +116,4 @@ pub trait Analytics: Sync + Send {
        index_creation: bool,
        request: &HttpRequest,
    );
    // this method should be called to aggregate the get tasks requests.
    fn get_tasks(&self, query: &TasksFilterQuery, request: &HttpRequest);
    // this method should be called to aggregate a add documents request
    fn health_seen(&self, request: &HttpRequest);
 }
--- a/meilisearch/src/analytics/segment_analytics.rs
+++ b/meilisearch/src/analytics/segment_analytics.rs
@@ -33,7 +33,6 @@ use crate::option::{
 };
 use crate::routes::indexes::documents::UpdateDocumentsQuery;
 use crate::routes::indexes::facet_search::FacetSearchQuery;
 use crate::routes::tasks::TasksFilterQuery;
 use crate::routes::{create_all_stats, Stats};
 use crate::search::{
    FacetSearchResult, MatchingStrategy, SearchQuery, SearchQueryWithIndex, SearchResult,
@@ -81,8 +80,6 @@ pub enum AnalyticsMsg {
    AggregateUpdateDocuments(DocumentsAggregator),
    AggregateGetFetchDocuments(DocumentsFetchAggregator),
    AggregatePostFetchDocuments(DocumentsFetchAggregator),
    AggregateTasks(TasksAggregator),
    AggregateHealth(HealthAggregator),
 }
 pub struct SegmentAnalytics {
@@ -152,8 +149,6 @@ impl SegmentAnalytics {
            update_documents_aggregator: DocumentsAggregator::default(),
            get_fetch_documents_aggregator: DocumentsFetchAggregator::default(),
            post_fetch_documents_aggregator: DocumentsFetchAggregator::default(),
            get_tasks_aggregator: TasksAggregator::default(),
            health_aggregator: HealthAggregator::default(),
        });
        tokio::spawn(segment.run(index_scheduler.clone(), auth_controller.clone()));
@@ -231,16 +226,6 @@ impl super::Analytics for SegmentAnalytics {
        let aggregate = DocumentsFetchAggregator::from_query(documents_query, request);
        let _ = self.sender.try_send(AnalyticsMsg::AggregatePostFetchDocuments(aggregate));
    }
    fn get_tasks(&self, query: &TasksFilterQuery, request: &HttpRequest) {
        let aggregate = TasksAggregator::from_query(query, request);
        let _ = self.sender.try_send(AnalyticsMsg::AggregateTasks(aggregate));
    }
    fn health_seen(&self, request: &HttpRequest) {
        let aggregate = HealthAggregator::from_query(request);
        let _ = self.sender.try_send(AnalyticsMsg::AggregateHealth(aggregate));
    }
 }
 /// This structure represent the `infos` field we send in the analytics.
@@ -394,8 +379,6 @@ pub struct Segment {
    update_documents_aggregator: DocumentsAggregator,
    get_fetch_documents_aggregator: DocumentsFetchAggregator,
    post_fetch_documents_aggregator: DocumentsFetchAggregator,
    get_tasks_aggregator: TasksAggregator,
    health_aggregator: HealthAggregator,
 }
 impl Segment {
@@ -458,8 +441,6 @@ impl Segment {
                        Some(AnalyticsMsg::AggregateUpdateDocuments(agreg)) => self.update_documents_aggregator.aggregate(agreg),
                        Some(AnalyticsMsg::AggregateGetFetchDocuments(agreg)) => self.get_fetch_documents_aggregator.aggregate(agreg),
                        Some(AnalyticsMsg::AggregatePostFetchDocuments(agreg)) => self.post_fetch_documents_aggregator.aggregate(agreg),
                        Some(AnalyticsMsg::AggregateTasks(agreg)) => self.get_tasks_aggregator.aggregate(agreg),
                        Some(AnalyticsMsg::AggregateHealth(agreg)) => self.health_aggregator.aggregate(agreg),
                        None => (),
                    }
                }
@@ -513,8 +494,6 @@ impl Segment {
            update_documents_aggregator,
            get_fetch_documents_aggregator,
            post_fetch_documents_aggregator,
            get_tasks_aggregator,
            health_aggregator,
        } = self;
        if let Some(get_search) =
@@ -562,12 +541,6 @@ impl Segment {
        {
            let _ = self.batcher.push(post_fetch_documents).await;
        }
        if let Some(get_tasks) = take(get_tasks_aggregator).into_event(user, "Tasks Seen") {
            let _ = self.batcher.push(get_tasks).await;
        }
        if let Some(health) = take(health_aggregator).into_event(user, "Health Seen") {
            let _ = self.batcher.push(health).await;
        }
        let _ = self.batcher.flush().await;
    }
 }
@@ -1503,176 +1476,6 @@ impl DocumentsDeletionAggregator {
    }
 }
 #[derive(Default, Serialize)]
 pub struct TasksAggregator {
    #[serde(skip)]
    timestamp: Option<OffsetDateTime>,
    // context
    #[serde(rename = "user-agent")]
    user_agents: HashSet<String>,
    filtered_by_uid: bool,
    filtered_by_index_uid: bool,
    filtered_by_type: bool,
    filtered_by_status: bool,
    filtered_by_canceled_by: bool,
    filtered_by_before_enqueued_at: bool,
    filtered_by_after_enqueued_at: bool,
    filtered_by_before_started_at: bool,
    filtered_by_after_started_at: bool,
    filtered_by_before_finished_at: bool,
    filtered_by_after_finished_at: bool,
    total_received: usize,
 }
 impl TasksAggregator {
    pub fn from_query(query: &TasksFilterQuery, request: &HttpRequest) -> Self {
        let TasksFilterQuery {
            limit: _,
            from: _,
            uids,
            index_uids,
            types,
            statuses,
            canceled_by,
            before_enqueued_at,
            after_enqueued_at,
            before_started_at,
            after_started_at,
            before_finished_at,
            after_finished_at,
        } = query;
        Self {
            timestamp: Some(OffsetDateTime::now_utc()),
            user_agents: extract_user_agents(request).into_iter().collect(),
            filtered_by_uid: uids.is_some(),
            filtered_by_index_uid: index_uids.is_some(),
            filtered_by_type: types.is_some(),
            filtered_by_status: statuses.is_some(),
            filtered_by_canceled_by: canceled_by.is_some(),
            filtered_by_before_enqueued_at: before_enqueued_at.is_some(),
            filtered_by_after_enqueued_at: after_enqueued_at.is_some(),
            filtered_by_before_started_at: before_started_at.is_some(),
            filtered_by_after_started_at: after_started_at.is_some(),
            filtered_by_before_finished_at: before_finished_at.is_some(),
            filtered_by_after_finished_at: after_finished_at.is_some(),
            total_received: 1,
        }
    }
    /// Aggregate one [TasksAggregator] into another.
    pub fn aggregate(&mut self, other: Self) {
        let Self {
            timestamp,
            user_agents,
            total_received,
            filtered_by_uid,
            filtered_by_index_uid,
            filtered_by_type,
            filtered_by_status,
            filtered_by_canceled_by,
            filtered_by_before_enqueued_at,
            filtered_by_after_enqueued_at,
            filtered_by_before_started_at,
            filtered_by_after_started_at,
            filtered_by_before_finished_at,
            filtered_by_after_finished_at,
        } = other;
        if self.timestamp.is_none() {
            self.timestamp = timestamp;
        }
        // we can't create a union because there is no `into_union` method
        for user_agent in user_agents {
            self.user_agents.insert(user_agent);
        }
        self.filtered_by_uid |= filtered_by_uid;
        self.filtered_by_index_uid |= filtered_by_index_uid;
        self.filtered_by_type |= filtered_by_type;
        self.filtered_by_status |= filtered_by_status;
        self.filtered_by_canceled_by |= filtered_by_canceled_by;
        self.filtered_by_before_enqueued_at |= filtered_by_before_enqueued_at;
        self.filtered_by_after_enqueued_at |= filtered_by_after_enqueued_at;
        self.filtered_by_before_started_at |= filtered_by_before_started_at;
        self.filtered_by_after_started_at |= filtered_by_after_started_at;
        self.filtered_by_before_finished_at |= filtered_by_before_finished_at;
        self.filtered_by_after_finished_at |= filtered_by_after_finished_at;
        self.filtered_by_after_finished_at |= filtered_by_after_finished_at;
        self.total_received = self.total_received.saturating_add(total_received);
    }
    pub fn into_event(self, user: &User, event_name: &str) -> Option<Track> {
        // if we had no timestamp it means we never encountered any events and
        // thus we don't need to send this event.
        let timestamp = self.timestamp?;
        Some(Track {
            timestamp: Some(timestamp),
            user: user.clone(),
            event: event_name.to_string(),
            properties: serde_json::to_value(self).ok()?,
            ..Default::default()
        })
    }
 }
 #[derive(Default, Serialize)]
 pub struct HealthAggregator {
    #[serde(skip)]
    timestamp: Option<OffsetDateTime>,
    // context
    #[serde(rename = "user-agent")]
    user_agents: HashSet<String>,
    #[serde(rename = "requests.total_received")]
    total_received: usize,
 }
 impl HealthAggregator {
    pub fn from_query(request: &HttpRequest) -> Self {
        Self {
            timestamp: Some(OffsetDateTime::now_utc()),
            user_agents: extract_user_agents(request).into_iter().collect(),
            total_received: 1,
        }
    }
    /// Aggregate one [HealthAggregator] into another.
    pub fn aggregate(&mut self, other: Self) {
        let Self { timestamp, user_agents, total_received } = other;
        if self.timestamp.is_none() {
            self.timestamp = timestamp;
        }
        // we can't create a union because there is no `into_union` method
        for user_agent in user_agents {
            self.user_agents.insert(user_agent);
        }
        self.total_received = self.total_received.saturating_add(total_received);
    }
    pub fn into_event(self, user: &User, event_name: &str) -> Option<Track> {
        // if we had no timestamp it means we never encountered any events and
        // thus we don't need to send this event.
        let timestamp = self.timestamp?;
        Some(Track {
            timestamp: Some(timestamp),
            user: user.clone(),
            event: event_name.to_string(),
            properties: serde_json::to_value(self).ok()?,
            ..Default::default()
        })
    }
 }
 #[derive(Default, Serialize)]
 pub struct DocumentsFetchAggregator {
    #[serde(skip)]
--- a/meilisearch/src/option.rs
+++ b/meilisearch/src/option.rs
@@ -13,6 +13,7 @@ use byte_unit::{Byte, ByteError};
 use clap::Parser;
 use meilisearch_types::features::InstanceTogglableFeatures;
 use meilisearch_types::milli::update::IndexerConfig;
 use meilisearch_types::milli::ThreadPoolNoAbortBuilder;
 use rustls::server::{
    AllowAnyAnonymousOrAuthenticatedClient, AllowAnyAuthenticatedClient, ServerSessionMemoryCache,
 };
@@ -666,7 +667,7 @@ impl TryFrom<&IndexerOpts> for IndexerConfig {
    type Error = anyhow::Error;
    fn try_from(other: &IndexerOpts) -> Result<Self, Self::Error> {
-        let thread_pool = rayon::ThreadPoolBuilder::new()
+        let thread_pool = ThreadPoolNoAbortBuilder::new()
            .thread_name(|index| format!("indexing-thread:{index}"))
            .num_threads(*other.max_indexing_threads)
            .build()?;
--- a/meilisearch/src/routes/indexes/mod.rs
+++ b/meilisearch/src/routes/indexes/mod.rs
@@ -269,12 +269,8 @@ impl From<index_scheduler::IndexStats> for IndexStats {
 pub async fn get_index_stats(
    index_scheduler: GuardedData<ActionPolicy<{ actions::STATS_GET }>, Data<IndexScheduler>>,
    index_uid: web::Path<String>,
    req: HttpRequest,
    analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
    let index_uid = IndexUid::try_from(index_uid.into_inner())?;
    analytics.publish("Stats Seen".to_string(), json!({ "per_index_uid": true }), Some(&req));
    let stats = IndexStats::from(index_scheduler.index_stats(&index_uid)?);
    debug!(returns = ?stats, "Get index stats");
--- a/meilisearch/src/routes/indexes/settings.rs
+++ b/meilisearch/src/routes/indexes/settings.rs
@@ -137,10 +137,8 @@ macro_rules! make_setting_route {
                let settings = settings(&index, &rtxn, meilisearch_types::settings::SecretPolicy::HideSecrets)?;
                debug!(returns = ?settings, "Update settings");
                let mut json = serde_json::json!(&settings);
                let val = json[$camelcase_attr].take();
-                Ok(HttpResponse::Ok().json(val))
+                Ok(HttpResponse::Ok().json(settings.$attr))
            }
            pub fn resources() -> Resource {
--- a/meilisearch/src/routes/mod.rs
+++ b/meilisearch/src/routes/mod.rs
@@ -8,11 +8,9 @@ use meilisearch_types::error::{Code, ResponseError};
 use meilisearch_types::settings::{Settings, Unchecked};
 use meilisearch_types::tasks::{Kind, Status, Task, TaskId};
 use serde::{Deserialize, Serialize};
 use serde_json::json;
 use time::OffsetDateTime;
 use tracing::debug;
 use crate::analytics::Analytics;
 use crate::extractors::authentication::policies::*;
 use crate::extractors::authentication::GuardedData;
 use crate::search_queue::SearchQueue;
@@ -296,10 +294,7 @@ pub struct Stats {
 async fn get_stats(
    index_scheduler: GuardedData<ActionPolicy<{ actions::STATS_GET }>, Data<IndexScheduler>>,
    auth_controller: GuardedData<ActionPolicy<{ actions::STATS_GET }>, Data<AuthController>>,
    req: HttpRequest,
    analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
    analytics.publish("Stats Seen".to_string(), json!({ "per_index_uid": false }), Some(&req));
    let filters = index_scheduler.filters();
    let stats = create_all_stats((*index_scheduler).clone(), (*auth_controller).clone(), filters)?;
@@ -355,11 +350,7 @@ struct VersionResponse {
 async fn get_version(
    _index_scheduler: GuardedData<ActionPolicy<{ actions::VERSION }>, Data<IndexScheduler>>,
    req: HttpRequest,
    analytics: web::Data<dyn Analytics>,
 ) -> HttpResponse {
    analytics.publish("Version Seen".to_string(), json!(null), Some(&req));
    let build_info = build_info::BuildInfo::from_build();
    HttpResponse::Ok().json(VersionResponse {
@@ -377,14 +368,10 @@ async fn get_version(
 }
 pub async fn get_health(
    req: HttpRequest,
    index_scheduler: Data<IndexScheduler>,
    auth_controller: Data<AuthController>,
    search_queue: Data<SearchQueue>,
    analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
    analytics.health_seen(&req);
    search_queue.health().unwrap();
    index_scheduler.health().unwrap();
    auth_controller.health().unwrap();
--- a/meilisearch/src/routes/tasks.rs
+++ b/meilisearch/src/routes/tasks.rs
@@ -270,12 +270,8 @@ pub struct AllTasks {
 async fn get_tasks(
    index_scheduler: GuardedData<ActionPolicy<{ actions::TASKS_GET }>, Data<IndexScheduler>>,
    params: AwebQueryParameter<TasksFilterQuery, DeserrQueryParamError>,
    req: HttpRequest,
    analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
    let mut params = params.into_inner();
    analytics.get_tasks(&params, &req);
    // We +1 just to know if there is more after this "page" or not.
    params.limit.0 = params.limit.0.saturating_add(1);
    let limit = params.limit.0;
@@ -298,8 +294,6 @@ async fn get_tasks(
 async fn get_task(
    index_scheduler: GuardedData<ActionPolicy<{ actions::TASKS_GET }>, Data<IndexScheduler>>,
    task_uid: web::Path<String>,
    req: HttpRequest,
    analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
    let task_uid_string = task_uid.into_inner();
@@ -310,8 +304,6 @@ async fn get_task(
        }
    };
    analytics.publish("Tasks Seen".to_string(), json!({ "per_task_uid": true }), Some(&req));
    let query = index_scheduler::Query { uids: Some(vec![task_uid]), ..Query::default() };
    let filters = index_scheduler.filters();
    let (tasks, _) = index_scheduler.get_tasks_from_authorized_indexes(query, filters)?;
--- a/meilisearch/src/search.rs
+++ b/meilisearch/src/search.rs
@@ -1,3 +1,4 @@
 use core::fmt;
 use std::cmp::min;
 use std::collections::{BTreeMap, BTreeSet, HashSet};
 use std::str::FromStr;
@@ -39,7 +40,7 @@ pub const DEFAULT_HIGHLIGHT_PRE_TAG: fn() -> String = || "<em>".to_string();
 pub const DEFAULT_HIGHLIGHT_POST_TAG: fn() -> String = || "</em>".to_string();
 pub const DEFAULT_SEMANTIC_RATIO: fn() -> SemanticRatio = || SemanticRatio(0.5);
-#[derive(Debug, Clone, Default, PartialEq, Deserr)]
+#[derive(Clone, Default, PartialEq, Deserr)]
 #[deserr(error = DeserrJsonError, rename_all = camelCase, deny_unknown_fields)]
 pub struct SearchQuery {
    #[deserr(default, error = DeserrJsonError<InvalidSearchQ>)]
@@ -88,6 +89,110 @@ pub struct SearchQuery {
    pub attributes_to_search_on: Option<Vec<String>>,
 }
 // Since this structure is logged A LOT we're going to reduce the number of things it logs to the bare minimum.
 // - Only what IS used, we know everything else is set to None so there is no need to print it
 // - Re-order the most important field to debug first
 impl fmt::Debug for SearchQuery {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        let Self {
            q,
            vector,
            hybrid,
            offset,
            limit,
            page,
            hits_per_page,
            attributes_to_retrieve,
            attributes_to_crop,
            crop_length,
            attributes_to_highlight,
            show_matches_position,
            show_ranking_score,
            show_ranking_score_details,
            filter,
            sort,
            facets,
            highlight_pre_tag,
            highlight_post_tag,
            crop_marker,
            matching_strategy,
            attributes_to_search_on,
        } = self;
        let mut debug = f.debug_struct("SearchQuery");
        // First, everything related to the number of documents to retrieve
        debug.field("limit", &limit).field("offset", &offset);
        if let Some(page) = page {
            debug.field("page", &page);
        }
        if let Some(hits_per_page) = hits_per_page {
            debug.field("hits_per_page", &hits_per_page);
        }
        // Then, everything related to the queries
        if let Some(q) = q {
            debug.field("q", &q);
        }
        if let Some(v) = vector {
            if v.len() < 10 {
                debug.field("vector", &v);
            } else {
                debug.field(
                    "vector",
                    &format!("[{}, {}, {}, ... {} dimensions]", v[0], v[1], v[2], v.len()),
                );
            }
        }
        if let Some(hybrid) = hybrid {
            debug.field("hybrid", &hybrid);
        }
        if let Some(attributes_to_search_on) = attributes_to_search_on {
            debug.field("attributes_to_search_on", &attributes_to_search_on);
        }
        if let Some(filter) = filter {
            debug.field("filter", &filter);
        }
        if let Some(sort) = sort {
            debug.field("sort", &sort);
        }
        if let Some(facets) = facets {
            debug.field("facets", &facets);
        }
        debug.field("matching_strategy", &matching_strategy);
        // Then everything related to the formatting
        debug.field("crop_length", &crop_length);
        if *show_matches_position {
            debug.field("show_matches_position", show_matches_position);
        }
        if *show_ranking_score {
            debug.field("show_ranking_score", show_ranking_score);
        }
        if *show_ranking_score_details {
            debug.field("self.show_ranking_score_details", show_ranking_score_details);
        }
        debug.field("crop_length", &crop_length);
        if let Some(facets) = facets {
            debug.field("facets", &facets);
        }
        if let Some(attributes_to_retrieve) = attributes_to_retrieve {
            debug.field("attributes_to_retrieve", &attributes_to_retrieve);
        }
        if let Some(attributes_to_crop) = attributes_to_crop {
            debug.field("attributes_to_crop", &attributes_to_crop);
        }
        if let Some(attributes_to_highlight) = attributes_to_highlight {
            debug.field("attributes_to_highlight", &attributes_to_highlight);
        }
        debug.field("highlight_pre_tag", &highlight_pre_tag);
        debug.field("highlight_post_tag", &highlight_post_tag);
        debug.field("crop_marker", &crop_marker);
        debug.finish()
    }
 }
 #[derive(Debug, Clone, Default, PartialEq, Deserr)]
 #[deserr(error = DeserrJsonError<InvalidHybridQuery>, rename_all = camelCase, deny_unknown_fields)]
 pub struct HybridQuery {
@@ -370,7 +475,7 @@ pub struct SearchHit {
    pub ranking_score_details: Option<serde_json::Map<String, serde_json::Value>>,
 }
-#[derive(Serialize, Debug, Clone, PartialEq)]
+#[derive(Serialize, Clone, PartialEq)]
 #[serde(rename_all = "camelCase")]
 pub struct SearchResult {
    pub hits: Vec<SearchHit>,
@@ -393,6 +498,46 @@ pub struct SearchResult {
    pub used_negative_operator: bool,
 }
 impl fmt::Debug for SearchResult {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        let SearchResult {
            hits,
            query,
            processing_time_ms,
            hits_info,
            facet_distribution,
            facet_stats,
            semantic_hit_count,
            degraded,
            used_negative_operator,
        } = self;
        let mut debug = f.debug_struct("SearchResult");
        // The most important thing when looking at a search result is the time it took to process
        debug.field("processing_time_ms", &processing_time_ms);
        debug.field("hits", &format!("[{} hits returned]", hits.len()));
        debug.field("query", &query);
        debug.field("hits_info", &hits_info);
        if *used_negative_operator {
            debug.field("used_negative_operator", used_negative_operator);
        }
        if *degraded {
            debug.field("degraded", degraded);
        }
        if let Some(facet_distribution) = facet_distribution {
            debug.field("facet_distribution", &facet_distribution);
        }
        if let Some(facet_stats) = facet_stats {
            debug.field("facet_stats", &facet_stats);
        }
        if let Some(semantic_hit_count) = semantic_hit_count {
            debug.field("semantic_hit_count", &semantic_hit_count);
        }
        debug.finish()
    }
 }
 #[derive(Serialize, Debug, Clone, PartialEq)]
 #[serde(rename_all = "camelCase")]
 pub struct SearchResultWithIndex {
--- a/meilisearch/tests/settings/get_settings.rs
+++ b/meilisearch/tests/settings/get_settings.rs
@@ -113,7 +113,8 @@ async fn secrets_are_hidden_in_settings() {
                "default": {
                    "source": "rest",
                    "url": "https://localhost:7777",
-                    "apiKey": "My super secret value you will never guess"
+                    "apiKey": "My super secret value you will never guess",
                    "dimensions": 4,
                }
            }
        }))
@@ -184,6 +185,7 @@ async fn secrets_are_hidden_in_settings() {
        "default": {
          "source": "rest",
          "apiKey": "My suXXXXXX...",
          "dimensions": 4,
          "documentTemplate": "{% for field in fields %} {{ field.name }}: {{ field.value }}\n{% endfor %}",
          "url": "https://localhost:7777",
          "query": null,
@@ -211,6 +213,7 @@ async fn secrets_are_hidden_in_settings() {
        "default": {
          "source": "rest",
          "apiKey": "My suXXXXXX...",
          "dimensions": 4,
          "url": "https://localhost:7777"
        }
      }
--- a/milli/Cargo.toml
+++ b/milli/Cargo.toml
@@ -17,7 +17,7 @@ bincode = "1.3.3"
 bstr = "1.9.0"
 bytemuck = { version = "1.14.0", features = ["extern_crate_alloc"] }
 byteorder = "1.5.0"
-charabia = { version = "0.8.8", default-features = false }
+charabia = { version = "0.8.10", default-features = false }
 concat-arrays = "0.1.2"
 crossbeam-channel = "0.5.11"
 deserr = "0.6.1"
@@ -115,6 +115,7 @@ lmdb-posix-sem = ["heed/posix-sem"]
 # allow chinese specialized tokenization
 chinese = ["charabia/chinese"]
 chinese-pinyin = ["chinese", "charabia/chinese-normalization-pinyin"]
 # allow hebrew specialized tokenization
 hebrew = ["charabia/hebrew"]
@@ -135,7 +136,11 @@ greek = ["charabia/greek"]
 # allow khmer specialized tokenization
 khmer = ["charabia/khmer"]
 # allow vietnamese specialized tokenization
 vietnamese = ["charabia/vietnamese"]
 # force swedish character recomposition
 swedish-recomposition = ["charabia/swedish-recomposition"]
 # allow CUDA support, see <https://github.com/meilisearch/meilisearch/issues/4306>
 cuda = ["candle-core/cuda"]
--- a/milli/src/error.rs
+++ b/milli/src/error.rs
@@ -9,6 +9,7 @@ use serde_json::Value;
 use thiserror::Error;
 use crate::documents::{self, DocumentsBatchCursorError};
 use crate::thread_pool_no_abort::PanicCatched;
 use crate::{CriterionError, DocumentId, FieldId, Object, SortError};
 pub fn is_reserved_keyword(keyword: &str) -> bool {
@@ -39,17 +40,19 @@ pub enum InternalError {
    Fst(#[from] fst::Error),
    #[error(transparent)]
    DocumentsError(#[from] documents::Error),
-    #[error("Invalid compression type have been specified to grenad.")]
+    #[error("Invalid compression type have been specified to grenad")]
    GrenadInvalidCompressionType,
-    #[error("Invalid grenad file with an invalid version format.")]
+    #[error("Invalid grenad file with an invalid version format")]
    GrenadInvalidFormatVersion,
-    #[error("Invalid merge while processing {process}.")]
+    #[error("Invalid merge while processing {process}")]
    IndexingMergingKeys { process: &'static str },
    #[error("{}", HeedError::InvalidDatabaseTyping)]
    InvalidDatabaseTyping,
    #[error(transparent)]
    RayonThreadPool(#[from] ThreadPoolBuildError),
    #[error(transparent)]
    PanicInThreadPool(#[from] PanicCatched),
    #[error(transparent)]
    SerdeJson(#[from] serde_json::Error),
    #[error(transparent)]
    Serialization(#[from] SerializationError),
@@ -57,9 +60,9 @@ pub enum InternalError {
    Store(#[from] MdbError),
    #[error(transparent)]
    Utf8(#[from] str::Utf8Error),
-    #[error("An indexation process was explicitly aborted.")]
+    #[error("An indexation process was explicitly aborted")]
    AbortedIndexation,
-    #[error("The matching words list contains at least one invalid member.")]
+    #[error("The matching words list contains at least one invalid member")]
    InvalidMatchingWords,
    #[error(transparent)]
    ArroyError(#[from] arroy::Error),
--- a/milli/src/index.rs
+++ b/milli/src/index.rs
@@ -678,6 +678,23 @@ impl Index {
            .get(rtxn, main_key::USER_DEFINED_SEARCHABLE_FIELDS_KEY)
    }
    /// Identical to `user_defined_searchable_fields`, but returns ids instead.
    pub fn user_defined_searchable_fields_ids(&self, rtxn: &RoTxn) -> Result<Option<Vec<FieldId>>> {
        match self.user_defined_searchable_fields(rtxn)? {
            Some(fields) => {
                let fields_ids_map = self.fields_ids_map(rtxn)?;
                let mut fields_ids = Vec::new();
                for name in fields {
                    if let Some(field_id) = fields_ids_map.id(name) {
                        fields_ids.push(field_id);
                    }
                }
                Ok(Some(fields_ids))
            }
            None => Ok(None),
        }
    }
    /* filterable fields */
    /// Writes the filterable fields names in the database.
@@ -824,11 +841,11 @@ impl Index {
    /// Identical to `user_defined_faceted_fields`, but returns ids instead.
    pub fn user_defined_faceted_fields_ids(&self, rtxn: &RoTxn) -> Result<HashSet<FieldId>> {
-        let fields = self.faceted_fields(rtxn)?;
+        let fields = self.user_defined_faceted_fields(rtxn)?;
        let fields_ids_map = self.fields_ids_map(rtxn)?;
        let mut fields_ids = HashSet::new();
-        for name in fields.into_iter() {
+        for name in fields {
            if let Some(field_id) = fields_ids_map.id(&name) {
                fields_ids.insert(field_id);
            }
--- a/milli/src/lib.rs
+++ b/milli/src/lib.rs
@@ -21,6 +21,7 @@ pub mod prompt;
 pub mod proximity;
 pub mod score_details;
 mod search;
 mod thread_pool_no_abort;
 pub mod update;
 pub mod vector;
@@ -42,6 +43,7 @@ pub use search::new::{
    SearchLogger, VisualSearchLogger,
 };
 use serde_json::Value;
 pub use thread_pool_no_abort::{PanicCatched, ThreadPoolNoAbort, ThreadPoolNoAbortBuilder};
 pub use {charabia as tokenizer, heed};
 pub use self::asc_desc::{AscDesc, AscDescError, Member, SortError};
@@ -128,7 +130,7 @@ impl fmt::Debug for TimeBudget {
 impl Default for TimeBudget {
    fn default() -> Self {
-        Self::new(std::time::Duration::from_millis(150))
+        Self::new(std::time::Duration::from_millis(1500))
    }
 }
--- a/milli/src/search/facet/facet_distribution.rs
+++ b/milli/src/search/facet/facet_distribution.rs
@@ -97,6 +97,7 @@ impl<'a> FacetDistribution<'a> {
    ) -> heed::Result<()> {
        match facet_type {
            FacetType::Number => {
                let mut lexicographic_distribution = BTreeMap::new();
                let mut key_buffer: Vec<_> = field_id.to_be_bytes().to_vec();
                let distribution_prelength = distribution.len();
@@ -111,14 +112,17 @@ impl<'a> FacetDistribution<'a> {
                    for result in iter {
                        let ((_, _, value), ()) = result?;
-                        *distribution.entry(value.to_string()).or_insert(0) += 1;
+                        *lexicographic_distribution.entry(value.to_string()).or_insert(0) += 1;
-                        if distribution.len() - distribution_prelength == self.max_values_per_facet
+                        if lexicographic_distribution.len() - distribution_prelength
                            == self.max_values_per_facet
                        {
                            break;
                        }
                    }
                }
                distribution.extend(lexicographic_distribution);
            }
            FacetType::String => {
                let mut normalized_distribution = BTreeMap::new();
--- a/milli/src/thread_pool_no_abort.rs
+++ b/milli/src/thread_pool_no_abort.rs
@@ -0,0 +1,69 @@
 use std::sync::atomic::{AtomicBool, Ordering};
 use std::sync::Arc;
 use rayon::{ThreadPool, ThreadPoolBuilder};
 use thiserror::Error;
 /// A rayon ThreadPool wrapper that can catch panics in the pool
 /// and modifies the install function accordingly.
 #[derive(Debug)]
 pub struct ThreadPoolNoAbort {
    thread_pool: ThreadPool,
    /// Set to true if the thread pool catched a panic.
    pool_catched_panic: Arc<AtomicBool>,
 }
 impl ThreadPoolNoAbort {
    pub fn install<OP, R>(&self, op: OP) -> Result<R, PanicCatched>
    where
        OP: FnOnce() -> R + Send,
        R: Send,
    {
        let output = self.thread_pool.install(op);
        // While reseting the pool panic catcher we return an error if we catched one.
        if self.pool_catched_panic.swap(false, Ordering::SeqCst) {
            Err(PanicCatched)
        } else {
            Ok(output)
        }
    }
    pub fn current_num_threads(&self) -> usize {
        self.thread_pool.current_num_threads()
    }
 }
 #[derive(Error, Debug)]
 #[error("A panic occured. Read the logs to find more information about it")]
 pub struct PanicCatched;
 #[derive(Default)]
 pub struct ThreadPoolNoAbortBuilder(ThreadPoolBuilder);
 impl ThreadPoolNoAbortBuilder {
    pub fn new() -> ThreadPoolNoAbortBuilder {
        ThreadPoolNoAbortBuilder::default()
    }
    pub fn thread_name<F>(mut self, closure: F) -> Self
    where
        F: FnMut(usize) -> String + 'static,
    {
        self.0 = self.0.thread_name(closure);
        self
    }
    pub fn num_threads(mut self, num_threads: usize) -> ThreadPoolNoAbortBuilder {
        self.0 = self.0.num_threads(num_threads);
        self
    }
    pub fn build(mut self) -> Result<ThreadPoolNoAbort, rayon::ThreadPoolBuildError> {
        let pool_catched_panic = Arc::new(AtomicBool::new(false));
        self.0 = self.0.panic_handler({
            let catched_panic = pool_catched_panic.clone();
            move |_result| catched_panic.store(true, Ordering::SeqCst)
        });
        Ok(ThreadPoolNoAbort { thread_pool: self.0.build()?, pool_catched_panic })
    }
 }
--- a/milli/src/update/del_add.rs
+++ b/milli/src/update/del_add.rs
@@ -71,8 +71,8 @@ pub enum DelAddOperation {
 /// putting each deletion obkv's keys under an DelAdd::Deletion
 /// and putting each addition obkv's keys under an DelAdd::Addition
 pub fn del_add_from_two_obkvs<K: obkv::Key + PartialOrd + Ord>(
-    deletion: obkv::KvReader<K>,
+    deletion: &obkv::KvReader<K>,
-    addition: obkv::KvReader<K>,
+    addition: &obkv::KvReader<K>,
    buffer: &mut Vec<u8>,
 ) -> Result<(), std::io::Error> {
    use itertools::merge_join_by;
--- a/milli/src/update/index_documents/extract/extract_docid_word_positions.rs
+++ b/milli/src/update/index_documents/extract/extract_docid_word_positions.rs
@@ -1,4 +1,4 @@
-use std::collections::{HashMap, HashSet};
+use std::collections::HashMap;
 use std::convert::TryInto;
 use std::fs::File;
 use std::io::BufReader;
@@ -12,6 +12,7 @@ use serde_json::Value;
 use super::helpers::{create_sorter, keep_latest_obkv, sorter_into_reader, GrenadParameters};
 use crate::error::{InternalError, SerializationError};
 use crate::update::del_add::{del_add_from_two_obkvs, DelAdd, KvReaderDelAdd};
 use crate::update::settings::{InnerIndexSettings, InnerIndexSettingsDiff};
 use crate::{FieldId, Result, MAX_POSITION_PER_ATTRIBUTE, MAX_WORD_LENGTH};
 pub type ScriptLanguageDocidsMap = HashMap<(Script, Language), (RoaringBitmap, RoaringBitmap)>;
@@ -25,10 +26,7 @@ pub type ScriptLanguageDocidsMap = HashMap<(Script, Language), (RoaringBitmap, R
 pub fn extract_docid_word_positions<R: io::Read + io::Seek>(
    obkv_documents: grenad::Reader<R>,
    indexer: GrenadParameters,
-    searchable_fields: &Option<HashSet<FieldId>>,
+    settings_diff: &InnerIndexSettingsDiff,
    stop_words: Option<&fst::Set<Vec<u8>>>,
    allowed_separators: Option<&[&str]>,
    dictionary: Option<&[&str]>,
    max_positions_per_attributes: Option<u32>,
 ) -> Result<(grenad::Reader<BufReader<File>>, ScriptLanguageDocidsMap)> {
    puffin::profile_function!();
@@ -36,6 +34,7 @@ pub fn extract_docid_word_positions<R: io::Read + io::Seek>(
    let max_positions_per_attributes = max_positions_per_attributes
        .map_or(MAX_POSITION_PER_ATTRIBUTE, |max| max.min(MAX_POSITION_PER_ATTRIBUTE));
    let max_memory = indexer.max_memory_by_thread();
    let force_reindexing = settings_diff.reindex_searchable();
    // initialize destination values.
    let mut documents_ids = RoaringBitmap::new();
@@ -56,8 +55,37 @@ pub fn extract_docid_word_positions<R: io::Read + io::Seek>(
    let mut value_buffer = Vec::new();
    // initialize tokenizer.
-    let mut builder = tokenizer_builder(stop_words, allowed_separators, dictionary, None);
+    let old_stop_words = settings_diff.old.stop_words.as_ref();
-    let tokenizer = builder.build();
+    let old_separators: Option<Vec<_>> = settings_diff
        .old
        .allowed_separators
        .as_ref()
        .map(|s| s.iter().map(String::as_str).collect());
    let old_dictionary: Option<Vec<_>> =
        settings_diff.old.dictionary.as_ref().map(|s| s.iter().map(String::as_str).collect());
    let mut del_builder = tokenizer_builder(
        old_stop_words,
        old_separators.as_deref(),
        old_dictionary.as_deref(),
        None,
    );
    let del_tokenizer = del_builder.build();
    let new_stop_words = settings_diff.new.stop_words.as_ref();
    let new_separators: Option<Vec<_>> = settings_diff
        .new
        .allowed_separators
        .as_ref()
        .map(|s| s.iter().map(String::as_str).collect());
    let new_dictionary: Option<Vec<_>> =
        settings_diff.new.dictionary.as_ref().map(|s| s.iter().map(String::as_str).collect());
    let mut add_builder = tokenizer_builder(
        new_stop_words,
        new_separators.as_deref(),
        new_dictionary.as_deref(),
        None,
    );
    let add_tokenizer = add_builder.build();
    // iterate over documents.
    let mut cursor = obkv_documents.into_cursor()?;
@@ -69,7 +97,7 @@ pub fn extract_docid_word_positions<R: io::Read + io::Seek>(
        let obkv = KvReader::<FieldId>::new(value);
        // if the searchable fields didn't change, skip the searchable indexing for this document.
-        if !searchable_fields_changed(&KvReader::<FieldId>::new(value), searchable_fields) {
+        if !force_reindexing && !searchable_fields_changed(&obkv, settings_diff) {
            continue;
        }
@@ -85,11 +113,8 @@ pub fn extract_docid_word_positions<R: io::Read + io::Seek>(
                // deletions
                lang_safe_tokens_from_document(
                    &obkv,
-                    searchable_fields,
+                    &settings_diff.old,
-                    &tokenizer,
+                    &del_tokenizer,
                    stop_words,
                    allowed_separators,
                    dictionary,
                    max_positions_per_attributes,
                    DelAdd::Deletion,
                    &mut del_buffers,
@@ -99,11 +124,8 @@ pub fn extract_docid_word_positions<R: io::Read + io::Seek>(
                // additions
                lang_safe_tokens_from_document(
                    &obkv,
-                    searchable_fields,
+                    &settings_diff.new,
-                    &tokenizer,
+                    &add_tokenizer,
                    stop_words,
                    allowed_separators,
                    dictionary,
                    max_positions_per_attributes,
                    DelAdd::Addition,
                    &mut add_buffers,
@@ -118,8 +140,8 @@ pub fn extract_docid_word_positions<R: io::Read + io::Seek>(
        // transforming two KV<FieldId, KV<u16, String>> into one KV<FieldId, KV<DelAdd, KV<u16, String>>>
        value_buffer.clear();
        del_add_from_two_obkvs(
-            KvReader::<FieldId>::new(del_obkv),
+            &KvReader::<FieldId>::new(del_obkv),
-            KvReader::<FieldId>::new(add_obkv),
+            &KvReader::<FieldId>::new(add_obkv),
            &mut value_buffer,
        )?;
@@ -160,8 +182,9 @@ pub fn extract_docid_word_positions<R: io::Read + io::Seek>(
 /// Check if any searchable fields of a document changed.
 fn searchable_fields_changed(
    obkv: &KvReader<FieldId>,
-    searchable_fields: &Option<HashSet<FieldId>>,
+    settings_diff: &InnerIndexSettingsDiff,
 ) -> bool {
    let searchable_fields = &settings_diff.new.searchable_fields_ids;
    for (field_id, field_bytes) in obkv.iter() {
        if searchable_fields.as_ref().map_or(true, |sf| sf.contains(&field_id)) {
            let del_add = KvReaderDelAdd::new(field_bytes);
@@ -206,14 +229,10 @@ fn tokenizer_builder<'a>(
 /// Extract words mapped with their positions of a document,
 /// ensuring no Language detection mistakes was made.
 #[allow(clippy::too_many_arguments)] // FIXME: consider grouping arguments in a struct
 fn lang_safe_tokens_from_document<'a>(
    obkv: &KvReader<FieldId>,
-    searchable_fields: &Option<HashSet<FieldId>>,
+    settings: &InnerIndexSettings,
    tokenizer: &Tokenizer,
    stop_words: Option<&fst::Set<Vec<u8>>>,
    allowed_separators: Option<&[&str]>,
    dictionary: Option<&[&str]>,
    max_positions_per_attributes: u32,
    del_add: DelAdd,
    buffers: &'a mut Buffers,
@@ -222,7 +241,7 @@ fn lang_safe_tokens_from_document<'a>(
    tokens_from_document(
        obkv,
-        searchable_fields,
+        &settings.searchable_fields_ids,
        tokenizer,
        max_positions_per_attributes,
        del_add,
@@ -246,12 +265,15 @@ fn lang_safe_tokens_from_document<'a>(
        // then we don't rerun the extraction.
        if !script_language.is_empty() {
            // build a new temporary tokenizer including the allow list.
-            let mut builder = tokenizer_builder(
+            let stop_words = settings.stop_words.as_ref();
-                stop_words,
+            let separators: Option<Vec<_>> = settings
-                allowed_separators,
+                .allowed_separators
-                dictionary,
+                .as_ref()
-                Some(&script_language),
+                .map(|s| s.iter().map(String::as_str).collect());
-            );
+            let dictionary: Option<Vec<_>> =
                settings.dictionary.as_ref().map(|s| s.iter().map(String::as_str).collect());
            let mut builder =
                tokenizer_builder(stop_words, separators.as_deref(), dictionary.as_deref(), None);
            let tokenizer = builder.build();
            script_language_word_count.clear();
@@ -259,7 +281,7 @@ fn lang_safe_tokens_from_document<'a>(
            // rerun the extraction.
            tokens_from_document(
                obkv,
-                searchable_fields,
+                &settings.searchable_fields_ids,
                &tokenizer,
                max_positions_per_attributes,
                del_add,
@@ -276,7 +298,7 @@ fn lang_safe_tokens_from_document<'a>(
 /// Extract words mapped with their positions of a document.
 fn tokens_from_document<'a>(
    obkv: &KvReader<FieldId>,
-    searchable_fields: &Option<HashSet<FieldId>>,
+    searchable_fields: &Option<Vec<FieldId>>,
    tokenizer: &Tokenizer,
    max_positions_per_attributes: u32,
    del_add: DelAdd,
--- a/milli/src/update/index_documents/extract/extract_facet_number_docids.rs
+++ b/milli/src/update/index_documents/extract/extract_facet_number_docids.rs
@@ -10,6 +10,7 @@ use crate::heed_codec::facet::{
    FacetGroupKey, FacetGroupKeyCodec, FieldDocIdFacetF64Codec, OrderedF64Codec,
 };
 use crate::update::del_add::{KvReaderDelAdd, KvWriterDelAdd};
 use crate::update::settings::InnerIndexSettingsDiff;
 use crate::Result;
 /// Extracts the facet number and the documents ids where this facet number appear.
@@ -20,6 +21,7 @@ use crate::Result;
 pub fn extract_facet_number_docids<R: io::Read + io::Seek>(
    fid_docid_facet_number: grenad::Reader<R>,
    indexer: GrenadParameters,
    _settings_diff: &InnerIndexSettingsDiff,
 ) -> Result<grenad::Reader<BufReader<File>>> {
    puffin::profile_function!();
--- a/milli/src/update/index_documents/extract/extract_facet_string_docids.rs
+++ b/milli/src/update/index_documents/extract/extract_facet_string_docids.rs
@@ -15,6 +15,7 @@ use crate::update::del_add::{DelAdd, KvReaderDelAdd, KvWriterDelAdd};
 use crate::update::index_documents::helpers::{
    merge_deladd_btreeset_string, merge_deladd_cbo_roaring_bitmaps,
 };
 use crate::update::settings::InnerIndexSettingsDiff;
 use crate::{FieldId, Result, MAX_FACET_VALUE_LENGTH};
 /// Extracts the facet string and the documents ids where this facet string appear.
@@ -25,6 +26,7 @@ use crate::{FieldId, Result, MAX_FACET_VALUE_LENGTH};
 pub fn extract_facet_string_docids<R: io::Read + io::Seek>(
    docid_fid_facet_string: grenad::Reader<R>,
    indexer: GrenadParameters,
    _settings_diff: &InnerIndexSettingsDiff,
 ) -> Result<(grenad::Reader<BufReader<File>>, grenad::Reader<BufReader<File>>)> {
    puffin::profile_function!();
--- a/milli/src/update/index_documents/extract/extract_fid_docid_facet_values.rs
+++ b/milli/src/update/index_documents/extract/extract_fid_docid_facet_values.rs
@@ -1,5 +1,5 @@
 use std::borrow::Cow;
-use std::collections::{BTreeMap, HashSet};
+use std::collections::BTreeMap;
 use std::convert::TryInto;
 use std::fs::File;
 use std::io::{self, BufReader};
@@ -20,6 +20,7 @@ use crate::error::InternalError;
 use crate::facet::value_encoding::f64_into_bytes;
 use crate::update::del_add::{DelAdd, KvWriterDelAdd};
 use crate::update::index_documents::{create_writer, writer_into_reader};
 use crate::update::settings::InnerIndexSettingsDiff;
 use crate::{CboRoaringBitmapCodec, DocumentId, Error, FieldId, Result, MAX_FACET_VALUE_LENGTH};
 /// The length of the elements that are always in the buffer when inserting new values.
@@ -43,7 +44,7 @@ pub struct ExtractedFacetValues {
 pub fn extract_fid_docid_facet_values<R: io::Read + io::Seek>(
    obkv_documents: grenad::Reader<R>,
    indexer: GrenadParameters,
-    faceted_fields: &HashSet<FieldId>,
+    settings_diff: &InnerIndexSettingsDiff,
    geo_fields_ids: Option<(FieldId, FieldId)>,
 ) -> Result<ExtractedFacetValues> {
    puffin::profile_function!();
@@ -82,7 +83,9 @@ pub fn extract_fid_docid_facet_values<R: io::Read + io::Seek>(
        let obkv = obkv::KvReader::new(value);
        for (field_id, field_bytes) in obkv.iter() {
-            if faceted_fields.contains(&field_id) {
+            let delete_faceted = settings_diff.old.faceted_fields_ids.contains(&field_id);
            let add_faceted = settings_diff.new.faceted_fields_ids.contains(&field_id);
            if delete_faceted || add_faceted {
                numbers_key_buffer.clear();
                strings_key_buffer.clear();
@@ -99,11 +102,12 @@ pub fn extract_fid_docid_facet_values<R: io::Read + io::Seek>(
                strings_key_buffer.extend_from_slice(docid_bytes);
                let del_add_obkv = obkv::KvReader::new(field_bytes);
-                let del_value = match del_add_obkv.get(DelAdd::Deletion) {
+                let del_value = match del_add_obkv.get(DelAdd::Deletion).filter(|_| delete_faceted)
                {
                    Some(bytes) => Some(from_slice(bytes).map_err(InternalError::SerdeJson)?),
                    None => None,
                };
-                let add_value = match del_add_obkv.get(DelAdd::Addition) {
+                let add_value = match del_add_obkv.get(DelAdd::Addition).filter(|_| add_faceted) {
                    Some(bytes) => Some(from_slice(bytes).map_err(InternalError::SerdeJson)?),
                    None => None,
                };
--- a/milli/src/update/index_documents/extract/extract_fid_word_count_docids.rs
+++ b/milli/src/update/index_documents/extract/extract_fid_word_count_docids.rs
@@ -10,6 +10,7 @@ use super::helpers::{
 use crate::error::SerializationError;
 use crate::index::db_name::DOCID_WORD_POSITIONS;
 use crate::update::del_add::{DelAdd, KvReaderDelAdd, KvWriterDelAdd};
 use crate::update::settings::InnerIndexSettingsDiff;
 use crate::Result;
 const MAX_COUNTED_WORDS: usize = 30;
@@ -23,6 +24,7 @@ const MAX_COUNTED_WORDS: usize = 30;
 pub fn extract_fid_word_count_docids<R: io::Read + io::Seek>(
    docid_word_positions: grenad::Reader<R>,
    indexer: GrenadParameters,
    _settings_diff: &InnerIndexSettingsDiff,
 ) -> Result<grenad::Reader<BufReader<File>>> {
    puffin::profile_function!();
--- a/milli/src/update/index_documents/extract/extract_vector_points.rs
+++ b/milli/src/update/index_documents/extract/extract_vector_points.rs
@@ -17,8 +17,9 @@ use crate::error::UserError;
 use crate::prompt::Prompt;
 use crate::update::del_add::{DelAdd, KvReaderDelAdd, KvWriterDelAdd};
 use crate::update::index_documents::helpers::try_split_at;
 use crate::update::settings::InnerIndexSettingsDiff;
 use crate::vector::Embedder;
-use crate::{DocumentId, FieldsIdsMap, InternalError, Result, VectorOrArrayOfVectors};
+use crate::{DocumentId, InternalError, Result, ThreadPoolNoAbort, VectorOrArrayOfVectors};
 /// The length of the elements that are always in the buffer when inserting new values.
 const TRUNCATE_SIZE: usize = size_of::<DocumentId>();
@@ -71,12 +72,15 @@ impl VectorStateDelta {
 pub fn extract_vector_points<R: io::Read + io::Seek>(
    obkv_documents: grenad::Reader<R>,
    indexer: GrenadParameters,
-    field_id_map: &FieldsIdsMap,
+    settings_diff: &InnerIndexSettingsDiff,
    prompt: &Prompt,
    embedder_name: &str,
 ) -> Result<ExtractedVectorPoints> {
    puffin::profile_function!();
    let old_fields_ids_map = &settings_diff.old.fields_ids_map;
    let new_fields_ids_map = &settings_diff.new.fields_ids_map;
    // (docid, _index) -> KvWriterDelAdd -> Vector
    let mut manual_vectors_writer = create_writer(
        indexer.chunk_compression_type,
@@ -98,8 +102,6 @@ pub fn extract_vector_points<R: io::Read + io::Seek>(
        tempfile::tempfile()?,
    );
    let vectors_fid = field_id_map.id("_vectors");
    let mut key_buffer = Vec::new();
    let mut cursor = obkv_documents.into_cursor()?;
    while let Some((key, value)) = cursor.move_on_next()? {
@@ -116,15 +118,29 @@ pub fn extract_vector_points<R: io::Read + io::Seek>(
        // lazily get it when needed
        let document_id = || -> Value { from_utf8(external_id_bytes).unwrap().into() };
-        let vectors_field = vectors_fid
+        // the vector field id may have changed
        let old_vectors_fid = old_fields_ids_map.id("_vectors");
        // filter the old vector fid if the settings has been changed forcing reindexing.
        let old_vectors_fid = old_vectors_fid.filter(|_| !settings_diff.reindex_vectors());
        let new_vectors_fid = new_fields_ids_map.id("_vectors");
        let vectors_field = {
            let del = old_vectors_fid
                .and_then(|vectors_fid| obkv.get(vectors_fid))
                .map(KvReaderDelAdd::new)
-            .map(|obkv| to_vector_maps(obkv, document_id))
+                .map(|obkv| to_vector_map(obkv, DelAdd::Deletion, &document_id))
-            .transpose()?;
+                .transpose()?
                .flatten();
            let add = new_vectors_fid
                .and_then(|vectors_fid| obkv.get(vectors_fid))
                .map(KvReaderDelAdd::new)
                .map(|obkv| to_vector_map(obkv, DelAdd::Addition, &document_id))
                .transpose()?
                .flatten();
            (del, add)
        };
-        let (del_map, add_map) = vectors_field.unzip();
+        let (del_map, add_map) = vectors_field;
        let del_map = del_map.flatten();
        let add_map = add_map.flatten();
        let del_value = del_map.and_then(|mut map| map.remove(embedder_name));
        let add_value = add_map.and_then(|mut map| map.remove(embedder_name));
@@ -155,7 +171,7 @@ pub fn extract_vector_points<R: io::Read + io::Seek>(
                    VectorStateDelta::NowGenerated(prompt.render(
                        obkv,
                        DelAdd::Addition,
-                        field_id_map,
+                        new_fields_ids_map,
                    )?)
                } else {
                    VectorStateDelta::NowRemoved
@@ -182,10 +198,16 @@ pub fn extract_vector_points<R: io::Read + io::Seek>(
                if document_is_kept {
                    // Don't give up if the old prompt was failing
-                    let old_prompt =
+                    let old_prompt = Some(prompt)
-                        prompt.render(obkv, DelAdd::Deletion, field_id_map).unwrap_or_default();
+                        // TODO: this filter works because we erase the vec database when a embedding setting changes.
-                    let new_prompt = prompt.render(obkv, DelAdd::Addition, field_id_map)?;
+                        // When vector pipeline will be optimized, this should be removed.
-                    if old_prompt != new_prompt {
+                        .filter(|_| !settings_diff.reindex_vectors())
                        .map(|p| {
                            p.render(obkv, DelAdd::Deletion, old_fields_ids_map).unwrap_or_default()
                        });
                    let new_prompt = prompt.render(obkv, DelAdd::Addition, new_fields_ids_map)?;
                    if old_prompt.as_ref() != Some(&new_prompt) {
                        let old_prompt = old_prompt.unwrap_or_default();
                        tracing::trace!(
                            "🚀 Changing prompt from\n{old_prompt}\n===to===\n{new_prompt}"
                        );
@@ -207,6 +229,7 @@ pub fn extract_vector_points<R: io::Read + io::Seek>(
            &mut manual_vectors_writer,
            &mut key_buffer,
            delta,
            settings_diff,
        )?;
    }
@@ -220,15 +243,6 @@ pub fn extract_vector_points<R: io::Read + io::Seek>(
    })
 }
 fn to_vector_maps(
    obkv: KvReaderDelAdd,
    document_id: impl Fn() -> Value,
 ) -> Result<(Option<serde_json::Map<String, Value>>, Option<serde_json::Map<String, Value>>)> {
    let del = to_vector_map(obkv, DelAdd::Deletion, &document_id)?;
    let add = to_vector_map(obkv, DelAdd::Addition, &document_id)?;
    Ok((del, add))
 }
 fn to_vector_map(
    obkv: KvReaderDelAdd,
    side: DelAdd,
@@ -256,10 +270,15 @@ fn push_vectors_diff(
    manual_vectors_writer: &mut Writer<BufWriter<File>>,
    key_buffer: &mut Vec<u8>,
    delta: VectorStateDelta,
    settings_diff: &InnerIndexSettingsDiff,
 ) -> Result<()> {
    puffin::profile_function!();
    let (must_remove, prompt, (mut del_vectors, mut add_vectors)) = delta.into_values();
-    if must_remove {
+    if must_remove
    // TODO: the below condition works because we erase the vec database when a embedding setting changes.
    // When vector pipeline will be optimized, this should be removed.
    && !settings_diff.reindex_vectors()
    {
        key_buffer.truncate(TRUNCATE_SIZE);
        remove_vectors_writer.insert(&key_buffer, [])?;
    }
@@ -287,6 +306,9 @@ fn push_vectors_diff(
        match eob {
            EitherOrBoth::Both(_, _) => (), // no need to touch anything
            EitherOrBoth::Left(vector) => {
                // TODO: the below condition works because we erase the vec database when a embedding setting changes.
                // When vector pipeline will be optimized, this should be removed.
                if !settings_diff.reindex_vectors() {
                    // We insert only the Del part of the Obkv to inform
                    // that we only want to remove all those vectors.
                    let mut obkv = KvWriterDelAdd::memory();
@@ -294,6 +316,7 @@ fn push_vectors_diff(
                    let bytes = obkv.into_inner()?;
                    manual_vectors_writer.insert(&key_buffer, bytes)?;
                }
            }
            EitherOrBoth::Right(vector) => {
                // We insert only the Add part of the Obkv to inform
                // that we only want to remove all those vectors.
@@ -339,7 +362,7 @@ pub fn extract_embeddings<R: io::Read + io::Seek>(
    prompt_reader: grenad::Reader<R>,
    indexer: GrenadParameters,
    embedder: Arc<Embedder>,
-    request_threads: &rayon::ThreadPool,
+    request_threads: &ThreadPoolNoAbort,
 ) -> Result<grenad::Reader<BufReader<File>>> {
    puffin::profile_function!();
    let n_chunks = embedder.chunk_count_hint(); // chunk level parallelism
--- a/milli/src/update/index_documents/extract/extract_word_docids.rs
+++ b/milli/src/update/index_documents/extract/extract_word_docids.rs
@@ -1,20 +1,23 @@
-use std::collections::{BTreeSet, HashSet};
+use std::collections::BTreeSet;
 use std::fs::File;
 use std::io::{self, BufReader};
-use heed::BytesDecode;
+use heed::{BytesDecode, BytesEncode};
 use obkv::KvReaderU16;
 use roaring::RoaringBitmap;
 use super::helpers::{
-    create_sorter, create_writer, merge_deladd_cbo_roaring_bitmaps, sorter_into_reader,
+    create_sorter, create_writer, merge_deladd_cbo_roaring_bitmaps, try_split_array_at,
-    try_split_array_at, writer_into_reader, GrenadParameters,
+    writer_into_reader, GrenadParameters,
 };
 use crate::error::SerializationError;
 use crate::heed_codec::StrBEU16Codec;
 use crate::index::db_name::DOCID_WORD_POSITIONS;
 use crate::update::del_add::{is_noop_del_add_obkv, DelAdd, KvReaderDelAdd, KvWriterDelAdd};
 use crate::update::index_documents::helpers::sorter_into_reader;
 use crate::update::settings::InnerIndexSettingsDiff;
 use crate::update::MergeFn;
-use crate::{DocumentId, FieldId, Result};
+use crate::{CboRoaringBitmapCodec, DocumentId, FieldId, Result};
 /// Extracts the word and the documents ids where this word appear.
 ///
@@ -27,7 +30,7 @@ use crate::{DocumentId, FieldId, Result};
 pub fn extract_word_docids<R: io::Read + io::Seek>(
    docid_word_positions: grenad::Reader<R>,
    indexer: GrenadParameters,
-    exact_attributes: &HashSet<FieldId>,
+    settings_diff: &InnerIndexSettingsDiff,
 ) -> Result<(
    grenad::Reader<BufReader<File>>,
    grenad::Reader<BufReader<File>>,
@@ -43,7 +46,7 @@ pub fn extract_word_docids<R: io::Read + io::Seek>(
        indexer.chunk_compression_type,
        indexer.chunk_compression_level,
        indexer.max_nb_chunks,
-        max_memory.map(|x| x / 3),
+        max_memory.map(|m| m / 3),
    );
    let mut key_buffer = Vec::new();
    let mut del_words = BTreeSet::new();
@@ -85,13 +88,19 @@ pub fn extract_word_docids<R: io::Read + io::Seek>(
        add_words.clear();
    }
    let mut word_fid_docids_writer = create_writer(
        indexer.chunk_compression_type,
        indexer.chunk_compression_level,
        tempfile::tempfile()?,
    );
    let mut word_docids_sorter = create_sorter(
        grenad::SortAlgorithm::Unstable,
        merge_deladd_cbo_roaring_bitmaps,
        indexer.chunk_compression_type,
        indexer.chunk_compression_level,
        indexer.max_nb_chunks,
-        max_memory.map(|x| x / 3),
+        max_memory.map(|m| m / 3),
    );
    let mut exact_word_docids_sorter = create_sorter(
@@ -100,31 +109,45 @@ pub fn extract_word_docids<R: io::Read + io::Seek>(
        indexer.chunk_compression_type,
        indexer.chunk_compression_level,
        indexer.max_nb_chunks,
-        max_memory.map(|x| x / 3),
+        max_memory.map(|m| m / 3),
    );
    let mut word_fid_docids_writer = create_writer(
        indexer.chunk_compression_type,
        indexer.chunk_compression_level,
        tempfile::tempfile()?,
    );
    let mut iter = word_fid_docids_sorter.into_stream_merger_iter()?;
-    // TODO: replace sorters by writers by accumulating values into a buffer before inserting them.
+    let mut buffer = Vec::new();
    // NOTE: replacing sorters by bitmap merging is less efficient, so, use sorters.
    while let Some((key, value)) = iter.next()? {
        // only keep the value if their is a change to apply in the DB.
        if !is_noop_del_add_obkv(KvReaderDelAdd::new(value)) {
            word_fid_docids_writer.insert(key, value)?;
        }
-        let (word, fid) = StrBEU16Codec::bytes_decode(key)
+        let (w, fid) = StrBEU16Codec::bytes_decode(key)
            .map_err(|_| SerializationError::Decoding { db_name: Some(DOCID_WORD_POSITIONS) })?;
-        // every words contained in an attribute set to exact must be pushed in the exact_words list.
+        // merge all deletions
-        if exact_attributes.contains(&fid) {
+        let obkv = KvReaderDelAdd::new(value);
-            exact_word_docids_sorter.insert(word.as_bytes(), value)?;
+        if let Some(value) = obkv.get(DelAdd::Deletion) {
            let delete_from_exact = settings_diff.old.exact_attributes.contains(&fid);
            buffer.clear();
            let mut obkv = KvWriterDelAdd::new(&mut buffer);
            obkv.insert(DelAdd::Deletion, value)?;
            if delete_from_exact {
                exact_word_docids_sorter.insert(w, obkv.into_inner().unwrap())?;
            } else {
-            word_docids_sorter.insert(word.as_bytes(), value)?;
+                word_docids_sorter.insert(w, obkv.into_inner().unwrap())?;
            }
        }
        // merge all additions
        if let Some(value) = obkv.get(DelAdd::Addition) {
            let add_in_exact = settings_diff.new.exact_attributes.contains(&fid);
            buffer.clear();
            let mut obkv = KvWriterDelAdd::new(&mut buffer);
            obkv.insert(DelAdd::Addition, value)?;
            if add_in_exact {
                exact_word_docids_sorter.insert(w, obkv.into_inner().unwrap())?;
            } else {
                word_docids_sorter.insert(w, obkv.into_inner().unwrap())?;
            }
        }
    }
@@ -178,3 +201,45 @@ fn words_into_sorter(
    Ok(())
 }
 #[tracing::instrument(level = "trace", skip_all, target = "indexing::extract")]
 fn docids_into_writers<W>(
    word: &str,
    deletions: &RoaringBitmap,
    additions: &RoaringBitmap,
    writer: &mut grenad::Writer<W>,
 ) -> Result<()>
 where
    W: std::io::Write,
 {
    if deletions == additions {
        // if the same value is deleted and added, do nothing.
        return Ok(());
    }
    // Write each value in the same KvDelAdd before inserting it in the final writer.
    let mut obkv = KvWriterDelAdd::memory();
    // deletions:
    if !deletions.is_empty() && !deletions.is_subset(additions) {
        obkv.insert(
            DelAdd::Deletion,
            CboRoaringBitmapCodec::bytes_encode(deletions).map_err(|_| {
                SerializationError::Encoding { db_name: Some(DOCID_WORD_POSITIONS) }
            })?,
        )?;
    }
    // additions:
    if !additions.is_empty() {
        obkv.insert(
            DelAdd::Addition,
            CboRoaringBitmapCodec::bytes_encode(additions).map_err(|_| {
                SerializationError::Encoding { db_name: Some(DOCID_WORD_POSITIONS) }
            })?,
        )?;
    }
    // insert everything in the same writer.
    writer.insert(word.as_bytes(), obkv.into_inner().unwrap())?;
    Ok(())
 }
--- a/milli/src/update/index_documents/extract/extract_word_pair_proximity_docids.rs
+++ b/milli/src/update/index_documents/extract/extract_word_pair_proximity_docids.rs
@@ -11,8 +11,9 @@ use super::helpers::{
 };
 use crate::error::SerializationError;
 use crate::index::db_name::DOCID_WORD_POSITIONS;
-use crate::proximity::{index_proximity, MAX_DISTANCE};
+use crate::proximity::{index_proximity, ProximityPrecision, MAX_DISTANCE};
 use crate::update::del_add::{DelAdd, KvReaderDelAdd, KvWriterDelAdd};
 use crate::update::settings::InnerIndexSettingsDiff;
 use crate::{DocumentId, Result};
 /// Extracts the best proximity between pairs of words and the documents ids where this pair appear.
@@ -23,8 +24,21 @@ use crate::{DocumentId, Result};
 pub fn extract_word_pair_proximity_docids<R: io::Read + io::Seek>(
    docid_word_positions: grenad::Reader<R>,
    indexer: GrenadParameters,
    settings_diff: &InnerIndexSettingsDiff,
 ) -> Result<grenad::Reader<BufReader<File>>> {
    puffin::profile_function!();
    let any_deletion = settings_diff.old.proximity_precision == ProximityPrecision::ByWord;
    let any_addition = settings_diff.new.proximity_precision == ProximityPrecision::ByWord;
    // early return if the data shouldn't be deleted nor created.
    if !any_deletion && !any_addition {
        let writer = create_writer(
            indexer.chunk_compression_type,
            indexer.chunk_compression_level,
            tempfile::tempfile()?,
        );
        return writer_into_reader(writer);
    }
    let max_memory = indexer.max_memory_by_thread();
@@ -77,6 +91,10 @@ pub fn extract_word_pair_proximity_docids<R: io::Read + io::Seek>(
        let (del, add): (Result<_>, Result<_>) = rayon::join(
            || {
                if !any_deletion {
                    return Ok(());
                }
                // deletions
                if let Some(deletion) = KvReaderDelAdd::new(value).get(DelAdd::Deletion) {
                    for (position, word) in KvReaderU16::new(deletion).iter() {
@@ -106,6 +124,10 @@ pub fn extract_word_pair_proximity_docids<R: io::Read + io::Seek>(
                Ok(())
            },
            || {
                if !any_addition {
                    return Ok(());
                }
                // additions
                if let Some(addition) = KvReaderDelAdd::new(value).get(DelAdd::Addition) {
                    for (position, word) in KvReaderU16::new(addition).iter() {
--- a/milli/src/update/index_documents/extract/extract_word_position_docids.rs
+++ b/milli/src/update/index_documents/extract/extract_word_position_docids.rs
@@ -11,6 +11,7 @@ use super::helpers::{
 use crate::error::SerializationError;
 use crate::index::db_name::DOCID_WORD_POSITIONS;
 use crate::update::del_add::{DelAdd, KvReaderDelAdd, KvWriterDelAdd};
 use crate::update::settings::InnerIndexSettingsDiff;
 use crate::update::MergeFn;
 use crate::{bucketed_position, DocumentId, Result};
@@ -22,6 +23,7 @@ use crate::{bucketed_position, DocumentId, Result};
 pub fn extract_word_position_docids<R: io::Read + io::Seek>(
    docid_word_positions: grenad::Reader<R>,
    indexer: GrenadParameters,
    _settings_diff: &InnerIndexSettingsDiff,
 ) -> Result<grenad::Reader<BufReader<File>>> {
    puffin::profile_function!();
--- a/milli/src/update/index_documents/extract/mod.rs
+++ b/milli/src/update/index_documents/extract/mod.rs
@@ -9,9 +9,9 @@ mod extract_word_docids;
 mod extract_word_pair_proximity_docids;
 mod extract_word_position_docids;
 use std::collections::HashSet;
 use std::fs::File;
 use std::io::BufReader;
 use std::sync::Arc;
 use crossbeam_channel::Sender;
 use rayon::prelude::*;
@@ -30,9 +30,8 @@ use self::extract_word_pair_proximity_docids::extract_word_pair_proximity_docids
 use self::extract_word_position_docids::extract_word_position_docids;
 use super::helpers::{as_cloneable_grenad, CursorClonableMmap, GrenadParameters};
 use super::{helpers, TypedChunk};
-use crate::proximity::ProximityPrecision;
+use crate::update::settings::InnerIndexSettingsDiff;
-use crate::vector::EmbeddingConfigs;
+use crate::{FieldId, Result, ThreadPoolNoAbortBuilder};
 use crate::{FieldId, FieldsIdsMap, Result};
 /// Extract data for each databases from obkv documents in parallel.
 /// Send data in grenad file over provided Sender.
@@ -43,18 +42,10 @@ pub(crate) fn data_from_obkv_documents(
    flattened_obkv_chunks: impl Iterator<Item = Result<grenad::Reader<BufReader<File>>>> + Send,
    indexer: GrenadParameters,
    lmdb_writer_sx: Sender<Result<TypedChunk>>,
    searchable_fields: Option<HashSet<FieldId>>,
    faceted_fields: HashSet<FieldId>,
    primary_key_id: FieldId,
    geo_fields_ids: Option<(FieldId, FieldId)>,
-    field_id_map: FieldsIdsMap,
+    settings_diff: Arc<InnerIndexSettingsDiff>,
    stop_words: Option<fst::Set<Vec<u8>>>,
    allowed_separators: Option<&[&str]>,
    dictionary: Option<&[&str]>,
    max_positions_per_attributes: Option<u32>,
    exact_attributes: HashSet<FieldId>,
    proximity_precision: ProximityPrecision,
    embedders: EmbeddingConfigs,
 ) -> Result<()> {
    puffin::profile_function!();
@@ -67,8 +58,7 @@ pub(crate) fn data_from_obkv_documents(
                        original_documents_chunk,
                        indexer,
                        lmdb_writer_sx.clone(),
-                        field_id_map.clone(),
+                        settings_diff.clone(),
                        embedders.clone(),
                    )
                })
                .collect::<Result<()>>()
@@ -81,13 +71,9 @@ pub(crate) fn data_from_obkv_documents(
                        flattened_obkv_chunks,
                        indexer,
                        lmdb_writer_sx.clone(),
                        &searchable_fields,
                        &faceted_fields,
                        primary_key_id,
                        geo_fields_ids,
-                        &stop_words,
+                        settings_diff.clone(),
                        &allowed_separators,
                        &dictionary,
                        max_positions_per_attributes,
                    )
                })
@@ -100,13 +86,12 @@ pub(crate) fn data_from_obkv_documents(
                        run_extraction_task::<_, _, grenad::Reader<BufReader<File>>>(
                            docid_word_positions_chunk.clone(),
                            indexer,
                            settings_diff.clone(),
                            lmdb_writer_sx.clone(),
                            extract_fid_word_count_docids,
                            TypedChunk::FieldIdWordCountDocids,
                            "field-id-wordcount-docids",
                        );
                        let exact_attributes = exact_attributes.clone();
                        run_extraction_task::<
                            _,
                            _,
@@ -118,10 +103,9 @@ pub(crate) fn data_from_obkv_documents(
                        >(
                            docid_word_positions_chunk.clone(),
                            indexer,
                            settings_diff.clone(),
                            lmdb_writer_sx.clone(),
-                            move |doc_word_pos, indexer| {
+                            extract_word_docids,
                                extract_word_docids(doc_word_pos, indexer, &exact_attributes)
                            },
                            |(
                                word_docids_reader,
                                exact_word_docids_reader,
@@ -139,6 +123,7 @@ pub(crate) fn data_from_obkv_documents(
                        run_extraction_task::<_, _, grenad::Reader<BufReader<File>>>(
                            docid_word_positions_chunk.clone(),
                            indexer,
                            settings_diff.clone(),
                            lmdb_writer_sx.clone(),
                            extract_word_position_docids,
                            TypedChunk::WordPositionDocids,
@@ -152,6 +137,7 @@ pub(crate) fn data_from_obkv_documents(
                        >(
                            fid_docid_facet_strings_chunk.clone(),
                            indexer,
                            settings_diff.clone(),
                            lmdb_writer_sx.clone(),
                            extract_facet_string_docids,
                            TypedChunk::FieldIdFacetStringDocids,
@@ -161,23 +147,23 @@ pub(crate) fn data_from_obkv_documents(
                        run_extraction_task::<_, _, grenad::Reader<BufReader<File>>>(
                            fid_docid_facet_numbers_chunk.clone(),
                            indexer,
                            settings_diff.clone(),
                            lmdb_writer_sx.clone(),
                            extract_facet_number_docids,
                            TypedChunk::FieldIdFacetNumberDocids,
                            "field-id-facet-number-docids",
                        );
                        if proximity_precision == ProximityPrecision::ByWord {
                        run_extraction_task::<_, _, grenad::Reader<BufReader<File>>>(
                            docid_word_positions_chunk.clone(),
                            indexer,
                            settings_diff.clone(),
                            lmdb_writer_sx.clone(),
                            extract_word_pair_proximity_docids,
                            TypedChunk::WordPairProximityDocids,
                            "word-pair-proximity-docids",
                        );
                    }
                    }
                    Ok(())
                })
@@ -195,12 +181,17 @@ pub(crate) fn data_from_obkv_documents(
 fn run_extraction_task<FE, FS, M>(
    chunk: grenad::Reader<CursorClonableMmap>,
    indexer: GrenadParameters,
    settings_diff: Arc<InnerIndexSettingsDiff>,
    lmdb_writer_sx: Sender<Result<TypedChunk>>,
    extract_fn: FE,
    serialize_fn: FS,
    name: &'static str,
 ) where
-    FE: Fn(grenad::Reader<CursorClonableMmap>, GrenadParameters) -> Result<M>
+    FE: Fn(
            grenad::Reader<CursorClonableMmap>,
            GrenadParameters,
            &InnerIndexSettingsDiff,
        ) -> Result<M>
        + Sync
        + Send
        + 'static,
@@ -213,7 +204,7 @@ fn run_extraction_task<FE, FS, M>(
        let child_span = tracing::trace_span!(target: "indexing::extract::details", parent: &current_span, "extract_multiple_chunks");
        let _entered = child_span.enter();
        puffin::profile_scope!("extract_multiple_chunks", name);
-        match extract_fn(chunk, indexer) {
+        match extract_fn(chunk, indexer, &settings_diff) {
            Ok(chunk) => {
                let _ = lmdb_writer_sx.send(Ok(serialize_fn(chunk)));
            }
@@ -230,8 +221,7 @@ fn send_original_documents_data(
    original_documents_chunk: Result<grenad::Reader<BufReader<File>>>,
    indexer: GrenadParameters,
    lmdb_writer_sx: Sender<Result<TypedChunk>>,
-    field_id_map: FieldsIdsMap,
+    settings_diff: Arc<InnerIndexSettingsDiff>,
    embedders: EmbeddingConfigs,
 ) -> Result<()> {
    let original_documents_chunk =
        original_documents_chunk.and_then(|c| unsafe { as_cloneable_grenad(&c) })?;
@@ -239,17 +229,19 @@ fn send_original_documents_data(
    let documents_chunk_cloned = original_documents_chunk.clone();
    let lmdb_writer_sx_cloned = lmdb_writer_sx.clone();
-    let request_threads = rayon::ThreadPoolBuilder::new()
+    let request_threads = ThreadPoolNoAbortBuilder::new()
        .num_threads(crate::vector::REQUEST_PARALLELISM)
        .thread_name(|index| format!("embedding-request-{index}"))
        .build()?;
    if settings_diff.reindex_vectors() || !settings_diff.settings_update_only() {
        let settings_diff = settings_diff.clone();
        rayon::spawn(move || {
-        for (name, (embedder, prompt)) in embedders {
+            for (name, (embedder, prompt)) in settings_diff.new.embedding_configs.clone() {
                let result = extract_vector_points(
                    documents_chunk_cloned.clone(),
                    indexer,
-                &field_id_map,
+                    &settings_diff,
                    &prompt,
                    &name,
                );
@@ -288,6 +280,7 @@ fn send_original_documents_data(
                }
            }
        });
    }
    // TODO: create a custom internal error
    let _ = lmdb_writer_sx.send(Ok(TypedChunk::Documents(original_documents_chunk)));
@@ -306,13 +299,9 @@ fn send_and_extract_flattened_documents_data(
    flattened_documents_chunk: Result<grenad::Reader<BufReader<File>>>,
    indexer: GrenadParameters,
    lmdb_writer_sx: Sender<Result<TypedChunk>>,
    searchable_fields: &Option<HashSet<FieldId>>,
    faceted_fields: &HashSet<FieldId>,
    primary_key_id: FieldId,
    geo_fields_ids: Option<(FieldId, FieldId)>,
-    stop_words: &Option<fst::Set<Vec<u8>>>,
+    settings_diff: Arc<InnerIndexSettingsDiff>,
    allowed_separators: &Option<&[&str]>,
    dictionary: &Option<&[&str]>,
    max_positions_per_attributes: Option<u32>,
 ) -> Result<(
    grenad::Reader<CursorClonableMmap>,
@@ -341,10 +330,7 @@ fn send_and_extract_flattened_documents_data(
                    extract_docid_word_positions(
                        flattened_documents_chunk.clone(),
                        indexer,
-                        searchable_fields,
+                        &settings_diff,
                        stop_words.as_ref(),
                        *allowed_separators,
                        *dictionary,
                        max_positions_per_attributes,
                    )?;
@@ -367,7 +353,7 @@ fn send_and_extract_flattened_documents_data(
                } = extract_fid_docid_facet_values(
                    flattened_documents_chunk.clone(),
                    indexer,
-                    faceted_fields,
+                    &settings_diff,
                    geo_fields_ids,
                )?;
--- a/milli/src/update/index_documents/mod.rs
+++ b/milli/src/update/index_documents/mod.rs
@@ -6,9 +6,9 @@ mod typed_chunk;
 use std::collections::{HashMap, HashSet};
 use std::io::{Read, Seek};
 use std::iter::FromIterator;
 use std::num::NonZeroU32;
 use std::result::Result as StdResult;
 use std::sync::Arc;
 use crossbeam_channel::{Receiver, Sender};
 use grenad::{Merger, MergerBuilder};
@@ -33,6 +33,7 @@ use self::helpers::{grenad_obkv_into_chunks, GrenadParameters};
 pub use self::transform::{Transform, TransformOutput};
 use crate::documents::{obkv_to_object, DocumentsBatchReader};
 use crate::error::{Error, InternalError, UserError};
 use crate::thread_pool_no_abort::ThreadPoolNoAbortBuilder;
 pub use crate::update::index_documents::helpers::CursorClonableMmap;
 use crate::update::{
    IndexerConfig, UpdateIndexingStep, WordPrefixDocids, WordPrefixIntegerDocids, WordsPrefixesFst,
@@ -259,21 +260,6 @@ where
            .expect("Invalid document addition state")
            .output_from_sorter(self.wtxn, &self.progress)?;
        let new_facets = output.compute_real_facets(self.wtxn, self.index)?;
        self.index.put_faceted_fields(self.wtxn, &new_facets)?;
        // in case new fields were introduced we're going to recreate the searchable fields.
        if let Some(faceted_fields) = self.index.user_defined_searchable_fields(self.wtxn)? {
            // we can't keep references on the faceted fields while we update the index thus we need to own it.
            let faceted_fields: Vec<String> =
                faceted_fields.into_iter().map(str::to_string).collect();
            self.index.put_all_searchable_fields_from_fields_ids_map(
                self.wtxn,
                &faceted_fields.iter().map(String::as_ref).collect::<Vec<_>>(),
                &output.fields_ids_map,
            )?;
        }
        let indexed_documents = output.documents_count as u64;
        let number_of_documents = self.execute_raw(output)?;
@@ -296,32 +282,35 @@ where
        let TransformOutput {
            primary_key,
-            fields_ids_map,
+            mut settings_diff,
            field_distribution,
            documents_count,
            original_documents,
            flattened_documents,
        } = output;
-        // The fields_ids_map is put back to the store now so the rest of the transaction sees an
+        // update the internal facet and searchable list,
-        // up to date field map.
+        // because they might have changed due to the nested documents flattening.
-        self.index.put_fields_ids_map(self.wtxn, &fields_ids_map)?;
+        settings_diff.new.recompute_facets(self.wtxn, self.index)?;
        settings_diff.new.recompute_searchables(self.wtxn, self.index)?;
        let settings_diff = Arc::new(settings_diff);
        let backup_pool;
        let pool = match self.indexer_config.thread_pool {
            Some(ref pool) => pool,
            #[cfg(not(test))]
            None => {
-                // We initialize a bakcup pool with the default
+                // We initialize a backup pool with the default
                // settings if none have already been set.
-                backup_pool = rayon::ThreadPoolBuilder::new().build()?;
+                #[allow(unused_mut)]
-                &backup_pool
+                let mut pool_builder = ThreadPoolNoAbortBuilder::new();
-            }
+
                #[cfg(test)]
-            None => {
+                {
-                // We initialize a bakcup pool with the default
+                    pool_builder = pool_builder.num_threads(1);
-                // settings if none have already been set.
+                }
-                backup_pool = rayon::ThreadPoolBuilder::new().num_threads(1).build()?;
+
                backup_pool = pool_builder.build()?;
                &backup_pool
            }
        };
@@ -333,13 +322,8 @@ where
        ) = crossbeam_channel::unbounded();
        // get the primary key field id
-        let primary_key_id = fields_ids_map.id(&primary_key).unwrap();
+        let primary_key_id = settings_diff.new.fields_ids_map.id(&primary_key).unwrap();
        // get searchable fields for word databases
        let searchable_fields =
            self.index.searchable_fields_ids(self.wtxn)?.map(HashSet::from_iter);
        // get filterable fields for facet databases
        let faceted_fields = self.index.faceted_fields_ids(self.wtxn)?;
        // get the fid of the `_geo.lat` and `_geo.lng` fields.
        let mut field_id_map = self.index.fields_ids_map(self.wtxn)?;
@@ -362,12 +346,6 @@ where
            None => None,
        };
        let stop_words = self.index.stop_words(self.wtxn)?;
        let separators = self.index.allowed_separators(self.wtxn)?;
        let dictionary = self.index.dictionary(self.wtxn)?;
        let exact_attributes = self.index.exact_attributes_ids(self.wtxn)?;
        let proximity_precision = self.index.proximity_precision(self.wtxn)?.unwrap_or_default();
        let pool_params = GrenadParameters {
            chunk_compression_type: self.indexer_config.chunk_compression_type,
            chunk_compression_level: self.indexer_config.chunk_compression_level,
@@ -400,8 +378,6 @@ where
        let max_positions_per_attributes = self.indexer_config.max_positions_per_attributes;
        let cloned_embedder = self.embedders.clone();
        let mut final_documents_ids = RoaringBitmap::new();
        let mut databases_seen = 0;
        let mut word_position_docids = None;
@@ -410,7 +386,6 @@ where
        let mut exact_word_docids = None;
        let mut chunk_accumulator = ChunkAccumulator::default();
        let mut dimension = HashMap::new();
        let stop_words = stop_words.map(|sw| sw.map_data(Vec::from).unwrap());
        let current_span = tracing::Span::current();
@@ -428,10 +403,6 @@ where
                let flattened_chunk_iter =
                    grenad_obkv_into_chunks(flattened_documents, pool_params, documents_chunk_size);
                let separators: Option<Vec<_>> =
                    separators.as_ref().map(|x| x.iter().map(String::as_str).collect());
                let dictionary: Option<Vec<_>> =
                    dictionary.as_ref().map(|x| x.iter().map(String::as_str).collect());
                let result = original_chunk_iter.and_then(|original_chunk| {
                    let flattened_chunk = flattened_chunk_iter?;
                    // extract all databases from the chunked obkv douments
@@ -440,18 +411,10 @@ where
                        flattened_chunk,
                        pool_params,
                        lmdb_writer_sx.clone(),
                        searchable_fields,
                        faceted_fields,
                        primary_key_id,
                        geo_fields_ids,
-                        field_id_map,
+                        settings_diff.clone(),
                        stop_words,
                        separators.as_deref(),
                        dictionary.as_deref(),
                        max_positions_per_attributes,
                        exact_attributes,
                        proximity_precision,
                        cloned_embedder,
                    )
                });
@@ -571,7 +534,7 @@ where
            }
            Ok(())
-        })?;
+        }).map_err(InternalError::from)??;
        // We write the field distribution into the main database
        self.index.put_field_distribution(self.wtxn, &field_distribution)?;
@@ -600,7 +563,8 @@ where
                    writer.build(wtxn, &mut rng, None)?;
                }
                Result::Ok(())
-            })?;
+            })
            .map_err(InternalError::from)??;
        }
        self.execute_prefix_databases(
--- a/milli/src/update/index_documents/transform.rs
+++ b/milli/src/update/index_documents/transform.rs
@@ -1,12 +1,11 @@
 use std::borrow::Cow;
 use std::collections::btree_map::Entry as BEntry;
 use std::collections::hash_map::Entry as HEntry;
-use std::collections::{HashMap, HashSet};
+use std::collections::HashMap;
 use std::fs::File;
 use std::io::{Read, Seek};
 use fxhash::FxHashMap;
 use heed::RoTxn;
 use itertools::Itertools;
 use obkv::{KvReader, KvReaderU16, KvWriter};
 use roaring::RoaringBitmap;
@@ -21,14 +20,17 @@ use super::{IndexDocumentsMethod, IndexerConfig};
 use crate::documents::{DocumentsBatchIndex, EnrichedDocument, EnrichedDocumentsBatchReader};
 use crate::error::{Error, InternalError, UserError};
 use crate::index::{db_name, main_key};
-use crate::update::del_add::{into_del_add_obkv, DelAdd, DelAddOperation, KvReaderDelAdd};
+use crate::update::del_add::{
    del_add_from_two_obkvs, into_del_add_obkv, DelAdd, DelAddOperation, KvReaderDelAdd,
 };
 use crate::update::index_documents::GrenadParameters;
-use crate::update::{AvailableDocumentsIds, ClearDocuments, UpdateIndexingStep};
+use crate::update::settings::{InnerIndexSettings, InnerIndexSettingsDiff};
 use crate::update::{AvailableDocumentsIds, UpdateIndexingStep};
 use crate::{FieldDistribution, FieldId, FieldIdMapMissingEntry, FieldsIdsMap, Index, Result};
 pub struct TransformOutput {
    pub primary_key: String,
-    pub fields_ids_map: FieldsIdsMap,
+    pub settings_diff: InnerIndexSettingsDiff,
    pub field_distribution: FieldDistribution,
    pub documents_count: usize,
    pub original_documents: File,
@@ -282,7 +284,9 @@ impl<'a, 'i> Transform<'a, 'i> {
                    self.original_sorter
                        .insert(&document_sorter_key_buffer, &document_sorter_value_buffer)?;
                    let base_obkv = KvReader::new(base_obkv);
-                    if let Some(flattened_obkv) = self.flatten_from_fields_ids_map(base_obkv)? {
+                    if let Some(flattened_obkv) =
                        Self::flatten_from_fields_ids_map(&base_obkv, &mut self.fields_ids_map)?
                    {
                        // we recreate our buffer with the flattened documents
                        document_sorter_value_buffer.clear();
                        document_sorter_value_buffer.push(Operation::Addition as u8);
@@ -315,7 +319,9 @@ impl<'a, 'i> Transform<'a, 'i> {
                    .insert(&document_sorter_key_buffer, &document_sorter_value_buffer)?;
                let flattened_obkv = KvReader::new(&obkv_buffer);
-                if let Some(obkv) = self.flatten_from_fields_ids_map(flattened_obkv)? {
+                if let Some(obkv) =
                    Self::flatten_from_fields_ids_map(&flattened_obkv, &mut self.fields_ids_map)?
                {
                    document_sorter_value_buffer.clear();
                    document_sorter_value_buffer.push(Operation::Addition as u8);
                    into_del_add_obkv(
@@ -524,7 +530,9 @@ impl<'a, 'i> Transform<'a, 'i> {
        // flatten it and push it as to delete in the flattened_sorter
        let flattened_obkv = KvReader::new(base_obkv);
-        if let Some(obkv) = self.flatten_from_fields_ids_map(flattened_obkv)? {
+        if let Some(obkv) =
            Self::flatten_from_fields_ids_map(&flattened_obkv, &mut self.fields_ids_map)?
        {
            // we recreate our buffer with the flattened documents
            document_sorter_value_buffer.clear();
            document_sorter_value_buffer.push(Operation::Deletion as u8);
@@ -541,8 +549,15 @@ impl<'a, 'i> Transform<'a, 'i> {
    // Flatten a document from the fields ids map contained in self and insert the new
    // created fields. Returns `None` if the document doesn't need to be flattened.
-    #[tracing::instrument(level = "trace", skip(self, obkv), target = "indexing::transform")]
+    #[tracing::instrument(
-    fn flatten_from_fields_ids_map(&mut self, obkv: KvReader<FieldId>) -> Result<Option<Vec<u8>>> {
+        level = "trace",
        skip(obkv, fields_ids_map),
        target = "indexing::transform"
    )]
    fn flatten_from_fields_ids_map(
        obkv: &KvReader<FieldId>,
        fields_ids_map: &mut FieldsIdsMap,
    ) -> Result<Option<Vec<u8>>> {
        if obkv
            .iter()
            .all(|(_, value)| !json_depth_checker::should_flatten_from_unchecked_slice(value))
@@ -563,7 +578,7 @@ impl<'a, 'i> Transform<'a, 'i> {
        // all the raw values get inserted directly in the `key_value` vec.
        for (key, value) in obkv.iter() {
            if json_depth_checker::should_flatten_from_unchecked_slice(value) {
-                let key = self.fields_ids_map.name(key).ok_or(FieldIdMapMissingEntry::FieldId {
+                let key = fields_ids_map.name(key).ok_or(FieldIdMapMissingEntry::FieldId {
                    field_id: key,
                    process: "Flatten from fields ids map.",
                })?;
@@ -581,7 +596,7 @@ impl<'a, 'i> Transform<'a, 'i> {
        // Once we have the flattened version we insert all the new generated fields_ids
        // (if any) in the fields ids map and serialize the value.
        for (key, value) in flattened.into_iter() {
-            let fid = self.fields_ids_map.insert(&key).ok_or(UserError::AttributeLimitReached)?;
+            let fid = fields_ids_map.insert(&key).ok_or(UserError::AttributeLimitReached)?;
            let value = serde_json::to_vec(&value).map_err(InternalError::SerdeJson)?;
            key_value.push((fid, value.into()));
        }
@@ -792,9 +807,19 @@ impl<'a, 'i> Transform<'a, 'i> {
            fst_new_external_documents_ids_builder.insert(key, value)
        })?;
        let old_inner_settings = InnerIndexSettings::from_index(self.index, wtxn)?;
        let mut new_inner_settings = old_inner_settings.clone();
        new_inner_settings.fields_ids_map = self.fields_ids_map;
        let settings_diff = InnerIndexSettingsDiff {
            old: old_inner_settings,
            new: new_inner_settings,
            embedding_configs_updated: false,
            settings_update_only: false,
        };
        Ok(TransformOutput {
            primary_key,
-            fields_ids_map: self.fields_ids_map,
+            settings_diff,
            field_distribution,
            documents_count: self.documents_count,
            original_documents: original_documents.into_inner().map_err(|err| err.into_error())?,
@@ -804,6 +829,44 @@ impl<'a, 'i> Transform<'a, 'i> {
        })
    }
    /// Rebind the field_ids of the provided document to their values
    /// based on the field_ids_maps difference between the old and the new settings,
    /// then fill the provided buffers with delta documents using KvWritterDelAdd.
    fn rebind_existing_document(
        old_obkv: KvReader<FieldId>,
        settings_diff: &InnerIndexSettingsDiff,
        original_obkv_buffer: &mut Vec<u8>,
        flattened_obkv_buffer: &mut Vec<u8>,
    ) -> Result<()> {
        let mut old_fields_ids_map = settings_diff.old.fields_ids_map.clone();
        let mut new_fields_ids_map = settings_diff.new.fields_ids_map.clone();
        let mut obkv_writer = KvWriter::<_, FieldId>::memory();
        // We iterate over the new `FieldsIdsMap` ids in order and construct the new obkv.
        for (id, name) in new_fields_ids_map.iter() {
            if let Some(val) = old_fields_ids_map.id(name).and_then(|id| old_obkv.get(id)) {
                obkv_writer.insert(id, val)?;
            }
        }
        let data = obkv_writer.into_inner()?;
        let new_obkv = KvReader::<FieldId>::new(&data);
        // take the non-flattened version if flatten_from_fields_ids_map returns None.
        let old_flattened = Self::flatten_from_fields_ids_map(&old_obkv, &mut old_fields_ids_map)?;
        let old_flattened =
            old_flattened.as_deref().map_or_else(|| old_obkv, KvReader::<FieldId>::new);
        let new_flattened = Self::flatten_from_fields_ids_map(&new_obkv, &mut new_fields_ids_map)?;
        let new_flattened =
            new_flattened.as_deref().map_or_else(|| new_obkv, KvReader::<FieldId>::new);
        original_obkv_buffer.clear();
        flattened_obkv_buffer.clear();
        del_add_from_two_obkvs(&old_obkv, &new_obkv, original_obkv_buffer)?;
        del_add_from_two_obkvs(&old_flattened, &new_flattened, flattened_obkv_buffer)?;
        Ok(())
    }
    /// Clear all databases. Returns a `TransformOutput` with a file that contains the documents
    /// of the index with the attributes reordered accordingly to the `FieldsIdsMap` given as argument.
    ///
@@ -811,8 +874,7 @@ impl<'a, 'i> Transform<'a, 'i> {
    pub fn prepare_for_documents_reindexing(
        self,
        wtxn: &mut heed::RwTxn<'i>,
-        old_fields_ids_map: FieldsIdsMap,
+        settings_diff: InnerIndexSettingsDiff,
        mut new_fields_ids_map: FieldsIdsMap,
    ) -> Result<TransformOutput> {
        // There already has been a document addition, the primary key should be set by now.
        let primary_key = self
@@ -848,78 +910,27 @@ impl<'a, 'i> Transform<'a, 'i> {
            self.indexer_settings.max_memory.map(|mem| mem / 2),
        );
-        let mut obkv_buffer = Vec::new();
+        let mut original_obkv_buffer = Vec::new();
        let mut flattened_obkv_buffer = Vec::new();
        let mut document_sorter_key_buffer = Vec::new();
        let mut document_sorter_value_buffer = Vec::new();
        for result in self.index.external_documents_ids().iter(wtxn)? {
            let (external_id, docid) = result?;
-            let obkv = self.index.documents.get(wtxn, &docid)?.ok_or(
+            let old_obkv = self.index.documents.get(wtxn, &docid)?.ok_or(
                InternalError::DatabaseMissingEntry { db_name: db_name::DOCUMENTS, key: None },
            )?;
-            obkv_buffer.clear();
+            Self::rebind_existing_document(
-            let mut obkv_writer = KvWriter::<_, FieldId>::new(&mut obkv_buffer);
+                old_obkv,
-
+                &settings_diff,
-            // We iterate over the new `FieldsIdsMap` ids in order and construct the new obkv.
+                &mut original_obkv_buffer,
-            for (id, name) in new_fields_ids_map.iter() {
+                &mut flattened_obkv_buffer,
-                if let Some(val) = old_fields_ids_map.id(name).and_then(|id| obkv.get(id)) {
+            )?;
                    obkv_writer.insert(id, val)?;
                }
            }
            let buffer = obkv_writer.into_inner()?;
            document_sorter_key_buffer.clear();
            document_sorter_key_buffer.extend_from_slice(&docid.to_be_bytes());
            document_sorter_key_buffer.extend_from_slice(external_id.as_bytes());
-            document_sorter_value_buffer.clear();
+            original_sorter.insert(&document_sorter_key_buffer, &original_obkv_buffer)?;
-            into_del_add_obkv(
+            flattened_sorter.insert(docid.to_be_bytes(), &flattened_obkv_buffer)?;
                KvReaderU16::new(buffer),
                DelAddOperation::Addition,
                &mut document_sorter_value_buffer,
            )?;
            original_sorter.insert(&document_sorter_key_buffer, &document_sorter_value_buffer)?;
            // Once we have the document. We're going to flatten it
            // and insert it in the flattened sorter.
            let mut doc = serde_json::Map::new();
            let reader = obkv::KvReader::new(buffer);
            for (k, v) in reader.iter() {
                let key = new_fields_ids_map.name(k).ok_or(FieldIdMapMissingEntry::FieldId {
                    field_id: k,
                    process: "Accessing field distribution in transform.",
                })?;
                let value = serde_json::from_slice::<serde_json::Value>(v)
                    .map_err(InternalError::SerdeJson)?;
                doc.insert(key.to_string(), value);
            }
            let flattened = flatten_serde_json::flatten(&doc);
            // Once we have the flattened version we can convert it back to obkv and
            // insert all the new generated fields_ids (if any) in the fields ids map.
            let mut buffer: Vec<u8> = Vec::new();
            let mut writer = KvWriter::new(&mut buffer);
            let mut flattened: Vec<_> = flattened.into_iter().collect();
            // we reorder the field to get all the known field first
            flattened.sort_unstable_by_key(|(key, _)| {
                new_fields_ids_map.id(key).unwrap_or(FieldId::MAX)
            });
            for (key, value) in flattened {
                let fid =
                    new_fields_ids_map.insert(&key).ok_or(UserError::AttributeLimitReached)?;
                let value = serde_json::to_vec(&value).map_err(InternalError::SerdeJson)?;
                writer.insert(fid, &value)?;
            }
            document_sorter_value_buffer.clear();
            into_del_add_obkv(
                KvReaderU16::new(&buffer),
                DelAddOperation::Addition,
                &mut document_sorter_value_buffer,
            )?;
            flattened_sorter.insert(docid.to_be_bytes(), &document_sorter_value_buffer)?;
        }
        let grenad_params = GrenadParameters {
@@ -934,22 +945,14 @@ impl<'a, 'i> Transform<'a, 'i> {
        let flattened_documents = sorter_into_reader(flattened_sorter, grenad_params)?;
-        let output = TransformOutput {
+        Ok(TransformOutput {
            primary_key,
            fields_ids_map: new_fields_ids_map,
            field_distribution,
            settings_diff,
            documents_count,
            original_documents: original_documents.into_inner().into_inner(),
            flattened_documents: flattened_documents.into_inner().into_inner(),
-        };
+        })
        let new_facets = output.compute_real_facets(wtxn, self.index)?;
        self.index.put_faceted_fields(wtxn, &new_facets)?;
        // We clear the full database (words-fst, documents ids and documents content).
        ClearDocuments::new(wtxn, self.index).execute()?;
        Ok(output)
    }
 }
@@ -964,20 +967,6 @@ fn drop_and_reuse<U, T>(mut vec: Vec<U>) -> Vec<T> {
    vec.into_iter().map(|_| unreachable!()).collect()
 }
 impl TransformOutput {
    // find and insert the new field ids
    pub fn compute_real_facets(&self, rtxn: &RoTxn, index: &Index) -> Result<HashSet<String>> {
        let user_defined_facets = index.user_defined_faceted_fields(rtxn)?;
        Ok(self
            .fields_ids_map
            .names()
            .filter(|&field| crate::is_faceted(field, &user_defined_facets))
            .map(|field| field.to_string())
            .collect())
    }
 }
 #[cfg(test)]
 mod test {
    use super::*;
--- a/milli/src/update/indexer_config.rs
+++ b/milli/src/update/indexer_config.rs
@@ -1,5 +1,6 @@
 use grenad::CompressionType;
-use rayon::ThreadPool;
+
 use crate::thread_pool_no_abort::ThreadPoolNoAbort;
 #[derive(Debug)]
 pub struct IndexerConfig {
@@ -9,7 +10,7 @@ pub struct IndexerConfig {
    pub max_memory: Option<usize>,
    pub chunk_compression_type: CompressionType,
    pub chunk_compression_level: Option<u32>,
-    pub thread_pool: Option<ThreadPool>,
+    pub thread_pool: Option<ThreadPoolNoAbort>,
    pub max_positions_per_attributes: Option<u32>,
    pub skip_index_budget: bool,
 }
--- a/milli/src/update/settings.rs
+++ b/milli/src/update/settings.rs
@@ -20,7 +20,7 @@ use crate::update::index_documents::IndexDocumentsMethod;
 use crate::update::{IndexDocuments, UpdateIndexingStep};
 use crate::vector::settings::{check_set, check_unset, EmbedderSource, EmbeddingSettings};
 use crate::vector::{Embedder, EmbeddingConfig, EmbeddingConfigs};
-use crate::{FieldsIdsMap, Index, Result};
+use crate::{FieldId, FieldsIdsMap, Index, Result};
 #[derive(Debug, Clone, PartialEq, Eq, Copy)]
 pub enum Setting<T> {
@@ -385,14 +385,14 @@ impl<'a, 't, 'i> Settings<'a, 't, 'i> {
    #[tracing::instrument(
        level = "trace"
-        skip(self, progress_callback, should_abort, old_fields_ids_map),
+        skip(self, progress_callback, should_abort, settings_diff),
        target = "indexing::documents"
    )]
    fn reindex<FP, FA>(
        &mut self,
        progress_callback: &FP,
        should_abort: &FA,
-        old_fields_ids_map: FieldsIdsMap,
+        settings_diff: InnerIndexSettingsDiff,
    ) -> Result<()>
    where
        FP: Fn(UpdateIndexingStep) + Sync,
@@ -400,7 +400,6 @@ impl<'a, 't, 'i> Settings<'a, 't, 'i> {
    {
        puffin::profile_function!();
        let fields_ids_map = self.index.fields_ids_map(self.wtxn)?;
        // if the settings are set before any document update, we don't need to do anything, and
        // will set the primary key during the first document addition.
        if self.index.number_of_documents(self.wtxn)? == 0 {
@@ -416,14 +415,7 @@ impl<'a, 't, 'i> Settings<'a, 't, 'i> {
        )?;
        // We clear the databases and remap the documents fields based on the new `FieldsIdsMap`.
-        let output = transform.prepare_for_documents_reindexing(
+        let output = transform.prepare_for_documents_reindexing(self.wtxn, settings_diff)?;
            self.wtxn,
            old_fields_ids_map,
            fields_ids_map,
        )?;
        let embedder_configs = self.index.embedding_configs(self.wtxn)?;
        let embedders = self.embedders(embedder_configs)?;
        // We index the generated `TransformOutput` which must contain
        // all the documents with fields in the newly defined searchable order.
@@ -436,32 +428,11 @@ impl<'a, 't, 'i> Settings<'a, 't, 'i> {
            &should_abort,
        )?;
        let indexing_builder = indexing_builder.with_embedders(embedders);
        indexing_builder.execute_raw(output)?;
        Ok(())
    }
    fn embedders(
        &self,
        embedding_configs: Vec<(String, EmbeddingConfig)>,
    ) -> Result<EmbeddingConfigs> {
        let res: Result<_> = embedding_configs
            .into_iter()
            .map(|(name, EmbeddingConfig { embedder_options, prompt })| {
                let prompt = Arc::new(prompt.try_into().map_err(crate::Error::from)?);
                let embedder = Arc::new(
                    Embedder::new(embedder_options.clone())
                        .map_err(crate::vector::Error::from)
                        .map_err(crate::Error::from)?,
                );
                Ok((name, (embedder, prompt)))
            })
            .collect();
        res.map(EmbeddingConfigs::new)
    }
    fn update_displayed(&mut self) -> Result<bool> {
        match self.displayed_fields {
            Setting::Set(ref fields) => {
@@ -1038,6 +1009,13 @@ impl<'a, 't, 'i> Settings<'a, 't, 'i> {
            }
            Setting::NotSet => false,
        };
        // if any changes force a reindexing
        // clear the vector database.
        if update {
            self.index.vector_arroy.clear(self.wtxn)?;
        }
        Ok(update)
    }
@@ -1066,20 +1044,10 @@ impl<'a, 't, 'i> Settings<'a, 't, 'i> {
    {
        self.index.set_updated_at(self.wtxn, &OffsetDateTime::now_utc())?;
-        // Note: this MUST be before `update_sortable` so that we can get the old value to compare with the updated value afterwards
+        let old_inner_settings = InnerIndexSettings::from_index(self.index, self.wtxn)?;
        let existing_fields: HashSet<_> = self
            .index
            .field_distribution(self.wtxn)?
            .into_iter()
            .filter_map(|(field, count)| (count != 0).then_some(field))
            .collect();
        let old_faceted_fields = self.index.user_defined_faceted_fields(self.wtxn)?;
        let old_fields_ids_map = self.index.fields_ids_map(self.wtxn)?;
        // never trigger re-indexing
        self.update_displayed()?;
        self.update_filterable()?;
        self.update_sortable()?;
        self.update_distinct_field()?;
        self.update_criteria()?;
        self.update_primary_key()?;
@@ -1089,16 +1057,19 @@ impl<'a, 't, 'i> Settings<'a, 't, 'i> {
        self.update_max_values_per_facet()?;
        self.update_sort_facet_values_by()?;
        self.update_pagination_max_total_hits()?;
        self.update_search_cutoff()?;
-        let faceted_updated = self.update_faceted(existing_fields, old_faceted_fields)?;
+        // could trigger re-indexing
-        let stop_words_updated = self.update_stop_words()?;
+        self.update_filterable()?;
-        let non_separator_tokens_updated = self.update_non_separator_tokens()?;
+        self.update_sortable()?;
-        let separator_tokens_updated = self.update_separator_tokens()?;
+        self.update_stop_words()?;
-        let dictionary_updated = self.update_dictionary()?;
+        self.update_non_separator_tokens()?;
-        let synonyms_updated = self.update_synonyms()?;
+        self.update_separator_tokens()?;
-        let searchable_updated = self.update_searchable()?;
+        self.update_dictionary()?;
-        let exact_attributes_updated = self.update_exact_attributes()?;
+        self.update_synonyms()?;
-        let proximity_precision = self.update_proximity_precision()?;
+        self.update_searchable()?;
        self.update_exact_attributes()?;
        self.update_proximity_precision()?;
        // TODO: very rough approximation of the needs for reindexing where any change will result in
        // a full reindexing.
        // What can be done instead:
@@ -1107,53 +1078,193 @@ impl<'a, 't, 'i> Settings<'a, 't, 'i> {
        // 3. Keep the old vectors but reattempt indexing on a prompt change: only actually changed prompt will need embedding + storage
        let embedding_configs_updated = self.update_embedding_configs()?;
-        // never trigger re-indexing
+        let new_inner_settings = InnerIndexSettings::from_index(self.index, self.wtxn)?;
-        self.update_search_cutoff()?;
+        let inner_settings_diff = InnerIndexSettingsDiff {
            old: old_inner_settings,
            new: new_inner_settings,
            embedding_configs_updated,
            settings_update_only: true,
        };
-        if stop_words_updated
+        if inner_settings_diff.any_reindexing_needed() {
-            || non_separator_tokens_updated
+            self.reindex(&progress_callback, &should_abort, inner_settings_diff)?;
            || separator_tokens_updated
            || dictionary_updated
            || faceted_updated
            || synonyms_updated
            || searchable_updated
            || exact_attributes_updated
            || proximity_precision
            || embedding_configs_updated
        {
            self.reindex(&progress_callback, &should_abort, old_fields_ids_map)?;
        }
        Ok(())
    }
    fn update_faceted(
        &self,
        existing_fields: HashSet<String>,
        old_faceted_fields: HashSet<String>,
    ) -> Result<bool> {
        if existing_fields.iter().any(|field| field.contains('.')) {
            return Ok(true);
 }
 pub struct InnerIndexSettingsDiff {
    pub(crate) old: InnerIndexSettings,
    pub(crate) new: InnerIndexSettings,
    // TODO: compare directly the embedders.
    pub(crate) embedding_configs_updated: bool,
    pub(crate) settings_update_only: bool,
 }
 impl InnerIndexSettingsDiff {
    pub fn any_reindexing_needed(&self) -> bool {
        self.reindex_searchable() || self.reindex_facets() || self.reindex_vectors()
    }
    pub fn reindex_searchable(&self) -> bool {
        self.old
            .fields_ids_map
            .iter()
            .zip(self.new.fields_ids_map.iter())
            .any(|(old, new)| old != new)
            || self.old.stop_words.as_ref().map(|set| set.as_fst().as_bytes())
                != self.new.stop_words.as_ref().map(|set| set.as_fst().as_bytes())
            || self.old.allowed_separators != self.new.allowed_separators
            || self.old.dictionary != self.new.dictionary
            || self.old.user_defined_searchable_fields != self.new.user_defined_searchable_fields
            || self.old.exact_attributes != self.new.exact_attributes
            || self.old.proximity_precision != self.new.proximity_precision
    }
    pub fn reindex_facets(&self) -> bool {
        let existing_fields = &self.new.existing_fields;
        if existing_fields.iter().any(|field| field.contains('.')) {
            return true;
        }
        let old_faceted_fields = &self.old.user_defined_faceted_fields;
        if old_faceted_fields.iter().any(|field| field.contains('.')) {
-            return Ok(true);
+            return true;
        }
        // If there is new faceted fields we indicate that we must reindex as we must
        // index new fields as facets. It means that the distinct attribute,
        // an Asc/Desc criterion or a filtered attribute as be added or removed.
-        let new_faceted_fields = self.index.user_defined_faceted_fields(self.wtxn)?;
+        let new_faceted_fields = &self.new.user_defined_faceted_fields;
        if new_faceted_fields.iter().any(|field| field.contains('.')) {
-            return Ok(true);
+            return true;
        }
        let faceted_updated =
-            (&existing_fields - &old_faceted_fields) != (&existing_fields - &new_faceted_fields);
+            (existing_fields - old_faceted_fields) != (existing_fields - new_faceted_fields);
-        Ok(faceted_updated)
+        self.old
            .fields_ids_map
            .iter()
            .zip(self.new.fields_ids_map.iter())
            .any(|(old, new)| old != new)
            || faceted_updated
    }
    pub fn reindex_vectors(&self) -> bool {
        self.embedding_configs_updated
    }
    pub fn settings_update_only(&self) -> bool {
        self.settings_update_only
    }
 }
 #[derive(Clone)]
 pub(crate) struct InnerIndexSettings {
    pub stop_words: Option<fst::Set<Vec<u8>>>,
    pub allowed_separators: Option<BTreeSet<String>>,
    pub dictionary: Option<BTreeSet<String>>,
    pub fields_ids_map: FieldsIdsMap,
    pub user_defined_faceted_fields: HashSet<String>,
    pub user_defined_searchable_fields: Option<Vec<String>>,
    pub faceted_fields_ids: HashSet<FieldId>,
    pub searchable_fields_ids: Option<Vec<FieldId>>,
    pub exact_attributes: HashSet<FieldId>,
    pub proximity_precision: ProximityPrecision,
    pub embedding_configs: EmbeddingConfigs,
    pub existing_fields: HashSet<String>,
 }
 impl InnerIndexSettings {
    pub fn from_index(index: &Index, rtxn: &heed::RoTxn) -> Result<Self> {
        let stop_words = index.stop_words(rtxn)?;
        let stop_words = stop_words.map(|sw| sw.map_data(Vec::from).unwrap());
        let allowed_separators = index.allowed_separators(rtxn)?;
        let dictionary = index.dictionary(rtxn)?;
        let fields_ids_map = index.fields_ids_map(rtxn)?;
        let user_defined_searchable_fields = index.user_defined_searchable_fields(rtxn)?;
        let user_defined_searchable_fields =
            user_defined_searchable_fields.map(|sf| sf.into_iter().map(String::from).collect());
        let user_defined_faceted_fields = index.user_defined_faceted_fields(rtxn)?;
        let searchable_fields_ids = index.searchable_fields_ids(rtxn)?;
        let faceted_fields_ids = index.faceted_fields_ids(rtxn)?;
        let exact_attributes = index.exact_attributes_ids(rtxn)?;
        let proximity_precision = index.proximity_precision(rtxn)?.unwrap_or_default();
        let embedding_configs = embedders(index.embedding_configs(rtxn)?)?;
        let existing_fields: HashSet<_> = index
            .field_distribution(rtxn)?
            .into_iter()
            .filter_map(|(field, count)| (count != 0).then_some(field))
            .collect();
        Ok(Self {
            stop_words,
            allowed_separators,
            dictionary,
            fields_ids_map,
            user_defined_faceted_fields,
            user_defined_searchable_fields,
            faceted_fields_ids,
            searchable_fields_ids,
            exact_attributes,
            proximity_precision,
            embedding_configs,
            existing_fields,
        })
    }
    // find and insert the new field ids
    pub fn recompute_facets(&mut self, wtxn: &mut heed::RwTxn, index: &Index) -> Result<()> {
        let new_facets = self
            .fields_ids_map
            .names()
            .filter(|&field| crate::is_faceted(field, &self.user_defined_faceted_fields))
            .map(|field| field.to_string())
            .collect();
        index.put_faceted_fields(wtxn, &new_facets)?;
        self.faceted_fields_ids = index.faceted_fields_ids(wtxn)?;
        Ok(())
    }
    // find and insert the new field ids
    pub fn recompute_searchables(&mut self, wtxn: &mut heed::RwTxn, index: &Index) -> Result<()> {
        // in case new fields were introduced we're going to recreate the searchable fields.
        if let Some(searchable_fields) = self.user_defined_searchable_fields.as_ref() {
            let searchable_fields =
                searchable_fields.iter().map(String::as_ref).collect::<Vec<_>>();
            index.put_all_searchable_fields_from_fields_ids_map(
                wtxn,
                &searchable_fields,
                &self.fields_ids_map,
            )?;
            let searchable_fields_ids = index.searchable_fields_ids(wtxn)?;
            self.searchable_fields_ids = searchable_fields_ids;
        }
        Ok(())
    }
 }
 fn embedders(embedding_configs: Vec<(String, EmbeddingConfig)>) -> Result<EmbeddingConfigs> {
    let res: Result<_> = embedding_configs
        .into_iter()
        .map(|(name, EmbeddingConfig { embedder_options, prompt })| {
            let prompt = Arc::new(prompt.try_into().map_err(crate::Error::from)?);
            let embedder = Arc::new(
                Embedder::new(embedder_options.clone())
                    .map_err(crate::vector::Error::from)
                    .map_err(crate::Error::from)?,
            );
            Ok((name, (embedder, prompt)))
        })
        .collect();
    res.map(EmbeddingConfigs::new)
 }
 fn validate_prompt(
@@ -1643,6 +1754,70 @@ mod tests {
            .unwrap()
            .count();
        assert_eq!(count, 4);
        // Set the filterable fields to be the age and the name.
        index
            .update_settings(|settings| {
                settings.set_filterable_fields(hashset! { S("age"),  S("name") });
            })
            .unwrap();
        // Check that the displayed fields are correctly set.
        let rtxn = index.read_txn().unwrap();
        let fields_ids = index.filterable_fields(&rtxn).unwrap();
        assert_eq!(fields_ids, hashset! { S("age"),  S("name") });
        let rtxn = index.read_txn().unwrap();
        // Only count the field_id 0 and level 0 facet values.
        let count = index
            .facet_id_f64_docids
            .remap_key_type::<Bytes>()
            .prefix_iter(&rtxn, &[0, 1, 0])
            .unwrap()
            .count();
        assert_eq!(count, 4);
        let rtxn = index.read_txn().unwrap();
        // Only count the field_id 0 and level 0 facet values.
        let count = index
            .facet_id_string_docids
            .remap_key_type::<Bytes>()
            .prefix_iter(&rtxn, &[0, 0])
            .unwrap()
            .count();
        assert_eq!(count, 5);
        // Remove the age from the filterable fields.
        index
            .update_settings(|settings| {
                settings.set_filterable_fields(hashset! { S("name") });
            })
            .unwrap();
        // Check that the displayed fields are correctly set.
        let rtxn = index.read_txn().unwrap();
        let fields_ids = index.filterable_fields(&rtxn).unwrap();
        assert_eq!(fields_ids, hashset! { S("name") });
        let rtxn = index.read_txn().unwrap();
        // Only count the field_id 0 and level 0 facet values.
        let count = index
            .facet_id_f64_docids
            .remap_key_type::<Bytes>()
            .prefix_iter(&rtxn, &[0, 1, 0])
            .unwrap()
            .count();
        assert_eq!(count, 0);
        let rtxn = index.read_txn().unwrap();
        // Only count the field_id 0 and level 0 facet values.
        let count = index
            .facet_id_string_docids
            .remap_key_type::<Bytes>()
            .prefix_iter(&rtxn, &[0, 0])
            .unwrap()
            .count();
        assert_eq!(count, 5);
    }
    #[test]
--- a/milli/src/vector/error.rs
+++ b/milli/src/vector/error.rs
@@ -3,6 +3,7 @@ use std::path::PathBuf;
 use hf_hub::api::sync::ApiError;
 use crate::error::FaultSource;
 use crate::PanicCatched;
 #[derive(Debug, thiserror::Error)]
 #[error("Error while generating embeddings: {inner}")]
@@ -80,6 +81,8 @@ pub enum EmbedErrorKind {
    OpenAiUnexpectedDimension(usize, usize),
    #[error("no embedding was produced")]
    MissingEmbedding,
    #[error(transparent)]
    PanicInThreadPool(#[from] PanicCatched),
 }
 impl EmbedError {
--- a/milli/src/vector/mod.rs
+++ b/milli/src/vector/mod.rs
@@ -7,6 +7,7 @@ use serde::{Deserialize, Serialize};
 use self::error::{EmbedError, NewEmbedderError};
 use crate::prompt::{Prompt, PromptData};
 use crate::ThreadPoolNoAbort;
 pub mod error;
 pub mod hf;
@@ -254,7 +255,7 @@ impl Embedder {
    pub fn embed_chunks(
        &self,
        text_chunks: Vec<Vec<String>>,
-        threads: &rayon::ThreadPool,
+        threads: &ThreadPoolNoAbort,
    ) -> std::result::Result<Vec<Vec<Embeddings<f32>>>, EmbedError> {
        match self {
            Embedder::HuggingFace(embedder) => embedder.embed_chunks(text_chunks),
--- a/milli/src/vector/ollama.rs
+++ b/milli/src/vector/ollama.rs
@@ -3,6 +3,8 @@ use rayon::iter::{IntoParallelIterator as _, ParallelIterator as _};
 use super::error::{EmbedError, EmbedErrorKind, NewEmbedderError, NewEmbedderErrorKind};
 use super::rest::{Embedder as RestEmbedder, EmbedderOptions as RestEmbedderOptions};
 use super::{DistributionShift, Embeddings};
 use crate::error::FaultSource;
 use crate::ThreadPoolNoAbort;
 #[derive(Debug)]
 pub struct Embedder {
@@ -71,11 +73,16 @@ impl Embedder {
    pub fn embed_chunks(
        &self,
        text_chunks: Vec<Vec<String>>,
-        threads: &rayon::ThreadPool,
+        threads: &ThreadPoolNoAbort,
    ) -> Result<Vec<Vec<Embeddings<f32>>>, EmbedError> {
-        threads.install(move || {
+        threads
            .install(move || {
                text_chunks.into_par_iter().map(move |chunk| self.embed(chunk)).collect()
            })
            .map_err(|error| EmbedError {
                kind: EmbedErrorKind::PanicInThreadPool(error),
                fault: FaultSource::Bug,
            })?
    }
    pub fn chunk_count_hint(&self) -> usize {
--- a/milli/src/vector/openai.rs
+++ b/milli/src/vector/openai.rs
@@ -4,7 +4,9 @@ use rayon::iter::{IntoParallelIterator, ParallelIterator as _};
 use super::error::{EmbedError, NewEmbedderError};
 use super::rest::{Embedder as RestEmbedder, EmbedderOptions as RestEmbedderOptions};
 use super::{DistributionShift, Embeddings};
 use crate::error::FaultSource;
 use crate::vector::error::EmbedErrorKind;
 use crate::ThreadPoolNoAbort;
 #[derive(Debug, Clone, Hash, PartialEq, Eq, serde::Deserialize, serde::Serialize)]
 pub struct EmbedderOptions {
@@ -241,11 +243,16 @@ impl Embedder {
    pub fn embed_chunks(
        &self,
        text_chunks: Vec<Vec<String>>,
-        threads: &rayon::ThreadPool,
+        threads: &ThreadPoolNoAbort,
    ) -> Result<Vec<Vec<Embeddings<f32>>>, EmbedError> {
-        threads.install(move || {
+        threads
            .install(move || {
                text_chunks.into_par_iter().map(move |chunk| self.embed(chunk)).collect()
            })
            .map_err(|error| EmbedError {
                kind: EmbedErrorKind::PanicInThreadPool(error),
                fault: FaultSource::Bug,
            })?
    }
    pub fn chunk_count_hint(&self) -> usize {
--- a/milli/src/vector/rest.rs
+++ b/milli/src/vector/rest.rs
@@ -2,9 +2,12 @@ use deserr::Deserr;
 use rayon::iter::{IntoParallelIterator as _, ParallelIterator as _};
 use serde::{Deserialize, Serialize};
 use super::error::EmbedErrorKind;
 use super::{
    DistributionShift, EmbedError, Embedding, Embeddings, NewEmbedderError, REQUEST_PARALLELISM,
 };
 use crate::error::FaultSource;
 use crate::ThreadPoolNoAbort;
 // retrying in case of failure
@@ -158,11 +161,16 @@ impl Embedder {
    pub fn embed_chunks(
        &self,
        text_chunks: Vec<Vec<String>>,
-        threads: &rayon::ThreadPool,
+        threads: &ThreadPoolNoAbort,
    ) -> Result<Vec<Vec<Embeddings<f32>>>, EmbedError> {
-        threads.install(move || {
+        threads
            .install(move || {
                text_chunks.into_par_iter().map(move |chunk| self.embed(chunk)).collect()
            })
            .map_err(|error| EmbedError {
                kind: EmbedErrorKind::PanicInThreadPool(error),
                fault: FaultSource::Bug,
            })?
    }
    pub fn chunk_count_hint(&self) -> usize {
--- a/milli/src/vector/settings.rs
+++ b/milli/src/vector/settings.rs
@@ -301,10 +301,14 @@ impl From<EmbeddingConfig> for EmbeddingSettings {
    fn from(value: EmbeddingConfig) -> Self {
        let EmbeddingConfig { embedder_options, prompt } = value;
        match embedder_options {
-            super::EmbedderOptions::HuggingFace(options) => Self {
+            super::EmbedderOptions::HuggingFace(super::hf::EmbedderOptions {
                model,
                revision,
                distribution,
            }) => Self {
                source: Setting::Set(EmbedderSource::HuggingFace),
-                model: Setting::Set(options.model),
+                model: Setting::Set(model),
-                revision: options.revision.map(Setting::Set).unwrap_or_default(),
+                revision: revision.map(Setting::Set).unwrap_or_default(),
                api_key: Setting::NotSet,
                dimensions: Setting::NotSet,
                document_template: Setting::Set(prompt.template),
@@ -314,14 +318,19 @@ impl From<EmbeddingConfig> for EmbeddingSettings {
                path_to_embeddings: Setting::NotSet,
                embedding_object: Setting::NotSet,
                input_type: Setting::NotSet,
-                distribution: options.distribution.map(Setting::Set).unwrap_or_default(),
+                distribution: distribution.map(Setting::Set).unwrap_or_default(),
            },
-            super::EmbedderOptions::OpenAi(options) => Self {
+            super::EmbedderOptions::OpenAi(super::openai::EmbedderOptions {
                api_key,
                embedding_model,
                dimensions,
                distribution,
            }) => Self {
                source: Setting::Set(EmbedderSource::OpenAi),
-                model: Setting::Set(options.embedding_model.name().to_owned()),
+                model: Setting::Set(embedding_model.name().to_owned()),
                revision: Setting::NotSet,
-                api_key: options.api_key.map(Setting::Set).unwrap_or_default(),
+                api_key: api_key.map(Setting::Set).unwrap_or_default(),
-                dimensions: options.dimensions.map(Setting::Set).unwrap_or_default(),
+                dimensions: dimensions.map(Setting::Set).unwrap_or_default(),
                document_template: Setting::Set(prompt.template),
                url: Setting::NotSet,
                query: Setting::NotSet,
@@ -329,29 +338,37 @@ impl From<EmbeddingConfig> for EmbeddingSettings {
                path_to_embeddings: Setting::NotSet,
                embedding_object: Setting::NotSet,
                input_type: Setting::NotSet,
-                distribution: options.distribution.map(Setting::Set).unwrap_or_default(),
+                distribution: distribution.map(Setting::Set).unwrap_or_default(),
            },
-            super::EmbedderOptions::Ollama(options) => Self {
+            super::EmbedderOptions::Ollama(super::ollama::EmbedderOptions {
                embedding_model,
                url,
                api_key,
                distribution,
            }) => Self {
                source: Setting::Set(EmbedderSource::Ollama),
-                model: Setting::Set(options.embedding_model.to_owned()),
+                model: Setting::Set(embedding_model),
                revision: Setting::NotSet,
-                api_key: Setting::NotSet,
+                api_key: api_key.map(Setting::Set).unwrap_or_default(),
                dimensions: Setting::NotSet,
                document_template: Setting::Set(prompt.template),
-                url: Setting::NotSet,
+                url: url.map(Setting::Set).unwrap_or_default(),
                query: Setting::NotSet,
                input_field: Setting::NotSet,
                path_to_embeddings: Setting::NotSet,
                embedding_object: Setting::NotSet,
                input_type: Setting::NotSet,
-                distribution: options.distribution.map(Setting::Set).unwrap_or_default(),
+                distribution: distribution.map(Setting::Set).unwrap_or_default(),
            },
-            super::EmbedderOptions::UserProvided(options) => Self {
+            super::EmbedderOptions::UserProvided(super::manual::EmbedderOptions {
                dimensions,
                distribution,
            }) => Self {
                source: Setting::Set(EmbedderSource::UserProvided),
                model: Setting::NotSet,
                revision: Setting::NotSet,
                api_key: Setting::NotSet,
-                dimensions: Setting::Set(options.dimensions),
+                dimensions: Setting::Set(dimensions),
                document_template: Setting::NotSet,
                url: Setting::NotSet,
                query: Setting::NotSet,
@@ -359,7 +376,7 @@ impl From<EmbeddingConfig> for EmbeddingSettings {
                path_to_embeddings: Setting::NotSet,
                embedding_object: Setting::NotSet,
                input_type: Setting::NotSet,
-                distribution: options.distribution.map(Setting::Set).unwrap_or_default(),
+                distribution: distribution.map(Setting::Set).unwrap_or_default(),
            },
            super::EmbedderOptions::Rest(super::rest::EmbedderOptions {
                api_key,
--- a/tracing-trace/src/processor/firefox_profiler.rs
+++ b/tracing-trace/src/processor/firefox_profiler.rs
@@ -217,9 +217,7 @@ fn add_memory_samples(
    memory_counters: &mut Option<MemoryCounterHandles>,
    last_memory: &mut MemoryStats,
 ) -> Option<MemoryStats> {
-    let Some(stats) = memory else {
+    let stats = memory?;
        return None;
    };
    let memory_counters =
        memory_counters.get_or_insert_with(|| MemoryCounterHandles::new(profile, main));
--- a/workloads/movies-subset-hf-embeddings.json
+++ b/workloads/movies-subset-hf-embeddings.json
@@ -0,0 +1,68 @@
 {
  "name": "movies-subset-hf-embeddings",
  "run_count": 5,
  "extra_cli_args": [
    "--max-indexing-threads=4"
  ],
  "assets": {
    "movies-100.json": {
      "local_location": null,
      "remote_location": "https://milli-benchmarks.fra1.digitaloceanspaces.com/bench/datasets/movies-100.json",
      "sha256": "d215e395e4240f12f03b8f1f68901eac82d9e7ded5b462cbf4a6b8efde76c6c6"
    }
  },
  "commands": [
    {
      "route": "experimental-features",
      "method": "PATCH",
      "body": {
        "inline": {
          "vectorStore": true
        }
      },
      "synchronous": "DontWait"
    },
    {
      "route": "indexes/movies/settings",
      "method": "PATCH",
      "body": {
        "inline": {
          "searchableAttributes": [
            "title",
            "overview"
          ],
          "filterableAttributes": [
            "genres",
            "release_date"
          ],
          "sortableAttributes": [
            "release_date"
          ]
        }
      },
      "synchronous": "WaitForTask"
    },
    {
      "route": "indexes/movies/settings",
      "method": "PATCH",
      "body": {
        "inline": {
          "embedders": {
            "default": {
              "source": "huggingFace"
            }
          }
        }
      },
      "synchronous": "WaitForTask"
    },
    {
      "route": "indexes/movies/documents",
      "method": "POST",
      "body": {
        "asset": "movies-100.json"
      },
      "synchronous": "WaitForTask"
    }
  ]
 }
--- a/workloads/settings-add-embeddings.json
+++ b/workloads/settings-add-embeddings.json
@@ -0,0 +1,72 @@
 {
  "name": "settings-add-embeddings-hf",
  "run_count": 5,
  "extra_cli_args": [
    "--max-indexing-threads=4"
  ],
  "assets": {
    "movies-100.json": {
      "local_location": null,
      "remote_location": "https://milli-benchmarks.fra1.digitaloceanspaces.com/bench/datasets/movies-100.json",
      "sha256": "d215e395e4240f12f03b8f1f68901eac82d9e7ded5b462cbf4a6b8efde76c6c6"
    }
  },
  "commands": [
    {
      "route": "experimental-features",
      "method": "PATCH",
      "body": {
        "inline": {
          "vectorStore": true
        }
      },
      "synchronous": "DontWait"
    },
    {
      "route": "indexes/movies/settings",
      "method": "PATCH",
      "body": {
        "inline": {
          "searchableAttributes": [
            "title",
            "overview"
          ],
          "filterableAttributes": [
            "genres",
            "release_date"
          ],
          "sortableAttributes": [
            "release_date"
          ]
        }
      },
      "synchronous": "DontWait"
    },
    {
      "route": "indexes/movies/documents",
      "method": "POST",
      "body": {
        "asset": "movies-100.json"
      },
      "synchronous": "WaitForTask"
    },
    {
      "route": "indexes/movies/settings",
      "method": "PATCH",
      "body": {
        "inline": {
          "embedders": {
            "default": {
              "source": "huggingFace",
              "model": null,
              "revision": null,
              "documentTemplate": null,
              "distribution": null
            }
          }
        }
      },
      "synchronous": "WaitForTask"
    }
  ]
 }
--- a/workloads/settings-add-remove-filters.json
+++ b/workloads/settings-add-remove-filters.json
@@ -1,6 +1,6 @@
 {
  "name": "settings-add-remove-filters.json",
-  "run_count": 2,
+  "run_count": 5,
  "extra_cli_args": [
    "--max-indexing-threads=4"
  ],
--- a/workloads/settings-proximity-precision.json
+++ b/workloads/settings-proximity-precision.json
@@ -1,6 +1,6 @@
 {
  "name": "settings-proximity-precision.json",
-  "run_count": 2,
+  "run_count": 5,
  "extra_cli_args": [
    "--max-indexing-threads=4"
  ],
--- a/workloads/settings-remove-add-swap-searchable.json
+++ b/workloads/settings-remove-add-swap-searchable.json
@@ -1,6 +1,6 @@
 {
  "name": "settings-remove-add-swap-searchable.json",
-  "run_count": 2,
+  "run_count": 5,
  "extra_cli_args": [
    "--max-indexing-threads=4"
  ],
--- a/workloads/settings-typo.json
+++ b/workloads/settings-typo.json
@@ -1,6 +1,6 @@
 {
  "name": "settings-typo.json",
-  "run_count": 2,
+  "run_count": 5,
  "extra_cli_args": [
    "--max-indexing-threads=4"
  ],