2444: add boilerplate for dump v5 r=MarinPostma a=MarinPostma
add the boilerplate files for dump v5
Co-authored-by: ad hoc <postma.marin@protonmail.com>
Co-authored-by: ManyTheFish <many@meilisearch.com>
2410: Make dump a task r=Kerollmops a=MarinPostma
This PR transforms the dump task into a proper task.
The `GET /dumps/:dump_uid` is removed.
Some changes were made to make this work, and a bit a refactoring was necessary.
- The `dump_actor` module has been renamed do `dumps` and moved to the root
- There isn't a `DumpActor` anymore, and the dump process is handled by the `DumpHandler`.
- The `TaskPerformer` is renamed to `BatchHandler`
- The `BatchHandler` trait no longer has a `perform_job` method, but instead has a `accept` method returning whether a handler can proccess a batch
- The scheduler now accept a list of `BatchHandler`, and iterates trhough them until it finds one to accept the current batch.
- `Job` doesn't exist anymore, and everything in now inside of the `BatchContent` enum.
- The `Vec<TaskId>` from `Batch` is replaced with a `BatchContent` enum which hints at the content.
- The Scheduler is slightly modified to accept batch, and prioritize them before regular tasks.
- The `TaskList` are not identified by a `String` representing the index uid anymore, but by a `TaskListIdentifier` which also works for dumps which are not targeting any specific indexes.
- The `GET /dump/:dump_id` no longer exists
- `DumpActorError` is renamed to `DumpError`
close#2410
Co-authored-by: ad hoc <postma.marin@protonmail.com>
2429: Send the analytics to `telemetry.meilisearch.com` instead of segment r=MarinPostma a=irevoire
Fix#2425
Co-authored-by: Irevoire <tamo@meilisearch.com>
538: speedup exact words r=Kerollmops a=MarinPostma
This PR make `exact_words` return an `Option` instead of an empty set, since set creation is costly, as noticed by `@kerollmops.`
I was not convinces that this was the cause for all of the performance drop we measured, and then realized that methods that initialized it were called recursively which caused initialization times to add up. While the first fix solves the issue when not using exact words, using exact word remained way more expensive that it should be. To address this issue, the exact words are cached into the `Context`, so they are only initialized once.
Co-authored-by: ad hoc <postma.marin@protonmail.com>
536: Improves ranking rules error message r=Kerollmops a=matthias-wright
This PR improves the ranking rules error message to properly reflect the case sensitivity.
The issue was highlighted in [meilisearch/issues/2407](https://github.com/meilisearch/meilisearch/issues/2407).
Cheers!
Co-authored-by: Matthias Wright <matthias.s.wright@gmail.com>
2357: chore(dump): add dump tests r=Kerollmops a=irevoire
Add tests on the import of dump v1, v2, v3 and v4.
Since the dumps are slow to decompress, I made the `flate2` crate always compile in optimized.
And since they're also slow to index, I also made the `milli` crate always compile in optimized. What do you think of this `@MarinPostma?`
Should we keep milli unoptimized in case it could help us debug some things? 👀
Co-authored-by: Tamo <tamo@meilisearch.com>
