469: add authorize typo setting r=Kerollmops a=MarinPostma
This PR adds support for an authorize typo settings. This makes is possible to disable typos for a whole index. Typos are enabled by default.
Co-authored-by: ad hoc <postma.marin@protonmail.com>
480: Increase benchmarks (push) CI timeout r=Kerollmops a=Kerollmops
This PR fixes the fact that the benchmarks CI on push were [canceled by GitHub](https://github.com/meilisearch/milli/actions/runs/2028844132) because they reached the default timeout of 6h. This PR changes the timeout to 72h, the same setting as the manually triggered benchmark one.
Co-authored-by: Kerollmops <clement@meilisearch.com>
479: Update version (v0.24.1) r=Kerollmops a=curquiza
From v0.23.1 to v0.24.1 since we had an issue with the versionning for the previous release
Co-authored-by: Clémentine Urquizar <clementine@meilisearch.com>
475: Bump tokenizer r=Kerollmops a=irevoire
This PR bump the tokenizer in v0.2.9 which fixes an issue we had with lindera where reqwest was used with openssl (which was breaking our benchmarks).
Co-authored-by: Irevoire <tamo@meilisearch.com>
476: Rollback meilisearch-tokenizer version r=Kerollmops a=irevoire
Lindera often fails to download some data from google drive we can’t compile consistently meilisearch / milli.
We can’t bump to the latest version (that moved out of google drive) either because lindera uses reqwest with openssl with no way of configuring it our benchmarks were not able to run. The latter issue should be fixed by https://github.com/lindera-morphology/lindera/pull/164.
Co-authored-by: Irevoire <tamo@meilisearch.com>
472: Remove useless variables in proximity r=Kerollmops a=ManyTheFish
Was passing by plane sweep algorithm to find some inspiration, and I discover that we have useless variables that were not detected because of the recursive function.
Co-authored-by: ManyTheFish <many@meilisearch.com>
468: Add a new error message when the filterableAttributes are empty r=Kerollmops a=brunoocasali
Fixes https://github.com/meilisearch/meilisearch/issues/2140
Is there a good way to reduce de duplication here? Maybe adding a shared function? I don't know the best and idiomatic way to do that, I appreciate any tip!
Another doubt is related to the duplication of the calling:
```rs
// filter.rs:373
FilterError::AttributeNotFilterable {
attribute,
filterable: filterable_fields.into_iter().collect::<Vec<_>>().join(" "),
},
```
and
```rs
// filter.rs:424
return Err(point[0].as_external_error(FilterError::AttributeNotFilterable {
attribute: "_geo",
filterable: filterable_fields.into_iter().collect::<Vec<_>>().join(" "),
}))?;
```
I think we could make the `filterable_fields.into_iter().collect::<Vec<_>>().join(" ")` directly into the error handling like the sortable error. I made it into the last commit, if this is something to avoid, let me know and I can remove it :)
Co-authored-by: Bruno Casali <brunoocasali@gmail.com>
466: Bump version to 0.23.1 r=curquiza a=Kerollmops
This PR bumps the crate versions to 0.23.1. Nothing seems to be breaking in the next release.
Co-authored-by: Kerollmops <clement@meilisearch.com>
467: optimize prefix database r=Kerollmops a=MarinPostma
This pr introduces two optimizations that greatly improve the speed of computing prefix databases.
- The time that it takes to create the prefix FST has been divided by 5 by inverting the way we iterated over the words FST.
- We unconditionally and needlessly checked for documents to remove in `word_prefix_pair`, which caused an iteration over the whole database.
Co-authored-by: ad hoc <postma.marin@protonmail.com>
465: Update dependencies r=ManyTheFish a=Kerollmops
This PR upgrade and updates this crate's dependencies but first, it removes three dependencies that we don't use anymore. I used [cargo udeps](https://github.com/est31/cargo-udeps) to upgrade them ⬆️
Co-authored-by: Kerollmops <clement@meilisearch.com>
Co-authored-by: Clément Renault <clement@meilisearch.com>
464: exporting heed to avoid having different versions of Heed in Meilisearch r=curquiza a=psvnlsaikumar
# Pull Request
## What does this PR do?
Fixes the issue in meilisearch https://github.com/meilisearch/meilisearch/issues/2210
## PR checklist
Please check if your PR fulfills the following requirements:
- [x] Does this PR fix an existing issue?
- [x] Have you read the contributing guidelines?
- [x] Have you made sure that the title is accurate and descriptive of the changes?
Thank you so much for contributing to Meilisearch!
Co-authored-by: psvnl sai kumar <psvnlsaikumar@gmail.com>
457: Avoid iterating on big databases when useless r=Kerollmops a=Kerollmops
This PR makes the prefix database updates to avoid iterating on big grenad files when it is unnecessary. We introduced this regression in #436 but it went unnoticed.
---
According to the following benchmark results, we take more time when we index documents in one run than before #436. It looks like it is probably due to the fact that, now, instead of computing the prefixes database by iterating on the LMDB we directly iterate on the grenad file. Those could be slower to iterate on and could be the slowdown cause.
I just pushed a commit that tests this branch with the new unreleased version of grenad where some work was done to speed up the iteration on grenad files. [The benchmarks for this last commit](https://github.com/meilisearch/milli/actions/runs/1927187408) are currently running. You can [see the diff](https://github.com/meilisearch/grenad/compare/v0.4.1...main) between the v0.4 and the unreleased v0.5 version of grenad.
```diff
group indexing_benchmark-multi-batch-indexing-before-speed-up_45f52620 indexing_stop-iterating-on-big-grenad-files_ac8b85c4
----- ---------------------------------------------------------------- ----------------------------------------------------
+ indexing/Indexing songs in three batches with default settings 1.12 57.7±2.14s ? ?/sec 1.00 51.3±2.76s ? ?/sec
- indexing/Indexing wiki 1.00 917.3±30.01s ? ?/sec 1.10 1008.4±38.27s ? ?/sec
+ indexing/Indexing wiki in three batches 1.10 1091.2±32.73s ? ?/sec 1.00 995.5±24.33s ? ?/sec
```
Co-authored-by: Kerollmops <clement@meilisearch.com>
461: Add a new error message when the `valid_fields` is empty r=curquiza a=brunoocasali
I've created a test case to handle the new error formatting behavior, but I'm not sure if:
- this is the right place to add the test?
- this is the best way to test this behavior?
And I'm not sure also regarding the `match` implementation, is this something required? Or maybe just an `if` statement is ok as well?
I left the two messages literally without "reusing the prefix" in the implementation because I think this could help the "searchability" of the error in the future.
# Pull Request
## What does this PR do?
Fixes https://github.com/meilisearch/meilisearch/issues/2140
## PR checklist
Please check if your PR fulfills the following requirements:
- [x] Does this PR fix an existing issue?
- [ ] Have you read the contributing guidelines?
- [x] Have you made sure that the title is accurate and descriptive of the changes?
Thank you so much for contributing to Meilisearch!
Co-authored-by: Bruno Casali <brunoocasali@gmail.com>
462: cli improvements r=Kerollmops a=MarinPostma
a few improvements:
- use bufreader to load documents, so the loading of the document doesn't appear on flamegraphs
- set default db path to current directory so the `-i` flag can be omitted.
Co-authored-by: ad hoc <postma.marin@protonmail.com>
> "Attribute `{}` is not sortable. This index doesn't have configured sortable attributes."
> "Attribute `{}` is not sortable. Available sortable attributes are: `{}`."
coexist in the error handling
459: Update heed link in cargo toml r=Kerollmops a=curquiza
Since grenad and heed have been moved to the meilisearch orga, this PR changes the link.
This is a minor change since GitHub handles automatically the redirection. This PR is only for consisitency.
Co-authored-by: Clémentine Urquizar <clementine@meilisearch.com>
456: Remove useless grenad merging r=Kerollmops a=Kerollmops
This PR must be merged after #454.
This PR removes the part of code that was merging all of the grenad Readers merging that we don't need as the indexer should have merged them and, therefore, we should only have one final grenad Reader. We reduce the amount of CPU usage and memory pressure we were doing uselessly.
`@ManyTheFish` are you sure I can skip merging the `word_docids` database?
Here is the benchmark comparison with the previously merged PR #454:
```
group indexing_reintroduce-appending-sorted-values_c05e42a8 indexing_remove-useless-grenad-merging_d5b8b5a2
----- ----------------------------------------------------- -----------------------------------------------
indexing/Indexing movies with default settings 1.06 16.6±1.04s ? ?/sec 1.00 15.7±0.93s ? ?/sec
indexing/Indexing songs with default settings 1.16 60.1±7.07s ? ?/sec 1.00 51.7±5.98s ? ?/sec
indexing/Indexing songs without faceted numbers 1.06 55.4±6.14s ? ?/sec 1.00 52.2±4.13s ? ?/sec
```
And the comparison with multi-batch indexing before #436, we can see that we gain time for benchmarks that index datasets in multiple batches but there is _so much_ variance that it's not clear.
```
group indexing_benchmark-multi-batch-indexing-before-speed-up_45f52620 indexing_remove-useless-grenad-merging_d5b8b5a2
----- ---------------------------------------------------------------- -----------------------------------------------
indexing/Indexing geo_point 1.07 6.6±0.08s ? ?/sec 1.00 6.2±0.11s ? ?/sec
indexing/Indexing songs in three batches with default settings 1.12 57.7±2.14s ? ?/sec 1.00 51.5±3.80s ? ?/sec
indexing/Indexing songs with default settings 1.00 47.5±2.52s ? ?/sec 1.09 51.7±5.98s ? ?/sec
indexing/Indexing songs without any facets 1.00 43.5±1.43s ? ?/sec 1.12 48.8±3.73s ? ?/sec
indexing/Indexing songs without faceted numbers 1.00 47.1±2.23s ? ?/sec 1.11 52.2±4.13s ? ?/sec
indexing/Indexing wiki 1.00 917.3±30.01s ? ?/sec 1.09 998.7±38.92s ? ?/sec
indexing/Indexing wiki in three batches 1.09 1091.2±32.73s ? ?/sec 1.00 996.5±15.70s ? ?/sec
```
What do you think `@irevoire?` Should we change the benchmarks to make them do more runs?
Co-authored-by: Kerollmops <clement@meilisearch.com>
