[TST] Property Test Generation Fixes #2383
Open
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…nt in favor of point test
Reviewer ChecklistPlease leverage this checklist to ensure your code review is thorough before approving Testing, Bugs, Errors, Logs, Documentation
System Compatibility
Quality
|
This stack of pull requests is managed by Graphite. Learn more about stacking. |
) Closes #2377 #2379 ## Description of changes *Summarize the changes made by this PR.* - Improvements & Bug fixes - Making dimension and version lookup optional in the Collection model creation in fastapi client ## Test plan *How are these changes tested?* - [x] Tests pass locally with `pytest` for python, `yarn test` for js, `cargo test` for rust ## Documentation Changes N/A
1 task
## Description of changes
Fix a typo in comment section in chromadb/db/system.py
```
"""
Create a new collection any (-> and) associated resources
in the SysDB.
"""
```
## Test plan
Do not need test
## Documentation Change
Not public facing API documentation change.
Merging into stacked branch
…nt in favor of point test
HammadB
changed the title
HammadB
commented
Jun 20, 2024
HammadB
commented
Jun 20, 2024
HammadB
commented
Jun 20, 2024
codetheweb
reviewed
Jun 20, 2024
| metadatas = [] | ||
| for i in range(len(ids)): | ||
| metadatas.append(generated_metadatas[i % len(generated_metadatas)]) | ||
|
|
atroyn
approved these changes
Jun 20, 2024
| @@ -75,6 +118,8 @@ def test_add( | |||
| ) | |||
|
|
|||
|
|
|||
| # Hypothesis struggles to generate large record sets so we explicitly create | |||
| # a large record set | |||
| def create_large_recordset( | |||
There was a problem hiding this comment.
Feel like we could add some more randomization in here. For example, all embeddings are the same - this is guaranteed to produce a bad HNSW graph. Unrelated to the focus of this PR However.
There was a problem hiding this comment.
Agreed, I tried to replace this with hypothesis but still need to do some munging, I cut myself a task
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description of changes
Summarize the changes made by this PR.
The primary intent of this PR is to remove the is_metadata_valid invariant which was a workaround for our metadata strategy generating faulty metadata and then us special casing all uses of the record set strategy to handle invalid generations. This PR patches the metadata generation to not generate invalid metadata.
Adds modes in test_add to add a medium sized record set. This was initially timing out in hypothesis's generation. Hypothesis bounds the buffer size of the bytes it uses to do random generation, so generating larger metadata was resulting in examples being marked at OVERRUN by conjecture (gleaned from issues like Tests fail with StopTest (OVERRUN) when generating a random integer (strategies.randoms) HypothesisWorks/hypothesis#3999 + reading hypothesis code + stepping through it). This PR adds the ability to generate N fixed metadata entries and uniformly distribute them over the record set, reducing the overall entropy.
Fixes a bug that test_embeddings was not handling None as a possible metadata state, since this state was never generated. Added an explicit test for this.
Fixes a bug in the reference filtering implementation in test_filtering that did not handle None metadata since that state was never generated.
This PR is forced to touch types related to metadata, which are incorrect and cause typing errors. I ignored the errors to minimize the surface area of this change and defer those changes to the pass mentioned in #2292.
Test plan
How are these changes tested?
These changes are covered by existing tests, and
pytestfor python,yarn testfor js,cargo testfor rustDocumentation Changes
No external changes required.