fix(datatypes): infer the correct shape for many existing ops #9334
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
cpcloud
requested changes
Jun 8, 2024
cpcloud
changed the title
cpcloud
added
bug
cpcloud
approved these changes
Jul 16, 2024
Usually when using eg duckdb's REGEXP_REPLACE() function, the pattern and replacement are always scalar constants. But actually it can accept columnar for all three of it's arguments. These sort of assumptions appeared to be all over the place in the code base. I found these fixs by grepping for "shape_like" and manually looking at all the instances. I didn't add any test cases, that felt like a monumental thing to do I wasn't willing to put the time in for. But IDK, I'm not sure I would even want to put in all the tests, it would be SO much boilerplate. To deal with Nodes sometimes having optional args, I modified shape_like() to be more flexible. IDK, that was sort of a lazy approach, we could be much more verbose and have each op do this filtering individually, but I didn't that was worth it.
|
@cpcloud hopefully that was OK for me to hit the merge button, the CI passed! Thank you! |
NickCrews
added a commit
to NickCrews/ibis
that referenced
this pull request
Jul 16, 2024
This is pinning down the expected behavior for cases before tackling the case() to cases() switch in ibis-project#9096 so that PR can be simpler I move the validation for comparable-ness down into the operation so that the logic is consolidated to one place. in ibis-project#9096 there might be multiple places that construct an ops.SimpleCase, and we don't want to have to implement the validation in all calling locations. We could consider relaxing the limitation for non-empty cases later, but for now lets be strict. I already fixed the shape of ops.SearchedCase in ibis-project#9334, but it looks like in that PR I forgot to also fix ops.SimpleCase, so I do that fix here.
cpcloud
pushed a commit
to NickCrews/ibis
that referenced
this pull request
Jul 23, 2024
This is pinning down the expected behavior for cases before tackling the case() to cases() switch in ibis-project#9096 so that PR can be simpler I move the validation for comparable-ness down into the operation so that the logic is consolidated to one place. in ibis-project#9096 there might be multiple places that construct an ops.SimpleCase, and we don't want to have to implement the validation in all calling locations. We could consider relaxing the limitation for non-empty cases later, but for now lets be strict. I already fixed the shape of ops.SearchedCase in ibis-project#9334, but it looks like in that PR I forgot to also fix ops.SimpleCase, so I do that fix here.
cpcloud
pushed a commit
to NickCrews/ibis
that referenced
this pull request
Jul 24, 2024
This is pinning down the expected behavior for cases before tackling the case() to cases() switch in ibis-project#9096 so that PR can be simpler I move the validation for comparable-ness down into the operation so that the logic is consolidated to one place. in ibis-project#9096 there might be multiple places that construct an ops.SimpleCase, and we don't want to have to implement the validation in all calling locations. We could consider relaxing the limitation for non-empty cases later, but for now lets be strict. I already fixed the shape of ops.SearchedCase in ibis-project#9334, but it looks like in that PR I forgot to also fix ops.SimpleCase, so I do that fix here.
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.
Found this when looking into #9333
Usually when using eg duckdb's REGEXP_REPLACE() function, the pattern and replacement are always scalar constants. But actually it can accept columnar for all three of it's arguments.
These sort of assumptions appeared to be all over the place in the code base. I found these fixs by grepping for "shape_like" and manually looking at all the instances.
I didn't add any test cases, that felt like a monumental thing to do I wasn't willing to put the time in for. But IDK, I'm not sure I would even want to put in all the tests, it would be SO much boilerplate.
To deal with Nodes sometimes having optional args, I modified shape_like() to be more flexible.
IDK, that was sort of a lazy approach, we could be much more verbose and have each op do this filtering individually, but I didn't that was worth it.