Fix pairing of function parameters. #5225
Conversation
|
Reviewing the changes to those test files was a bear. Maybe someone should write a program that helps one ignore extraneous differences in... never mind. |
|
@ShabbyX If you could approve the CI run for this PR, then I can address whatever problems it shows up. |
|
The CI-macos-clang-release-bazel and CI-shaderc-smoketest are not your problem. Please run your change through clang format. You can use git clang-format if you are not sure how to do that. |
jimblandy
force-pushed
the
use-type-pairings-for-parameters
branch
from
09888a1
to
9d5cd69
Compare
|
Not sure if you missed a push, I don't see any diff w.r.t the resolved comments 🤔 |
|
Sorry, that was just the formatting, I'm working on the comments right now. (Thought I'd resolved them, and then wanted to tweak them a bit more.) |
- Consider prior type pairings when attempting to pair function parameters by type. - Pair all parameters that have matching types, not just the first. - Update diff tests. Fixes KhronosGroup#5218.
jimblandy
force-pushed
the
use-type-pairings-for-parameters
branch
from
9d5cd69
to
6b50a18
Compare
|
Okay, I think 6b50a18 should address all comments. |
|
There's a general didactic principle, "concrete before abstract", so I changed the comment to start with the concrete example of pairing buckets by type. It's supposed to be easier to follow the example when holding that specific use case in mind. Then I used the verb 'pair' for what one does with buckets, and 'match' for what one does with ids. |
|
Nice, thanks. Eventually, it would be nice to have the id "namespace" (src or dst) find its way to the type (as opposed to just |
|
@s-perron This is ready to merge. |
Fixes #5218.