Cherry Pick for release 1.9: [NFC][HLSLExec][LongVectors] Make derivative test literals explicitly typed#8532
Open
ruiji-shi wants to merge 1 commit into
Conversation
Wrap the bare integer literals in the DerivativeDdx/DerivativeDdy expected-result expressions in T(...) so the arithmetic resolves cleanly under clang's stricter overload resolution. Previously: `((A + 2) - (A + 0))` — `A` is of template parameter type `T` (e.g. `HLSLHalf_t`, `HLSLMin16Float_t`). `T + int` has no unambiguous overload for those wrapper types and clang rejects it; MSVC accepted via implicit-conversion ranking. Now: `((A + T(2.0f)) - (A + T(0.0f)))` — the literal is constructed as `T` first, so `operator+` resolves to the type's own implementation. The value is bit-identical to the previous code path on MSVC. No behavior change. Affects only the 4 `DEFAULT_OP_1(OpType::DerivativeDd[xy][Fine?], ...)` rows around line 1352.
Collaborator
|
Can you explain why you need this? These tests are only for testing DXC not Clang and we generally don't mutate shipped releases for things that aren't critical bug fixes. |
alsepkow
changed the title
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Cherry pick 505629f from #8516
Wrap the bare integer literals in the DerivativeDdx/DerivativeDdy expected-result expressions in T(...) so the arithmetic resolves cleanly under clang's stricter overload resolution.
Previously:
((A + 2) - (A + 0))—Ais of template parameter typeT(e.g.HLSLHalf_t,HLSLMin16Float_t).T + inthas no unambiguous overload for those wrapper types and clang rejects it; MSVC accepted via implicit-conversion ranking.Now:
((A + T(2.0f)) - (A + T(0.0f)))— the literal is constructed asTfirst, sooperator+resolves to the type's own implementation. The value is bit-identical to the previous code path on MSVC.No behavior change. Affects only the 4
DEFAULT_OP_1(OpType::DerivativeDd[xy][Fine?], ...)rows around line 1352.