diff options
| author | 许杰友 Jieyou Xu (Joe) <39484203+jieyouxu@users.noreply.github.com> | 2024-06-09 19:16:21 +0100 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-06-09 19:16:21 +0100 |
| commit | 3720757d21180224d22376cee8d927d4d52760cd (patch) | |
| tree | 3d59b0cec7569d174c1f583f5b3ae4febbce13d4 /compiler/rustc_codegen_llvm/src | |
| parent | 74408b186247daa869689c6c0647a99f0a93f4ef (diff) | |
| parent | fb6fbf02a65ad1fc766aeee7c7698907d56f8e12 (diff) | |
| download | rust-3720757d21180224d22376cee8d927d4d52760cd.tar.gz rust-3720757d21180224d22376cee8d927d4d52760cd.zip | |
Rollup merge of #125963 - workingjubilee:remove-hashes-from-codegen-tests, r=Mark-Simulacrum
Remove hard-coded hashes from codegen tests This removes hard-coded hashes from the codegen and assembly tests. These use FileCheck, which supports eliding part of the pattern being matched, including by capturing it as a pattern parameter for later matching-on. This is much more appropriate than asking contributors to engage with deliberately-opaque identifier schemes. In order to reduce the likelihood of error, every hash-coded segment I've touched now expects a certain length. This correctly represents these cases, as our hash outputs have a predetermined amount of entropy attached to them. This is not done for the UI test suite as those are comparatively easy to simply `--bless`, whereas that would be inappropriate for codegen tests. It is also not done for debuginfo tests as those tests do not support such elision in a correct and useful way.
Diffstat (limited to 'compiler/rustc_codegen_llvm/src')
0 files changed, 0 insertions, 0 deletions