Splitup std.regex testsuite to avoid spurious failures on the auto-tester#6164
Splitup std.regex testsuite to avoid spurious failures on the auto-tester#6164wilzbach wants to merge 1 commit intodlang:masterfrom
Conversation
|
Thanks for your pull request, @wilzbach! Bugzilla referencesYour PR doesn't reference any Bugzilla issue. If your PR contains non-trivial changes, please reference a Bugzilla issue or create a manual changelog. |
f1ee990 to
f85f2ef
Compare
f85f2ef to
1a1a55d
Compare
Another idea is to deprecate ctRegex and stop testing it. It's 100% obvious that it doesn't work on anything significant in practice due to our compiler limitations. Then we enable it again once CTFE works with std.regex. ATM it's not even remotely fully tested in CI and I feel guilty for advertising things we do not test. |
Sounds very reasonably to me, though I'm not sure on whether we should actually trigger the deprecation message or just remove it from the docs. |
|
Undocumenting it, perhaps, but removing it might be too much. Move it to its separate module, maybe? The compiler is very likely to only keep improving. As a semi-aside, and I might be completely wrong about this, but one thing I've noticed to be receiving a lot of attention is Unicode-correctness in |
|
The rest of the language is unicode correct (mostly). I don't think we can justify not supporting it when D is built around it, and our higher level competitors support it as well |
|
Closing this as it's not the real fix, but we really should solve the problem of ctRegex or at least deprecate it for now. |
I know that this feels wrong (it is!), but we frequently get this error on auto-tester:
See also: braddr/d-tester#66, braddr/d-tester#69
newCTFE still seems far away from becoming reality (CC @UplinkCoder)
We already tried splitting it up into two pieces, but
tests.dseems to be the big part (Split-up regex tests to please auto-tester #6061)At the moment only a part of the testsuite is actually run at CT:
=> I guess we are left to splitting it up again. As currently it already fails with only 35 elements of the queue, I went for 10 elements per file.
Disclaimer: I don't want to do this, but we gotta fight the CI errors somehow.
Better ideas are more than welcome!