Build the compiler with PCRE2 #13084
Conversation
|
Converted to draft. There are a couple of CI issues with PCRE2 that need to be resolved first. |
straight-shoota
force-pushed
the
feature/compiler-pcre2
branch
2 times, most recently
from
8be84d6
to
cdf9edc
Compare
straight-shoota
force-pushed
the
feature/compiler-pcre2
branch
from
f8ba190
to
c3a2949
Compare
|
CI is all green. This is ready for review now. |
.github/workflows/linux.yml
Outdated
| @@ -15,13 +15,18 @@ jobs: | |||
| runs-on: ubuntu-latest | |||
| strategy: | |||
| matrix: | |||
| crystal_bootstrap_version: [1.2.2, 1.3.2, 1.4.1, 1.5.1, 1.6.2, 1.7.2] | |||
| crystal_bootstrap_version: [1.2.2, 1.3.2, 1.4.1, 1.5.1, 1.6.2] | |||
| env: ["USE_PCRE1=true"] | |||
There was a problem hiding this comment.
Hash notation seems more natural:
| env: ["USE_PCRE1=true"] | |
| env: | |
| USE_PCRE1: true |
There was a problem hiding this comment.
Oh shit, I've noticed just now that's not jobs.<job_id>.env but rather a matrix variable called env. In this case using hash notation rather won't work, sorry for the mixup. btw, I'd suggest renaming this variable to something more descriptive, like flags, so it gives better understanding what is it for and won't mixup with the env key, which is a thing for GHA.
There was a problem hiding this comment.
@straight-shoota This most likely broke the CI :(
There was a problem hiding this comment.
Oh shit. Thanks for noticing.
What's really, really bad about this is that the GitHub Actions summary on the PR and commit doesn't mention this. The workflow file failed to parse but that's not reported as a failure? WTF.
Co-authored-by: Sijawusz Pur Rahnama <sija@sija.pl>
The explicit selection of PCRE2 via
-D use_pcre2is intentional. We want the compiler to built with PCRE2 and not silently fall back to PCRE.