Bug 233821 – [libpas] make build.sh, test.sh, build_and_test.sh, and clean.sh work on Linux

• Component: bmalloc
• Status: RESOLVED FIXED

[Filip Pizlo]

...

[Zan Dobersek]

Created attachment 445984 [details]
WIP patch

This integrates CMake invocation into the build.sh script. The patch just needs some minor refinements before it's reviewable. Test invocation works fine as it's currently done.

This covers invoking the build. There's still different build problems like missing includes or Darwin-specific code.

[Zan Dobersek]

Created attachment 446008 [details]
Patch

[Filip Pizlo]

Comment on attachment 446008 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=446008&action=review

Thank you for doing this!  I just think that build_variant, in its current form, would break us.

> Source/bmalloc/libpas/build.sh:98
> +    case $sdk in
> +        macosx.internal)
> +            build_variant_xcodebuild $@
> +            ;;
> +    esac

This isn't adequate for us - we can have $sdk be:

- macosx
- macosx.internal
- iphoneos
- iphoneos.internal
- maybe other things too, but definitely not Linux ;-)

Also, the build_variant_linux seems like it would probably work on other non-Xcode systems, so "Linux" might be a bad name.

What if you set $sdk to "cmake" instead of "linux" and this was just "case $sdk in cmake) build_variant_cmake ;; *) build_variant_xcodebuild esac" or something like that?

[Zan Dobersek]

I see. "cmake" as an SDK makes sense, and other variants can default to build_variant_xcodebuild.

[Zan Dobersek]

Created attachment 446128 [details]
Patch

[Radar WebKit Bug Importer]

<rdar://problem/86325940>

[Yusuke Suzuki]

Comment on attachment 446128 [details]
Patch

r=me

[EWS]

Committed r286934 (245160@main): <https://commits.webkit.org/245160@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 446128 [details].

[Yusuke Suzuki]

Committed r287016 (245221@trunk): <https://commits.webkit.org/245221@trunk>