-
Notifications
You must be signed in to change notification settings - Fork 54
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
feat: upgrade to v8 11.6 #66
Conversation
"The system cannot execute the specified program." But it's not saying what program and I can't really divine it from the logs... I'm guessing it's builtins-generated from cmake/GenerateBuiltinsList.cmake but not 100% sure. |
More like the long path issue on windows. Need a new way to handle torque. |
Long path or long argument list? torque.exe ends up D:/a/v8-cmake/build, right? |
long argument list |
That's what I thought. CMake has a bunch of response file variables (like Torque doesn't support reading from response files but that's easy to add. |
Forgot to mention the obvious: invoking torque multiple times with shorter argument lists - but I don't know if that actually works. Easy enough to test though. |
Edited: ah, I did not notice, it has been on CI at least. I will give it a go locally then. |
macOS and linux should works. windows still has the path issues. |
@gengjiawen Great. P. S. Then I just need to find someone who can help sorting out powerpc assembler a bit :) I have 8.3.110.13 building on macOS ppc, but a) something still needs to be fixed for it to work correctly and b) from 8.4.x+ upstream has broken ppc somewhat further. |
How did you get around the long argument list issue? FWIW, the internet suggests long paths are enabled on GHA Windows runners but git needs to opt in, presumably like so: diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 89498826..cfed0378 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -18,7 +18,9 @@ jobs:
steps:
- uses: actions/checkout@v2
- name: setup
- run: cmake -E make_directory ${{runner.workspace}}/build
+ run: |
+ git config core.longpaths true
+ cmake -E make_directory ${{runner.workspace}}/build
- name: configure
run: cmake ${{runner.workspace}}/v8-cmake
working-directory: ${{runner.workspace}}/build Maybe cmake needs to opt in too, somehow. |
Kinda on a side-note, but on macOS long paths can also be a problem. Example: AlphaSparse/Library#19 |
I turned on long-path support on my windows. Not sure if it's GHA Windows revert long path support ? |
Hm, still failing with this:
Unclear what exit code 9020 signifies. |
Can this done per file (sacrifice perf) on windows ? |
Hm, that's weird. https://github.com/actions/runner-images/blob/62a46d0fd8/images/windows/scripts/build/Configure-BaseImage.ps1#L73 suggests it's (still?) enabled on 2022 runners. It was definitely enabled on 2019 runners. |
Looks still a messy for windows 2019. They break it for somehow. |
@bnoordhuis The problem part is add_custom_command(
COMMAND
torque
-o ${PROJECT_BINARY_DIR}/torque-generated
-v8-root ${PROJECT_SOURCE_DIR}/v8
${torque_files}
DEPENDS
torque
${torque_dirs}
${torque_files_abs}
OUTPUT
${torque-outputs}
${torque_outputs}
)
foreach(file IN LISTS torque_outputs)
add_custom_command(
OUTPUT ${file}
COMMAND ${CMAKE_COMMAND} -E touch ${file}
DEPENDS
torque
${torque_dirs}
${torque_files_abs}
)
endforeach() |
How do you figure? I went through the CI results for the last four commits but I'm not sure how you came to that conclusion. I do see a lot of warnings like these:
Culminating - eventually - in this error:
|
Normally win build failed in 2m. When I skip |
@gengjiawen can you rebase on top of master? I just merged a commit that - with some luck - fixes the Windows build. |
@bnoordhuis Looks like build passed! |
@bnoordhuis Let's merge this ? |
Yes sorry, lots of things going on at the moment :) |
@bnoordhuis Could this be released in tags? |
Thank you! |
fix #65
fix #42