The V8 port hasn't been updated since 2013 it seems.
There have been *very* important improvements in between, including API changes and many performance and stability improvements on all platforms.
Maybe lang/v8 should be at v5.1 and lang/v8-devel at v5.5 (or at least v8-devel should be at v5.1)
Thank you very much.
The problem is the build structure, so someone needs to provide a patch
for the update.
I think, maintainers of www/chromium or devel/electron4 can help with that. Current release of v8 uses gn build system, which is also used by chromium and electron.
The 5.1 port referenced here fails with a gyp error, but maybe it's an easy fix.
I have a wip port: https://github.com/MikaelUrankar/v8-ports
Created attachment 215743 [details]
Created attachment 215746 [details]
Created attachment 215773 [details]
I have testbuilt in poudriere on cur-amd, cur-i386, 11.4-amd and it was fine.
It somehow failed on 12.1 ?
(In reply to Kurt Jaeger from comment #9)
It requires llvm10, I've uploaded the wrong patch :(
Created attachment 215995 [details]
(In reply to Mikael Urankar from comment #11)
Thanks, built was fine on cur-a, cur-i, 12.1a and 11.4a.
Created attachment 216145 [details]
Only install llvm10 on 12.1 (11.4 and 13 already ship clang10 in base)
Add cppgc include
testbuilds are fine after I added on pkg-plist line for:
Error: Orphaned: include/cppgc/internal/caged-heap-local-data.h
Created attachment 216165 [details]
Created attachment 216175 [details]
put distfiles on freefall
testbuilds are fine, thanks!
Thanks, Mikael. I'll test it ASAP.
BTW, lang/v8 should track the stable release (8.3) which is 184.108.40.206 now.
I also received a build error for lang/v8 after updating from FreeBSD 11.3 to FreeBSD 11.4. Could someone point me in the direction of applying this patch while using Poudriere. I have downloaded the files from Mikael and uploaded them to my server. I have tried a few different ways to implement this patch with Poudriere but all have failed. Any help is appreciated.
You can do that:
git clone -b 8.3 email@example.com:MikaelUrankar/v8-ports.git
mv /usr/local/poudriere/ports/default/lang/v8 /usr/local/poudriere/ports/default/lang/v8.orig
mv v8-ports /usr/local/poudriere/ports/default/lang/v8
Thanks for the assistance. I was able to move the files to the necessary directory and began the build again. We then received a new error with the chromium-gn build. Poudriere output:
[00:01:26] Failed ports: devel/chromium-gn:extract
Reviewed logs and found "Write to restore size failed." at the end of nearly every line.
chromium-83.0.4103.116/storage/common/file_system/file_system_mount_option.h: Write to restore size failed
chromium-83.0.4103.116/storage/common/file_system/OWNERS: Write to restore size failed
Here was the output from the end of file:
tar: Error exit delayed from previous errors.
*** Error code 1
make: stopped in /usr/ports/devel/chromium-gn
=>> Cleaning up wrkdir
===> Cleaning for chromium-gn-83.0.4103.116_1
build of devel/chromium-gn | chromium-gn-83.0.4103.116_1 ended at Mon Jul 27 13:35:24 CDT 2020
build time: 00:02:54
!!! build failure encountered !!!
Here is the FreeBSD version and arch:
Jail Version : 11.4-RELEASE-p1 amd64
Thanks again for the assistance.
(In reply to Ryan D from comment #23)
Weird error, never saw that before, what filesystem / mount option do you use?
Is it the only port failing?
Using the ZFS file system. It is the only port that fails. Once it fails, lang/v8 and databases/postgresql-plv8js is skipped due to its failing . There is plenty of space on the file system. Unsure which info you need about the mount options but here is the output relevant to mounting when running a Poudriere build.
[00:00:01] Mounting system devices for freebsd_11-4x64-HEAD
[00:00:01] Mounting ports/packages/distfiles
[00:00:01] Stashing existing package repository
[00:00:01] Mounting packages from: /usr/local/poudriere/data/packages/freebsd_11-4x64-HEAD
This system is running on a bare metal server.
(In reply to Ryan D from comment #25)
Can you try to untar the distfiles with archivers/gtar?
(In reply to Mikael Urankar from comment #26)
This worked. I was able to untar the files in a test directory.