Summary: | net-libs/webkit-gtk-2.411-r1 JSInternalSettingsGenerated.h: No such file or directory when MAKEOPTS > -j1 | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Sergiusz M <protryt> |
Component: | [OLD] GNOME | Assignee: | Gentoo Linux Gnome Desktop Team <gnome> |
Status: | RESOLVED OBSOLETE | ||
Severity: | major | CC: | ajacoutot, aurelien, brakarov, gentoo, gottlieb, maratbn, Martin.vGagern, nbowler, rossi.f, z23 |
Priority: | Normal | Keywords: | NeedPatch |
Version: | unspecified | ||
Hardware: | AMD64 | ||
OS: | Linux | ||
See Also: | https://bugs.gentoo.org/show_bug.cgi?id=598102 | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: |
Build log of webkit-gtk-2.4.3
2.4.7-r200 emerge-pqv.txt 2.4.7-r200 emerge-info.txt force_makeopts_j1 |
Description
Sergiusz M
2014-07-01 07:21:38 UTC
Created attachment 380030 [details]
Build log of webkit-gtk-2.4.3
Could you retry with MAKEOPTS="-j1"? (In reply to Pacho Ramos from comment #2) > Could you retry with MAKEOPTS="-j1"? I have tried and after 3 hours, compilation hanged out on: ./Source/WebKit/gtk/webkit/*.cpp Source/WebKit/gtk/webkit/webkitversion.h:37: Warning: WebKit: symbol='WEBKITGTK_API_VERSION': Unknown namespace for symbol 'WEBKITGTK_API_VERSION' I have switched from nvidia to xorg-x11 and everything goes well. I experienced the same error, so I don't see this as FIXED. Please reopen. (In reply to Pacho Ramos from comment #2) > Could you retry with MAKEOPTS="-j1"? Did solve the issue for me. So this seems to be a parallel build issue. make_settings.pl is listed in line 316/4229 as well as line 439 of the failed build, so I'd guess that JSInternalSettingsGenerated.idl should exist pretty early. I have no clue why it is apparently called twice. The call to generate-bindings.pl which should build JSInternalSettingsGenerated.h from this does not appear in the log of either the failed nor the succeeding build. Not sure what to conclude from all this. (In reply to Sergiusz M from comment #4) > I have switched from nvidia to xorg-x11 and everything goes well. You switched WHAT? OpenGL implementation? In any case, this seems to have nothing to do with the originally reported issue, and I did no such switch. Also having this problem when nvidia OpenGL enabled. Running into same error with net-libs/webkit-gtk-2.4.7-r200 since using hardened profile + pax kernel. Will try with -j1. Created attachment 390200 [details]
2.4.7-r200 emerge-pqv.txt
Created attachment 390202 [details]
2.4.7-r200 emerge-info.txt
build.log here: https://snscltt.info/bugzilla/webkit-gtk/build.log as it is far beyond the 1000kB (In reply to Alexander Hof from comment #7) > Running into same error with net-libs/webkit-gtk-2.4.7-r200 since using > hardened profile + pax kernel. Will try with -j1. Building with -j1 succeded. I can confirm this on webkit-gtk-2.4.8; "-j1" fixed it for me as well. Created attachment 400758 [details, diff]
force_makeopts_j1
I've added -j1 to MAKEOPTS in src_compile, I guess this would fix this bug.
Is this an acceptable fix for the issue? Can a dev confirm this is according to ebuild policy?
Hitting this with 2.4.9 -j1 would result in unbearable build for me. Better add proper dependency in the Makefiles. It can't be that hard. Switching opengl away from nvidia works for me in 2.4.9 Switching back and logging into GNOME... compilation still works :/ Hitting the same on a pure Intel system with mesa and whatnot. Issues like these are purely hit by luck, when the race condition from lacking Makefile dependencies gets hit. Not sure if to try again and maybe succeed this time, or to use MAKEOPTS=-j1 and go 4 times slower. As far as this upgrade goes. Can't really work on a patch of any sort at this time, but commenting to confirm the bug outside graphics or something. This is with net-libs/webkit-gtk-2.4.9::gentoo now Same problem with webkit-gtk-2.4.11-r1. MAKEOPTS=-j1 solves the issue. (Intel + Mesa) Still hitting this with 2.4.11-r1 Same here with 2.4.11-r1 Bug still present with net-libs/webkit-gtk-2.4.11-r200, passed with -J1 So this is a 3 years old bug, hitting people randomly. The build system for this software is broken but there is a known workaround: why isn't this workaround implemented? Or at least an error message pointing people to this? Sorry, this version should have left the tree long ago due to over a hundred security issues, but there are still consumers that are slow to port to the newer webkit-gtk API or will never do so. Priority has been elsewhere for me than getting rid of webkit-gtk-2.4, but you should try to avoid it if at all possible. glsa-check -l affected should show it as well afaik. This old SLOT has been p.masked, please remove it if you still have it. # Mart Raudsepp <leio@gentoo.org> (23 Feb 2018) # Old net-libs/webkit-gtk SLOTs have hundreds of known security issues. # Use the security safe net-libs webkit-gtk SLOT=4 instead via # libraries and applications ported to gtk3 and webkit2gtk API. # Masked for removal in 30 days. Bug #577068. # Please keep this package.mask entry until at least 25th May 2018 for # extra notification of the security vulnerabilities. net-libs/webkit-gtk:2 net-libs/webkit-gtk:3 |