see attachments Reproducible: Always
Created attachment 252479 [details] build.log
Created attachment 252481 [details] emerge --info
Had the same issue. MAKEOPTS="-j1" did it for me (had -j2).
YEAH! With MAKEOPTS="-j1" it compiles. How can i force this options in the ebuild?
Hm. Some with MAKEOPTS="-j4" it compiles.
Maybe upstream should be aware of this parallel make issue :-/
Hmm. Successful assembly with MAKEOPTS = "-j4" with gcc-4.5.1 (AMD Phenon x3), and failed with MAKEOPTS = "-j3" with gcc-4.4.4-r1 (Intel Core Duo)...
*** Bug 344121 has been marked as a duplicate of this bug. ***
(In reply to comment #7) > Hmm. Successful assembly with MAKEOPTS = "-j4" with gcc-4.5.1 (AMD Phenon x3), > > and failed with MAKEOPTS = "-j3" with gcc-4.4.4-r1 (Intel Core Duo)... > Indeed, this is most likely not "always reproducible", as I can successfully link the modules on my laptop with MAKEOPTS="-j2" (AMD Turion x2 in long mode), also with MAKEOPTS="-j4" on my desktop computer (AMD Phenom II x4 in long mode), although both computers have gcc-4.5.1 and make-3.82, which are still flagged as unstable in the Portage main tree. Maybe comrade b1254633 should first install gcc-4.5.1, switch to it, and then try it again, before me make further fuss about this.
Created attachment 253705 [details, diff] changeset_r71535.diff Upstream patch that could solve this
(In reply to comment #10) > Created an attachment (id=253705) [details] > changeset_r71535.diff > > Upstream patch that could solve this > This patch didn't make any difference for me, but appending -j1 to MAKEOPTS worked.
*** Bug 347035 has been marked as a duplicate of this bug. ***
Failed with MAKEOPTS="-j3" on x86_64-pc-linux-gnu-4.5.1. Could it be an issue with odd threads somehow? It seems that the people reporting success are using even thread numbers.
I am experiencing this problem also. Using MAKEOPTS="-j4", the build seems to succeed about 25% of the time.
I have exactly the same problem here, with gcc-4.5.1-r1 libtool-2.4-r1 MAKEOPTS="-j4" using MAKEOPTS="-j1" it does compile fine.
Here, the error message is always: libtool: link: cannot find the library `libwebkit-1.0.la' or unhandled argument `libwebkit-1.0.la' (Mentioning for documentary purposes, since I could not find a matching report before, apart from a dupe of this.)
same issue for me and MAKEOPTS="-j1" help
-j1 also worked for me. With -j2 it didn't build (tried several times).
My build is falling over at a different point to the others I think in that it's unable to find libwebkit-1.0.1a . My MAKEOPTS="-j2". will try again tommorrow as it's 01:00 and bed is a calling, with opts=-j1 to see if a difference is made. I will attach my build log environment and info as well tonight.
Created attachment 257030 [details] my emerge info + -pqv result appended
Created attachment 257033 [details] buildlog for webkit-gtk-1.2.5
Created attachment 257034 [details] environment file for x86
Ok -j1 worked for me also. edited /etc/portage/package.use to reflect the change in conditions,so that further builds of webkit-gtk get the same environment to work in.
(In reply to comment #23) > Ok -j1 worked for me also. edited /etc/portage/package.use to reflect the > change in conditions,so that further builds of webkit-gtk get the same > environment to work in. > package.use is not for modifying arbitrary portage variables, it's for modifying USE. /etc/portage/env/<category>/<package-name>[-package-version][.env] will allow you to modify other variables, such as MAKEOPTS and CFLAGS on a per-package basis. Note that, last I read, this was undocumented and not officially supported behavior.
The problem would most likely by solved by the simple fact of not building unittest programs at make time but at make check as it is supposed to happen.
+ 22 Dec 2010; Gilles Dartiguelongue <eva@gentoo.org> webkit-gtk-1.2.5.ebuild, + +files/webkit-gtk-1.2.5-tests-build.patch: + Make sure tests are built only when needed, bug #343249. Pin slotted + dependencies to needed slots. Replace addpredict by appropriate env variable + adjustement and do the same for tests. + Building tests while building the rest of the program being stupid for downstream and only justified by lazyness upstream, I've patched the ebuild to only build tests as required which fixes the problem altogether. If you still get parallel build failures but don't see the exact same error message, please open a new bug. Thanks all.
I can't reopen but I'm getting this again with version 1.2.7: CC WebKitTools/GtkLauncher/Programs_GtkLauncher-main.o CXX WebKitTools/DumpRenderTree/Programs_DumpRenderTree-AccessibilityController.o CXX WebKitTools/DumpRenderTree/Programs_DumpRenderTree-AccessibilityUIElement.o CXX WebKitTools/DumpRenderTree/Programs_DumpRenderTree-GCController.o CXX WebKitTools/DumpRenderTree/Programs_DumpRenderTree-LayoutTestController.o CXX WebKitTools/DumpRenderTree/Programs_DumpRenderTree-WorkQueue.o CXX WebKitTools/DumpRenderTree/gtk/Programs_DumpRenderTree-AccessibilityControllerGtk.o CXX WebKitTools/DumpRenderTree/gtk/Programs_DumpRenderTree-AccessibilityUIElementGtk.o CXX WebKitTools/DumpRenderTree/gtk/Programs_DumpRenderTree-DumpRenderTree.o CXX WebKitTools/DumpRenderTree/gtk/Programs_DumpRenderTree-EventSender.o CXX WebKitTools/DumpRenderTree/gtk/Programs_DumpRenderTree-GCControllerGtk.o CXX WebKitTools/DumpRenderTree/gtk/Programs_DumpRenderTree-LayoutTestControllerGtk.o CXX WebKitTools/DumpRenderTree/gtk/Programs_DumpRenderTree-WorkQueueItemGtk.o /bin/mkdir -p ./.deps/DerivedSources CXXLD libJavaScriptCore.la CXXLD TestNetscapePlugin/libtestnetscapeplugin.la CXXLD Programs/jsc CCLD Programs/minidom CCLD Programs/GtkLauncher libtool: link: cannot find the library `libwebkit-1.0.la' or unhandled argument `libwebkit-1.0.la' make[1]: *** [Programs/GtkLauncher] Error 1 make[1]: *** Waiting for unfinished jobs.... make[1]: Leaving directory `/var/tmp/portage/net-libs/webkit-gtk-1.2.7/work/webkit-1.2.7' make: *** [all] Error 2 emake failed * ERROR: net-libs/webkit-gtk-1.2.7 failed (compile phase): * Compile failed * * Call stack: * ebuild.sh, line 56: Called src_compile * environment, line 3078: Called die * The specific snippet of code: * emake XDG_DATA_HOME="${T}/.local" || die "Compile failed" * * If you need support, post the output of 'emerge --info =net-libs/webkit-gtk-1.2.7', * the complete build log and the output of 'emerge -pqv =net-libs/webkit-gtk-1.2.7'. * The complete build log is located at '/var/tmp/portage/net-libs/webkit-gtk-1.2.7/temp/build.log'. * The ebuild environment file is located at '/var/tmp/portage/net-libs/webkit-gtk-1.2.7/temp/environment'. * S: '/var/tmp/portage/net-libs/webkit-gtk-1.2.7/work/webkit-1.2.7'
(In reply to comment #27) > I can't reopen but I'm getting this again with version 1.2.7: Me too.
Looks like all reports are related with 1.2.x and I cannot reproduce neither with 1.4 => please retry with 1.4.2-r200
I had the same problem on x86_64-pc-linux-gnu-4.4.5 (with MAKEOPTS="-j3") I could solve it by downgrading make from 3.82 to 3.81-r2 leaving MAKEOPTS untouched.
*** Bug 378891 has been marked as a duplicate of this bug. ***
dirtyepic, can you please commit the "MAKEOPTS=-j1" hack for 1.2.x ebuilds yourself? Leave 1.4.x as-is because it looks to fix the parallel build problem (and upstream won't fix 1.2.x series anymore... we should stabilize webkit-1.4.x soon but until then and since this looks to hit a lot of people per comments and duplicates...) Thanks a lot
Um, I just reopened it because people kept complaining. I don't think I even have webkit installed anymore.
+ 31 Aug 2011; Pacho Ramos <pacho@gentoo.org> webkit-gtk-1.2.7.ebuild: + Parallel build fails for 1.2.x, seems fixed in 1.4.x (bug #343249). +
*** Bug 384527 has been marked as a duplicate of this bug. ***