wine-1.1.4 fails to emerge for me. Looks like a parallel build issue to me, but giving it a try with MAKEOPTS=-j1 will have to wait till tomorrow. make[2]: Entering directory `/var/tmp/portage/app-emulation/wine-1.1.4/work/wine-1.1.4/dlls/winscard' /usr/bin/install -c winscard.dll.so /var/tmp/portage/app-emulation/wine-1.1.4/image//usr/lib/wine/winscard.dll.so make[2]: Entering directory `/var/tmp/portage/app-emulation/wine-1.1.4/work/wine-1.1.4/dlls/wintab32' /usr/bin/install -c wintab32.dll.so /var/tmp/portage/app-emulation/wine-1.1.4/image//usr/lib/wine/wintab32.dll.so make[2]: Entering directory `/var/tmp/portage/app-emulation/wine-1.1.4/work/wine-1.1.4/dlls/wintab32' /usr/bin/install -c wintab32.dll.so /var/tmp/portage/app-emulation/wine-1.1.4/image//usr/lib/wine/wintab32.dll.so /usr/bin/install: cannot create regular file `/var/tmp/portage/app-emulation/wine-1.1.4/image//usr/lib/wine/wintab32.dll.so': File exists make[2]: *** [install] Error 1 make[2]: Leaving directory `/var/tmp/portage/app-emulation/wine-1.1.4/work/wine-1.1.4/dlls/wintab32' make[1]: *** [wintab32/__install__] Error 2 make[1]: *** Waiting for unfinished jobs....
I had the same issue in an amd64 system. With the "-j1" it works.
same issue in ~x86, with MAKEOPTS=-j1 worked.
Please attach a full build log. From the excerpt, it looks like there is a problem with install stage, which is not what I would expect from a parallel build issue (common problem is compile failures when make targets are missing dependencies). A successful build log with '-j1' could be useful too, for the sake of comparison
Created attachment 164876 [details] Full Build log
Created attachment 164878 [details] The environment configuration file
Hi all, Attached are the requested file. Best regards, Michel
One intermediate cause of this issue seems to be the fact that make enters the dlls directory twice, once for target "install" and once for target "install-lib". This leads to the install commands being invoked multiple times. Useful debugging command: "make --dry-run --debug=j install". I suspect this dependency in the main Makefile to be responsible for the multiple runs: programs/__install__ programs/__install-lib__: \ libs tools include dlls/__install-lib__ While it may well be true that the installations of programs only requires the libs to be installed, I think that make has no way of knowing that install implies install-lib in a subdirectory, and therefore schedules both install and install-lib for the dlls subdir. One workaround might be to habe programs/__install__ depend on dlls/__install__, which won't hurt us as we want to install everything recursively in any case. Another workaround might be to drop that line altogether, hoping that the dependency is only a runtime dependency, and not really needed at instalation time. I'll continue to investigate this, giving preference to the first workaround. As parallel build issues are sometimes difficult to reproduce, I'll have it build quite a few times before I'll be satisified, so this might take a while yet.
This here seems to be the upstream commit which introduced the issue: http://source.winehq.org/git/wine.git/?a=commit;h=ada5e6521bfb0ca5 I'm a bit surprised that this commit dates from 2002 and the relevant dependency was already included in previous wine releases but the issue just surfaced now. Once things are resolved in Gentoo, I guess we should file a bug report upstream.
(In reply to comment #8) > This here seems to be the upstream commit which introduced the issue: > http://source.winehq.org/git/wine.git/?a=commit;h=ada5e6521bfb0ca5 I was wrong: back then, install-libs was the /only/ dependency, and the install target was not made in the dlls subdir. This was still the case in 1.1.3. Only recently, install got run in the dlls dir as well, along with the ancient install-libs from 2002, thus resulting in the issue at hand. Introduced in http://source.winehq.org/git/wine.git/?a=commit;h=b3ec2e5f95e36aa9
Created attachment 164886 [details, diff] depend on install not install-libs for install target This patch seems to work, as it survived 10 install phases in a row, ranging from -j2 to -j128. Please include in the portage tree. No revbump needed.
(In reply to comment #8) > I guess we should file a bug report upstream. Reported as http://bugs.winehq.org/show_bug.cgi?id=15179
Martin, thanks for all your work tracking down the problem and providing a patch. Assigning to maintainers.
Upstream already applied my patch, they were really fast: http://source.winehq.org/git/wine.git/?a=commitdiff;h=78c79ba349deea39 So I'd say that patch the way for Gentoo as well, until the next release comes along or someone reports any issues with it.
can you apply to portage?
Thank you for report and digging the patch. Patch was added to CVS. FIXED.
(In reply to comment #15) > Thank you for report and digging the patch. Patch was added to CVS. FIXED. > I know this bug is resolved fixed, but 1.1.5 was released today. Please remember to remove this patch when you bump the ebuild =) Thanks for everything.
