Release notes for 1.7.22: http://www.winehq.org/announce/1.7.22 Release notes for 1.7.23: http://www.winehq.org/announce/1.7.23 Reproducible: Always
Release notes for 1.7.24: http://www.winehq.org/announce/1.7.24
Renaming latest wine ebuild from portage still works, provided you don't need pulseaudio patchset.
(In reply to Andreas Sturmlechner from comment #2) > Renaming latest wine ebuild from portage still works, provided you don't > need pulseaudio patchset. That breaks USE=pipelight. In addition, USE=pipelight is broken on the 9999 ebuild. Updated patches are available here: https://github.com/compholio/wine-compholio https://github.com/compholio/wine-compholio-daily/archive/v1.7.22.tar.gz https://github.com/compholio/wine-compholio-daily/archive/v1.7.23.tar.gz https://github.com/compholio/wine-compholio-daily/archive/v1.7.24.tar.gz They include the pulseaudio patches as a convenience. The pipelight developers would like to see us update our wine package. If the wine package maintainers do not respond to this bug within 2 weeks, I will commit updated ebuilds on their behalf for the sake of good relations with upstream.
Created attachment 383004 [details] Updated winepulse patchset for wine-1.7.24 I took a moment to rebase the winepulse patchset on wine-1.7.24. I attached it here, in case that helps. :) There were no merge conflicts or anything, and it seems to work for me.
*** Bug 520096 has been marked as a duplicate of this bug. ***
I wanted to mention a few additional things. * First of all to the wine-pulse patches - there is no need to rebase them manually, this could lead to additional errors which can easily be avoided. They are updated by the maintainer (Maarten Lankhorst) on each new release. We're also updating them in our branch (which will always apply on the current wine git tree) whenenver Maarten or wine upstream changes anything: https://github.com/compholio/wine-compholio/tree/master/patches/winepulse-PulseAudio_Support * If Gentoo decides to add release builds for the versions < 1.7.25, then the following additional workarounds should be used when the COMPHOLIO use-flag is enabled. The gitapply.sh script internally runs "awk", which will try to print the binary data as utf8 encoded string (in new versions)... this leads to patch apply failures. An easy workaround is to set some environment variables at appropriate places, and the error is gone. > export LANG=C > export LC_ALL=C This was already fixed in an updated compholio patchset, and is not necessary anymore for >= 1.7.25, see: https://github.com/compholio/wine-compholio/commit/4c1c303b623a3c6a3f82eaa52a16439c3cad4f29 * The method to apply the compholio patchset manually was deprecated some versions ago, because it was too difficult to maintain the patches this way. We're now using a Makefile, which is automatically generated based on dependencies and relations between individual patchsets. The ebuild should also use the new method. Please note that its still possible to exclude patches / apply a subset of the patches. You can find more information in developer guide: https://github.com/compholio/wine-compholio/blob/master/DEVELOPER.md Don't hesitate to ask in case of problems. * Last but not least, tomorrow is wine release day for 1.7.25, and gentoo will be 4 versions behind upstream... ;)
New release 1.7.25 is available. Release notes: https://www.winehq.org/announce/1.7.25
Wine ebuild didn't update already 2 months. If main Wine maintainer Alexandre Rostovtsev is retired and no longer wants to maintain this package, maybe Gentoo devs should consider to change Wine maintainer?
Release notes for 1.7.26: https://www.winehq.org/announce/1.7.26
Somewhere between 1.7.21 and 1.7.26, libpcap support was added to wine. It might be prudent to add '--without-pcap' to configure options for now, since Gentoo apparently doesn't provide a multilib libpcap yet, and introduce a use flag later.
i want to play diablo with newest wine! :) please make an ebuild already.
I made a wine 1.7.26 ebuild with pulseaudio, pipelight (compholio) and gstreamer support. You can find it in my wine overlay: http://www.linuxsystems.it/2014/09/new-linuxsystems-overlay-wine/
(In reply to darkbasic from comment #12) > I made a wine 1.7.26 ebuild with pulseaudio, pipelight (compholio) and > gstreamer support. You can find it in my wine overlay: > http://www.linuxsystems.it/2014/09/new-linuxsystems-overlay-wine/ Just installed it from your overlay. It works for me without a problem so far. Thanks. [ebuild R ~] app-emulation/wine-1.7.26::wine USE="X alsa cups fontconfig gecko gphoto2 jpeg lcms ldap mono mp3 ncurses nls opencl opengl perl png realtime run-exes samba scanner ssl threads truetype udisks v4l xcomposite xinerama xml -capi -custom-cflags -dos -gsm -gstreamer -netapi -odbc -openal -osmesa -oss -pipelight (-prelink) -pulseaudio (-selinux) {-test}" ABI_X86="32 64 (-x32)" LINGUAS="de en -ar -bg -ca -cs -da -el -en_US -eo -es -fa -fi -fr -he -hi -hr -hu -it -ja -ko -lt -ml -nb_NO -nl -or -pa -pl -pt_BR -pt_PT -rm -ro -ru -sk -sl -sr_RS@cyrillic -sr_RS@latin -sv -te -th -tr -uk -wa -zh_CN -zh_TW" 0 kB
(In reply to darkbasic from comment #12) > I made a wine 1.7.26 ebuild with pulseaudio, pipelight (compholio) and > gstreamer support. You can find it in my wine overlay: > http://www.linuxsystems.it/2014/09/new-linuxsystems-overlay-wine/ Thank you very much! Everything works for me very well!
This ebuild have multislot support http://data.gpo.zugaina.org/eroen/app-emulation/wine/
Release notes for 1.7.27: https://www.winehq.org/announce/1.7.27
New ebuilds wine-1.7.27 and wine-1.7.27 with gallium nine available: http://www.linuxsystems.it/overlay/
Created attachment 385336 [details, diff] wine-1.7.27.ebuild.diff My attempt: - let´s use pulseaudio patchset from compholio, but not depend on pipelight - unpack in src_unpack, apply in src_prepare - add --without-pcap - works as well for 1.7.26, 1.7.25 (configure-tested), built successfully with 1.7.27 Pulseaudio users please step up: I can´t build/run test it, but at least the patches applied fine.
Created attachment 385338 [details, diff] wine-1.7.24.ebuild.over1.7.27.diff small fix for 1.7.24 over my 1.7.27 ebuild to make sure there's no awk trouble
Created attachment 385340 [details, diff] wine-1.7.23.ebuild.over1.7.27.diff small fix for 1.7.23 on top of 1.7.27 (awk, no pcap support, manually remove pulseaudio patchset from Makefile)
Created attachment 385342 [details, diff] wine-1.7.22.ebuild.diff patch over 1.7.21 from portage; still uses manual patch method for compholio, and also takes the pulseaudio patchset from there. I've stuck with winepulse-patches-1.7.21 in all my ebuilds because that is already on the mirrors and most wine users' distfiles dir, but in fact only use 0038-TESTING-override-pthreads-to-fix-gstreamer-v2.patch
Created attachment 385348 [details] winepulse-patches-1.7.22.tar.bz2 winepulse-patches minus those from compholio, after minor editing
Created attachment 385350 [details] winepulse-patches-1.7.24.tar.bz2 winepulse-patches minus those from compholio, after minor editing
Created attachment 385352 [details, diff] wine-1.7.22.ebuild.diff update winepulse-patches
Created attachment 385354 [details, diff] wine-1.7.23.ebuild.over1.7.27.diff update winepulse-patches
Created attachment 385356 [details, diff] wine-1.7.27.ebuild.diff update winepulse-patches
some official statement from some $dev would look professional. is vapier still active or is anybody else in the wine@gentoo.org alias? soon this bug is 2 months old and already 6 wine releases were skipped. imho this doesn't have to happen with such an important/popular package. thanks to everybody contributing updates in here and to ryao, who is the only dev giving a comment so far.
I guess the obvious answer is that more users should do quizzes and become devs. *hides*
(In reply to Marcel Unbehaun from comment #27) > some official statement from some $dev would look professional. > is vapier still active or is anybody else in the wine@gentoo.org alias? > soon this bug is 2 months old and already 6 wine releases were skipped. > imho this doesn't have to happen with such an important/popular package. > thanks to everybody contributing updates in here and to ryao, who is the > only dev giving a comment so far. I have a long list of things to do and doing proper review takes time. With my current schedule, the earliest at which I will be able to devote an evening to doing review is Tuesday. The work by Andreas will likely reduce the amount of time that I need to spend and it is much appreciated.
(In reply to Richard Yao from comment #29) > I have a long list of things to do and doing proper review takes time. With > my current schedule, the earliest at which I will be able to devote an > evening to doing review is Tuesday. The work by Andreas will likely reduce > the amount of time that I need to spend and it is much appreciated. I know and understand that reviewing/creating ebuilds takes its time and that this time is free time. I was one of the main contributors of the gamerlay overlay until it was almost killed by some gentoo $dev instead of helped. so $dev is not dev necessarily. I don't use wine very often anymore, but was a bit anxious about the missing updates and communication. So thank you again Richard for being present.
Hies! I guess it might be worth noting here as well (already did at bug 516006), for those who are making them new ebuilds: I got some new versions of the patch again via e-mail while we were looking into a memory leak I discovered this patch introducing. It looks like the latest one is also available here now: http://repo.or.cz/w/wine/multimedia.git/commit/9e081cd4a04e3326d4927aa082695f15432590e2 I've been using it for some time without new issues creeping up (without the PulseAudio patches), although the gstreamer-using application (Echoes) I have been testing it with now fails with my git-builds (but not with the Portage build, even if building the same version). Probably some local issue I'm not thinking of...
Release notes for 1.7.28: https://www.winehq.org/announce/1.7.28
Built wine-1.7.28 right now, works using my 1.7.27 ebuild. There's trouble though, which I haven't thought of before - my ebuilds only work for pulseaudio if pipelight is set at the same time, because compholio carries a dsound patch outside of its pulseaudio patches dir. I'm also wondering about gstreamer - should it really be inside winepulse-patches (even if it depends on them) when we need to touch all those winepulse-tarballs in case of a gstreamer-patch update, as it seemed to have happened now?
Created attachment 386042 [details] winepulse-patches-1.7.24.tar.bz2 updated winepulse-patches-1.7.24 tarball to include gstreamer patch 'v4' - again, at least it configures fine, gstreamer users will find out more. Chiitoo, how far back in wine history should it apply instead of v2?
Created attachment 386044 [details, diff] wine-1.7.28.ebuild.diff 1.7.28 version bump + fix for USE="pulseaudio -pipelight" build same file also works for wine-1.7.2{5,6,7}.ebuild
Created attachment 386046 [details, diff] wine-1.7.23.ebuild.over1.7.28.diff attaching remixed patch for wine-1.7.23.ebuild based on wine-1.7.28.ebuild - last version of wine-compholio-daily that didn't contain the offending dsound patch, so doesn't need the fix. To get wine-1.7.24.ebuild, the following wee patch is enough: --- wine-1.7.28.ebuild 2014-10-04 16:45:56.130052648 +0200 +++ wine-1.7.24.ebuild 2014-10-04 16:45:33.479335822 +0200 @@ -329,6 +329,9 @@ # First of all, don't run autoreconf and tools/make_requests twice sed -i 's/.*cat.*sort.*patchlist.*APPLY.*/&\n\n.PHONY: postinstall\npostinstall:/' \ "../wine-compholio-${COMPHOLIOV}"/patches/Makefile || die + # See bug #518792: fix possible awk trouble + sed -i 's/# Decode base85 git data.*/export LANG=C\nexport LC_ALL=C\n\n&/' \ + "../wine-compholio-${COMPHOLIOV}"/debian/tools/gitapply.sh || die # Use Makefile instead of manually applying patches # ...exclude pulseaudio patchset, we apply it conditionally # ...also exclude dsound-Fast_Mixer (conflicts with PULSE_PATCHES)
(In reply to Andreas Sturmlechner from comment #33) > I'm also wondering about gstreamer - should it really be inside > winepulse-patches (even if it depends on them) when we need to touch all > those winepulse-tarballs in case of a gstreamer-patch update, as it seemed > to have happened now? I feel it should be separate, which is more or less the reason for the bug 516006. When I first asked Lankhorst if it depends on the PulseAudio patches, the reply was: “As far as I know, this specific patch does not, but most of the other patched pulled in probably do.” (In reply to Andreas Sturmlechner from comment #34) > Chiitoo, how far back in wine history should it apply instead of v2? Hm, I'm not entirely sure. I haven't thought of asking about it from Lankhorst, but for what it's worth, I just tried myself: without any modifications, wine-1.7.12 is where the patch still applies fine, but failures start at 1.7.11. Building both x32 and x32_64 works too. During a short game-play test I didn't notice anything obvious going astray, although the test-app fails on a 64-bit Wine prefix with 1.7.12 build (possibly not related to the patch at all). I might also be messing something obvious up again though (not ccache this time!). Anyblue, it seems to work from what I can tell!
The patches included in our compholio repository ( ./patches/winepulse-PulseAudio_Support ) are fully sufficient for pulseaudio support. The name for the use flag "pulseaudio" might be misleading, as it also includes all other patches from wine-multimedia, including various hacks. I know, there are also some more distributions that include them all, but as we're not always sure what they are supposed to fix, we decided not to include them in the compholio branch (yet). The gstreamer fix is definitely separate and unrelated, so I would also suggest to create a separate use flag for that. Since some people reported regressions when using it, I'm not sure if its a good decision when its always applied together with the pulseaudio patchset. @Andreas: Thanks for all your effort. One suggestion to Andreas, which could probably simplify maintaining all the patchsets, and to avoid apply failures for situations which were not tested yet: For compholio we use a special patch system, which is also included in our repository. This patch system needs a path to the wine git repo, and will (when a git precommit hook is installed) automatically prevent committing stuff, if a subset of patches don't apply properly. Lets assume you add patch X, and it conflicts with patch Y, then you get a chance to fix it, before someone accidentially runs into this issue. When you hit such a problem, one possible solution is to reduce the context size (recreate patch with "git diff -U1" or -U2, or manually replace some hunks with a smaller context size), to add dependencies between patchsets (if patch Y always needs patch X applied before, the definition file can be used to specify that), and so on ... If you want to use that, you could just fork compholio, add your additional patchsets into ./patches/ (don't forget to create definition files, this is the information used for wine --patches), then setup the patch system: ln -s ~/path/to/wine ./debian/tools/wine ln -s ../../precommit-hook.sh .git/hooks/pre-commit And commit it. If it fails, you know that the patches do not apply properly. All patches in our repository are enforced to apply strict (--fuzz=0), but feel free to change that, if you don't want to have the additional effort to fix those issues, too ( patch to change it -> http://ix.io/eCA ). Once you have all the patches committed, you can simplify the EBUILD file a lot: Just translate use-flags into corresponding arguments to the Makefile, to leave our / apply specific patchsets. You can set the PATCHLIST variable to do that, for example: # Generate PATCHLIST based on use flags ... make -C ... DESTDIR="..." PATCHLIST="${PATCHLIST}" install If you like this idea, and need more information about the patch system, feel free to talk to us in #wine-compholio on Freenode. Regards, Sebastian
Created attachment 386060 [details, diff] wine-1.7.28.ebuild.diff (In reply to Sebastian Lackner from comment #38) > The patches included in our compholio repository ( > ./patches/winepulse-PulseAudio_Support ) are fully sufficient for pulseaudio > support. The name for the use flag "pulseaudio" might be misleading, as it > also includes all other patches from wine-multimedia, including various > hacks. I like the idea of not having to reinvent the wheel when compholio has already done the work; I just left the rest of winepulse-patches in place to retain whatever users were expecting from it before. Gentoo ebuilds usually aren't known for lots of conditional patching so I suspect before there is going to be another USE flag those patches would be dropped. > I know, there are also some more distributions that include them all, but as > we're not always sure what they are supposed to fix, we decided not to > include them in the compholio branch (yet). The gstreamer fix is definitely > separate and unrelated, so I would also suggest to create a separate use > flag for that. Since some people reported regressions when using it, I'm not > sure if its a good decision when its always applied together with the > pulseaudio patchset. Thanks for reaffirming that the gstreamer fix could be separated; that's what I did in a new version of wine-1.7.28.ebuild. It was already applied separately since 1.7.21, it only wasn't possible to enable gstreamer without pulseaudio so far. And thanks again for answering at length to the bug; I was merely trying to ease the work to get those versions into the tree, while not actually using gstreamer/pipelight/pulseaudio myself; so small diffs win.
Created attachment 386062 [details, diff] wine-1.7.23.ebuild.over1.7.28.ebuild.diff same changes for 1.7.23, while 1.7.24 patch stays the same as in comment #36
Created attachment 386064 [details, diff] wine-1.7.22.ebuild.diff diffs for 1.7.20 and .21 can be found in bug 516006, or the whole bunch in https://github.com/a17r/a17rgentoo/tree/master/app-emulation/wine
Example how using our patch system could look like (applies on compholio 1.7.28): http://ix.io/eCL This patch adds both Multimedia and the gstreamer hack into ./patches/ and makes it available as Makefile patch targets. It is guaranteed to apply always (ensured by the patch system algorithms), no matter which combination of patches (gstreamer, Multimedia, pipelight, pulseaudio) is selected. Nevertheless I wasn't really able to test them all, so it doesn't necessarily mean that it will also compile fine. To get rid of the apply conflicts between Multimedia <-> dsound-Fast-Mixer, I removed some changes from the Multimedia patch. An unnecessary variable will still be present, but shouldn't do any harm. Please note that ryao is atm also working again on updating the ebuild, so I'm not sure if you should really spend the time to clean all that up, if it will probably be unnecessary soon. But thats your decision of course ;) Regards, Sebastian
app-emulation/wine-1.7.28 is in the tree. I initially based it on preliminary work from 2014-08-09, but later assimilated Andreas Sturmlechner's work into it. I consider this to have created an ebuild better than what either of us had done on our own. Thanks Andreas. :)
/me diffs and learns - agree, much better than what I was capable of. I could only spot two leftover comments wrt winepulse that can be deleted with next revision. ;)
