15:45:45 <itchka> #startmeeting
15:45:45 <chwido> Woof! Let's start the meeting. It's Wed Oct 11 15:45:45 2017 UTC. The chair is itchka. Information about me at https://wiki.openmandriva.org/en/Chwido.
15:45:45 <chwido> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:45:45 <chwido> You should add extra chair(s) just in case, with #chair nick.
15:45:45 <chwido> Have a good meeting, don't bark too much!
15:47:08 <itchka> Who's here _TPG: ben79: christann: cc^mate fedya: HisShadow jbj:  mandian: Pharaoh_Atem
15:47:29 <itchka> #item 1
15:47:36 <itchka> # Release briefing
15:47:56 <itchka> ~topic Release Briefing
15:48:13 <itchka> oh fsck I'm having a bad day
15:48:28 <itchka> #topic Release Briefing
15:50:26 <ben79> ¬°Ay, caramba!
15:51:13 <itchka> Ok Everything is pretty much ready with the exception of a rendering problem in a number of applications. The affected apps are as far as we know firefox, chromium-browser-stable, om-welcome (qtwebengine) and wine. There may be others
15:52:44 <ben79> Is there a ready fix or patch?
15:53:00 <itchka> We have two choices we can release with crappy rendering and because it occurs in our start-up application it will be there for all to see. It is very evident in chromiium-browser-stable.
15:53:51 <itchka> There are patches for all of these applications.
15:53:57 <christann> thunderbird is also affected, but probably for the same reason as firefox.
15:54:23 <itchka> christann: Almost ceratinly.
15:55:10 <ben79> I'm thinking if we could just fix qtwebengine then release
15:55:28 <mandian> why this happens?
15:55:44 <ben79> that way OM-Welcome will be fixed
15:56:27 <itchka> These applications acrry their own versions of skia and they don't respect the freetype api.
15:56:43 <_TPG> i'm trying to update rust and cargo, and these are needed for firefox 56.x
15:57:04 <_TPG> unfortunately rust somehow wants to compile with clang rather with gcc
15:57:11 <_TPG> and it simply fails
15:58:05 * ben79 mandian: http://lists.gnu.org/archive/html/freetype-devel/2017-09/msg00081.html
15:58:45 * ben79 mandian: problem didn't hit us till Oct. 5 as far as I know
15:59:11 <itchka> _TPG: It's very strange. I can compile the beta branch of rust outside of rpmbuild and it builds fine using the same flags to configure as there are in the spec file. I can't work out why this should be.
15:59:14 <_TPG> fast stolution is to disable SUBPIXEL_RENDERING
15:59:37 <_TPG> in freetype
16:01:28 <itchka> A few bug reporters on other platforms have tried this and report that  fonts look dreadful so I suspect they will for us too.
16:01:50 * ben79 mandian: see tab for more
16:02:04 <cc^mate> !
16:03:11 <itchka> cc^mate: hi
16:04:04 <itchka> _TPG: Is rebuilding qtwebengine a big problem?
16:05:16 <_TPG> itchka: it needs a patch
16:05:31 <itchka> I have the patch
16:05:55 <itchka> I'll give it a try
16:06:21 <_TPG> itchka: in 3.0 there is a qtwebengine 5.8.0
16:06:32 <_TPG> do you have a patch for 5.8.0 ?
16:06:43 <itchka> Will check
16:09:25 <ben79> Is disabling anti-aliasing in systemSettings the same thing as disable SUBPIXEL_RENDERING?
16:11:52 <itchka> _TPG: No patch is for 5.9.1 (qtwebengine)
16:12:39 <itchka> _TPG: How different is it likely to be?
16:17:03 <_TPG> itchka: try to apply it on 5.8.0 and see if it does fit
16:17:25 <itchka> Just downloading the source now
16:17:51 <itchka> So I'll try it.
16:20:43 <itchka> _TPG: rust 1.2.0 will build with clang outside of rpmbuild so I'm trying the CFLAGS from that build with an rpmbuild of 1.20.0. rust 1.21.0 should be released from beta tomorrow.
16:24:34 <_TPG> itchka: ok, i get a little bit further, it now fails on binutils module
16:24:46 <itchka> qtwebengine sources are a gigabyte
16:24:53 <_TPG> :)
16:25:04 <cc^mate> .
16:25:23 <cc^mate> Computers cost but don't save time and money!
16:26:05 <itchka> Mind you building rust locally needs 38Gig!!!!!
16:29:19 * ben79 If I go to: SystemSettings>Appearance>Fonts and set anti-aliasing to Enabled then set Configure>Sub-Pixel rendering type to None and Hinting style to System default then fonts look OK including in OM-Welcome
16:29:40 * ben79 possible work around
16:32:21 <_TPG> ben79: well seems like you found a workaround
16:33:00 <ben79> _TPG: Um, maybe
16:33:50 <itchka> _TPG: The patch applies to qtwebengine 5.8 without a proble.
16:33:57 <itchka> problem
16:34:15 <ben79> I guess it may be an adequate work around
16:34:59 <itchka> _TPG: Shall I rediff it for 5.8
16:35:32 <itchka> or jsy hack the patch
16:35:37 <itchka> just
16:39:16 <_TPG> itchka: well that's a good  news
16:39:35 <_TPG> itchka: apply that patch to qt5-qtwebeengine a 3.0 branch
16:39:53 <itchka> _TPG: Will do
16:40:17 <_TPG> itchka: anyways this should not hold us with a release
16:40:34 <_TPG> itchka: as ben79 has a workaround
16:43:24 <itchka> _TPG: Are you suggesting we put something in the release notes  or make a change to the config before shipping?
16:45:29 <itchka> We are in a position to fix chromium and qtwebengine
16:48:45 <ben79> I *think* I'd feel better about work around if it were tested on some other hardware, here it seems to work OK so far on my  desktop and notebook
16:49:43 <ben79> with Intel graphics
16:50:39 <itchka> ben79: I'll try it here once I haveqtwebenging building. I have nVidia
16:51:07 <ben79> OK,
16:52:57 <_TPG> itchkta: commit that patch and let's see if qt5-qtwebengine build on 3.0
16:53:41 <itchka> _TPG: In progress
16:54:41 <_TPG> itchka: without rust we can not fix firefox
16:56:37 <itchka> I got as far as a link error using different cflags
17:05:50 <itchka> _TPG: qt5-webkit for 3.0 started on ABF
17:06:09 <_TPG> good
17:18:15 <itchka> Ok tring ben79 workaround
17:22:25 <itchka> _TPG: qt5-webengine build failed bu I can't see why...
17:23:07 <itchka> I did abb buildp here to make sure that the pacth applied cleanly.
17:25:32 <itchka> I've restarted it.
17:27:36 <_TPG> ok
17:35:16 <itchka> It's on 10 cores so it should get done fairly quickly:)
17:37:54 <itchka> _TPG: Regards rust I hoed that there is an --enable-clang in the the configure so I'm try a local build with clang rather than gcc and with the --enable-clang flag passed to configure.
17:38:23 <mandian> at ABF ji.openmandriva.org and tanmad.openmandriva.org are out of order
17:39:55 <itchka> I think they are probably fedyas boxes he has no connection through his ISP to Europe apparently.
17:40:01 <mandian> I mean all build on these hosts systematically fail
17:40:51 <itchka> Right thanks for the heads up. They accept the builds but just don't work.
17:42:52 <christann> docker appears to be running properly here - tanmad. Should I restart it?
17:43:05 <_TPG> mandian: ji is jbj box
17:43:11 <_TPG> christann: please restart it
17:43:29 <_TPG> anyways i've asked fedya to take care of this issue
17:43:44 <_TPG> he said these two needs our docker update
17:44:50 <christann> docker restarted.
17:44:58 <christann> Where do I get the docker update?
17:45:55 <mandian> _TPG: thank you
17:48:52 <_TPG> christann: you need to update this https://hub.docker.com/r/openmandriva/builder/
17:50:09 <christann> Do I have to remove the current builders first?
17:52:19 <christann> I think that I have updated the builders here.
17:55:35 <_TPG> good
17:55:45 <_TPG> just run docker pull
17:55:47 <itchka> ben79: Your workaround looks ok here on om-welcome. I can't test firefox at the moment all my processors are flat out building stuff and it won't start.
17:56:41 <christann> Done.
17:59:49 <ben79> OK
18:07:21 <_TPG> itchka: you should build qt5-webengine instead of qt5-webkit
18:08:30 <_TPG> fixing
18:08:59 <itchka> Thought I was... sorry
18:09:21 <_TPG> no release bump
18:10:08 <itchka> Blast I always forget to do that!
18:10:16 <_TPG> jbj: hi, can you restart ji ?
18:12:45 <itchka> Alas --enable-clang in rust didn't fix things.
18:14:58 <_TPG> :<
18:16:54 <_TPG> fedya: jbj can you guys fix ji ?
18:26:01 <itchka> Trying another plan
18:44:36 <_TPG> itchka: qt5-qtwebengine failed due to v8
18:46:52 <itchka> Starnge patch applied locally my install defaults to 3.0 so I should have had the correct source.
18:50:33 <_TPG> itchka: i think patch have nothing to do with that build fail
18:51:22 <itchka> _TYPG: Ah I see now that I've looked at the log.
18:52:13 <itchka> _TPG: ^^
18:56:43 <_TPG> ok
18:58:09 <itchka> persumably it built before?
19:04:17 <itchka> _TPG: Perhaps this? https://patchwork.openembedded.org/patch/141727/
19:10:38 <itchka> _TPG: or perhaps this may be the one https://github.com/OSSystems/meta-browser/blob/master/recipes-browser/chromium/files/0001-v8-fix-build-with-gcc7.patch
19:12:28 <itchka> _TPG: I think thats the one. Can you see it or shal I post it in a tab.
19:18:54 <itchka> ben79: Your workaround looks ok here on om-welcome, and chromium. I do't have Thunderbird installed but I guess if firefox is ok then so will thinderbird.
19:29:50 <itchka> Looks like rust-1.21.0 has been released.
19:30:20 <itchka> https://github.com/rust-lang/rust/blob/master/RELEASES.md
19:31:45 <_TPG> let's try it
19:33:45 <_TPG> nope still not released
19:35:08 <itchka> They must just be getting ready. It does say 12-10-2017 in the README
19:40:44 <_TPG> what about 3.03 release ?
19:40:59 <_TPG> again deylayed or we use workaround ?
19:41:45 <_TPG> btw, our mesa support S3TC :)
19:42:11 <itchka> _TPG: How do we apply workaround? Throgh release notes or with a fixup package?
19:42:30 <itchka> Through
19:43:10 <_TPG> release notes ?
19:43:38 <itchka> Tell people to change the settings if they don't like the fonts.
19:46:10 <_TPG> itchka: sounds easier than fixing tons of package
19:46:12 <_TPG> s
19:48:40 <itchka> _TPG: So do we have a config package that we use to configure these setting? If so can we rebuild that with the needed config change and then ship it. Then build some new isos and release?
19:52:05 <itchka> _TPG: Whats the command to switch to the bfd linker?
19:53:50 <_TPG> itchka: %global optflags %{optflags} -fuse-ld=bfd
19:54:13 <itchka> Thanks :)
20:10:14 <itchka> So _TPG are you saying you just want to put something in the release notes?
20:10:20 <_TPG> yes
20:11:41 <itchka> What do you think ben79?
20:17:14 <_TPG> time for a sleep
20:17:15 <_TPG> bbl
20:17:22 <itchka> Looks like ben79 is having his tea. I feel it's not the best solution as it will give a bad impression. I'd be happier if we changed the defaults and shipped new isos but if you think that means rebuilding loads of packages then reluctantly I'll go with the release notes solution.
20:17:36 <itchka> Ok
20:17:55 <itchka> #endmeeting