**** BEGIN LOGGING AT Tue Oct 06 02:59:58 2015 Oct 06 05:17:38 morning Oct 06 05:30:12 iscgar|Pre: morning Oct 06 05:34:05 Herrie: I think I'm ready to submit the rewrite I did to pmcertificatemgr. the only problem is that for some reason it refuses to compile, complaining about 'open' and the O_* constants being undeclared even though I included fcntl.h Oct 06 05:35:16 I couldn't find a solution online. Have you encountered something like that? Oct 06 05:44:50 iscgar|Pre: I don't touch C++ stuff ;) But Tofe, morphis, JaMa, ka6sox and nizovn do for sure ;) Oct 06 05:45:11 iscgar|Pre: pong Oct 06 05:45:30 iscgar|Pre: on which system are you building it? Oct 06 05:46:14 morphis: Ubuntu 15.04 Oct 06 05:46:24 iscgar|Pre: can you show me the code? Oct 06 05:46:32 but the original version builds fine. Oct 06 05:47:19 morphis: Unfortunately no. haven't pushed to github and I'm not at home right now. Oct 06 05:47:40 ok Oct 06 06:52:49 iscgar: when you'll have pushed it to github, I can help too Oct 06 11:16:45 Anyone here interested in helping me analyzing some pretty bizarre C++ crash? I'm a bit out of ideas right now Oct 06 11:18:10 The crash occurs here: https://github.com/qtproject/qtdeclarative/blob/dev/src/quick/scenegraph/util/qsgtexturematerial.cpp#L99 (the calls to bind() leads to a 0x000 address), and the texture object is created here: https://github.com/Tofee/qtwebengine/blob/tofe/work/src/core/delegated_frame_node.cpp#L348 Oct 06 11:18:38 Hint: we never go into ~QGSTexture destructor. Oct 06 11:19:46 I'm trying to figure out some scenario where the vtable of the object is incorrect, but I can't find any. Oct 06 12:57:09 JaMa: morning Oct 06 12:58:35 morning Oct 06 12:59:44 Was just preparing release and was wondering why we're not using fido,HEAD for openembedded-core and meta-openembedded? Oct 06 13:30:37 we don't want oe-core and meta-oe to be automatically upgraded when there are some new commits in them Oct 06 13:30:49 we (I) always need to test and verify them first Oct 06 13:31:40 JaMa: Ah OK, because we were either equal to head or very close (i.e. 2-3 commits behind, hence I was wondering) Oct 06 13:51:56 Herrie|Veer: we're always relatively close to HEAD Oct 06 13:52:10 it only depends on how often I find time to update it Oct 06 13:52:20 and Fido is currently updated more offten than old Dizzy Oct 06 13:52:31 Jethro will be release soon :) Oct 06 13:59:55 That has Qt 5.5 built in already? Or Qt is not part of Yocto? Oct 06 14:05:21 JaMa: do you know if we are compiling Qt with -fno-rtti ? Oct 06 14:05:46 I couldn't find that information, don't know exactly where it is defined Oct 06 14:09:27 Herrie|Veer: it's part of Yocto, but we can use different branches as we do now (Fido for some layers, master for meta-qt5) Oct 06 14:09:38 as long as we're able to resolve possible incompatibilities Oct 06 14:09:51 Tofe: it rings the bell, but I don't remember the details Oct 06 14:10:00 Tofe: isn't it added by default by Qt mkspecs? Oct 06 14:10:51 JaMa: I don't know, I'll have to check Oct 06 14:11:31 I did very quick check and don't see it there Oct 06 14:11:34 except mingw Oct 06 14:12:55 ok, so maybe it is built with RTTI; I would be surprised though. Oct 06 14:42:56 Hi Oct 06 14:50:26 I'm finally back home and I pushed one huge commit to github (there was no point to commit it in small chunks because the changes are all over and need to be reviewed at once). Would anyone care to check why cert_db.c fails to compile because of undeclared identifiers which should be pulled from fcntl.h? Oct 06 14:52:18 it's over at https://github.com/iscgar/pmcertificatemgr Oct 06 14:52:38 the branch is "massive_code_cleanup" Oct 06 15:48:08 scratch that. Apparently long coding sessions can make you see things that aren't there. cert_db.c compiles fine but there are a bunch of other errors I'm currently fixing. Oct 06 15:52:31 iscgar: Can happen. Seems you indeed did quite some thorough work :P Once it compiles you can test if all still works :) Oct 06 15:54:37 Herrie|Veer: I can tell you right now that it won't work because I removed functions that were deprecated for a long time (like getCertInfo and CertReadX509TimeProperty). Once certmgrd is adjusted it should work though. Oct 06 16:27:19 Herrie|Veer: It's building now. Just need to implement one more function before I submit it. Oct 06 16:27:51 but I'd love reviews of what's already there. morphis? Tofe? Oct 06 16:41:01 iscgar: nizovn knows quite some c++ too ;) Oct 06 16:41:47 So nizovn, when you've got some time feel free to have a look at https://github.com/iscgar/pmcertificatemgr/tree/massive_code_cleanup Oct 06 16:42:04 Herrie: thanks Oct 06 16:42:15 If someone feels like testing some upcoming release images: http://build.webos-ports.org/luneos-stable-staging/wip/images/ Oct 06 16:42:31 I'll be quickly testing N4, TP and emu, but others are welcome to test too Oct 06 16:43:08 Herrie: I can test on N7 I have lying around and on TP, if that helps. Oct 06 16:44:19 That Oct 06 16:44:26 would be good, I don't have a N7 Oct 06 16:52:31 iscgar: one moment, I need to finish something before looking at your commit :) Oct 06 17:01:05 iscgar: i will have a look too :) Oct 06 17:04:27 Thanks Tofe and nizovn :) Oct 06 17:05:18 Herrie: then I'll try to get to testing it later tonight. Oct 06 17:43:52 iscgar: apart from some minor remarks that I've put on GH, it looks good to me Oct 06 17:44:03 Quite some huge work here ! Oct 06 17:58:35 iscgar: thnx. I'm testing N4 now, looks good Oct 06 17:58:39 Will be flashing TP shortly Oct 06 17:59:02 Running various browser benchmarks to see how it stacks up against qtwebkit Oct 06 17:59:05 Tofe: thanks. I appreciate it. Oct 06 18:01:44 Herrie: I've solved my puzzling issue, the issue was that the problem was not here :) Oct 06 18:14:56 Tofe: Ah OK Oct 06 18:15:05 So any progress :P ? Oct 06 18:15:13 OR still stuck at non working stuff? Oct 06 18:16:52 Testing various images here now for release ;) Oct 06 18:32:08 VirtualBox & N4 seem stable and OK. Now testing TP, but so far so good Oct 06 18:34:21 Herrie: still stuck, seems it's a problem with the Qt OpenGL wrapper functions somehow Oct 06 18:36:06 Hmmz Oct 06 18:36:16 Let me know when you want some distraction :P Oct 06 18:36:27 Did you try upgrading QT to 5.5.1 or 5.6? Oct 06 18:36:46 I might have some minor QML stuff for you to look at when you want a change from C++ Oct 06 18:41:13 TP seems OK too Oct 06 18:41:27 iscgar: Let me know once you tested N7 and I can push release out Oct 06 18:41:53 Actually I can do anyway, it might be broken beyond repair after we updated from dizzy to fido ;) Oct 06 18:42:34 But would be good to know if it works, or we need to look into fixing it again going fwd Oct 06 18:52:10 Herrie: still on 5.5.1 :) Oct 06 18:52:32 Tofe: Ah OK Oct 06 18:52:36 Maybe that's the issue? Oct 06 18:52:52 I know 5.6 upgrade will give other headaches that are more major probably Oct 06 18:53:40 no, for the current crash I think we'll have the same issue with 5.6; it looks like the GL context isn't correct initialized or something like that Oct 06 18:53:53 so platform specific, unlikely to be changed in 5.6 Oct 06 18:54:01 5.6 drops WebKit, doesn't it? Oct 06 18:56:09 Tofe: I think I read it will still be included in 5.6 but dropped afterwards? Oct 06 18:56:48 ah ok, well, good for me Oct 06 18:58:12 Let me double check just to be sure Oct 06 19:00:30 Tofe seems you're partly right: With Qt 5.6 the following modules are no longer part of the release packages, but users can still build them from source: Qt WebKit, Qt Declarative (Qt Quick 1) Oct 06 19:00:46 So we'd need some extra recipe to have it build I guess in case we stillneed it? Oct 06 19:00:53 As per https://wiki.qt.io/New_Features_in_Qt_5.6 Oct 06 19:02:46 ok, and that's a new reason to tix the spurious dependency on qtquick1 :) Oct 06 20:00:01 Herrie: you did try the WebEngine browser on device, right? Oct 06 20:00:36 I'm getting some weird output now, when I launch it, saying that wayland-egl isn't yet supported Oct 06 20:17:33 Herrie: ah no, my fault, missing env variable Oct 06 20:21:04 Herrie: sorry for the delay. My N7 doesn't want to power up currently. I'll update when it decides to boot up so I can install the test image. Oct 06 20:23:20 mmmh I think I should do a clean rebuild. Something's fishy with my build, it looks like qtwebengine is not correctly linked against libhybris... Oct 06 20:30:07 Tofe: Yeah it's OK on TP and N4, did html5test, kraken 1.1, octane 2.0 tests on both without issues Oct 06 21:37:26 iscgar: https://github.com/iscgar/pmcertificatemgr/blob/massive_code_cleanup/src/cert_utils.c#L1149 seems suspicious. probably there should be "len" instead of the first "added_mem" **** ENDING LOGGING AT Wed Oct 07 02:59:58 2015