**** BEGIN LOGGING AT Fri Nov 06 02:59:58 2015 Nov 06 08:38:15 Morning Nov 06 08:38:42 Herrie: do you still have the link to the "TypeError" you posted some time ago? Nov 06 08:38:57 ah it must be in the logs of irc Nov 06 08:40:05 Tofe: that rebased webengine doesn't build (fails silently), I'm still looking into that Nov 06 08:41:50 JaMa: oh, that's sad Nov 06 08:42:03 Tell if you need help Nov 06 08:44:09 it fails with vanilla qtwebengine already, so it's not in LuneOS specific patches Nov 06 08:44:29 mornin Nov 06 08:44:44 Tofe: TypeError for some apps in WebEngine? or... ? Nov 06 08:44:51 Herrie|Veer: yes Nov 06 08:45:00 the "great screenshot" you took Nov 06 08:45:12 well, sort of (I think there are 2 errors, one in yasm in vanilla, one in LuneOS gyp changes), once I fix the first I'll probably ask you to check the second Nov 06 08:45:54 I'm not so surprised my gyp changes aren't valid anymore, they refactored a lot Nov 06 08:46:47 http://bpaste.net/show/0a71f2b8adbf Nov 06 08:46:58 That's one for email but a bit old already Nov 06 08:48:02 Newer ones: http://bpaste.net/show/43f3fe63ff59 Nov 06 08:48:04 it's this kind of error, but the screenshot with the WebInspector was even better Nov 06 08:48:41 http://tinypic.com/r/1z3864z9 Nov 06 08:49:28 ah, they erased it Nov 06 08:49:33 well, nevermind Nov 06 08:51:10 Sorry my bad, fat fingered there :P http://tinypic.com/r/1z3864z/9 Nov 06 08:53:07 perfect, thx Nov 06 08:58:21 Herrie|Veer: I may have a clue where the issue is Nov 06 08:58:35 :) Nov 06 08:58:53 You reckon it will be an easy fix? Or a tough one? Nov 06 09:08:15 if it is what I think, it should be easy Nov 06 09:12:09 https://github.com/Tofee/qtwebengine/blob/tofe/work/src/core/web_channel_ipc_transport_host.cpp#L66 <-- here, we should also check that message.id==0 Nov 06 09:13:45 Ah :) That's why it doesn't show the list you reckon? Nov 06 09:16:29 The issue seems to be that the flow of messages gets confused: we send an async message followed by a sync message, but the response of the async message comes first and is sent as the answer to the sync message; then the response of the sync message comes, but there is not sync slot left, so an async response is sent, but it has an id of zero, which is invalid Nov 06 09:17:28 I knew my sync response handling was a bit weak, so I'm not very surprised Nov 06 09:18:55 Yeah that sounds like a reasonable explanation to me :) Nov 06 09:19:30 Might also be the cause of the macaw one but could be something completely different. You want me to send you a log for the window.open crash in Testr? Nov 06 09:19:45 BTW seems ADB still no good on my new Dell laptop :s Nov 06 09:20:01 Must be some of the USB write protection software we have :P Nov 06 09:20:15 Funny thing is that Novacom on legacy works fine :D Nov 06 09:20:18 usually the logs for crashes aren't very useful, unfortunately. Nov 06 09:21:02 It's the usual open test from Testr for window management ? Nov 06 09:21:14 Yeah Nov 06 09:21:22 mmh that used to work Nov 06 09:21:29 In windowing there's a window.open test Nov 06 09:21:58 I must have introduced a regression with one of my latest fixes Nov 06 09:24:10 I'm confused seems to work on Maguro Nov 06 09:24:49 Let me check builds Nov 06 09:25:49 Seems OK on Mako now too :s Nov 06 09:29:55 Seems pre-last build on both. Last build only had your recipe fix and rotatable calculator, but no underlying qtwebengine/chromium changes so should be OK Nov 06 09:30:25 I guess it was some other condition causing me to think it was crashing (i.e. very slow or another abnormal termination elsewhere) Nov 06 09:33:26 ok Nov 06 14:07:31 Tofe: http://pastebin.com/DfJt6Jk6 rings a bell? Nov 06 14:18:33 JaMa: only with our commits? Nov 06 14:23:22 JaMa: for qtwebengine, Qt has made a bunch of python scripts that generate gyp build files from the qmake build files Nov 06 14:23:32 gyp: name 'enable_palmbridge' is not defined while evaluating condition 'enable_palmbridge==1' in /OE/build/owpb/webos-ports/tmp-glibc/work/i586-webos-linux/qtwebengine/5.5.99+5.6.0-alpha1+gitAUTOINC+6ded929500_3f6319f4fa-r0/git/src/3rdparty/chromium/third_party/WebKit/Source/web/web.gyp Nov 06 14:23:45 ah, that one is interesting :) Nov 06 14:23:48 yes only with our commits now (fixed other 7 issues in upstream qtwebengine alreado) Nov 06 14:23:58 well this is with only the first commit Nov 06 14:24:09 I'm trying to bisect our commits to narrow it down a bit Nov 06 14:24:34 so this could be one of the cases where squash would be useful :) Nov 06 14:25:18 or it's some cache issue, because git grep enable_palmbridge in current code doesn't show it in web.gyp Nov 06 14:25:55 damn.. different repo Nov 06 14:26:11 the definition should come from the gyp file generated from this: https://github.com/webOS-ports/qtwebengine/commit/f570e472d2953ba97d2f0caff05e29d0fb1905b1 Nov 06 14:27:36 and is used in file 2 and 3 of this commit: https://github.com/webOS-ports/qtwebengine-chromium/commit/995816ffe7b9bb2ab44de2622bdc9c85cdbf1172 Nov 06 15:06:09 Tofe: it's triggered by modules.gyp change in https://github.com/webOS-ports/qtwebengine-chromium/commit/995816ffe7b9bb2ab44de2622bdc9c85cdbf1172 Nov 06 15:30:15 JaMa: yes, but it works with 5.5... Nov 06 15:30:35 Tofe: i have something which works, but I don't have any idea why :) Nov 06 15:31:42 :p Nov 06 16:15:44 Tofe: hard to stop difference :) https://github.com/webOS-ports/qtwebengine-chromium/commit/8e2898974d46d4bb9b41766cb0887720b4ebdd61 Nov 06 16:16:27 the problem is that upstream added another 'conditions' element above "# Disable c4267 warnings until we fix size_t to int truncations." Nov 06 16:16:48 which sets the 'type' Nov 06 16:17:25 so my humble understanding of gyp is that this second 'conditions' was overwritting the first -> that's why gyp was complaining that there isn't 'type' set for modules Nov 06 17:48:12 JaMa: yes, that makes sense Nov 06 17:48:27 there is only one set of conditions per target Nov 06 17:57:11 JaMa: after having slightly lost my eyes reading this gyp file, I think your change is the right one Nov 06 17:58:14 I had to add one more at the end Nov 06 17:58:19 just another include Nov 06 17:58:26 now waiting for another do_compile run Nov 06 18:02:51 Chromium should proving a gyp editor... it's just impossible to understand all this 8+ depths of json brackets Nov 06 18:03:00 provide* Nov 06 18:31:47 true :) Nov 06 18:48:02 elvispre: hi Nov 06 18:48:21 Already managed to get your QtCreator setup? Nov 06 18:51:25 Herrie: I installed it and downloaded the telephone app and luneos-components this afternoon, but it didn't Just Work so I figure I need to RTFM just a little bit. Nov 06 18:52:36 elvispre: phone app is a bit of a different animal. Better start with luna-next-cardshell Nov 06 18:52:42 You're on Windows or *nix? Nov 06 18:52:54 I have problems with luneos-components on my Windows machines Nov 06 18:53:03 I need to disable 1 line of code to get it working always Nov 06 18:53:14 Or FirstUse maybe Nov 06 18:54:12 Tofe: PR tested? Nov 06 18:54:17 Does it solve email bug? Nov 06 18:55:46 Herrie: I am doing LuneOS work in an Ubuntu VM (hosted on Windows). Nov 06 18:56:05 elvispre: Ah OK Nov 06 18:56:21 Herrie: Thanks for the tips. I'll try out those simpler apps first. Nov 06 18:56:24 I usually do QML/QT stuff on my Windows machine and have QtCreator on there Nov 06 18:56:30 Phone has quite some integration Nov 06 18:56:48 If you get cannot find plugin LuneOSComponents ro something I know the solution :P Nov 06 18:57:29 Herrie: It does give me quite a list of missing things for the phone app. I'll ask if I need to. Nov 06 18:57:38 elvispre: Hmmmz Nov 06 18:57:46 Post a log, we might be able to help out Nov 06 18:59:29 Herrie: Let me try luna-next-cardshell and FirstUse... (but first I have to go and buy some birthday wrapping paper) Nov 06 18:59:58 :D Nov 06 19:00:13 This is my folder structure http://tinypic.com/r/2zr47m8/9 Nov 06 19:17:06 Herrie: Thanks for the clarity, FirstUse does Just Work at first. It lets me pick a language then starts getting into trouble with content for the rest of the screens. Nov 06 19:28:53 Herrie: I am trying "Qt Creator 3.0.1 based on Qt 5.2.1". Does that seem about right? Nov 06 19:29:29 That's old Nov 06 19:29:53 I use 3.5.0 from QT 5.5 Nov 06 19:30:00 We use quite some features from 5.5 Nov 06 19:30:05 And that one has QtWebEngine Nov 06 19:30:14 And QtWebEngine is in LuneOS Components ;) Nov 06 19:30:21 So you might want to upgrade Nov 06 19:31:50 Herrie: Yeah. This one was just what was offered by the Ubuntu Software Center so it was trivial to install. I'll go fetch the proper one. Nov 06 19:35:09 Herrie: no, and no Nov 06 19:35:36 I'm preparing a qtwebengine PR, which should solve issues Nov 06 19:36:37 Herrie: note that luneos-components doesn't work on desktop anymore, due to our additional font settings Nov 06 19:37:33 Tofe: Ah ok Nov 06 19:37:48 But that should be fairly easy to fix by disabling them in the QML right? Nov 06 19:38:05 yes, but it can't be committed Nov 06 19:39:13 Yeah Nov 06 19:39:20 Any ideas to work around this? Nov 06 19:39:25 Or will be difficult/ Nov 06 19:39:26 ? Nov 06 19:39:46 I'm working on it Nov 06 19:40:41 :D Nov 06 19:43:17 https://github.com/webOS-ports/luneos-components/pull/21 something like this, should be tested on device first Nov 06 19:46:40 Herrie: what doesn't work with email, exactly? Nov 06 19:51:43 the TypeError is still there... Nov 06 20:25:09 Herrie: this.$.body.setRedirects <-- setRedirects is undefined on chromium, is it a standard API ? Nov 06 20:25:54 (in MessageDisplay.js:906) Nov 06 20:27:51 https://github.com/webOS-ports/core-apps/blob/master/com.palm.app.email/mail/source/MessageDisplay.js#L906 Nov 06 20:30:59 https://github.com/webOS-ports/qtwebengine/pull/10 <-- luna-webappmanager PR should be merged before this one Nov 06 20:40:01 Tofe: Basically you setup and email, 1st pane has your acocunts, folders, favorites. 2nd pane the list of emails Nov 06 20:40:18 This list doesn't show, though first pane shows that there are 9 unread emails Nov 06 20:40:25 I simply added my gmail account Nov 06 20:41:54 So you also cannot tap any emails and see their content Nov 06 20:43:03 I have the same issue; however my PRs get rid of the TypeError thingies - apart the one from mail app itself, which looks like a DOM error Nov 06 20:43:33 Herrie: I am (essentially) up and running with Qt Creator 3.5.1. FirstUse trips up on passwordCharacter, when run. Cardshell runs well but gets confused after a while... but it does that on device too so that might be normal? Nov 06 20:44:13 gets confused? Nov 06 20:44:15 Tofe: I don't see luna-webappmanager pr? Nov 06 20:44:25 Herrie: it's the same one as before Nov 06 20:44:33 Tofe: Ah Nov 06 20:44:38 All tested now? Nov 06 20:44:49 yes Nov 06 20:44:59 elvispre: Yeah passwordchar should work on device only I think Nov 06 20:45:02 Not sure why Nov 06 20:45:11 You can disable that line on desktop I guess Nov 06 20:45:17 Or Tofe might have a workaround Nov 06 20:46:32 Herrie: OK. Sounds like I am up and running. I'll have a play and learn some QML. Nov 06 20:46:37 where is it used? Nov 06 20:47:13 Tofe: firstuse for wifi pw I think Nov 06 20:47:28 Tofe: NetworkConnectPage.qml Nov 06 20:49:49 Tofe: Does teh Enyo 1 appmenu work now? Nov 06 20:49:54 I.e. Display ;) Nov 06 20:49:59 Backend looked OK already Nov 06 20:52:06 no Nov 06 20:55:55 Herrie: it looks like Mojo.relaunch is called twice Nov 06 21:10:54 Tofe: Hmmmz Nov 06 21:11:27 Herrie: on what app should it show a menu ? tweaks ? Nov 06 21:13:04 Tweaks, email Nov 06 21:13:32 Mojo.relaunch should be called on the main application frame, right? Nov 06 21:13:38 I guess so Nov 06 21:13:40 i.e. the headless one Nov 06 21:14:21 mmmh Nov 06 21:16:04 I'm not sure here tbh Nov 06 21:16:15 Tweaks doesn't have headless right? Nov 06 21:16:23 it does Nov 06 21:16:41 on the setRedirects I guess it's something webkit specific somewhere Nov 06 21:16:52 On 3.0.5. doctor I only see it in JS files Nov 06 21:16:55 No references anywhere Nov 06 21:16:58 Checking further Nov 06 21:17:25 I didn't fiind on google, quite strange Nov 06 21:17:54 I only have js sources Nov 06 21:18:36 I mean I would have thought it would be somewhere also in a Webkit source file Nov 06 21:18:42 in /usr/palm/frameworks/enyo/1.0/framework/source/palm/controls/WebView.js there's redirectsChanged: function(inOldRedirects) { this.$.view.setRedirects(this.redirects); }, Nov 06 21:19:58 But that doesn't help much Nov 06 21:22:26 The appmenu confuses me... Should be working... Nov 06 22:02:32 Tofe: Maybe it's on the wrong fram eafter all for Enyo 1? Nov 06 22:05:44 Tofe: I pushed old Calculator and Memos app (both not headless) and app menu works there Nov 06 22:05:59 https://github.com/webOS-ports/core-apps/tree/master/com.palm.app.calculator Nov 06 22:06:04 https://github.com/webOS-ports/core-apps/tree/master/com.palm.app.memos Nov 06 22:24:43 Tofe: so that might confirm my suggestion above that it's the frame where you attach it to, that's the problem **** ENDING LOGGING AT Sat Nov 07 02:59:59 2015