**** BEGIN LOGGING AT Tue Sep 29 02:59:58 2015 Sep 29 04:03:52 morphis ping? Sep 29 06:50:56 morning Sep 29 07:14:15 Morning Sep 29 07:16:32 Tofe: morning Sep 29 07:16:41 Nice progress on WebEngine! Sep 29 07:16:53 Tofe: nice! Sep 29 07:26:00 wow wow, not so fast, I didn't try to flash it yet :) Sep 29 07:27:42 ah, morphis, one question: is there a reason why you didn't use the simpler WebKit Bridge approach for communicating between html client and qml/C++ ? Sep 29 07:28:06 simpler WEbKit Bridge approach? Sep 29 07:28:36 It's just a matter of publishing Qt objects to the JS client side directly, and calls are then synchronous Sep 29 07:28:43 https://www.qtdeveloperdays.com/sites/default/files/webchannel.pdf Sep 29 07:29:15 ah Sep 29 07:29:28 Tofe: at the time I started with this qtwebchannel wasn't there Sep 29 07:29:41 or that far that it was usable for this Sep 29 07:29:59 The PDF isn't only about webchannel, it's actually a comparision of webchannel and bridge Sep 29 07:30:10 Tofe: if that works well now I don't have anything against using it Sep 29 07:30:12 Quite a good summary of different usages Sep 29 07:30:28 Well, I have: it doesn't exist anymore on WebEngine :) Sep 29 07:31:40 ? Sep 29 07:31:50 ah you mean the old bridge Sep 29 07:31:54 yep Sep 29 07:31:57 yeah, that one went away after webkit1 Sep 29 07:32:10 as it doesn't work anymore with the UI/Web process model Sep 29 07:32:21 Didn't we use webkit1 so far? Sep 29 07:32:53 ah ok, yes it makes sense that it doesn't work with webprocess Sep 29 07:33:32 I'm still trying to see how to achieve synchronous calls with WebEngine, but more investigation is needed, as the IPC system is made to be async Sep 29 07:33:45 Tofe: no, we're using webkit2 since the beginning Sep 29 07:34:08 Tofe: right, I am not sure if we can simulate a sync call with that somehow in pure js Sep 29 07:34:10 but doubt it Sep 29 07:34:14 morphis: ah ok, I misunderstood that bit. Anyway, I think I never really knew webkit1. Sep 29 07:34:57 morphis: there is a way to get QML properties synchronously, as they expose a cached value on client side Sep 29 07:35:11 But that doesn't solve the getResource() case Sep 29 07:35:20 Tofe: right Sep 29 07:35:31 getResource is the most critical one Sep 29 07:38:09 There seem to be an IPCSyncMessage in WebEngine, but usage is strongly restricted Sep 29 07:42:34 Used a bit here for instance: https://github.com/atom/electron/blob/master/atom/renderer/api/atom_api_renderer_ipc.cc Sep 29 08:27:57 Tofe: https://github.com/crosswalk-project/crosswalk is also worth a look Sep 29 08:28:03 they have a extension API Sep 29 09:39:37 Ah yes, right, interesting too Sep 29 13:16:12 morphis: I think I'll be able to hack WebChannel a bit and provide an IPC::SyncMessage there too. Sep 29 13:16:21 nice Sep 29 13:18:57 Indeed we can see in https://github.com/Tofee/qtwebengine-chromium/blob/33.0.1750.170-based/chromium/ipc/ipc_message_macros.h that IPC sync messages handling is already available; it's just that chromium avoid using it. Sep 29 13:43:37 morphis: You ok with both webengine and webkit for now? Sep 29 13:43:51 So we include both in the build until we can phase out webkit? Sep 29 13:44:45 JaMa: Tofe came up with this: https://github.com/webOS-ports/meta-webos-ports/commit/588c889b920252f4b779fb9af8c82f2c5b1cad0f , you happy with this or have another solution? Sep 29 13:46:02 morphis: So we can switch browser to WebEngine already for this month's release and then Tofe can iron out the other bits going forward for the apps? Sep 29 13:48:00 Herrie: it's ok, having $D in FILES variable was completely wrong from the beginning Sep 29 13:50:01 but I'm not sure how this is supposed to work in case we have multiple recipes providing fonts in the image Sep 29 13:50:32 they will all either install fonts in qt5 directory if qtbase is installed first Sep 29 13:50:59 or qtbase installation will fail, because opkg will refuse to replace fonts directory with symlink to qt5/fonts Sep 29 13:53:27 JaMa: Yeah we found out $D didn't work during build and fixed it Sep 29 13:53:39 If you have a better suggestion we're open for it of course :) Sep 29 13:55:14 Just whole WebEngine would crash because it couldn't find the fonts. THe symlink "fixes" that ;) Sep 29 14:12:55 JaMa: should we symlink the files individually ? Change Qt5 configuration ? Sep 29 14:31:15 I'm adding new variable in meta-qt5 which should allow to easily move fonts around Sep 29 14:32:59 other option would be to append another element to fontconfig's fonts.conf Sep 29 14:50:53 ok great Sep 29 15:25:12 morphis: actually, sync messages in chromium are quite great; it's even possible for the receiver to make a delayed answer, which basically means more common code between sync call and async call management Sep 29 17:30:15 Hi guys! Sep 29 17:31:00 I've got some free time on my hand in the near future and would like to get back to luneos development. Sep 29 17:33:13 So as a start I was thinking about taking something easy to get me going and settled on trying to finally close the long standing #633 (display last password char) Sep 29 17:34:39 Has anyone played with QPA before? Sep 29 17:58:51 iscgar: I think that one might have been solved already, not 100% sur. Seems QT 5.5 to the rescue there Sep 29 17:58:58 morphis: ping Sep 29 17:59:20 Herrie: pong Sep 29 18:01:45 Herrie: I'm currently struggling with building an image, so can't test this; but I'm curious to know if the cause was ever investigated. Sep 29 18:07:22 morphis: You OK for now to have both QtWK and QTWE in images? Memory wise footprint seems similar-ish. Little higher on QTWE but nothing too shocking. Image with both takes 920 or so MB on TP Sep 29 18:08:49 This way we can put browser with QTWE in monthly release while Tofe sorts out other bits for the webappmanager etc. So eventually maybe in 1-2 months time we can remove QTWK Sep 29 18:09:19 iscgar: Seems QT changed the behavior a couple of times with different releases Sep 29 18:10:34 We need to add http://doc.qt.io/qt-5/qml-qtquick-textinput.html#passwordMaskDelay-prop to the luna-webappmanager I think Sep 29 18:12:07 Herrie: That property is already there? And what about native apps? Sep 29 18:13:58 Seems it's actually: https://github.com/Tofee/qt5-qpa-hwcomposer-plugin/blob/938abe7537268227effa60920b162ac458a28115/hwcomposer/qeglfsintegration.cpp#L181 Sep 29 18:14:09 But that currently doesn't work so needs some debugging Sep 29 18:14:23 iscgar: THey introduced it with QT 5.4 ;) Sep 29 18:14:31 We moved there not too long ago Sep 29 18:14:35 Now on 5.5 Sep 29 18:14:52 This should deal with Enyo 1/2 Sep 29 18:15:02 For QML apps we can jsut put it in the QML file where appropriate Sep 29 18:17:18 Herrie: The enum value was there beginning with 5.0 (and 4.x had some other ugly hack to set it), only the QML property was introduced in 5.4 Sep 29 18:19:06 From what I gather it seems that qtwayland doesn't load the platform plugin, so it'll never work for wayland clients (which webappmanager is, if I understand correctly) Sep 29 18:21:47 Herrie: And so far it seems that other projects decided to just recode the general wayland theme and add it there. Would you like me to take a look at it? Sep 29 18:22:43 Sure, just check with Tofe to make sure it's not contraproductive with his migration of QtWebKit to QtWebEngine and the rework he's doing ;) Sep 29 18:23:05 Anyone helping out with C++ side of stuff I'm happy. I can read a bit but that's about it :P Sep 29 18:25:26 Herrie: sure thing, though I don't think that there should be any problem (given that there isn't an obvious connection between qtwayland and qtweb(kit|engine). Sep 29 18:27:11 Tofe: Are you working currently on something related to #633 and/or qtwayland theme? Can I mess with it to try and close #633? Sep 29 18:41:16 Herrie: Trying to check if it was really fixed, I can see that in browser the issue still persists, which means that Enyo1/2 apps are likely affected as well. Sep 29 19:16:27 iscgar: yeah it's still there ;) Sep 29 19:36:07 Herrie: well, I'll try a few things as soon as oe finishes building the first image on my VM (this is the third time I'm trying after my setup got messed up becuase of a stalled build). Sep 29 19:46:00 iscgar: hi! Sep 29 19:47:02 iscgar: I'm not currently playing with qtwayland-related things, only qtwebengine, luna-webappmanager and qtwebchannel. So if you want to fix things in qtwayland, please do :) Sep 29 19:47:16 iscgar: beware, we use our own patched fork of qtwayland Sep 29 19:47:42 (but we try to keep it as close to upstream as possible) Sep 29 19:49:03 Tofe: yeah, I was actually going to ask about that. Should the solution be webOS specific (that's what the Mer project decided to do) or be a generic one that can (hopefully) be submitted upstream? Sep 29 19:52:09 If it's hacky, it's unlikely to be accepted upstream; I'd say let's first try to have some specific working Sep 29 19:52:53 something* Sep 29 19:54:51 Sure. upstream itself is quite a hack when it comes to honoring the platform theme (it explicitely checks for KDE/Gnome and resors to GenericUnixTheme otherwise). Sep 29 19:55:08 *resorts Sep 29 19:56:17 iscgar: for the theme, don't we already have some code for our platform? Sep 29 19:56:27 I don't remember well that part Sep 29 19:57:03 Herrie: result of my latest flashing: luna-webappmanager is broken, but I didn't break webengine. Sep 29 19:57:33 Herrie: but I have some understandable logs, so it shouldn't be too hard Sep 29 19:59:35 iscgar: mmh no, doesn't seem so.... Maybe I'm confused with another code. Sep 29 20:00:09 Tofe: No that I know of. The only thing that might have been used in open-webos is qt5-webos-plugin (or something like that), but it was abandoned early on. Sep 29 20:01:49 But that might be an opportunity to actually create a platform theme to host the platform specific looks. Sep 29 20:02:06 (after I actually get it to work, that is) Sep 29 20:03:10 maybe yes Sep 29 21:02:08 morphis: roughly, this is what I had in mind: https://github.com/Tofee/qtwebengine/commit/bae880066dd3eeda466adc5b3894f725787056c6 + https://github.com/Tofee/qtwebchannel/commit/eb763e8bf5e8362985b5253e145e77f74248ded9 + https://github.com/Tofee/luna-webappmanager/commit/c91e6f688860e300e43232ce471f431ac77235d7 Sep 29 21:02:36 (take a deep breath before reading these...) Sep 29 21:03:29 There may be some compilation errors here and there, but the design should be ok **** ENDING LOGGING AT Wed Sep 30 02:59:59 2015