poetaster | piggz[m], I spoke too soon. I don't get it :) | 08:24 |
---|---|---|
poetaster | piggz, went back to a flashed .64 also solves this issue: https://forum.sailfishos.org/t/webview-produces-incorrect-screen-height-window-innerheight-values/13328 | 13:15 |
piggz | hmm, so my issue or jolla issue? | 13:16 |
poetaster | ah, it's a bloody good question. Both, I'm afraid. | 13:18 |
poetaster | The .72 10ii install has that issue I linked here. But none of the other issues which appear on your .72 volla/gs290/gs5 ... | 13:19 |
poetaster | html 6beta4, wasm, ecmascript93859-2 what could possibly go wrong. | 13:20 |
poetaster | piggz, I'll leave you alone until I've figured it out ;) | 13:21 |
mal | poetaster: I checked changelog of webview between .64 and .72 and there were couple of changes, one is this https://github.com/sailfishos/sailfish-components-webview/commit/df9ae3ced8e2a7c4e570ae58463127bec466f23b | 13:25 |
poetaster | thanks! | 13:26 |
poetaster | mal, I'll go through them all. tags? | 13:28 |
mal | 1.5.9-1.19.1.jolla -- 1.5.9.2-1.20.1.jolla | 13:28 |
poetaster | thank you! | 13:29 |
mal | btw, you can find those in forum https://forum.sailfishos.org/t/changelog-vanha-rauma-4-4-0/10655 | 13:29 |
poetaster | Changelog is your friend. | 13:30 |
mal | poetaster: any chance the app doesn't set component size or something as it should? | 13:41 |
HengYeDev[m] | on the pinephone, when i wake the phone it shuts down with the "goodbye" message after about 4 seconds | 14:56 |
poetaster | mal, it's possible but various. some cases webview is top level in Page, anchored to parent and sometimes it's in a SilcaFlickable. Generally anchors.fill: parent | 14:59 |
poetaster | mal, as a 'rule' a web component can't be but scaled to fill parent. In the case where I use webview, that's allways true. | 15:02 |
poetaster | HengYeDev[m], did your package arrive yet? | 16:40 |
HengYe[m] | nope poetaster | 16:42 |
poetaster | damn. it's only one ocean. | 16:42 |
poetaster | mal, piggz[m] for reference I posted some examples of .64 vs. .72, volla/gs and x10ii https://forum.sailfishos.org/t/webview-produces-incorrect-screen-height-window-innerheight-values/13328 | 16:59 |
poetaster | That ticket is reporting the self reporting of javascript running in a webview. demonstrable is that the values are correct in portairt, not in landscape. sigh. | 17:00 |
Thaodan | poetaster: Thanks for fixing tooter. | 17:29 |
Thaodan | Besides some small issues it feels really polished. | 17:29 |
poetaster | Thaodan, oh, it's just the beginning, I fear. | 17:33 |
Thaodan | Why? | 17:33 |
poetaster | Thaodan, ah, just lots of 'small' things. audio doesn't play, can't select server on start, edits not possible. but I like using it :) | 17:34 |
Thaodan | poetaster: hm for me server selection works just fine | 17:34 |
Thaodan | But yeah the small things I've noticed. | 17:35 |
Thaodan | for example there's no "me" page, it is hard to get to the navigation bar, I noticed the missing edits. | 17:35 |
Thaodan | The option to disable or enable images is a little missleading | 17:36 |
poetaster | Thaodan, yeah, profile you also can't edit :) | 17:36 |
Thaodan | notifications don't use the app icon | 17:36 |
poetaster | Ok, will you add them to the issues ;) | 17:36 |
Thaodan | thanks | 17:36 |
mal | poetaster: want to test something? no idea if it can affect anything | 17:47 |
poetaster | mal, sure | 17:48 |
mal | poetaster: try defining the physical size as done here https://github.com/mlehtima/droid-config-fp2-sibon/blob/master/sparse/var/lib/environment/compositor/droid-hal-device.conf#L6 the values are in millimeters | 17:49 |
mal | then reboot and see if anything changes | 17:49 |
poetaster | this is webview, not opengl, right :) I'm swapping too many devices and problems at once. | 17:50 |
mal | poetaster: yes but if something in the system doesn't know physical size it could make it think it's desktop device? | 17:51 |
mal | in theory at least | 17:51 |
poetaster | mal, the height I have is 130 or are you suggesting fo the 10ii? | 17:53 |
poetaster | ah, that's from the fairphone2. so, just a fixed arbirary width? | 17:53 |
mal | poetaster: I mean in the forum you said "the .72 GS290/Volla/GS5 are rendered as if the user agent where desktop" | 17:53 |
mal | for those devices | 17:53 |
poetaster | ok. rebooting. | 17:54 |
mal | measure the physical size of the screen, add the values to the config file and test | 17:54 |
poetaster | thankfully the GS5 is really fast. that does not have any effect on the rendering in webview. | 17:57 |
mal | ok | 17:57 |
poetaster | ist it really in mm? | 17:58 |
poetaster | so 65 mm x 130 mm in the GS5 case? | 17:58 |
mal | yes | 17:58 |
poetaster | so, who'se brilliant idea ..... | 17:59 |
mal | why is it not a good idea? | 17:59 |
poetaster | I'm just prepped by the issue of dpi in relation to printing on paper substrates. and 6 colour separations, mounting plates on presses ;) | 18:01 |
poetaster | so, according to spec. 67x145 ..... | 18:02 |
poetaster | I must admit I'm not up to speed on the meaning of ppi (per inch!) and it's relation to reality. | 18:03 |
mal | journal log might actually tell if it can find physical size | 18:03 |
poetaster | lipstick stil looks good :) | 18:04 |
mal | just strange that x10 II worked fine for you so it suggests some adaptation issue | 18:05 |
poetaster | well, the x10 has one problem in common with the GS290/GS5 ... in landscape. and this is not device but qtmozembed or the like. | 18:06 |
poetaster | So, I think it's two different issues. One of which is port specific and one of which is webview library specific. | 18:07 |
mal | I don't quite understand what that landscape issue is | 18:07 |
poetaster | the javascript run time get's wrong values back for screen height when the phone is in landscape mode. | 18:08 |
poetaster | I wrote a page that spits out what js thinks - something like this comes out: window.innerHeight = 780, screen.height = 780 | 18:09 |
poetaster | when the actual values are: window.innerHeight = 360, screen.height = 780 | 18:10 |
poetaster | BUT, this is not the same issue as with the desktop display. That's what's so infuriating. | 18:10 |
poetaster | At least I have ONE device now where everyting (excpet video recording) works ;) | 18:11 |
mal | there is also QT_QPA_EGLFS_WIDTH and height for size values in pixels if for some reason some low level part is not returning valid values, unlikely though | 18:11 |
poetaster | which logs should I check (the physical units check out). | 18:12 |
mal | journal | 18:12 |
poetaster | is there a keyword? | 18:13 |
mal | physical | 18:13 |
mal | obviously | 18:13 |
Thaodan | poetaster: journalctl -f -u user@100000.service | 18:14 |
Thaodan | that should limit the output to the user session. | 18:14 |
poetaster | Thaodan, but we're at hardware init or? | 18:15 |
poetaster | which should preceede the user foo? | 18:16 |
Thaodan | The qpa plugin is loaded per app. | 18:16 |
Thaodan | So you can change those while starting the app in the shell to test. | 18:16 |
poetaster | --dmesg | 18:17 |
poetaster | ah. ok. | 18:21 |
poetaster | Can't find symbol 'eglCreateDeviceANGLE'. and Release is missing hmmm. | 18:23 |
mal | afaik that is normal | 18:24 |
mal | the first one at least | 18:24 |
poetaster | neato. first time I've heard sound from this app in a year! | 18:24 |
poetaster | (sorry, otherwise broken, but it made a sound!) | 18:24 |
poetaster | it is however only broken in landscape | 18:25 |
poetaster | piggz[m], "/android/vendor/lib/hw/awinic.audio.effect.so" is 32-bit instead of 64-bit | 18:25 |
poetaster | Thaodan, mal there's nothing interesting in the logs. well, the audio errors are ALSO an issue. | 18:26 |
poetaster | mal, to see the common issue (only in landscape mode) https://forum.sailfishos.org/t/webview-produces-incorrect-screen-height-window-innerheight-values/13328 | 18:30 |
mal | poetaster: the values I mentioned were about the desktop mode issue | 18:30 |
mal | not about landscape | 18:30 |
poetaster | mal, sure, but the same app, moremahjong in Portrait mode on the GS5 with the adjusted values still behaves as if it were a desktop browser | 18:31 |
mal | just to be sure, test also setting the other values I mentioned | 18:32 |
mal | because desktop mode issue must be related to some part of the system not knowing some size of the display | 18:32 |
poetaster | mal, yes | 18:33 |
poetaster | mal, no difference. | 18:36 |
poetaster | strange. it has the correct values in portrait. | 18:36 |
poetaster | sorry, too many devices. | 18:37 |
mal | "it has the correct values in portrait" what does that mean | 18:38 |
poetaster | damn it. the gs5 network leaves something to be desired. | 18:46 |
poetaster | mal, piggz[m] in the functional volla those values are unset. hmmm | 19:58 |
poetaster | mal, when the device is in portrait, the webview displays content AND javacript returns height/width values which make sense (1080x1920 for insance). | 19:59 |
poetaster | It's all laid out in detail in the ticket link I posted. If you would prefer, I'll send you an app that demonstrates it with javscript writing to the display on orientation changes. | 20:00 |
poetaster | in short: | 20:02 |
poetaster | window.innerWidth = 780, screen.width = 360 | 20:02 |
poetaster | window.innerHeight = 780, screen.height = 780 | 20:02 |
poetaster | what's wrong with this picture? (orientation landscape). | 20:02 |
poetaster | 'correct values' would be 780/780 and 360/360 | 20:04 |
poetaster | in portrait orientation: | 20:04 |
poetaster | window.innerWidth = 360, screen.width = 360 | 20:04 |
poetaster | window.innerHeight = 780, screen.height = 780 | 20:04 |
poetaster | the correctness of which is self-evident from the proportions. | 20:05 |
EttoreChimenti[m] | <attah> "60 or 78 are your options if you..." <- Trying to compile on 60.9esr, I get some problems as system cairo 16.0 is not compiled with Qt bindings (getting errors on QPainter). | 20:33 |
EttoreChimenti[m] | In mozconfig I see `ac_add_options --enable-default-toolkit=cairo-qt`, is it possible to use a different toolkit? | 20:33 |
attah | Are you asking if QtMozembed can work without Qt? O.o | 20:35 |
EttoreChimenti[m] | attah: Of course not :) Just asking if there is an other way. If not, there is a known procedure to compile cairo bindings with qt5? | 20:37 |
attah | Recompile the system cairo with the needed bindings? | 20:38 |
EttoreChimenti[m] | On Yocto seems like there is no option https://git.yoctoproject.org/poky/plain/meta/recipes-graphics/cairo/cairo_1.16.0.bb | 20:38 |
EttoreChimenti[m] | Do you know which version is used to compile qtmozembed for esr60? | 20:39 |
attah | No idea, and no experience... but should be evident from the related repos | 20:39 |
attah | Presumably it is the very latest minor version, like what we have on 78 | 20:40 |
attah | Or which version of what? | 20:40 |
EttoreChimenti[m] | It's very strange as cairo 1.17 removed all qt bindings | 20:42 |
EttoreChimenti[m] | https://github.com/sailfishos-mirror/cairo/commit/64ece6d545df015377a1f289829eeea20f5ae019 | 20:42 |
Thaodan | I'm not sure if Qmozembedded is the same as the cairo qt backend that firefox had | 20:43 |
Thaodan | that was a nokia thing for on the desktop | 20:43 |
attah | So perhaps (at least) the 78 port has another means of rendering | 20:44 |
EttoreChimenti[m] | I don't see any switch for Qt even here https://github.com/sailfishos/cairo/blob/1.16.0/rpm/cairo.spec | 20:44 |
EttoreChimenti[m] | I'm very confused at this point | 20:44 |
attah | We should really pester flypig about this :=) | 20:48 |
EttoreChimenti[m] | Is it possible that cairo can be compiled against Qt4 and be still compatible with Qt5? | 20:48 |
Thaodan | no | 20:48 |
Thaodan | Qmozembbed talks to mozembedded and this is used in the browser | 20:48 |
Thaodan | the cairo thing was an enterily different thing | 20:49 |
EttoreChimenti[m] | Do you know why Qt complains about not finding references to QPainting (which are in the cairo package, but disabled by default and impossible to enable)? | 20:52 |
Thaodan | Idk I wouldn't waste time on it. | 20:52 |
* EttoreChimenti[m] sent a code block: https://matrix.org/_matrix/media/v3/download/matrix.org/HMIUsEQgBUVoZwhsODgAddNk | 20:53 | |
EttoreChimenti[m] | * ```... (full message at <https://matrix.org/_matrix/media/v3/download/matrix.org/bijFDEtgduRCqvFJgOwrzDNA>) | 20:53 |
EttoreChimenti[m] | Thaodan: Me neither, seems really difficult to make it work in this environment | 20:54 |
Thaodan | EttoreChimenti[m]: Here watch this if you want to know more about the api behind the sailfish browser: https://www.youtube.com/watch?v=fksXutzD5hk | 20:56 |
Thaodan | What you are trying is something else enterily | 20:56 |
EttoreChimenti[m] | My goal is to compile the Qtmozembed module to be embedded in a Qt application | 20:57 |
Thaodan | than you don't need cairo with qt | 20:58 |
EttoreChimenti[m] | Ah. | 20:58 |
Thaodan | You have to compile Qtmozembed and the gecko used by sfos | 20:58 |
EttoreChimenti[m] | It's what I'm trying to do, am I missing something? | 21:00 |
EttoreChimenti[m] | I am using the commands, patches and the configs found in the rpm folder | 21:01 |
EttoreChimenti[m] | <attah> "We should really pester flypig..." <- flypig: can you please advice? | 21:15 |
Thaodan | You have to wait till tomorrow | 21:19 |
Thaodan | About: Cairo I was wrong however the browser uses intree cairo I think. | 21:21 |
EttoreChimenti[m] | Thanks for today:) | 21:24 |
poetaster | anyone who wants to observ webview behaviour: install https://poetaster.de/media/webview-screensize-0.1-1.noarch.rpm and rotate through the orientations. boring. but boring. | 21:26 |
poetaster | goes to get beer. | 21:26 |
poetaster | I do think it has something todo with mozembeded. | 21:29 |
poetaster | Thaodan, I was just wondering you had mentioned being able to select which mastodon instance? | 21:34 |
Thaodan | poetaster: You mean for login? Yes. | 21:35 |
* Thaodan is away dinner | 21:35 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!