Monday, 2020-01-27

abransonWhat does that do?00:00
r0kk3rza compositor hack and maybe some event plumbing i think00:00
*** zbenjamin is now known as Guest4254302:20
*** zbenjamin_ is now known as zbenjamin02:20
*** SpeedEvil is now known as Guest9495302:22
*** BitEvil is now known as SpeedEvil02:53
rinigusr0kk3rz: flatpak-runner is also config app. you can set env vars, qt scaling, dpi (default and app-specific).06:29
rinigusas it also now generates gl extension on start for you, it has desktop file to make it simple06:30
rinigusabranson: flatpak does add desktop files in XDG, but its ignored by lipstick06:33
rinigusright now I am using it for our advantage by generating new desktop files under .local/share/applications and changing Exec to `flatpak-runner flatpak-app-name` . its rather crude, but works normally. could probably extract Exec command and just replace bits in it.06:35
rinigusabranson: flatpak-runner contains wayland compositor used by flatpak apps, ensures that it rotates with the device and assists with maliit keyboard required info (rotation and whether app is active)06:36
abransonrinigus: fixed teh XDG_DATA_DIRS in lipstick fairly recently:
abransonmight be more robust to somehow integrate that flatpak-runner functionality into the flatpak binary itself? then you could let flatpak look after the desktop files07:02
rinigusabranson: do you know whether that fix is already published in sfos 3.2.1.x?07:04
riniguswe probably still need a wrapper as we have to open new wayland compositor and rotate it. plus send data to maliit plugin inside flatpak07:05
rinigusre fix - as I don't have any extra icons coming on after flatpak app installs07:06
abransondoes it need to be a wrapper specifically, or just called during flatpak app init?07:06
rinigusright now its a wrapper. in addition to functions listed above - our compositor does not support minimizing => apps disappear from screen if you switch them or lock the screen07:08
abransonhmm that fix should be in there...07:08
rinigusxdg_data_dirs are set in my environment07:09
rinigusto point to flatpak exports, so should work already. but doesn't :(07:10
abransonset in /var/lib/environment/compositor/ ?07:11
rinigusabranson: don't know where. in terminal, the var is set07:18
rinigusabranson: in compositor, I don't think so07:18
rinigusabranson: do you know why single instance = no support is not working in lipstick? its a bummer right now as we can have only one flatpak app running in a time...07:20
abransonrinigus: no, but I bet no-one's used it for a long time07:35
abransondidn't the harmattan browser use that?07:35
abransonadd a conf file to that dir and lipstick should get it07:36
rinigusabranson: there is a patch around - can you folks somehow restore single-instance=no?07:37
abransonwill have a look07:37
rinigusabranson: thanks!07:38
abransonoh, it worked for r0kk3rz? might not be so simple07:38
rinigusabranson: I have no idea why it worked for r0kk3rz . I am not 100% sure whether he started them from app grid though07:39
rinigusabranson: do you know whether your lipstick fix also applied for icons search path? from looking into the code it seemed to me that it didn't07:40
abransonno, that's more complicated to fix because it involves the theme paths07:41
rinigusabranson: I better not distract you from single-instance=no fix. those smaller annoyances can come later :)07:43
abransonrinigus: looking at that single-instance thing, i'm not sure if that's what you want. you'd be launching multiple instances of the same flatpak apps.08:15
abransonlooking in that Switcher qml that the patch applies to, there's an 'isAndroidApplication' function that takes the first arg of the command as the identifier instead. that might be better for your use case as long as there's no other side effects.08:16
abransonit's a simple check for a 'X-apkd-apkfile' key in the desktop file by the looks of it08:16
rinigusabranson: if it is possible to make it more clever than plain single-instance=no - even better. so, what should I specify in `X-apkd-apkfile`?08:28
abransonrinigus: looks like it's returning: launcherItem.readValue("X-apkd-apkfile") != ""08:28
abransonso i guess whatever you like. might be nicer to put the package name there maybe08:29
abransonhopefully that won't cause any strangeness elsewhere. like trying to launch aliendalvik ;)08:29
rinigusabranson: I will try tonight ... although, don't have aliendalvik - so cannot test it08:30
rinigusabranson: although, its still not clear why different .desktop files (one per app) would lead to launcher to consider them the same app.08:34
abransoncheck the function windowIndexOf in that Switcher.qml08:38
rinigusabranson: will do, thanks08:39
abransonrinigus: looking a bit further down, you may also need to set window.surface.className in your flatpak apps. they are analogous to android apps, but it's still going to be a bit of a hack.08:49
rinigusabranson: where do I set that (window.surface.className)?08:50
rinigusin .desktop file?08:50
abransoni guess it's a property of the wayland surface, so in your runner?08:51
rinigusabranson: will have to look into how to set that08:56
abransonyeah sorry i'm guessing now :) good luck08:57
abransonvar window = Lipstick.compositor.windowForId(switcherModel.windowId(i))08:57
dcaliste@flypig, do you want some help on ? If you are busy with various things, I can address some of jpetrell comments ?09:59
flypig@dcaliste, thanks for the offer. It's awkward, because while it would be useful, I'm halfway through addressing a bunch of them. I think I'd have to commit my changes first before it would make sense.10:00
dcalisteSure, I would start from your work anyway. As you prefer.10:01
flypigOne thing that I've not looked at yet is this issue of the "initial state". When things are first set up, some account types have all folders with sync flag set, some have none.10:01
dcalisteAs commented on git @ SFOS, I've quickly tested the full stack on device, and it seems to work.10:01
flypigGreat, I didn't get to your comments yet, but that's good to hear :)10:02
dcalisteYes, that's right. I didn't look at it neither... I thought first that QMF was setting this to True by default.10:02
dcalisteOtherwise it would create now (before my patches) some issues of folder not synced.10:02
flypigThat would make sense, but apparently it's not always the case. Or maybe other things may have changed them for existing accounts.10:03
flypigI'm not sure Google and EAS were respecting the flags though.10:03
dcalisteSo, I let you have look to the full stack status when you have the time to and ping me if you need help on something.10:04
dcalisteI don't know about the status for me to access eas repo to look at your patches. Currently I'm not allowed to.10:05
flypigAh, okay, I'll follow that up.10:05
flypigSorry about that.10:05
dcalisteNo problem, not in a hurry from my side ;)10:05
flypigWell, I'm sorry, as I'm sure you must be keen to see it finalised (as am I) and it's frustrating if you have some time but I'm blocking things.10:06
dcalisteNo, no, you're not don't worry.10:07
flypigNevertheless, I'll try to get my current changes committed as a WIP ASAP :)10:07
dcalisteI plan this thing with synchronizationEnabled flag because I thought at first that it was designed for and not intrusive. But you showed my it was a bit more complex.10:07
dcalisteSo it needs more time to polish it, no problem.10:07
flypigWell, your version was fine when it was applied only to IMAP. It's the extension to the other types that seem to have complicated things. But we're getting there!10:08
dcaliste@flypig, for the record, the synchronization flag is set in imapclient.cpp#825 in ::mailboxListed(). For pop, (for what it's worth), it's in popclient.cpp#309.10:18
dcalisteI don't know for imap, if there can be some code paths that are creating a folder different from the one mentioned here though...10:19
flypigdcaliste, that's useful, thank you. Can POP3 have folders?10:35
dcalisteNo, but internally in QMF it has one, the Inbox one.10:35
dcalisteAnd it has flag set also.10:36
flypigOkay :) I've asked again about EAS, but it's not looking straightforward. I'd like to get you access, but it may not be possible.10:36
dcalisteBut it's no used anywhere (well, almost, grep is returning one match of usage in pop code).10:36
dcaliste@flypig, no problem, it was for review anyway.10:37
flypigIt would really help for you to have the full picture, in my view, but we'll see.10:37
dcalistenot that review are useless ;)10:37
dcalisteOn the contrary of course.10:37
*** frinring_ is now known as frinring10:40
rinigusabranson: summarized my experience so far in sfos-devel mailing list. in short, nothing works perfectly. while I can get several instances up, its a mess. didn't try android hack - that would require different cmd line, as I can guess.21:30
*** zama_ is now known as zama21:45
abransonrinigus: i don't think so, it just uses the first arg instead of the binary name21:46
rinigusabranson: cmd.split(' ')[2] is probably 3rd?21:47
abransonAnd you set that wayland classname field to whatever the flatpak packag name is21:47
abransonOk, maybe.21:48
riniguswill have to see how to set that classname and to which window...21:49
abransonBut we can probably slip in some support for flatpak desktop files21:49
rinigusI am off to sleep, will read logs/replies tomorrow21:49
Sailor3354_wich apk is working for YouTube on my device ?22:19
BenKerryWhats Up22:32

Generated by 2.17.1 by Marius Gedminas - find it at!