[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-panel-devel
Subject:    Minutes Monday Plasma Hangout
From:       Sebastian =?ISO-8859-1?Q?K=FCgler?= <sebas () kde ! org>
Date:       2018-01-22 12:09:15
Message-ID: 3606728.t6AUnWFE4L () miro
[Download RAW message or body]

Meeting minutes and log...
-- 
sebas

http://www.kde.org | http://vizZzion.org
["minutes-week-04.txt" (minutes-week-04.txt)]

Plasma Team Meeting, 22 january 2018

Present: Sho, notmart, d_ed, kbroulik, bshah, romangg, sebas

Sho:
* [Kicker] Fixed the search result that Enter/Return will run while still in the \
search field not being highlighted, giving no visual feedback you can just slam the \
                key
* [Kicker] Fixed menu closing unexpectedly when moving mouse cursor away from an item \
                that opened a subdialog (two seperate bugs, fixed in one approach)
* [Dashboard] Fixed a small regression from a recent refactoring breaking "Add to \
                Favorites"
* [FV] Almost ready to add requested unit test to \
https://codereview.qt-project.org/#/c/215939/ - was painful to figure out how to \
                write
* [Simple Menu] Released v1.0.8:
** Added user-requested option to only change categories on click (default: hover \
                switch)
** Synced custom panel button image handling to latest Kicker
** Synced config dialog UI to latest Kicker
** Mouse-hover-vs-keyboard-nav fix a la previous Dashboard fix
** Quite nifty speed optimizations a la previous Dashboard work
* [Simple Menu] Released v1.0.9 with a follow-up fix for a bug that causes system \
                action buttons to stay highlighted when the mouse pointer left the \
                area
* [Simple Menu] Have another bugfix release planned soon ...
* [Other] Did some GIO/GVFS remote filesystem tests to evaluate their approach to \
mounting vs. ours, as users seem to prefer it

notmart:
* first rough working kernel driver for dell tablet mode (it turns out it's a \
standard-ish feature of intel acpi, so may work on more laptops than dells), in talk \
                with some kernel people to make it into mainline
* big kirigami branch about static plugins has been merged, deployment story perhaps \
                a bit easier
* fixes build issues on Android
* started patch to expose to kwin wether we are in tablet mode D9944 need help how to \
correctly wire things together Diff 9944 "[WIP]expose if the tablet mode switch is \
                available" [Needs Review] https://phabricator.kde.org/D9944
* plasma: tablet mode experiments: show/hide panels and show/hide plasmoids
* pushed slimmer scrollbars in breeze just on master, will be in 5.13

d_ed:
I ported the minimize all applet to libtaskmanager to work on wayland
and get rid of an import whilst we're at it

kbroulik:
* Fixed memory leak in notifications history using a Repeater and creating all \
delegates at once (now uses ListView, it's rather ugly but the least invasive (read: \
                not a total rewrite) patch I could come up with that could still go \
                into 5.12)
* Tried to fix Kickoff user avatar not reliably updating, D9950 and D9951 but it \
still breaks in certain cases Diff 9950 "[KUser] Check whether .face.icon is actually \
readable before returning it" [Accepted] https://phabricator.kde.org/D9950 Diff 9951 \
"[KUserProxy] Adjust to accounts service" [Needs Review] \
                https://phabricator.kde.org/D9951
* Appmenu applet now underlines mnemonics only when Alt pressed to be in line with \
                Breeze style
* Fixed deleting and creating files ("New File" menu) in Folder View taking forever \
(it didn't signal KDirNotify properly), D9912 still open. Haven't figured out a fix \
for DropJob yet Diff 9912 "[KNewFileMenu] KDirNotify::emitFilesAdded after storedPut" \
                [Needs Review] https://phabricator.kde.org/D9912
* Looked into getting rid of those stupid "Loop device" we get in places for e.g. \
snappy mounts: D9895 Diff 9895 "[UDisks] Ignore non-user mounts" [Needs Review] \
                https://phabricator.kde.org/D9895
* Since we're getting quite a few crashes in libsqlite in the Bookmarksrunner lately, \
which pokes Firefox' database file to extract bookmarks (speaking of sqlite, could \
that be related to kactivities also breaking? that qsqlite has a regression?), I was \
looking into providing the bookmarks runner through plasma-browser-integration. It \
can just query the browser instead. However, after I did all of that for D9984 I \
figured that it obviously won't work if the  browser isn't running. So I don't know \
how to procede. We could also have p-b-i extract all bookmarks, save them into a json \
file somewhere and the search there if not running but that wouldn't make it much \
better than what we currently have, except for that we don't need SQL. Diff 9984 \
"WIP: Add bookmarks runner" [Needs Review] https://phabricator.kde.org/D9984

bshah:
- Mostly working on same stuff as last week
- Which includes testing the fixes for udev-android-networkmanager-simplelogin \
                startup race
- On popular request worked on Plasma Mobile ISO based on neon : \
                http://images.plasma-mobile.org/iso/
- Need help with that as plasmashell doesn't start on first start
- Will work on packaging newer libinput in neon
- will work on packaging qt 5.10 on neon

romangg:
well, currently (re)writing my Xwayland patches as I said.
d_ed and I added two KWin ideas for GSoC
https://community.kde.org/GSoC/2018/Ideas#KWin
there are still not many Plasma ideas.
Could there be a plasma workspace idea?

sebas:
* Testing kscreen osd branch by dvratil D9414 (X works nicely with a few patches on \
                top, but
* wayland hotplugging is broken on my system, so looking into that
* More kscreen bug triaging done and to do
* semi-related: Working on privacy goal stuff, collecting info, rallying people, PR \
                stuff around it
* want to get back onto the Plasma Mobile state of the art


[12:00:36] <sebas> Meeting time! Who's here?
[12:02:35] <bshah> I am here, but will take bit to preparte notes
[12:02:37] <Sho_> here
[12:02:43] <notmart> here
[12:03:06] <d_ed> I've not had any time for Plasma, so not really here.
[12:03:21] <notmart> the schroedinger david :)
[12:03:44] <kbroulik> jeez. bug against kdialog. I trace it, went from kdialog to \
plasma-integration, then went to KIO and now it's back in plasma-integration again \
[12:03:48] <kbroulik> here [12:04:59] <sebas> Alright ... let's start with Sho, \
notmart, d_ed, kbroulik, bshah, /me [12:05:00] <romangg> here, though it's like David \
(working on my Xwayland patches) [12:05:01] <sebas> GO!
[12:05:04] <sebas> +romangg :)
[12:05:18] <Sho_> * [Kicker] Fixed the search result that Enter/Return will run while \
still in the search field not being highlighted, giving no visual feedback you can \
just slam the key [12:05:19] <Sho_> * [Kicker] Fixed menu closing unexpectedly when \
moving mouse cursor away from an item that opened a subdialog (two seperate bugs, \
fixed in one approach) [12:05:21] <Sho_> * [Dashboard] Fixed a small regression from \
a recent refactoring breaking "Add to Favorites" [12:05:22] <Sho_> * [FV] Almost \
ready to add requested unit test to https://codereview.qt-project.org/#/c/215939/ - \
was painful to figure out how to write [12:05:24] <Sho_> * [Simple Menu] Released \
v1.0.8: [12:05:25] <Sho_> ** Added user-requested option to only change categories on \
click (default: hover switch) [12:05:27] <Sho_> ** Synced custom panel button image \
handling to latest Kicker [12:05:28] <Sho_> ** Synced config dialog UI to latest \
Kicker [12:05:30] <Sho_> ** Mouse-hover-vs-keyboard-nav fix a la previous Dashboard \
fix [12:05:31] <Sho_> ** Quite nifty speed optimizations a la previous Dashboard work
[12:05:33] <Sho_> * [Simple Menu] Released v1.0.9 with a follow-up fix for a bug that \
causes system action buttons to stay highlighted when the mouse pointer left the area \
[12:05:34] <Sho_> * [Simple Menu] Have another bugfix release planned soon ... \
[12:05:36] <Sho_> * [Other] Did some GIO/GVFS remote filesystem tests to evaluate \
their approach to mounting vs. ours, as users seem to prefer it [12:06:19] <Sho_> \
next() [12:06:43] <romangg> re Simple Menu: I can't click on "recent documents" from \
the context menu of a favorite [12:06:56] --> llucas \
(d4f4b91c@gateway/web/freenode/ip.212.244.185.28) has joined #plasma [12:07:03] \
<kbroulik> Sho_: did you look at kio fuse? [12:07:04] <Sho_> that's the * [Simple \
Menu] Have another bugfix release planned soon ... [12:07:08] <Sho_> kbroulik: no
[12:07:15] <kbroulik> https://techbase.kde.org/Projects/KioFuse you should
[12:07:25] <kbroulik> if you're into "remote filesystem tests"
[12:07:36] <Sho_> kbroulik: I didn't know someone resurrected it
[12:07:40] <Sho_> You have it working?
[12:07:40] <kbroulik> it wasn't
[12:07:51] <kbroulik> at one point I checked out the svn and tried to port it to qt5 \
and ecm but gave up [12:08:03] <kbroulik> but might be worth resurrecting it and \
completing it [12:08:21] <Sho_> kbroulik: My employer asked me to look into it so I \
asked David Faure to summarize issues with the GVFS approach and he had me run some \
testrs [12:08:21] <sebas> I once tried to mount a nepomuk kio slave, that was pretty \
hilariously broken :D [12:08:32] <Sho_> Which are inconclusive so far because I \
couldn't actually get anything to use a mount yet [12:08:40] <Sho_> apparently most \
things now just GIO directly, which is basically like our stuff [12:08:43] <kbroulik> \
not being able to easily browse smb shares in libreoffice and $third-party-IDE is a \
major pain point with plasma, same for not being able to stream video from a NAS \
easily [12:08:45] <sebas> idea was to do a fs-based backup tool of semantic data, \
crack smoking is awesome [12:08:53] <kbroulik> do a GIO-KIO bridge \o/
[12:09:07] <Sho_> So the question would be why does LibreOffice have GIO spport but \
not KIO support [12:09:19] <kbroulik> we should bester LiMux into having them do \
that, they sponsored the Qt 5 port after all [12:09:21] <kbroulik> *pester
[12:09:27] <sebas> b/c Libreoffice is run by GNOME fansboys
[12:09:31] <kbroulik> it would be in their interest, too
[12:09:47] <Sho_> sebas: well they have a project to do a qt port and we have our \
icons upstream etc. [12:09:48] <notmart> well, limux...
[12:09:50] <kbroulik> (though, that seems to be a common thing. Chromium also has \
more Gnome-specific codepaths than KDE stuff) [12:09:55] <notmart> is, you know...
[12:09:57] <Sho_> there's a history of successful upstreaming kde stuff with LO, I \
think, so ... [12:09:57] --> oc2k1__ (~oc2k1@62-193-39-77.as16211.net) has joined \
#plasma [12:10:04] <kbroulik> notmart: not dead until End of 2020!
[12:10:12] <kbroulik> yeah we should look into LO KIO :)
[12:10:29] <Sho_> kbroulik: but otoh I remember talking to a Chrome dev who \
explicitly reached out to us to add KDE codepaths, etc. [12:10:29] <notmart> if there \
is still room for meneuvering there, sure [12:10:30] --> Son_Goku \
(~King_InuY@fedora/ngompa) has joined #plasma [12:10:41] <Sho_> let's not always \
blame the others - are we screwing up somehow, too? [12:10:45] <-- oc2k1_ \
(~oc2k1@62-193-39-77.as16211.net) has quit (Ping timeout: 276 seconds) [12:10:48] \
<romangg> kbroulik: What is dead may never die! XD [12:10:51] <Sho_> food for thought
[12:10:59] <kbroulik> it was just an unbiased observation
[12:11:46] <romangg> well, the question is why app devs need to write two different \
backends on Linux Desktop at all. [12:11:49] <kbroulik> I wanted to read into how to \
contribute to chromium, though, to improve the kde integration there. Often times \
they have a gnome-specific codepath when there's actually a FDO standard we both \
support [12:12:00] --> s8321414 (~s8321414@150.117.137.90) has joined #plasma
[12:12:19] <kbroulik> (like the "open containnig folder" stuff, they check for \
nautilus and otherwise just open parent folder, rather than using the FileManager1 \
dbus iface we both supprot, etc) [12:12:19] <kbroulik> but let's not digress too much
[12:12:33] <sebas> notmart's turn!
[12:12:35] <Sho_> I think KDE often does adopt a mindset of "we're fixing all of our \
issues inside of our bubble and if others want stuff it's up to them to adopt it", \
but we should probably be more aggressive about writing patches and submitting them \
[12:12:41] <Sho_> and with that, notmart take it away indeed [12:12:50] <sebas> FWIW, \
I agree with Sho [12:13:02] <notmart> yeah, i kind of agree with that
[12:13:03] <notmart> aanyways
[12:13:42] <notmart> * first rough working kernel driver for dell tablet mode (it \
turns out it's a standard-ish feature of intel acpi, so may work on more laptops than \
dells), in talk with some kernel people to make it into mainline [12:13:44] <notmart> \
* big kirigami branch about static plugins has been merged, deployment story perhaps \
a bit easier [12:13:45] <notmart> * fixes build issues on Android
[12:13:47] <notmart> * started patch to expose to kwin wether we are in tablet mode \
D9944 need help how to correctly wire things together [12:13:48] <sKreamer> Diff 9944 \
"[WIP]expose if the tablet mode switch is available" [Needs Review] \
https://phabricator.kde.org/D9944 [12:13:48] <notmart> * plasma: tablet mode \
experiments: show/hide panels and show/hide plasmoids [12:13:50] <notmart> * pushed \
slimmer scrollbars in breeze just on master, will be in 5.13 [12:14:17] <notmart> \
(so, pushing a kernel patch to go outside our bubble, yay \o/) [12:14:33] <sebas> \
cool stuff, and awesome you're getting into kernel development [12:14:40] <sebas> \
(You know you'll regret that) [12:14:43] <sebas> d_ed's turn
[12:15:11] <d_ed> I ported the minimize all applet to libtaskmanager to work on \
wayland [12:15:13] <notmart> luckily it's a driver of a laptop thing, so at least i \
know it won't compromise half of world's servers anytime soon :p [12:15:22] <d_ed> \
and get rid of an import whilst we're at it [12:15:40] <d_ed> will be on review at \
lunch [12:15:47] <notmart> nice :)
[12:15:53] <d_ed> that's pretty much it on plasma
[12:15:59] <sebas> d_ed deletes all my code
[12:16:01] <sebas> awesome
[12:16:23] <sebas> kbroulik!
[12:16:30] <Sho_> "the process of getting Plasma to LTS involved rewriting sebas' \
code" [12:16:34] <Sho_> :P <3
[12:16:38] <sebas> I love you, too!
[12:16:44] <kbroulik> * Fixed memory leak in notifications history using a Repeater \
and creating all delegates at once (now uses ListView, it's rather ugly but the least \
invasive (read: not a total rewrite) patch I could come up with that could still go \
into 5.12) [12:16:45] <Sho_> :)
[12:16:54] <kbroulik> * Tried to fix Kickoff user avatar not reliably updating, D9950 \
and D9951 but it still breaks in certain cases [12:16:55] <sKreamer> Diff 9950 \
"[KUser] Check whether .face.icon is actually readable before returning it" \
[Accepted] https://phabricator.kde.org/D9950 [12:16:55] <sKreamer> Diff 9951 \
"[KUserProxy] Adjust to accounts service" [Needs Review] \
https://phabricator.kde.org/D9951 [12:17:02] <kbroulik> * Appmenu applet now \
underlines mnemonics only when Alt pressed to be in line with Breeze style [12:17:07] \
<kbroulik> * Fixed deleting and creating files ("New File" menu) in Folder View \
taking forever (it didn't signal KDirNotify properly), D9912 still open. Haven't \
figured out a fix for DropJob yet [12:17:07] <sKreamer> Diff 9912 "[KNewFileMenu] \
KDirNotify::emitFilesAdded after storedPut" [Needs Review] \
https://phabricator.kde.org/D9912 [12:17:18] <kbroulik> * Looked into getting rid of \
those stupid "Loop device" we get in places for e.g. snappy mounts: D9895 [12:17:19] \
<sKreamer> Diff 9895 "[UDisks] Ignore non-user mounts" [Needs Review] \
https://phabricator.kde.org/D9895 [12:17:25] <kbroulik> * Since we're getting quite a \
few crashes in libsqlite in the Bookmarksrunner lately, which pokes Firefox' database \
file to extract bookmarks (speaking of sqlite, could that be related to kactivities \
also breaking? that qsqlite has a regression?), I was looking into providing the \
bookmarks runner through plasma-browser-integration. It can just query the browser \
instead. However, after I did all of that for D9984 I figured that it obviously won't \
work if the [12:17:25] <kbroulik>  browser isn't running. So I don't know how to \
procede. We could also have p-b-i extract all bookmarks, save them into a json file \
somewhere and the search there if not running but that wouldn't make it much better \
than what we currently have, except for that we don't need SQL. [12:17:25] <sKreamer> \
Diff 9984 "WIP: Add bookmarks runner" [Needs Review] \
https://phabricator.kde.org/D9984 [12:18:30] <notmart> kbroulik: maybe is thread \
safeness and could be made an out of process runner like baloo? [12:18:30] <kbroulik> \
next [12:18:54] <kbroulik> yeah, was thinking about that too, but then had the \
brilliant idea with p-b-i which turned out to be not as brilliant after all :) \
[12:19:02] <sebas> bshah is next up [12:19:11] <bshah> okay
[12:19:31] <bshah> - Mostly working on same stuff as last week
[12:19:33] <bshah> - Which includes testing the fixes for \
udev-android-networkmanager-simplelogin startup race [12:19:35] <bshah> - On popular \
request worked on Plasma Mobile ISO based on neon : \
http://images.plasma-mobile.org/iso/ [12:19:37] <bshah> - Need help with that as \
plasmashell doesn't start on first start [12:19:39] <bshah> - Will work on packaging \
newer libinput in neon [12:19:41] <bshah> - will work on packaging qt 5.10 on neon
[12:20:05] <-- ceed^ (~vivaldi@c-73-16-30-241.hsd1.ma.comcast.net) has quit (Ping \
timeout: 240 seconds) [12:20:43] <bshah> notmart: can you take a look at the \
plasmashell not starting on iso bug? it seems panels show up but not containment \
[12:20:52] <bshah> if I restart session everything is fine then [12:21:09] <notmart> \
bshah: on mobile? pine? normal neon? [12:21:24] <bshah> notmart: plasma mobile iso
[12:21:27] --> Jucato (~jucato@kde/developer/jucato) has joined #plasma
[12:21:33] <notmart> so the panels are there, not the desktop... pretty much the \
opposite of what used to happen [12:21:39] <bshah> :D
[12:21:56] <bshah> anyway.. breakout I suppose :)
[12:21:57] <-- Jucato (~jucato@kde/developer/jucato) has quit (Client Quit)
[12:21:58] <bshah> next
[12:22:05] <notmart> i'll give a try
[12:24:21] <romangg> shall I go?
[12:24:23] <bshah> *awkward silence*
[12:24:33] <bshah> romangg: sure :)
[12:24:46] <romangg> well, currently (re)writing my Xwayland patches as I said.
[12:24:58] <romangg> d_ed and I added two KWin ideas for GSoC
[12:25:05] <romangg> https://community.kde.org/GSoC/2018/Ideas#KWin
[12:25:20] <romangg> there are still not many Plasma ideas.
[12:25:37] <romangg> Could there be a plasma workspace idea?
[12:25:51] <romangg> next
[12:28:02] --> Jucato (~jucato@kde/developer/jucato) has joined #plasma
[12:28:06] <romangg> notmart: Do you have a repo with your kernel work?
[12:28:07] <ilya_b> Could I break in with Plasma Mobile UX?
[12:28:11] <romangg> ilya_b: sure
[12:28:37] <romangg> we are waiting on sebas currently anyway
[12:28:40] <ilya_b> As I asaid, I really, really hope to propose some PlaMo projects \
to GSoC students [12:28:52] <ilya_b> And KCMs seem to be a good idea
[12:28:57] <ilya_b> But we. need. mentorw
[12:29:02] <ilya_b> But we. need. mentors*
[12:29:14] <notmart> romangg: yes (it's cloned from an ubuntu repo tough as i wanted \
to have everything else built with the exact same config as i'm a n00b :p) [12:29:15] \
<sebas> sorry ... I got distracted [12:29:20] <romangg> ilya_b: you know about the \
KCM redesign effort like one moth ago, right? [12:29:43] <notmart> romangg: do you \
have hardware which the patch could work on? [12:29:45] <ilya_b> romangg: I know \
there is effort around, but don't know where we are atm [12:29:54] <sebas> * Testing \
kscreen osd branch by dvratil D9414 (X works nicely with a few patches on top, but \
[12:29:54] <sebas> * wayland hotplugging is broken on my system, so looking into that \
[12:29:54] <sebas> * More kscreen bug triaging done and to do [12:29:54] <sebas> * \
semi-related: Working on privacy goal stuff, collecting info, rallying people, PR \
stuff around it [12:29:54] <sebas> * want to get back onto the Plasma Mobile state of \
the art [12:29:54] <sKreamer> Diff 9414 "Implement OSD to select action when unknown \
monitors is connected" [Needs Review] https://phabricator.kde.org/D9414 [12:29:57] \
<sebas> there we go [12:30:43] <romangg> sebas: regarding hotplugging: you use atomic \
mode setting ? [12:30:52] <ilya_b> So I think USB/MTP, Locale & Language, \
Notifications mobile-friendly KCMs are still not there? [12:31:01] <romangg> maybe \
try it with and without. [12:31:01] <sebas> romangg: 4.14 kernel, that should be \
default with intel graphics, no? [12:31:02] --> gpunk (~gpunk@80.215.210.51) has \
joined #plasma [12:31:16] <gpunk> hi
[12:31:41] <gpunk> the official Instant Messaging application for kde ?
[12:32:10] <sebas> for KDE, you mean used by the KDE developers?
[12:32:13] <romangg> sebas: I thought so. but just take a look in the wayland log if \
it says somewhere "atomic" [12:32:18] <sebas> or do you mean used in Plasma, gpunk?
[12:32:23] <gpunk> yes
[12:32:35] <sebas> romangg: "wayland log" ? :D
[12:32:40] <kbroulik> fvogt: also QProcess start (non-static) is inew in Qt 5.10 :D
[12:32:44] <kbroulik> but good tip nonetheless
[12:32:48] <sebas> gpunk: "yes" is not an answer to an either/or question :D
[12:33:10] <gpunk> "do you mean used in Plasma, gpunk?" yes
[12:33:28] <sebas> There's no official IM app to be used in Plasma
[12:33:45] <gpunk> ok
[12:33:48] <sebas> some devs use telegram, some people use kopete, many people use \
IRC [12:34:04] <gpunk> ok thnx :)
[12:34:06] <sebas> IME it mostly depends on who you want to talk to
[12:34:33] <sebas> The Plasma meeting, however, is closed, thanks all, will sent \
notes and log to the list now



[prev in list] [next in list] [prev in thread] [next in thread]