alan_g | duflu: any thoughts on the right way to handle the libmirclient deprecations for 16.04? Reworking all the downstreams to use the new APIs seems like risky behaviour. | 08:23 |
---|---|---|
duflu | alan_g: No immediate thoughts :P | 08:30 |
duflu | :/ I mean | 08:30 |
alan_g | :) | 08:34 |
xnox | hi, is there anybody who could advise me on qtmir and qtmir-gles | 13:05 |
xnox | are they going to be supported, developed and needed in artful and going forward? | 13:05 |
xnox | e.g. does kiosk use it? | 13:05 |
alan_g | xnox: the kiosk doesn't use it | 13:06 |
xnox | alan_g, thanks. what about maintainance and supporting qtmir in artful? no idea? | 13:07 |
alan_g | I can't give a definitive answer to that yet. But it isn't needed for IoT, so I don't forsee any active development | 13:10 |
alan_g | xnox: what usecase interest you? | 13:12 |
xnox | alan_g, i am interested to drop src:upstart, src:cgmanger, src:libnih from artful. And currently src:qtmir[-gles] depend on cgmanager libraries. I'm pondering if i need to touch/fix that | 13:15 |
xnox | or src:qtmir can simply be removed from artful | 13:15 |
alan_g | Saviq: can you give a better answer? ^^ | 13:17 |
Saviq | xnox, remove it, if we find that we need it, we'll revisit | 13:18 |
xnox | hm, that would mean removing src:unity8 and e.g. src:ubuntu-push | 13:21 |
xnox | but i thought ubuntu-push will be in snappy, no? | 13:21 |
xnox | or was ubuntu-push phone/touchy stuff? | 13:21 |
xnox | alan_g, Saviq - so ubuntu-app-launch has moved onto systemd, yet qtmir is using cgmanager/upstart cgroup paths to determine related app pids | 13:24 |
xnox | so my understanding is that fetchAssociatedPids is now broken. | 13:24 |
xnox | and effectively always returns "which is not app-specific" | 13:24 |
xnox | (e.g. does not find any related pids) | 13:24 |
Saviq | that might very well be, greyback ↑ ? | 13:25 |
alan_g | greyback: ^^ | 13:25 |
xnox | QSet<pid_t> DBusFocusInfo::fetchAssociatedPids(pid_t pid) always goes down the else path, as no things in zesty+ have 'upstart' in their cgroup | 13:26 |
greyback | xnox: I think you're right yes. However that code is only used to expose a pid list via dbus of focused apps, which is used only by autopilot I think. | 13:28 |
xnox | aha | 13:29 |
greyback | it is not important code, and we didn't notice this bug in the systemd transition | 13:29 |
greyback | because almost nobody uses that dbus api | 13:30 |
xnox | greyback, i think i will make the code to go down the else path, without changing any api/abi | 13:30 |
xnox | and drop build-dep and runtime dep on the cgmanager. | 13:30 |
xnox | meaning i'm not making things any worse. | 13:30 |
xnox | it is possible that autopilot uses ubuntu-app-launch apis directly for the systemd code path =/ | 13:30 |
greyback | xnox: I think that change won't hurt anyone that we care about | 13:31 |
xnox | running tests now, fingers crossed | 14:11 |
xnox | horay! | 14:12 |
xnox | onto gles variant | 14:13 |
=== JanC is now known as Guest16266 | ||
=== JanC_ is now known as JanC | ||
xnox | greyback, i've asked tedg if there is anyway to get all the pids from ubuntu-app-launch and here is his response | 14:37 |
xnox | <tedg> xnox: Yup, you can grab it from the instance object, pids() returns a vector. | 14:37 |
xnox | <tedg> xnox: You can also call ubuntu-app-launch-pids | 14:37 |
xnox | <xnox> nice. | 14:37 |
xnox | <tedg> xnox: But those will only really work on U8... | 14:37 |
xnox | so either autopilot should use that, or qtmir should re-expose that over db | 14:38 |
greyback | xnox: right. UAL does make that possible, we just kept neglecting that particular bit of code | 14:41 |
xnox | ah, ok. | 14:42 |
xnox | i shall open a bug report. | 14:42 |
xnox | at least | 14:42 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!