Wizzup | https://wizzup.org/bluetooth-car.mp4 think this is worth putting on yt/twitter? | 00:47 |
---|---|---|
humpelstilzchen[ | hide the phone number if you do.. | 06:14 |
buZz | yes Wizzup ! edit it a bit indeed | 08:17 |
buZz | i can edit it for you | 08:17 |
buZz | say, long ago i heard its easy to make a maemo gui app with python | 08:17 |
buZz | which python library should i use for that, and are there any examples? | 08:18 |
buZz | also lol, not sure whats happening, but pressing ctrl+shift+N gets hildon in | 08:26 |
buZz | also lol, not sure whats happening, but pressing ctrl+shift+N gets hildon in a weird state | 08:26 |
buZz | depends on what program has focus how weird, with chromium it seems to hang until you press ctrl+N to make a new window | 08:27 |
buZz | in xterm it opens the statusmenu at top, but, -behind- xterm | 08:27 |
buZz | hehe | 08:27 |
Wizzup | humpelstilzchen[: it's my own number, mostly unused :) | 10:14 |
Wizzup | humpelstilzchen[: but yeah I guess | 10:14 |
Wizzup | buZz: iirc ctrn/shift/n is turn off compositing or so | 10:14 |
Wizzup | buZz: do you want the raw/bigger video, or is the 720p enough? | 10:15 |
buZz | 720p is fine but raw is better | 10:33 |
buZz | throw it on a http, i'll sneakernet it later | 10:34 |
buZz | Wizzup: does ctrl shift N do something like that for you? the 'contextmenu behind xterm' in xterm is easiest, it'll close after you tap on it | 10:35 |
Wizzup | can you make an issue describing how to reproduce? | 10:36 |
buZz | yeah willdo | 10:37 |
uvos | this isent an issue | 10:39 |
uvos | or it kinda is but its something thats deeply rooted in how h-d works (and is broken) | 10:39 |
uvos | ctrl-shift-n disables composing | 10:39 |
uvos | while composing h-d renders the frame, without composing xorg renders the frame | 10:40 |
uvos | h-d makes no serious effort to sync up the real x11 windows with the clutter geometry it renders | 10:40 |
uvos | thus when xorg takes over it starts rendering based on the x11 windows, which are in the wrong place or dont even contain anything | 10:41 |
uvos | everything breaks | 10:41 |
uvos | it only works fully when some window is fullscreen | 10:42 |
uvos | thus hideing the mess that is h-d behind a spec abideing x client | 10:42 |
uvos | fixing this is euqal in complexity, and indeed is the same thing, as makeing h-d work without compositing | 10:46 |
buZz | hmm well, not sure why we would want that on a key | 10:55 |
buZz | if it breaks so much | 10:56 |
uvos | this is something they added to community ssu | 10:57 |
uvos | since h-d compositing is very very slow using it often boosts the speed of a 3d client by up to 2x | 10:58 |
uvos | this is presumably why it exists even though it is very broken | 10:58 |
buZz | ah hmm, i'll try it in cloudgps i guess? | 11:08 |
Wizzup | it's more about games | 11:09 |
buZz | sadly thats capturing ctrlN and shiftN seems impossible to trigger ctrlshiftN | 11:10 |
Wizzup | heh | 11:11 |
buZz | isnt cloudgps doing opengles? its so fast and does some '3d' | 11:12 |
uvos | cloudgps should be doing the same thing itself | 11:12 |
uvos | since its fullscren | 11:12 |
uvos | it needs to be a non fullscreen 3d app | 11:12 |
uvos | 2d rendering is also somewhat faster | 11:12 |
uvos | i gues a easy "fix" for ctrl-shift-n would be to make it also have h-d force whatever window is on top be fulllscreen, hideing the statusbar mess | 11:14 |
buZz | hmm i dont know a nonfullscreen 3d app offhand | 11:24 |
buZz | maybe es2gears or something? | 11:24 |
uvos | any qt app that scrolls should do | 11:53 |
uvos | maybe qsigstoped | 11:53 |
Wizzup | qtwebbrowser | 11:53 |
Wizzup | conversations | 11:53 |
Wizzup | https://baltimore.craigslist.org/ele/7550887106.html well at least they're on craigslist | 12:45 |
norayr | heh. pickup. | 15:01 |
Wizzup | so I'm still not seeing the reset/freeze with pm disabled on the d4 with psu | 15:46 |
Wizzup | I bet the moment I detach kernel console (and let it idle) it'll start happening again | 15:47 |
Wizzup | uvos__: I think if we just load the echo cancel module, it will auto unloda itself | 16:22 |
Wizzup | or rather become not active.. | 16:22 |
Wizzup | that will of course only apply if the right pulse props are requested | 16:23 |
Wizzup | but I think twinkle does this | 16:23 |
uvos | uvos__: why are you here twiche | 16:23 |
uvos | *twice | 16:23 |
uvos | Wizzup: erm ok | 16:23 |
uvos | Wizzup: needs benchmarking | 16:23 |
uvos | norayr: volentiering | 16:24 |
uvos | Wizzup: yeah problem with caigslist is the regional charecter | 16:24 |
Wizzup | uvos: what I mean is that it doesn't use cpu like I observed earlier | 16:24 |
uvos | Wizzup: ok | 16:25 |
Wizzup | the problem I think was that I made the default sink and source echo cancelled | 16:25 |
uvos | ok | 16:25 |
uvos | but then how will it be used | 16:25 |
Wizzup | applications request it through env variable | 16:25 |
uvos | the applications will need to explictly use its sources/sinks? | 16:25 |
Wizzup | no, just through env | 16:25 |
Wizzup | module-filter-heuristics and module-filter-apply should do the right things | 16:26 |
uvos | sure you can ovr the default sink for some application via env | 16:26 |
uvos | but then wont pulse use inordinate amounts of power | 16:26 |
uvos | just bacuse some application is running? | 16:26 |
uvos | regardless of if its really playing any sounds | 16:26 |
uvos | the problem i see here is that pa uses power for echo filtering if echo filtering is in use for any active context (or so it would seam) | 16:27 |
uvos | resitricting it to certain applications would then just move the problem to those applications | 16:28 |
uvos | unless im missing something | 16:28 |
Wizzup | so twinkle running doesn't start up the pulse load | 16:28 |
Wizzup | it's like 20-25% cpu when it is doing echo cnacel | 16:28 |
Wizzup | only when twinkle makes a call it starts up | 16:28 |
Wizzup | and when the call stops, it calms down after a few seconds | 16:28 |
uvos | its allso possible that cpu load here comes from not becasue pa is missbehaving | 16:29 |
uvos | but because a pa client is missbehaveing | 16:29 |
uvos | and activateing echo canceling is just makeing it more apperant | 16:29 |
uvos | like maybe mis keeps a stream active or something | 16:29 |
Wizzup | currently it seems more likely I just did it wrong | 16:29 |
Wizzup | it's possible mis does this I suppose | 16:29 |
uvos | mis is just an example | 16:29 |
uvos | could also be sphone for instance | 16:30 |
uvos | or hildon home | 16:30 |
uvos | any pa client thats allways running | 16:30 |
Wizzup | in any case, none of those clients should use the echo cancelled sink/source | 16:30 |
uvos | the allarm clock | 16:30 |
uvos | .... | 16:30 |
Wizzup | and I did make that effectively happen | 16:30 |
Wizzup | ah, I think I see what you're saying | 16:30 |
Wizzup | it's possible that what I did before technically was sound, but something else misbehaved | 16:31 |
Wizzup | in any case, I actually like the idea of pulse doing echo cancel based on requested application props, as this clearly is the 'upstream way' | 16:31 |
uvos | sure theres nothing wrong with that | 16:31 |
uvos | im just suspecting something is misbehaveing regardless | 16:31 |
uvos | (and wasteing some power, maybe the extra power leste uses vs plain debain) | 16:32 |
Wizzup | right, whatever keeps audio active | 16:36 |
Wizzup | well this should make itself evident quickly from basic searching | 16:37 |
uvos | linux 6.1 works fine now btw | 16:38 |
uvos | been using it today after fixing some remaining issues with some of our patches breaking | 16:38 |
uvos | ill push later today | 16:38 |
Wizzup | great, please do :) | 16:40 |
Wizzup | uvos: damnit looks like it still does something dumb with ppoll | 16:49 |
Wizzup | at least, I think so.. | 16:51 |
Wizzup | yeah it causes pa to do some polling | 16:52 |
Wizzup | argh. | 16:53 |
uvos | ok | 18:12 |
uvos | so its pa being dumb | 18:12 |
Wizzup | mhm | 18:36 |
Wizzup | I've been trying to update some of the milestones and labels: https://github.com/maemo-leste/bugtracker/milestones?direction=asc&sort=due_date&state=open | 21:03 |
Wizzup | Feedback very welcome, I've also deleted a few milestones that were kind of open ended and more like a label/tag | 21:04 |
uvos | i know, i need to disable gh sending emails, its spamms me inessently xD | 21:04 |
Wizzup | I'm hoping to make it so that it's easier for new people to see what's up, what we're working on, how they can join and help, but also for ourselves | 21:04 |
Wizzup | uvos: yeah, sorry about that | 21:04 |
Wizzup | I'm kind of glad lel doesn't work atm | 21:04 |
uvos | Wizzup: no worrys :) | 21:05 |
uvos | Wizzup: i gues we should go through the old bugs and sort them a bit | 21:05 |
uvos | i can do some of that when theres some downtime | 21:05 |
Wizzup | yeah, I've tried to streamline the labels a bit but I keep running into a wall when trying to decide what to do with all the current bugs | 21:05 |
uvos | imo | 21:06 |
Wizzup | most importantly it should be clear what the relevant bugs are, but we have many old feature requests | 21:06 |
uvos | we should delete old feature requests more agressively | 21:06 |
Wizzup | Hmmm | 21:06 |
uvos | imo for manny it dosent make mutch sense to have them as bugs | 21:06 |
uvos | since its pretty clear certin stuff dosent work/isent ready | 21:06 |
Wizzup | right, they can be feature requests in the bug tracker | 21:07 |
uvos | but maybe thats just me | 21:07 |
uvos | sure yeah that would be good | 21:07 |
uvos | but wadeing though lots of feature requests to find real bugs erm bugs me | 21:07 |
Wizzup | I've created a feature request label | 21:07 |
Wizzup | so we can just exclude them that way once we tag them | 21:07 |
uvos | thats a start | 21:07 |
uvos | can we make gh hide those by default? | 21:08 |
Wizzup | not sure, I assume so | 21:08 |
Wizzup | we also need to update our wiki tablets with dev tasks, maybe just deprecate them | 21:08 |
uvos | yeah | 21:08 |
uvos | the wiki is very out of date in generall | 21:09 |
uvos | but keeping it updated is a lot of work of | 21:09 |
uvos | c | 21:09 |
Wizzup | I've been trying to find time in my head to figure out how to reorganise it | 21:09 |
uvos | im bad at that so.. | 21:09 |
Wizzup | me too :) | 21:09 |
uvos | not much i can do there | 21:09 |
Wizzup | I think everyone is | 21:09 |
uvos | i mostly like how pmos is organized | 21:10 |
uvos | maybe we can copy some of that | 21:10 |
uvos | also how https://wiki.postmarketos.org/wiki/Devices pulls the info from the device pages | 21:11 |
uvos | is nice | 21:11 |
uvos | maybe we can copy that too | 21:11 |
uvos | but idk mediawiki stuff | 21:11 |
Wizzup | I used some of their templates in the past | 21:12 |
uvos | kernel is on its way | 21:17 |
Wizzup | uvos: we can also add a 'bug' label | 21:18 |
Wizzup | if that's easier | 21:18 |
uvos | nah id rather have bug be the default really | 21:18 |
Wizzup | ok | 21:19 |
Wizzup | I'll go through some items now | 21:19 |
Wizzup | I'm going to unassign parazyd from all the issues so that others can claim | 21:31 |
uvos | at some point i think it also makes sense to invalidate permissions too | 21:33 |
uvos | we can allways give them back if he resurfaces | 21:33 |
Wizzup | right, that will be a bit more involved, but agreed | 21:35 |
Wizzup | fyi https://github.com/maemo-leste/bugtracker/issues?q=is%3Aopen+-label%3A%22Feature+Request%22 issues except the ones that are marked as feature requests | 21:37 |
Wizzup | uvos: does the new kernel contain fmg's patches for usb/power ? | 21:43 |
uvos | Wizzup: partally, only the ones that where also in the previous kernel | 21:54 |
uvos | Wizzup: idk which new ones are ready/ he wants included so he needs to comment for those | 21:54 |
Wizzup | ah, ok | 21:57 |
Wizzup | so not the ones that for example keep usbnet alive on full battery | 21:57 |
Wizzup | :D | 21:57 |
uvos | no freemangordon care to comment what you want included? | 21:58 |
freemangordon | uvos: just went home after 850km drive, will ping you tomorrow | 22:34 |
uvos | freemangordon: no rush | 22:56 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!