=== mibofra is now known as Guest17057 | ||
=== marcusto_ is now known as marcustomlinson | ||
=== marcusto_ is now known as marcustomlinson | ||
duflu | greyback: Hey I spent a brief while relearning how QtMir does compositing today. And I suspect it's holding buffers longer than it should. Although I need to re-check Mir's logic before I can suggest something better. Are you sure you can't release your buffer pointer held in the texture earlier? | 08:51 |
---|---|---|
greyback | duflu: you're correct in your suspicion. We only release a buffer when we acquire a new one, which is not ideal. | 08:53 |
duflu | greyback: I think the phone will be noticeably smoother if we release earlier :) | 08:53 |
greyback | duflu: we could improve this by releasing the buffer if we're notified a new one has been created, and we've fiished rendering that frame | 08:54 |
greyback | duflu: I don't disagree | 08:54 |
duflu | greyback: I need to re-check Mir again. I know cachine just the GL texture is fine for software surfaces but hardware surface safety is dependent on the EGL extensions we use :P | 08:55 |
duflu | *caching | 08:55 |
greyback | duflu: I don't understand what you mean by caching here. | 08:56 |
greyback | caching the whole buffer while it's being composited, so buffer can be released? | 08:57 |
duflu | greyback: For traditional texture uploads at least, it gets copied to the GPU. So you can reclaim the buffer immediately | 08:57 |
duflu | Although I think our hardware surfaces don't get copied. So need more care | 08:58 |
duflu | Never mind. If I have a promising idea I'll propose it | 08:59 |
greyback | duflu: understood | 08:59 |
duflu | greyback: Are there any software buffers used in practice on a phone, or all hardware? | 09:05 |
greyback | duflu: sdl-based apps use software buffers I believe | 09:06 |
duflu | greyback: OK, yeah I know about those. Same on desktop | 09:06 |
greyback | yep | 09:06 |
duflu | Does image 263 work for anyone else or am I being over-ambitious? | 09:09 |
duflu | I guess that's why it's called "proposed" | 09:15 |
alan_g | alf_: I'm deciding what to do for the rest of the sprint. Is "Fix fd leaks in acceptance tests" something I could take on? | 09:22 |
alf_ | alan_g: yes, feel free to do so | 09:23 |
alf_ | anpok_: ^^ See Daniel's comment about image 263 | 10:08 |
alf_ | duflu: Did you find an image that boots? | 10:08 |
alan_g | duflu: lp:1477051 | 10:09 |
alf_ | alan_g: thanks | 10:10 |
anpok_ | alf_: I think I have an older image here too | 10:14 |
alf_ | anpok_: and it fails too? | 10:14 |
anpok_ | rc-proposed or devel-proposed | 10:15 |
anpok_ | hmm I flashed 280 it seems | 10:15 |
ogra_ | bug 1477051 FWIW | 10:15 |
ubot5 | bug 1477051 in live-build (Ubuntu) "Phones on devel-proposed do not boot - /bin/sh: 1: /bin/sh: initctl: not found" [Critical,Confirmed] https://launchpad.net/bugs/1477051 | 10:15 |
alf_ | anpok_: so that's not devel-proposed then | 10:16 |
alf_ | anpok_: ah, sorry, ubuntu-touch/devel-proposed/ubuntu has a 280 image for krillin | 10:18 |
alf_ | anpok_: which is the latest | 10:18 |
alf_ | anpok_: the latest for ubuntu-touch/devel-proposed/ubuntu mako is 263 | 10:19 |
alf_ | anpok_: I will try with 262 | 10:21 |
anpok_ | ok i am on krillin #277 and spinner spins now | 10:28 |
duflu | alf_: Yes, vivid works. But that doesn't help me right now | 10:30 |
* duflu wanders off | 10:31 | |
alf_ | anpok_: ubuntu-touch/devel-proposed/ubuntu mako 261 boots, will test silo with that | 10:35 |
ogra_ | only the latest images dont boot (see the bug i posted above) | 10:36 |
ogra_ | yesterdays builds should be fine | 10:36 |
alf_ | ogra_: ack | 10:36 |
=== Guest17057 is now known as mibofra | ||
=== dandrader is now known as dandrader|bbl | ||
camako | vogons, after latest wily upgrade, my wifi doesn't work on my laptop (doesn't wired connection). Can't join standup. | 15:07 |
camako | ... (doesn't have wired connection) | 15:07 |
=== chihchun is now known as chihchun_afk | ||
racarr | Morning! | 16:25 |
conyoo | um.. it's 19:38 o_O | 16:38 |
racarr | California \o/ | 16:39 |
conyoo | hehe :P | 16:52 |
=== alan_g is now known as alan_g|EOD | ||
=== dandrader|bbl is now known as dandrader | ||
=== dandrader is now known as dandrader|afk | ||
robert_ancell | kgunn, do you use a bluetooth mouse/keyboard in combination with a "slimport cable" (not sure of the actual term for it, i.e. the HDMI connector) | 21:03 |
kgunn | robert_ancell: yes | 21:03 |
kgunn | robert_ancell: yeah, it's "slimport" | 21:04 |
kgunn | soon it will be replaced by mhl | 21:04 |
kgunn | Mobile High-Definition Link | 21:04 |
robert_ancell | kgunn, do neither of those options allows audio/video AND USB? | 21:05 |
kgunn | robert_ancell: yeah, mhl is supposed to support audio i think...altho might depend on oem support/drivers | 21:06 |
robert_ancell | kgunn, yeah, but by switching to MHL/slimport mode does it no longer do USB on those pins? | 21:06 |
robert_ancell | So you could have a USB hub + HDMI connected | 21:06 |
robert_ancell | i.e. a complete docking station | 21:07 |
kgunn | robert_ancell: uh, that i don't know, altho at least with slimport there is a hub on a lot of cables.... | 21:07 |
kgunn | not sure if that's power only | 21:07 |
kgunn | never check | 21:08 |
kgunn | ed | 21:08 |
robert_ancell | It appears to be just for power (so you can charge the phone while using the slimport) | 21:08 |
robert_ancell | It appears an OTG cable will give you USB and a slimport gives you HDMI but you can't get both :( | 21:10 |
robert_ancell | I wonder if any phone supports using an OTG + a DisplayLink device | 21:11 |
anpok | robert_ancell: the mhl solutions should allow both | 21:12 |
anpok | robert_ancell: can get an android buffer displayed on a drm device...? | 21:14 |
robert_ancell | anpok, sounds difficult! I guess the Android driver spec doesn't require anything like this (yet) | 21:15 |
anpok | http://support.displaylink.com/knowledgebase/articles/557853-how-do-i-connect-my-displaylink-device-to-my-andro | 21:15 |
anpok | hmm | 21:15 |
anpok | robert_ancell: we could just copy the contents in usc? | 21:16 |
robert_ancell | anpok, yeah, I guess that's what the app is doing more or less | 21:17 |
=== dandrader|afk is now known as dandrader |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!