/srv/irclogs.ubuntu.com/2014/01/29/#ubuntu-mir.txt

=== WebbyIT is now known as rpadovani
=== rpadovani is now known as WebbyIT
=== chihchun_afk is now known as chihchun
RAOFduflu: Really? Generalise the X11 support so we can support all the other legacy display servers that litter the Linux ecosystem?04:20
dufluRAOF: I meant generalize for spawning other processes Mir "might need"04:35
RAOFWhich would be the process-spawner MP that it depends upon?04:35
dufluBut that's less important than ensuring we don't have a hard build-dep. Perhaps optional04:35
RAOFIt's just for the (disabled) acceptance test (at this point).04:36
dufluRAOF: Just worth thinking about how to avoiding mentioning "X" so much. There's always another solution no one has thought of04:37
dufluRAOF: So long as we don't try to compile code (or require it) that includes X11/*04:37
dufluMaybe optional04:37
dufluRAOF: I don't yet understand why Mir has to own the Spawner actually. I need educating04:39
RAOFIt doesn't *have* to, but it's convenient if it does.04:39
RAOFParticularly - shells which want to provide seamless X11 support will need to do a bunch of stuff between "I want to start an app that uses X11" and "I've got a Surface that comes from this X11 app".04:41
dufluMeanwhile, I finally hit the brick wall of bypass requires 4 buffers to work properly :P04:48
RAOFHm.06:32
* RAOF dislikes process::ProcessFactory06:32
dufluIn a way "ps" is more meaningful. Because "ps" is specifically always an OS process and not a generic term for a methodology :)06:37
* duflu ducks06:37
RAOFHow would you feel about process::Factory? :)06:41
dufluRAOF, mlankhorst: Anyone know if hardware cursor movement implicitly triggers another wait for vblank? I mean... how/when is it rendered if we're not flipping pages?08:22
mlankhorstduflu: no08:59
mlankhorstit's implemented as an overlay09:00
duflumlankhorst: I mean how does it fit into the physical display timing...?09:00
dufluHardware cursors seem to update much faster than anything else09:00
mlankhorstno they are updated on next frame, always tear free09:01
mlankhorstat least when you update the position09:02
duflumlankhorst: OK. Sounds like the "next frame" to DRM is sooner than the next frame Mir is able to flip09:04
dufluWeirtd09:04
dufluWeird09:04
mlankhorstduflu: it's an overlay, so it might be possible09:04
* duflu wonders if there's a wait for vblank hiding inside SwapBuffers09:04
dufluThat would explain it09:05
mlankhorstpossibly09:05
dufluOtherwise there's no logical reason (I can think of) why the hardware cursor would ever get ahead of compositing09:06
mlankhorstwell it can be updated while the screen is already drawing, cursor is special :P09:06
* mlankhorst guesses09:06
dufluThat's another explanation09:07
mlankhorstcursor is special, it's not part of a page flip09:07
dufluYeah there's obviously more display logic beyond my programmer's view09:08
mlankhorstah cursor has its own channel09:13
mlankhorstfigures :P]09:13
mlankhorstso page flipping has nothing to do with a cursor :)09:23
duflumlankhorst: What's a channel? :)09:31
mlankhorstduflu: sorry, it's used for command submission, probably some kind of fifo09:44
mlankhorstnvidia probably calls it different09:45
* duflu throws gmock out the window and gains karma11:35
anpok_not surface?11:36
* ogra_ hopes you checked if someone wlaked outside before doing that ... 11:37
ogra_*walked11:37
=== alan_g is now known as alan_g|afk
=== alan_g|afk is now known as alan_g
ricmmSaviq: ping12:35
Saviqricmm, pong12:35
ricmmSaviq: do you know if the shell is the one who currently does restore of old brightness12:36
ricmmif the screen has gone dim?12:36
Saviqricmm, no, powerd is doing all atm12:36
ricmmor is powerd doing the whole bit12:36
ricmmok12:36
Saviqricmm, while I have you here, did you have a think about making the N7 landscape?12:40
alf_alan_g: anpok_: Please take a look at lp:~afrantzis/mir/using-stub-client-platform when you get some time (mir-screencast-basic-client-api depends on this functionality landing).12:52
alan_galf_: ack12:52
=== alan_g is now known as alan_g|lunch
ricmmSaviq: I think it should all happen in the shell13:26
Saviqricmm, not for MWC it can't13:26
=== alan_g|lunch is now known as alan_g
fgintherkgunn, you pinged me yesterday I think?14:23
kgunnfginther: yep..no worries...i got helped14:24
fgintherkgunn, good to hear14:24
=== dandrader is now known as dandrader|afk
=== dandrader|afk is now known as dandrader
ricmmkdub_: :)15:20
=== alan_g is now known as alan_g|tea
=== alan_g|tea is now known as alan_g
dandradermaybe restarting xchat does the trick15:42
dandraderoops, wrong channel15:42
=== mhall119_ is now known as mhall119
=== mterry_ is now known as mterry
kdub_ricmm, Saviq for rotation, the mir compositor will rotate all the surfaces to be composited if the display is reconfigured with a different orientation16:32
Saviqkdub_, how about input?16:32
kdub_i'm unsure about input16:32
Saviqkdub_, yeah, if it doesn't transform input then we need to do it higher anyway16:33
=== chihchun is now known as chihchun_afk
=== dandrader is now known as dandrader|lunch
=== dandrader|lunch is now known as dandrader
bregmais there an easy way to take a screenshot when Mir is running as the system compositor (er, without using a camera on a tripod pointed at the screen)?19:37
kgunnbregma: i'll fwd you a mail alf sent racarr20:39
bregmakgunn, thanks20:40
kgunnbregma: sure, if you have any issues, feel free to hit up alf in the morning20:41
kgunnkdub_: (or anyone)....so if you update manta to 4.4.2 are you stuck using rsalveti's preview img's ?20:42
kgunne.g. i have to go back to 4.2.2 to use the phablet distro images ?20:43
* kgunn wants to avoid bricking20:43
=== dandrader_ is now known as dandrader|afk
=== seb128_ is now known as seb128
=== dandrader|afk is now known as dandrader
kdub_kgunn, you have to flash the radio21:46
kdub_but you can go back and forth between them21:46
kgunnkdub_: yeah...i see now...no biggie21:58

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!