| robert_ancell | Has anything changed with how to run Mir demo servers? I'm running sudo mir_demo_server ; chown me.me /run/usr/1000/mir_socket ; mir_demo_client_basic -m /run/user/1000/mir_socket and getting "Failed to connect to server `/run/user/1000/mir_socket': No such file or directory" | 01:56 |
|---|---|---|
| robert_ancell | /run/usr/1000/mir_socket definitely exists | 01:56 |
| duflu | robert_ancell: Not that I'm aware of. Everything should work as before.. | 01:57 |
| robert_ancell | duflu, any debugging tips? | 01:57 |
| robert_ancell | The client/server seem otherwise happy | 01:58 |
| robert_ancell | aha, strace suggest the failure is actually to open /usr/lib/x86_64-linux-gnu/mir/client-platform/ | 02:06 |
| duflu | robert_ancell: Try installing mir-graphics-drivers-desktop | 02:09 |
| duflu | I think there's a bug for that but it's also arguably a feature | 02:10 |
| robert_ancell | duflu, yeah, I'm looking at the Mir code so see how the error text could be clearer | 02:10 |
| duflu | I agree, our lack of useful error messages is an issue. There are one or two bugs for that. | 02:10 |
| robert_ancell | The issue is there's more than one thing that might not exist, so the error text needs to be clear | 02:10 |
| robert_ancell | ugh, I give up trawling the frontend code. Bug it is. | 02:11 |
| duflu | robert_ancell: A new bug is fine. Although I remember it was logged recently somewhere and ended up being won't fix/can't fix because you can't automatically determine the best driver package to have installed. We do however have: https://bugs.launchpad.net/mir/+bug/1381398 | 02:13 |
| ubot5 | Launchpad bug 1381398 in Mir "Mir error messages from exceptions are not user-friendly" [Medium,Triaged] | 02:13 |
| robert_ancell | yeah, they're the server side messages. | 02:14 |
| robert_ancell | The client side should be easier to fix | 02:14 |
| robert_ancell | bug 1473268 | 02:14 |
| ubot5 | bug 1473268 in Mir "Connection error when no client-platform installed is confusing" [Undecided,New] https://launchpad.net/bugs/1473268 | 02:14 |
| duflu | robert_ancell: Wait, what?! If your server is already running you must have graphics drivers installed... | 02:23 |
| duflu | You must have installed a working mir-platform-* to get the server driver but be missing the client driver package | 02:23 |
| robert_ancell | duflu, I had mir-platform-graphics-mesa2 installed but not mir-client-platform-mesa2 | 02:24 |
| duflu | robert_ancell: Yeah OK, no problem. | 02:24 |
| duflu | robert_ancell: At last discussion this was labelled "not an issue" as our desktop-next metapackage has the required deps | 02:24 |
| duflu | But we can do better | 02:24 |
| robert_ancell | It would be fine if the error message actually indicated the mismatch | 02:25 |
| duflu | robert_ancell: Totally agree. Should be a simple enhancement | 02:25 |
| duflu | robert_ancell: Also I have had thoughts that "driver" should be a single package, not separate client/server. | 02:28 |
| duflu | Even a single binary | 02:28 |
| robert_ancell | Yeah, it would be nicer to have one package though I guess there are potential small gains in splitting? | 02:28 |
| duflu | robert_ancell: We can keep two binaries and still have a single package. But also code reuse (and hence total system memory used) might be reduced by a single binary | 02:29 |
| duflu | -reuse +duplication | 02:30 |
| duflu | robert_ancell: Incidentally, this week I was focussing on "bugs that users can actually see" | 02:35 |
| === duflu_ is now known as duflu | ||
| robert_ancell | duflu, do you know of an issue running mir_demo_server on wily and it crashing when a client connects with "*** stack smashing detected ***"? | 03:34 |
| duflu | robert_ancell: I am aware of only one stack smashable issue. That is apparently fixed in 0.13.3 and later - https://bugs.launchpad.net/mir/+bug/1465883 | 03:36 |
| ubot5 | Launchpad bug 1465883 in Mir "libmirprotobuf's ABI can be broken when modifying protobuf message definitions" [High,Fix committed] | 03:36 |
| duflu | robert_ancell: Obviously first check if it's the client rather than Mir doing the smashing | 03:37 |
| robert_ancell | the message is on the servers stderr | 03:37 |
| robert_ancell | ah, I have a 0.14.0 version of libmirprotobuf it seems | 03:37 |
| duflu | robert_ancell: From memory you need the fix in libmirclient and libmirserver packages (0.13.3 and later) | 03:38 |
| duflu | Not libmirprotobuf, despite that being closely related | 03:38 |
| robert_ancell | duflu, I downgraded and that fixed it, thanks! | 03:39 |
| robert_ancell | btw fingerpaint is fun with a 10 fingers :) | 03:39 |
| duflu | robert_ancell: That's a worry if 0.14 caused issues | 03:39 |
| duflu | Yay, fingerpaint | 03:39 |
| robert_ancell | Should libmirclient8 0.13.3+15.10.20150617-0ubuntu1 with with libmirprotobuf0 0.14.0+15.10.20150701-0ubuntu1? | 03:40 |
| duflu | robert_ancell: I think so. But there might be bugs we're not aware of | 03:41 |
| duflu | If you can install them simultaneously then that's theoretically a supported combination. But not a guarantee that they will talk to each other (you might have multiple versions of a package installed too) | 03:42 |
| robert_ancell | RAOF, with DRI3 do you create a Pixmap then use DRI3BufferFromPixmap or do you create the buffer using some direct call to the driver and then use DRI3PixmapFromBuffer? | 04:18 |
| === chihchun_afk is now known as chihchun | ||
| duflu | robert_ancell: He's on holiday for a week now | 04:27 |
| robert_ancell | ah | 04:27 |
| duflu | Although reading email on the go | 04:27 |
| duflu | sometimes | 04:27 |
| === sturmflut_ is now known as sturmflut | ||
| UukGoblin | hi :-) | 10:02 |
| UukGoblin | is there a way to run applications written for X11 under Mir? | 10:02 |
| anpok_ | yes | 10:21 |
| anpok_ | in wily the new package is called xmir .. on vivid it was xserver-xorg-xmir | 10:21 |
| anpok_ | you bascially launch xmir with -mirSocket /path-to-your-mir-socket/ (i.e. /run/3201/mir_socket) and assign a display id and launch the x-application with that DISPLAY | 10:22 |
| anpok_ | UukGoblin: depending on the mir shell in use you might need additional steps.. I believe for unity8 you need to append a --desktop-file-hint <path-to-a-desktop-file> if in doubt ask in #ubuntu-unity | 10:23 |
| UukGoblin | cool, thanks :-) | 11:08 |
| === chihchun is now known as chihchun_afk | ||
| kgunn | alf_: i'm assumign the system setting for "lock screen timout" does not alter or change the dimming timer? | 16:22 |
| kgunn | (it's currently conflating lock with screen blank i think) | 16:22 |
| kgunn | so for 1 minute.... 50 bright, 10 sec dim | 16:23 |
| alf_ | kgunn: the dimming timout is set to power-off timout - 10s | 16:23 |
| kgunn | ah...so fixed | 16:23 |
| alf_ | kgunn: (by other components, we are not forcing it, our API is set_timeouts(dim, poweroff) | 16:24 |
| kgunn | alf_: for your ques 6 i think scenario 1 makes the most sense to me at least | 16:26 |
| alf_ | kgunn: note: there was mistake in the description of (6) (fixed now), the initial trigger is a power key event | 16:34 |
| kgunn | sure | 16:34 |
| === dandrader is now known as dandrader|afk | ||
| === mibofra is now known as Guest23067 | ||
| === Guest23067 is now known as mibofra | ||
| === dandrader|afk is now known as dandrader | ||
| * kdub figures out how to tease apart BufferQueue | 19:54 | |
| camako | \o/ | 20:05 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!