tjaalton | bryce: san jose has caltrain, not BART :) | 14:15 |
---|---|---|
tjaalton | public transport seems pretty cheap there, but caltrain is surprisingly slow | 14:50 |
tjaalton | ~2h from the hotel to the airport, but only $7.25 | 14:51 |
mvo | tjaalton: so you are comming? | 16:16 |
mvo | hm, so I read reports that compiz does hang on a i945 for intrepid | 16:17 |
mvo | anyone here with more information? | 16:17 |
tjaalton | mvo: yes | 16:18 |
tjaalton | no idea about the i945 issue | 16:26 |
mvo | tjaalton: is there someone familiar with intel I could talk to? do we have a contact at intel? I mean, blacklisting i830,845,945 does not leave that many cards that we don't blacklist | 16:27 |
mvo | bryce: ^--- ? | 16:29 |
tjaalton | mvo: bryce knows better, but I think he's been talking to jbarnes at least | 16:29 |
pcjc2 | hi guys | 17:42 |
pcjc2 | Anyone about who knows about the evdev vs. joysticks problem? | 17:42 |
tjaalton | o/ | 17:43 |
pcjc2 | https://help.ubuntu.com/community/Joystick_lshal_outputs suggests it is fixed with xserver-xorg-input-evdev 1:2.0.99+git20080912-0ubuntu6 | 17:43 |
pcjc2 | Not fixed for me, and 3Dconnextion space-navigator device | 17:43 |
pcjc2 | according to changelog for the evdev driver, it was only an Amd64 big which was fixed | 17:44 |
tjaalton | yes | 17:45 |
tjaalton | evdev should refuse to use them | 17:45 |
tjaalton | so it's still picked for you? | 17:45 |
tjaalton | pcjc2: ^^ | 17:51 |
bryce | heya | 17:58 |
bryce | (on holiday today) | 17:58 |
tjaalton | howdy | 17:59 |
bryce | tjaalton: ah, that's too bad, I thought bart went down that far | 17:59 |
tjaalton | not yet anyway, depends on how the ballot went :) | 17:59 |
bryce | too bad mvo's not online | 18:00 |
tjaalton | they are planning to extend it 16mi to san jose (the eastern route) | 18:00 |
bryce | tjaalton: if you can link up with enough other attendees, you can probably get a shuttle or taxi at a decent rate (and that will get you there faster) | 18:05 |
pcjc2 | hi, sorry, got distracted in the lab | 18:05 |
pcjc2 | tjaalton: Yes, lshal shows evdev being picked for the space navigator | 18:05 |
pcjc2 | want the output posting somewhere... pastebin? | 18:05 |
jcristau | lshal picking evdev is ok, i think | 18:06 |
tjaalton | pcjc2: logfile should show if it's actually being used | 18:06 |
pcjc2 | logfile does indeed show it being used | 18:06 |
pcjc2 | (II) 3Dconnexion SpaceNavigator: Configuring as mouse | 18:07 |
pcjc2 | (II) XINPUT: Adding extended input device "3Dconnexion SpaceNavigator" (type: MO | 18:07 |
pcjc2 | USE) | 18:07 |
tjaalton | bryce: well I think that I'll just go straight to downtown for a couple of hours before getting to the airport, that should minimize waiting times on stations :) | 18:07 |
bryce | ah, good idea | 18:09 |
pcjc2 | Seems that evdev's "is it a mouse" heuristic is being broken by the SpaceNavigator reporting as having mouse buttons | 18:11 |
pcjc2 | (as well as axis) | 18:11 |
tjaalton | well, maybe the buttons are within the permitted zone | 18:11 |
tjaalton | duh, I've forgotten the command to show the keyid's of an input device? | 18:14 |
pcjc2 | I have a driver I wrote / copy-pasted for it | 18:14 |
pcjc2 | and it is finding those buttons as BTN_0 from the kernel's /dev/input/ | 18:14 |
tjaalton | s/ten// | 18:15 |
pcjc2 | ten? | 18:15 |
tjaalton | ok, s/forgotten/forgot/ :) | 18:15 |
tjaalton | I'm tired or somthing.. | 18:16 |
pcjc2 | ah... was looking for a number - brain didn't register the word forgotten | 18:16 |
tjaalton | heh | 18:16 |
pcjc2 | ok, so its coming under the BTN_MISC rangeC | 18:16 |
pcjc2 | s/C// | 18:17 |
pcjc2 | But are there any devices which _are_ mice, which have no buttons under the BTN_MOUSE range? | 18:17 |
pcjc2 | If not, that suggests a patch to the evdev driver to only register something as a mouse if it has axes, and (some) buttons which are > BTN_MOUSE | 18:18 |
tjaalton | hm, dunno | 18:18 |
pcjc2 | s/>/>=/ | 18:18 |
tjaalton | you should probably ask whot on #xorg-devel, although it's rather early down under | 18:18 |
pcjc2 | My Hardy workaround, was to ensure I never installed the evdev driver, but going forward it would be nice to fix this properly | 18:18 |
pcjc2 | xorg developers are in Oz? | 18:19 |
jcristau | x input developers are | 18:19 |
pcjc2 | ok | 18:19 |
tjaalton | yes, whot and daniels both are :) | 18:19 |
pcjc2 | I'll knock up a patch which seems "right" here, and post that and an explanation on xorg-devel | 18:19 |
pcjc2 | (email) | 18:19 |
jcristau | but, you'll probably break something else if you change that check.. | 18:20 |
pcjc2 | No doubt... but as a local workaround, and a discussion point, it makes a good start | 18:20 |
jcristau | yup | 18:20 |
pcjc2 | It would be nice to have this device handled properly by XINPUT generally | 18:20 |
pcjc2 | I ended up writing a driver for the /dev/input/... device directly, rather than using XINPUT, due to various confusion about the way it represents its axis maxing it a bit screwey to use with xinput-joystic | 18:21 |
pcjc2 | k | 18:21 |
pcjc2 | its reporting relative axis, which IIRC, the joystic driver integrates | 18:22 |
pcjc2 | where you really just want the raw-numbers (perhaps scaled) back for a 6DOF controller navigating a 3D application | 18:22 |
pcjc2 | it makes no sense that the XINPUT system tries to return coordinates in X device space | 18:23 |
jcristau | doesn't XI give you the raw coordinates? | 18:24 |
pcjc2 | it has been a while since i played with the device.. | 18:27 |
pcjc2 | there are about "n" different ways you can interface withit | 18:28 |
pcjc2 | the proprioraty drivers open the USB device directly. You can use linux's input device support, X11... | 18:28 |
pcjc2 | its not picking up the right number of axis at the moment - I'd guess because evdev has grabbed it, and thinks it is a mouse | 18:29 |
tjaalton | pcjc2: have you filed a bug on launchpad about this? | 18:32 |
pcjc2 | not yet, just figured I'd come and ask if this was supposed to be fixed along with the joystick issues which came up relating to evdev | 18:39 |
pcjc2 | ah crap... | 18:39 |
pcjc2 | Xinput is now a microsoft directX "name" as well | 18:39 |
pcjc2 | Any idea what the easiest way to black-list the device is... so that evdev doesn't even try to use it? | 18:45 |
pcjc2 | My application (and the "official" propriotary drivers) expect to own the device | 18:45 |
jcristau | an fdi file to unset the input.x11_driver key for that device | 18:45 |
pcjc2 | Like in the comment here: https://lists.ubuntu.com/archives/ubuntu-devel-discuss/2008-October/005956.html ? | 18:46 |
jcristau | yeah | 18:47 |
tjaalton | pcjc2: people are reporting similar issues on bug 274203, so I'd like to redirect them to the new bug | 18:48 |
ubottu | Launchpad bug 274203 in xserver-xorg-input-evdev "Joystick detected as mouse, crashes X" [Undecided,Fix released] https://launchpad.net/bugs/274203 | 18:48 |
jcristau | i didn't know people made proprietary drivers for joysticks.. | 18:50 |
pcjc2 | this is a Ndof controller. The proprieraty driver communicates with the application with X messags | 18:51 |
pcjc2 | and their UI allows you to tune the 3D navigation preferences on a per-application basis. It is also their way of locking down using restrictive licensing... so I can buy a piece of hardware with no license to use the drivers for commercially related work | 18:52 |
pcjc2 | On the plus side, the "Personal edition" is much cheaper, and with an OSS driver, no different to the expensive version ;) | 18:52 |
jcristau | heh | 18:53 |
pcjc2 | didn't take too long for someone to clone the driver interface either... http://spacenav.sourceforge.net/ | 18:55 |
bryce | http://www.collegehumor.com/video:1886349 | 19:16 |
pcjc2 | that would be funny if my XP desktop were _slower_ than my Intrepid install on this laptop... | 19:18 |
pcjc2 | actually, it lots better since apt-get remove tracker* jockey* and landscape* | 19:26 |
pcjc2 | I was upset to see how much disk IO was spent checking for propriatory drivers I don't have on the system | 19:27 |
tjaalton | yeah, they are such resource hogs ;) | 19:27 |
pcjc2 | landscape is some kind of remote admin? | 19:27 |
jcristau | it's canonical's attempt to make money on support contracts aiui | 19:27 |
pcjc2 | thought as much, hence apt-get remove | 19:28 |
pcjc2 | also, the gdm prefetch list was pointing to some bad links | 19:28 |
pcjc2 | those readahead / prefetch files bitrot damned fast! | 19:28 |
tjaalton | the landscape client has no daemon, so it's harmless to have, unless you hate the stats when logging in from the console/ssh :) | 19:30 |
pcjc2 | landscape-sysin took about 11 seconds blocking on disk-io if I'm interperting this bootchart correctly | 19:32 |
tjaalton | ah | 19:32 |
pcjc2 | My laptop has a slow-spinning HDD, so I suffer from IO intensive boot processes | 19:33 |
pcjc2 | For that matter... can't we have a "cron" type replacement start with the desktop, so the computer doesn't immediately check for apt updates every time I login? | 19:33 |
pcjc2 | (It can wait a few seconds!) | 19:34 |
superm1 | bryce, has there been any thought put forth in deciding whether to do SRU's or backports for newer nvidia drivers in the same release series? | 20:09 |
superm1 | eg if NV were to do an updated 177 driver that fixed bugs | 20:10 |
tjaalton | duh, pcjc2 left.. http://cgit.freedesktop.org/xorg/driver/xf86-input-evdev/commit/?id=64554e4799a697d37dfd8be480f8eee636b9bea1 | 22:21 |
bryce | superm1: it's something that there's been discussion about from time to time | 23:22 |
bryce | esp. around when we were thinking of doing an sru for -fglrx | 23:22 |
bryce | in general the release team is very uncomfortable doing sru's for binaries and would prefer avoiding it | 23:23 |
bryce | however, in the case where we would be going from a 100% non-functional binary driver, to a functional one, the case would be pretty evident | 23:23 |
bryce | in cases where the driver works for some people already, then the case to do an sru is harder to make | 23:24 |
bryce | with backports, the release team would be fine with that | 23:25 |
bryce | the trouble is aiui, the backport team is fairly manpower limited | 23:26 |
bryce | however that may be less of an issue in the case of binary drivers since they come precompiled ;-) | 23:26 |
bryce | and of course there's also envyng for solving the need | 23:27 |
superm1 | bryce, okay well i expect a bug fix nvidia release very soon that is supposed to fix suspend regressions, so you'll see when it shows up i suppose | 23:33 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!