From Fedora Project Wiki

< SIGs‎ | Desktop

Desktop SIG Meeting of 2007-09-19

Present

  • MatthiasClasen (mclasen)
  • JesseKeating (f13)
  • JeremyKatz (jeremy)
  • RayStrode (halfline)
  • ColinWalters (walters)
  • DavidZeuthen (davidz)
  • DavidWoodhouse (dwmw2_gone)
  • WarrenTogami (warren)

Notes

  • xulrunner has been built, but untagged. Trying to sort out issues before test3
  • A working NM snapshot needs to be in rawhide tomorrow, or it won't be in F8
  • gdm root login will be reenabled
  • looking at a handful of serious bugs to fix by test3

Log


Sep 19 14:01:03 <jeremy>        yaaarr!
Sep 19 14:01:16 <mclasen>       ok, looks like we will be a bit light today, with various people on vacation or at home
Sep 19 14:01:33 *       mclasen has changed the topic to: Desktop SIG meeting
Sep 19 14:02:38 <mclasen>       so, we should perhaps start off with the status of the few outstanding F8 features that we still have
Sep 19 14:03:02 <mclasen>       xulrunner finally succumbed to repeated build attempts yesterday
Sep 19 14:03:11 <f13>   only to be backed out
Sep 19 14:03:17 <f13>   -ENOTREADY
Sep 19 14:03:18 <mclasen>       and caused havoc in rawhide today...
Sep 19 14:03:52 <mclasen>       with caillon out on vacation for the rest of the month, I'm not sure if we have the bandwidth to get that sorted out before test3
Sep 19 14:04:00 <dwmw2_gone>    it's building for ppc just fine with a one-line makefile change, btw. ppc64 is on the way; it looks also cosmetic
Sep 19 14:04:01 <jeremy>        it looks like dwmw2 got the ppc bits figured out
Sep 19 14:04:10 <dwmw2_gone>    untested as yet
Sep 19 14:04:12 <mclasen>       thats at least something
Sep 19 14:04:15 <f13>   there is still the odd Obsoletes in tehre
Sep 19 14:04:43 <mclasen>       so, in theory, if we remove the obsoletes
Sep 19 14:04:50 <mclasen>       can xulrunner and firefox coexist ?
Sep 19 14:05:09 <jeremy>        the -devel can't from what I gathered
Sep 19 14:05:28 <mclasen>       but xulrunner is supposed to replace the -devel bits anyway, right ?
Sep 19 14:05:31 <jeremy>        right
Sep 19 14:05:42 <mclasen>       ie we would rebuild everything (- ff and tb) against xulrunner
Sep 19 14:05:46 <mezcalero>     for when is the test3 freeze scheduled?
Sep 19 14:05:52 <mclasen>       Tuesday
Sep 19 14:05:56 <walters>       mezcalero: http://fedoraproject.org/wiki/Releases/8/Schedule
Sep 19 14:06:06 <blizzard>      mclasen: that's not a safe assumption - it's not clear that that is how ff or xulrunner work today?
Sep 19 14:06:07 <walters>       i happened to have that url open
Sep 19 14:06:19 <blizzard>      mclasen: need some caillon time
Sep 19 14:06:26 <mclasen>       thats not available right now
Sep 19 14:06:27 *       dwmw2_gone wonders if he'll be allowed to work again before then :)
Sep 19 14:06:35 <mezcalero>     walters: that schedule lacks the date for the test3 freeze...
Sep 19 14:06:59 <f13>   no it doesn't.
Sep 19 14:07:21 <f13>   but it'snot made obvious that the one date entry is for all the next items in the list
Sep 19 14:07:30 <f13>   multiple things happen on that date, so the date is given once.
Sep 19 14:07:38 <f13>   just like with test2
Sep 19 14:07:47 <mezcalero>     f13: ah, got it
Sep 19 14:08:05 <mclasen>       blizzard: in the light of the unavailability of caillon, does it make sense to keep trying to sort out firefox vs xulrunner ?
Sep 19 14:08:30 <mclasen>       I don't know if stransky is up to sorting that out on his own
Sep 19 14:09:15 <blizzard>      mclasen: no idea; I suspect not?
Sep 19 14:09:35 <blizzard>      mclasen: I haven't been involved in that so don't take my opinion as gospel
Sep 19 14:09:51 <mclasen>       I guess I'll poke stransky and see if caillon left him some more detailed instructions
Sep 19 14:09:53 <jeremy>        caillon seemed pretty hot for getting it done when we dropped it
Sep 19 14:10:00 <blizzard>      mclasen: I just know that in the past trying to build ff or tb as an app on top of a xulrunner framework hasn't been a supported method
Sep 19 14:10:03 <jeremy>        we only undropped it because he jumped up and down repeatedly
Sep 19 14:10:14 <walters>       blizzard: mclasen was saying not ff or tb
Sep 19 14:10:19 <jeremy>        blizzard: the idea was not to build ff/tb on top of xulrunner, just everything else
Sep 19 14:10:21 <blizzard>      ahh
Sep 19 14:10:26 <mclasen>       blizzard: I think the immediate rebuild plan is for everything else, but not ff and tb
Sep 19 14:10:27 <blizzard>      I thought someone just said 'ff on xulrunner'
Sep 19 14:10:31 <blizzard>      everything else seems fine
Sep 19 14:10:37 <blizzard>      sorry, misread
Sep 19 14:10:48 <blizzard>      ahh, (-ff and tb)
Sep 19 14:10:53 <blizzard>      that - was important!
Sep 19 14:11:02 <jeremy>        blizzard: indeed! :)
Sep 19 14:11:12 <blizzard>      :)
Sep 19 14:11:49 <mclasen>       ok, so what is the conclusion here ?
Sep 19 14:12:07 <mclasen>       1. fix up the xulrunner packages to build on all arches and have sane provides/obsoletes
Sep 19 14:12:21 <mclasen>       2. check that xulrunner and current ff/tb can nicely coexist
Sep 19 14:12:27 <mclasen>       3. rebuild the rest
Sep 19 14:12:43 *       davidz (n=davidz@c-24-147-123-95.hsd1.ma.comcast.net) has joined #fedora-meeting
Sep 19 14:12:55 <mclasen>       and reserve to roll all that back if it doesn't work out by early next week ?
Sep 19 14:13:16 <jeremy>        sounds good to me
Sep 19 14:13:42 <mclasen>       ok, I'll contact stransky and try to get hold of caillon
Sep 19 14:13:48 <f13>   works for me.
Sep 19 14:13:52 <mclasen>       next feature: NM
Sep 19 14:14:27 <blizzard>      dcbw is around somewhere
Sep 19 14:14:29 <sgarrity>      annoying late-comer asks: is there an agenda for this meeting?
Sep 19 14:14:30 <mclasen>       dcbw has been promising a working snapshot for this week, and he is working on integrating the new config gui bits today
Sep 19 14:14:39 <blizzard>      jay are bee
Sep 19 14:14:45 <mclasen>       sgarrity: just what I posted to fedora-desktop
Sep 19 14:15:02 <sgarrity>      mclasen: thx
Sep 19 14:15:03 <jeremy>        if we're going to get testing on the bits before test3, then they need to be in rawhide _tomorrow_ at the latest
Sep 19 14:15:17 *       mclasen walks over to dan
Sep 19 14:15:56 <jeremy>        we really should have had the bits in rawhide the day after test2 was released (if not when we unfroze) so that we could have had a testing live image out immediately for people
Sep 19 14:16:10 <f13>   and if they're not, he's probably going to have a pile of existing NM bugs to try and tackle if we're going with the current NM for F8
Sep 19 14:17:26 <mclasen>       dan says that whats currently in will work for unencrypted wireless
Sep 19 14:17:39 <f13>   any eta on encrypted wireless?
Sep 19 14:17:41 <mclasen>       he's waiting on ui bits for entering passwords, etc
Sep 19 14:17:50 <mclasen>       which were promised to land today
Sep 19 14:17:51 <f13>   it doesn't feel very good to enable by default a fairly useless NM :/
Sep 19 14:18:00 <mclasen>       right
Sep 19 14:18:01 <jeremy>        mclasen: to be blunt, that's basically useless
Sep 19 14:18:09 <dwmw2_gone>    if we're doing the bluetooth feature then it would be really nice to integrate the pan support in NM -- there are patches for that in gnome bugzilla
Sep 19 14:18:22 *       jrb drops in while on vacation
Sep 19 14:18:26 <jeremy>        frankly, I
Sep 19 14:18:41 <mclasen>       dwmw2_gone: bastian wants to tackle that for F9, I think
Sep 19 14:18:44 <jeremy>        'm not happy that we made multiple exceptions to the feature freeze for things that don't feel any closer now than they did two weeks ago :(
Sep 19 14:19:10 <mclasen>       jeremy: how are they not closer ? are you running upstream svn to say that ?
Sep 19 14:19:18 <dwmw2_gone>    mclasen: ok
Sep 19 14:19:44 <f13>   mclasen: basically they aren't in rawhide.
Sep 19 14:19:58 <mclasen>       f13: they were in rawhide before, but got pulled
Sep 19 14:20:13 <f13>   because they weren't even testable.
Sep 19 14:20:18 <f13>   it was completely broken
Sep 19 14:20:26 <f13>   and it hasn't gotten to a state that could be considered testable.
Sep 19 14:20:49 <f13>   and now we're looking at the test3 freeze and rushing in more stuff.
Sep 19 14:21:06 <f13>   which will probably have bugs, need more last minute fixes, and cause more delays in the test release and a longer freeze.
Sep 19 14:21:22 <mclasen>       just like the existing NM in rawhide has bugs, too
Sep 19 14:21:32 <mclasen>       i'm not saying it is a good situation
Sep 19 14:21:50 <f13>   at least the existing one functions for people
Sep 19 14:21:51 <mclasen>       but imo the "must be in rawhide tomorrow" deadline is a bit artificial
Sep 19 14:22:08 <jeremy>        mclasen: bugs are one thing.  I don't think "no UI exists to connect to encrypted wireless" is really a bug so much as "regression of the feature"
Sep 19 14:22:22 <f13>   mclasen: having at least a couple days of use by people before we freeze is absolutely necessary
Sep 19 14:22:35 <jeremy>        mclasen: NM was given an exception to the feature freeze on the condition that we would have testable bits in rawhide to actually build snapshots with prior to test3
Sep 19 14:22:52 <mclasen>       I don't recall that specific condition
Sep 19 14:22:57 <mclasen>       but it makes sense, sure
Sep 19 14:22:58 <jeremy>        test3 is ON TUESDAY.  if I don't have testable bits in rawhide tomorrow, how the hell are we supposed to have time to get testing feedback off of a snapshot from people?
Sep 19 14:23:16 <jeremy>        mclasen: dude, you were the one that SUGGESTED it.
Sep 19 14:23:22 <mclasen>       have we been doing regular snapshots since test2 ?
Sep 19 14:23:33 <jeremy>        <mclasen> we could do a one-off spin once a working snapshot is there, and ask people to test it
Sep 19 14:23:46 <mclasen>       right, I said that
Sep 19 14:23:53 <mclasen>       but we didn't do any snapshots
Sep 19 14:24:03 <jeremy>        mclasen: because I've been waiting on NetworkManager
Sep 19 14:24:12 <jeremy>        mclasen: since I've been told "one more day" every day this week
Sep 19 14:24:24 <dwmw2_gone>    who's "owning" xulrunning in caillon's absence? i.e. who should I cc this patch to?
Sep 19 14:24:31 <mclasen>       stransky
Sep 19 14:24:35 <dwmw2_gone>    ta
Sep 19 14:24:58 <jeremy>        (... and I've been doing snapshots every morning, just not putting them up on the torrent and discouraging our testers with another 700 megs to download every day)
Sep 19 14:25:08 <warren>        stransky seems to be gone for the day.
Sep 19 14:25:33 <notting>       warren: well, being in .cz...
Sep 19 14:27:00 <mclasen>       ok, so NM must appear in tomorrows rawhide or we do without
Sep 19 14:27:10 <mclasen>       next  topic ?
Sep 19 14:27:40 <mclasen>       ah, right, pre-test3 status, current breakages, etc
Sep 19 14:27:57 <f13>   gnome-keyring is still a disaster
Sep 19 14:28:16 <mclasen>       I thought I should perhaps just list the things that are on my radar for must-be-fixed-pronto
Sep 19 14:28:17 <warren>        NM is behaving really strange for me
Sep 19 14:28:30 <mclasen>       warren: thats the one you get to keep for F8...
Sep 19 14:28:41 <warren>        mclasen, ?
Sep 19 14:29:08 <mclasen>       warren: if we don't get the new NM in, F8 will stay with the old NM
Sep 19 14:29:12 <mclasen>       including strange behaviour
Sep 19 14:29:14 <f13>   warren: see above discussion about deadline for the NM rewrite.
Sep 19 14:29:21 <warren>        ah
Sep 19 14:29:24 <f13>   mclasen: wrong, you have from test3 to final to fix bugs in the existing NM
Sep 19 14:30:04 <mclasen>       f13: only I don't have that many nm hackers
Sep 19 14:30:32 <mclasen>       anyway, I wanted to list the critical things I am aware of
Sep 19 14:30:56 <mclasen>       - ice problem with gnome-session
Sep 19 14:31:09 <mclasen>       - eclipse has some gtk related crash in tooltip code
Sep 19 14:31:20 <mclasen>       - panel pulls in tomboy+mono
Sep 19 14:31:31 <mclasen>       - tomboy pops up start here on first login
Sep 19 14:31:55 <mclasen>       - gnome-keyring brokenness
Sep 19 14:32:13 <mclasen>       - mixer applet strange behaviour
Sep 19 14:32:36 <mclasen>       that is pretty much my short list of serious issues
Sep 19 14:32:47 <mclasen>       what am I missing ?
Sep 19 14:33:02 <f13>   mclasen: that seems pretty good.  Are there bugs on these?  We have a F8Test3 tracker now
Sep 19 14:33:14 <warren>        mclasen, may I propose that we turn off gnome-power-manager's handling of backlight brightness?  It seems to be far too buggy.
Sep 19 14:33:15 <notting>       mclasen: gdm theme is bust
Sep 19 14:33:29 <mclasen>       notting: i'll fix that up after the meeting
Sep 19 14:33:56 <mclasen>       warren: "far too buggy" is a bit vague
Sep 19 14:34:23 <warren>        mclasen, interferes and fights manual changes of brightness
Sep 19 14:34:28 <f13>   mclasen: often times the g-p-m and the physical hardware itself fights to set brightness
Sep 19 14:34:30 <warren>        mclasen, and the list of open bugs complaining about it is extensive.
Sep 19 14:34:45 <f13>   mclasen: and it stutters or chagnes for no apparent reason often.
Sep 19 14:34:56 <warren>        mclasen, sometimes it brightens or reduces brightness for seemingly no reason with no event
Sep 19 14:35:01 <warren>        It simply isn't ready.
Sep 19 14:35:09 <warren>        and you can't even disable it unless you kill gnome-power-manager entirely
Sep 19 14:35:43 <mclasen>       whats wrong with the "Dim display when idle" checkbox ?
Sep 19 14:35:49 <mclasen>       does that not do what I think it does ?
Sep 19 14:35:56 <warren>        mclasen, it would be fine if it were disabled by default, with a checkbox to enable.  That might allow the feature to be used by those who want it, and it can be further debugged.  But otherwise we're too close to release to deal with this ATM.
Sep 19 14:36:06 *       warren looks again...
Sep 19 14:36:20 <mdomsch>       df -H
Sep 19 14:36:22 <warren>        mclasen, I have it unchecked (not default) and it still interferes with manual changes.
Sep 19 14:36:24 <mclasen>       davidz: whats your take on backlight vs g-p-m ?
Sep 19 14:36:35 <dwmw2_gone>    warren: er, I think you're describing a feature which works perfectly for me in F7.
Sep 19 14:36:41 <dwmw2_gone>    or is that not done by g-p-m there?
Sep 19 14:36:45 <warren>        dwmw2_gone, YMMV
Sep 19 14:36:47 <warren>        dwmw2_gone, possibly
Sep 19 14:36:56 <warren>        dwmw2_gone, BIOS does it itself on many laptops
Sep 19 14:36:59 <davidz>        mclasen: ask richard
Sep 19 14:37:03 <f13>   dwmw2_gone: it seems to depend on hardware.
Sep 19 14:37:21 <f13>   dwmw2_gone: like IBMs and Dells do it in bios automatically
Sep 19 14:37:23 <warren>        Seems to be too variable and problematic, let's turn it off by default in F8
Sep 19 14:37:29 <davidz>        hey, let's wait
Sep 19 14:37:47 <warren>        Although, I unchecked the box and it still interferes.
Sep 19 14:37:53 <davidz>        and get to the bottom of the issue
Sep 19 14:37:57 <mclasen>       warren: you just claimed that turning it off does not help...
Sep 19 14:38:08 <warren>        yes, there is something else wrong
Sep 19 14:38:17 <davidz>        well
Sep 19 14:38:20 <walters>       i wouldn't call brightness a showstopper...networkmanager and keyring look far more important
Sep 19 14:38:20 <davidz>        point to good bug reports
Sep 19 14:38:33 <walters>       brightness could be fixed after the release easily enough if necessary
Sep 19 14:38:52 <davidz>        warren: fwiw, I don't recall seeing bug reports about this; maybe richard gets the g-p-m bugmail
Sep 19 14:39:04 <davidz>        warren: can you post about the issue to fedora-desktop-list?
Sep 19 14:39:08 <warren>        davidz, ok
Sep 19 14:39:10 <davidz>        warren: including pointers to bug reports
Sep 19 14:39:13 <davidz>        warren: great, thanks
Sep 19 14:39:18 <mclasen>       yeah, that would be good
Sep 19 14:39:21 <davidz>        then I'll look into it
Sep 19 14:39:36 <davidz>        who knows, maybe hughsie is using a SQLite database to store the things :-)
Sep 19 14:39:42 <davidz>        seems to be a sign of the times :-)
Sep 19 14:39:45 *       davidz bad joke
Sep 19 14:39:46 <davidz>        ok
Sep 19 14:39:59 <mclasen>       one more test3 item is the artwork; I've put the lock dialog and gdm theme in today
Sep 19 14:40:04 <mclasen>       though I messed up the gdm part
Sep 19 14:40:06 <f13>   warren: did you uncheck it for both on AC and on Battery?
Sep 19 14:40:12 <warren>        f13, yes
Sep 19 14:40:14 <f13>   k
Sep 19 14:40:36 <f13>   warren: and also slide the slider up to 100% on the on battery?
Sep 19 14:40:53 <warren>        f13, I think what I want is for it to stop interfering entirely
Sep 19 14:41:13 <f13>   warren: of course, n/m, I was just poking for more info, but not time/place.
Sep 19 14:41:22 <f13>   notting: what did you mean by gdm is stuffed?
Sep 19 14:41:32 <warren>        f13, there is currently no option for it to stop interfering.
Sep 19 14:41:34 <f13>   oh n/m I see that was addressed.
Sep 19 14:43:02 *       mclasen watches dan type like crazy...
Sep 19 14:43:34 <f13>   heh
Sep 19 14:43:57 <davidz>        warren: (my guess is that you are using a Dell that with it's own backlight stuff turned on - but post all details to the mailing list)
Sep 19 14:45:16 <mclasen>       is there some trick like making sure dcdbas is insmoded, or something ?
Sep 19 14:45:45 <f13>   mclasen: that has some of it's own issues.
Sep 19 14:45:46 <davidz>        mclasen: that's probably a separate bug
Sep 19 14:45:54 <f13>   but I thought warren was using a thinkpad now
Sep 19 14:45:54 <davidz>        mclasen: needs to be fixed in the upstream kernel
Sep 19 14:46:02 <davidz>        but lets start on the mailing list
Sep 19 14:46:08 <davidz>        to figure out what's wrong and how to fix it
Sep 19 14:46:10 <davidz>        much easier than IRC
Sep 19 14:46:13 <mclasen>       yeah
Sep 19 14:46:46 <mclasen>       do we want to talk about the desktop flamewars du jour at all ? root login and mono dependency ?
Sep 19 14:47:26 <warren>        davidz, thinkpad
Sep 19 14:48:26 <f13>   the root login one would be fun to talk about
Sep 19 14:48:46 <f13>   so far, the best suggestion I've seen is to have a limited root desktop env that gives you a few tools to fix things or add a user.
Sep 19 14:48:46 <notting>       obviously we should promote all user logins to root
Sep 19 14:49:14 <mclasen>       we have talked about allowing user creation from the login screen
Sep 19 14:49:16 <f13>   'course people are going to whine that this is the same as aliasing 'rm' to 'rm -i' but whatever.
Sep 19 14:49:23 <f13>   mclasen: that would work too
Sep 19 14:49:26 <walters>       imo, no login as root supported, because we don't have a root password that the user knows
Sep 19 14:49:33 <f13>   honestly I thought that if we had the guest account stuff this would be less of an issue.
Sep 19 14:49:35 <jeremy>        mclasen: I think once gnome-panel doesn't spew an error for missing applets and we drop the tomboy dep, the mono one fizzles out.  (and it's reasonable enough to do, so I don't see it as worthy of that much discussion really)
Sep 19 14:49:49 <mclasen>       I think so too
Sep 19 14:49:59 <mclasen>       I just need to find someone who writes that panel patch
Sep 19 14:50:14 <jeremy>        mclasen: halfline said he was going ot take a look on the bus yesterday ;)
Sep 19 14:50:45 <mclasen>       halfline: is that true ?
Sep 19 14:50:58 <halfline>      yes, i'm planning on looking at it
Sep 19 14:51:00 <jeremy>        for the root one -- we've talked about allowing adding users at the gdm screen.  but we don't have that functionality today, so that may not be the argument to make
Sep 19 14:51:06 <halfline>      i own panel, and it's on the blocker list after all...
Sep 19 14:51:17 <davidz>        remind me - why would we ever permit root login via gdm?
Sep 19 14:51:28 <mclasen>       jeremy: right, but I don't think we need a very strong argument
Sep 19 14:51:30 <davidz>        explain to me why such users can't switch to VT1 and startx?
Sep 19 14:51:35 <walters>       it's too late for this root discussion for f8 anyways
Sep 19 14:51:42 <davidz>        if they really want to shoot themselves in the head, that is
Sep 19 14:51:51 <walters>       for f9 we should sit down and really fix it (and imo, this means sudo by default and disabled root account)
Sep 19 14:51:52 <mclasen>       anybody who brings up unreliable nis as an argument for root login is better served by a local account anyway
Sep 19 14:52:12 <davidz>        uh, remind me - isn't this for the desktop spin? We can just make the change there
Sep 19 14:52:20 <walters>       for f8?
Sep 19 14:52:34 <davidz>        mainline Fedora; what they do... I don't care that much; they are for different users
Sep 19 14:52:38 <mclasen>       we did the change in the gdm config
Sep 19 14:52:39 <f13>   davidz: this change was already made, distro wide.
Sep 19 14:52:43 <davidz>        cool
Sep 19 14:52:47 <davidz>        so what's the issue?
Sep 19 14:52:51 <mclasen>       doing it just for the spin would be a bit painful in %post
Sep 19 14:52:52 <f13>   there was backlash
Sep 19 14:52:54 <halfline>      creating a user is optional
Sep 19 14:53:01 <jeremy>        mclasen: I think the argument for going back is that there were things that we going to get done to make it a) less likely to get you into the "no user created, you're doomed" scenario and b) things like adding users being discussed.  they haven't happened, so people are somewhat rightfully concerned
Sep 19 14:53:09 <davidz>        f13: so we change it to be only for the desktop spin?
Sep 19 14:53:10 <halfline>      so disabling root account means you end up in situation where you can't login
Sep 19 14:53:14 <halfline>      (besides from a vt)
Sep 19 14:53:44 <f13>   davidz: I think the thing is we want this distro wide, but we want a reasonable excape hatch for users that didn't create a login
Sep 19 14:53:50 <mclasen>       davidz: we can do that; I guess. it would be easier to do if the config were separate
Sep 19 14:53:54 <f13>   davidz: telling them to vt switch and useradd is not very friendly.
Sep 19 14:54:00 <halfline>      we should either make a local account required by firstboot instead of optional, or make gdm smart and detect if there is no account do something
Sep 19 14:54:03 <davidz>        f13: I think that's a mainline Fedora thing
Sep 19 14:54:08 <davidz>        f13: that discussion I mean
Sep 19 14:54:14 <halfline>      right we're discussing mainline
Sep 19 14:54:17 <davidz>        f13: so can we have that discussion elsewhere?
Sep 19 14:54:20 <walters>       f13: just because it's graphical doesn't mean it's understandable or sane
Sep 19 14:54:25 <davidz>        halfline: look at the topic
Sep 19 14:54:31 <f13>   davidz: dude, you guys /still/ provide the desktop for the mainline
Sep 19 14:54:39 <f13>   unless you want to just give that up too?
Sep 19 14:54:53 <halfline>      no we're not giving up on that
Sep 19 14:54:58 <davidz>        well
Sep 19 14:55:06 <f13>   I don't want to see a bunch of changes/enhancements be made outside the packages just ine one spin and then tell the rest of the users to just sod off.
Sep 19 14:55:08 <davidz>        one of the _main_ purposes behind this SIG
Sep 19 14:55:11 <davidz>        and this whole excersize
Sep 19 14:55:22 <davidz>        was to stop caring about mainline users who are not desktop users
Sep 19 14:55:45 <mclasen>       f13: honestly, some of the users in the root login thread could just sod off, imo
Sep 19 14:55:54 <davidz>        so I actually think it's off-topic here
Sep 19 14:55:55 <f13>   mclasen: some yes.
Sep 19 14:56:03 <f13>   but you're /still/ in a bad place for the desktop.
Sep 19 14:56:11 <f13>   where if you don't create a user, you get... tty1
Sep 19 14:56:12 <f13>   and shell.
Sep 19 14:56:13 <f13>   woo.
Sep 19 14:56:37 <halfline>      i mean bottom line root login for gdm isn't viable either
Sep 19 14:56:42 <mclasen>       firstboot already makes it hard to not create a user
Sep 19 14:56:47 <halfline>      neither tty1 or root login are nice options
Sep 19 14:56:48 <f13>   "hard"?
Sep 19 14:56:51 <mclasen>       it pops up a warning
Sep 19 14:56:53 <f13>   halfline: correct.
Sep 19 14:57:08 <f13>   mclasen: yea, but that warning doesn't tell you you won't be able to login next screen if you don't create one
Sep 19 14:57:13 <f13>   or enable network login
Sep 19 14:57:32 <mclasen>       f13: yeah, we should have updated the text
Sep 19 14:57:41 <f13>   what happened to the guest login stuff?
Sep 19 14:58:02 <f13>   where you could log in as guest, call the user tool and supply the root password you created during install?
Sep 19 14:58:17 <mclasen>       halfline: do you think the "allow root if no other users exist" gdm patch is doable ?
Sep 19 14:58:20 <walters>       f13: as currently implemented it requires a nasty gdm patch, i didn't get any comments on it and didn't have time to clean it up
Sep 19 14:58:27 <warren>        Does any OS disallow root login?
Sep 19 14:58:50 <walters>       f13: but we discussed how to do it better, i think we can probably move on it for f9
Sep 19 14:58:51 <halfline>      mclasen: yea probably, but i'm not sure that's a good idea either
Sep 19 14:59:02 <halfline>      i mean then you still end up in a "the user is logged in as root" situation
Sep 19 14:59:11 <mclasen>       right
Sep 19 14:59:26 <warren>        How did this become a priority given the other bigger issues?
Sep 19 14:59:26 <mclasen>       in the end, you just can't help people who absolutely want that
Sep 19 14:59:29 <halfline>      (i mean there is a slight complication, that /etc/passwd has a bunch of non-user accounts, but we can ignore those)
Sep 19 14:59:36 <f13>   warren: we're not talking about priorites now.
Sep 19 14:59:36 <mclasen>       because they will just turn it on in gdmsetup anyway
Sep 19 14:59:43 <f13>   warren: catch up with the topics (:
Sep 19 15:00:13 <halfline>      so we have a few solutions
Sep 19 15:00:22 <halfline>      1) make first boot require a user get created instead of optional
Sep 19 15:00:36 <halfline>      2) make firstboot tell the user that if they don't want to create an account, they're going to be boned
Sep 19 15:00:42 *       walters votes for 1)
Sep 19 15:00:55 <halfline>      3) make gdm detect if no viable account is available and allow root login
Sep 19 15:01:17 <halfline>      4) revert the "force non-root user" patch
Sep 19 15:01:17 <f13>   2 or 3
Sep 19 15:01:28 <warren>        #4
Sep 19 15:01:35 <halfline>      those are all possible short term fixes
Sep 19 15:01:36 <f13>   walters: avoiding 1 can give you more distrowide acceptance
Sep 19 15:01:45 <dwmw2_gone>    3 isn't particularly feasible
Sep 19 15:01:58 <f13>   dwmw2_gone: how's that?
Sep 19 15:02:08 <walters>       ok, 4) is fine too, i don't really care
Sep 19 15:02:13 <halfline>      dwmw2_gone: well you could do a pretty good heuristic
Sep 19 15:02:15 <f13>   dwmw2_gone: surely we use the same mechanism we use to fill in the face browser?
Sep 19 15:02:16 <dwmw2_gone>    all your users have kerberos passwords, the domain server is buggered
Sep 19 15:02:17 <dwmw2_gone>    for example
Sep 19 15:02:34 <dwmw2_gone>    the generic case of "I want network auth, but it isn't working right yet"
Sep 19 15:02:35 <mclasen>       halfline: even if we pick 4, gdm does warn if you log in as root, right ?
Sep 19 15:02:38 <halfline>      yea, you can't detect all values of "viable"
Sep 19 15:02:38 <f13>   dwmw2_gone: face browser doesn't get filled in there either
Sep 19 15:02:39 <notting>       #2/#1. create user or set up network info
Sep 19 15:02:41 <warren>        #4 means forget about disallowing root logins?
Sep 19 15:02:43 <halfline>      mclasen: yes
Sep 19 15:03:00 <warren>        Allow root but warn seems like the least problematic.
Sep 19 15:03:06 <warren>        Easiest to test.
Sep 19 15:03:07 *       walters thinks the entire concept of root is screwed and doesn't really care in which way it's screwed
Sep 19 15:03:09 <dwmw2_gone>    how about we permit root logins, but only if the user actually knows some secret known to permitted users only.... we can call it a 'password' :)
Sep 19 15:03:09 <mclasen>       dwmw2_gone: that is exactly the situation that was brought forward as an argument for allowing root login, no ?
Sep 19 15:03:42 <walters>       dwmw2_gone: right
Sep 19 15:03:46 <halfline>      the problem is root login is busted
Sep 19 15:03:50 <halfline>      things don't work right
Sep 19 15:03:51 <dwmw2_gone>    mclasen: I thought re-allowing root login was #4. Perhaps I'm missing something; sorry.
Sep 19 15:03:55 <halfline>      like you can't lock your screen
Sep 19 15:04:07 <mclasen>        disks don't get automounted
Sep 19 15:04:12 <halfline>      chat servers reject you, your path is different
Sep 19 15:04:15 <halfline>      right
Sep 19 15:04:36 <dwmw2_gone>    those don't prevent you from fixing up the problems which prevent proper users from logging in :)
Sep 19 15:04:44 <dwmw2_gone>    perhaps we should give root an xterm-only session?
Sep 19 15:04:47 <walters>       but as far as fedora bugs go, this one is hard to hit
Sep 19 15:04:50 <f13>   dwmw2_gone: that was suggested.
Sep 19 15:04:56 <halfline>      then again it's what we've had for forvever, so if we don't have a more baked solution, we cna just go with 4
Sep 19 15:04:59 <f13>   dwmw2_gone: but may not be easy to get to for F8
Sep 19 15:05:18 <f13>   dwmw2_gone: I think we're looking for a simple way to reduce the number of people that use X as root for F8, while investigating longer term solutions for F9
Sep 19 15:05:25 <notting>       well, by saying 'you can log in on a vt and fix it', how does that fit in with ajax's long-term KILL VT DEAD PLZ plan?
Sep 19 15:05:34 <halfline>      dwmw2_gone: though, what's so different between an xterm session and logging in from a vt?
Sep 19 15:05:45 <dwmw2_gone>    halfline: fair point.
Sep 19 15:05:59 <dwmw2_gone>    although I _can_ start X programs from an xterm
Sep 19 15:06:01 <halfline>      notting: doesn't fit in with long-term at all.  we are talking about short term solutions right now
Sep 19 15:06:03 <f13>   besides the people who don't realize they /have/ vts
Sep 19 15:06:04 <walters>       f13: do we have reports that people are really doing that?
Sep 19 15:06:07 <f13>   as seen onthe mailing list
Sep 19 15:06:24 <f13>   walters: really doing what?
Sep 19 15:06:28 <walters>       f13: logging in as root
Sep 19 15:06:29 <notting>       halfline: yeah, i'm just saying if we're going to have some sort of semi-gui 'administrator' logon then, we might not want to nuke it now
Sep 19 15:06:36 <f13>   walters: we see it all the time.
Sep 19 15:06:40 <warren>        Seriously, this is too hot and we're doing this too late in the cycle.
Sep 19 15:06:40 <halfline>      walters: oh yea, there's a really mean report i got the other day
Sep 19 15:06:41 <walters>       damn
Sep 19 15:07:03 <f13>   Ok, I suggest we just make the gdm warning a lot scarrier
Sep 19 15:07:10 <halfline>      warren: we made the change a long time ago
Sep 19 15:07:11 <mclasen>       warren: what do you mean, we are still before test3 and this wasn't a very recent change
Sep 19 15:07:14 <halfline>      just didn't do anything with it sense
Sep 19 15:07:29 <mclasen>       f13: string change
Sep 19 15:07:30 <f13>   so #4, but with scarrier warnings and more informative as "these things will break for you" lists.
Sep 19 15:07:43 <dwmw2_gone>    makes sense to me
Sep 19 15:07:45 <f13>   mclasen: we can probably get an exception from the translation team if we do it quickly
Sep 19 15:07:47 <walters>       i don't think developers should be spending time on admin guis at this stage
Sep 19 15:07:51 <halfline>      i don't really want to change the warning, given string freeze
Sep 19 15:07:58 <mclasen>       no, no admin gui
Sep 19 15:07:59 <walters>       so that leads us to #4
Sep 19 15:08:02 <halfline>      we'll lose upstream translations
Sep 19 15:08:10 <f13>   walters: they don't view it as 'admin gui'
Sep 19 15:08:11 <warren>        What exactly does #4 do?
Sep 19 15:08:12 <mclasen>       in particular no special root desktop
Sep 19 15:08:16 <f13>   walters: they just see no reason to run as a user instead of root
Sep 19 15:08:22 <mclasen>       with a bomb image on the background
Sep 19 15:08:32 <mclasen>       that feels so 90s
Sep 19 15:08:45 <walters>       f13: ok, well, we'll fix it right in f9, for now they stay boned
Sep 19 15:08:48 <halfline>      so it looks like most people are in favor of 4 (revert the change)
Sep 19 15:09:16 <mclasen>       looks like it
Sep 19 15:09:19 <halfline>      ok
Sep 19 15:09:27 <halfline>      we can do that, and get things better in f9
Sep 19 15:09:40 <halfline>      get a user automatically set up with sudo, get apps moved over to policykit, etc
Sep 19 15:09:50 <f13>   Guest account
Sep 19 15:09:54 <halfline>      yea
Sep 19 15:10:02 <f13>   I think that would quiet down most the haters out there.
Sep 19 15:10:10 <notting>       we already yell at you if you log in as root, right?
Sep 19 15:10:13 <halfline>      yes
Sep 19 15:10:18 <f13>   although I wonder if _A_ would freak over guest account stuff.
Sep 19 15:10:39 <f13>   notting: not loudly but yes.
Sep 19 15:10:47 <mclasen>       moving on
Sep 19 15:10:57 <mclasen>       anything else, or should we close for today ?
Sep 19 15:11:32 <ajax>  _A_ freaks out about random dust particles that float by
Sep 19 15:11:37 <ajax>  i wouldn't worry too much
Sep 19 15:11:58 <f13>   ajax: haha
Sep 19 15:12:00 <mclasen>       comes with the age
Sep 19 15:12:04 <dwmw2_gone>    I gave him a Mac Mini to play with last week, to get pmac_ide working with libata. That should keep him quiet for a while
Sep 19 15:12:05 *       mclasen gets more irritable, too
Sep 19 15:12:30 <mclasen>       anyway, lets come to an end. so we have
Sep 19 15:12:38 <mclasen>       warren send out mail about g-p-m brightness
Sep 19 15:12:51 <mclasen>       halfline changes root login and looks at panel applet warning
Sep 19 15:13:05 <halfline>      but not today, today is art stuff!
Sep 19 15:13:08 <mclasen>       I contact stransky about xulrunner things
Sep 19 15:13:22 <mclasen>       and all the other problems just magically resolve themselves...
Sep 19 15:13:37 <notting>       mclasen: NM AAARGH
Sep 19 15:13:42 <mclasen>       see you next week !
Sep 19 15:13:48 <mclasen>       right, nm too
Sep 19 15:14:07 <warren>        f13, did we already untag xulrunner?
Sep 19 15:14:48 <f13>   warren: yes.
Sep 19 15:14:52 <warren>        k