Ubuntu Traffic #2 For 2004/09/03 By Benjamin Mako Hill Table Of Contents • Standard Format • Text Format • XML Source • Introduction • Mailing List Stats For This Week • Threads Covered 1. 2004/08/29 - 2004/09/02 (46 posts) Install Reports 2. 2004/08/31 - 2004/09/02 (4 posts) Jidgo Files for Warty 3. 2004/08/31 - 2004/09/02 (7 posts) nVidia Drivers 4. 2004/08/31 - 2004/09/01 (3 posts) Discover 5. 2004/08/31 - 2004/09/02 (13 posts) Init Scripts 6. 2004/09/01 - 2004/09/02 (15 posts) pmount and Breakage 7. 2004/09/01 - 2004/09/02 (5 posts) Desktop/Computer 8. 2004/09/01 - 2004/09/02 (5 posts) mdadm Daemon 9. 2004/09/04 (2 posts) Bugzilla Introduction Welcome to the second edition of Ubuntu Traffic. This issue covers the week of August 28 - September 3 in 2004. Ubuntu Traffic summarizes the most important mailing list and IRC discussions involving the Ubuntu Linux distribution. Bits and pieces that didn't get a full story but are worth mentioning include: • Baptiste Mille-Mathias pointed out that the warthogs wiki is currently broken for French and suggested that the multi-language handler should be disabled in moin-moin on that wiki. • Matthias Klose argued for a rather late-in-the-game sync of Python2.3 from Debian unstable to fix a couple important issues. Mailing List Stats For This Week We looked at 182 posts in 759K. There were 41 different contributors. 28 posted more than once. 17 posted last week too. The top posters of the week were: • 35 posts in 160K by Matt Zimmerman • 17 posts in 69K by Jeff Waugh • 13 posts in 57K by Colin Watson • 11 posts in 45K by Fabio Massimo Di Nitto • 10 posts in 46K by Martin Pitt • Full Stats  1. Install Reports 2004/08/29 - 2004/09/02 (46 posts) Subject: "ubuntu install" People: Stuart Langridge, B, James Troup, Mary Gardiner This week saw five threads by people installing Ubuntu and each one generated a deal of feedback and discussion on the list. The threads were: Stuart Langridge's Install Report Stuart posted a more "stream of consciousness" report that is somewhat hard to summarize concisely. One particular problem he ran into was that his PCMCIA wireless card was not detected. It was unclear however, what card he was using. Stuart also reported that, "It asked me for a username, so I picked one, and then it asked me again to "Select a username for the new account", defaulting it to the username I'd just picked! So what was the first question, then, if it wasn't picking a username? I am confused. Ah, the first thing it asked for was the full name. That might confuse more experienced users, who expect to be asked for a username first. This is a lesson in reading dialog boxes, I admit." James Blackwell reported that he had done the same thing. B's Review B, had the following to report about their installation experience: Big problem. My Netgear WG511 802.11 PCMCIA cardbus. I attempted to setup a wireless network, but this 'died' as soon as I activated it. OK, Try wget the firmware, error message on boot, still no wifi. So far I can't get this working. For a, 'totally rad laptop support' it needs to find PCMCIA wireless cards and have them work first time. Suse found it and configured it with Yast easy. Impressions? Well I really like the look and feel of the Gnome system, if Ubuntu supports my wifi card I'll use it as my distro of choice. Thom May identified that this was a prism54 missing firmware issue but B thought the firmware was there. Matt Zimmerman suggested that the firmware might be incomplete or corrupt and B said they would try again later. James Troup's Install Report James Troup got his new powerbook and did the installation. He had the following issues to report: □ I was prompted for whether or not I wanted to use firewire ethernet? I thought we'd disabled that... □ base-config talks about setting root password in it's introductory dialog but our installation process doesn't do this anymore. □ After entering password for the new user a second time there's a noticeable pause as the CD rom spins up. Then random "scary" apt-get output as it preinstalls mdetect etc. Might be nice to prefix this with a note to say what's about to happen? □ There's a long pause after "Initializing package states... Done" (i.e. while it's trying to work out what to install for ubuntu-desktop). Might be nice if whatever this is (aptitude?) printed something while it was thinking? □ When there's an uninstallable package (as there is today), the dialog talks about going back up to select different packages, but a) the user didn't select the packages, b) the solution (use aptitiude) is non-obvious. □ One of the partitioning dialogs claims you'll be given a chance to edit the partitions, but you aren't really without resorting to 'go back' which then drops you out of the nice automation. □ ubuntu-artwork no longer depends on suede-icons leading to "Computer" and "Wastebasket" being generic 'paper' icons. (Thanks to seb128 for diagnosis) □ Wrong permission for /dev/pmu device in gnome still. □ Still no 2nd/3rd mouse button mappings for powerbooks. □ xscreensaver doesn't password-lock by default yet. There was a little disagreement on the list about whether or not xscreensaver should in fact lock by default. Mary Gardiner's Report Mary Gardiner installed onto a Fujitsu Lifebook S6210. She ran into the three problems: 1. This laptop has a memory stick slot, but nothing shows up in /var/log/ messages or in Nautilus when I insert a known good memory stick. 2. The desktop icons and icons in Nautilus are all blank pieces of paper, which I take it means "no icon for this". 3. No shocks here but: the wireless card (Centrino, 0000:01:0d.0 Network controller: Intel Corp. Intel(R) PRO/Wireless 2200BG (rev 05)) doesn't work. Looking forward to trying the new test kernels in a minute... Mary followed up her own message to report that the new test kernel posted fixed the problem. Matt Zimmerman informed Mary that we simply do no have a lot of support for those memory sticks at the moment regardless of the type.  2. Jidgo Files for Warty 2004/08/31 - 2004/09/02 (4 posts) Subject: "Jigdo for warty?" People: Cef, Colin Watson Cef asked, "Since not everything changes between test releases, is it possible to get jigdo files so we can build our own CD images? This makes a bit more sense for those of us that have local repositories of files or an apt-proxy, so that we don't download things multiple times." Colin Walters replied to this saying: I do agree that we should do this, but I'm not entirely sure how to set up the fallback mirror in a suitable way. The outside world doesn't have direct access to the machine I use for CD building, so I'd need support in our archive to get a hardlink tree produced at the point when the image was generated. It seemed like a three-banana problem when I last thought about it, although I'm happy to be corrected. :-) For now, rsync seems to work reasonably well, especially if you use the dailies.  3. nVidia Drivers 2004/08/31 - 2004/09/02 (7 posts) Subject: "installation of nvidia drivers" People: Lionel (Ploum) Dricot, Mark Shuttleworth Lionel (Ploum) Dricot posted that he was sad to see that the Warty was shipping with the free nVidia drivers which did not provide very good accelerated 3d support. He suggested the following alternative: "a package that will do all this job for the user, even downloading the binary driver on the nvidia website. The package will itself kill X, modify the conf and restart it." Mark Shuttleworth posted to say that, "We do have the "restricted" component, and the non-free drivers could live there. So getting them should be easy. Fabio, how easy could we make installing them? Will the X autodetection stuff use them if they are available?"  4. Discover 2004/08/31 - 2004/09/01 (3 posts) Subject: "discover v1/v2" People: Matt Zimmerman Matt Zimmerman posted asking anyone if they had a problem with removing discover from the supported seed altogether: "We need discover v1 for debian-installer and xfree86. However, I am not sure that we need to ship discover v2 (currently in SupportedSeed) at all. Is there any reason we should not remove it?" Nobody raised any major objections.  5. Init Scripts 2004/08/31 - 2004/09/02 (13 posts) Subject: "Initscripts and tput" People: Nathaniel McCallum, Matt Zimmerman, Scott James Remnant Nathaniel McCallum announced that: After a long session hacking initscripts today, I realized that we are relying on tput to get ansi control characters. This is not bad except that tput is in /usr/bin/ and /usr may not be available to use before mounting filesystems. There is a couple solutions as i see it: 1. .move tput from /usr/bin to /bin in the terminfo-bin package 2. .fall back to hardcoded ansi control characters (may not work on some strange terminals) 3. .other solution... What do you think? Scott James Remnant asked why Nathaniel needed control characters at all. He replied that it was necessary for right more beautiful boot process through a wrapper around each init script. Matt Zimmerman weighed in saying, "Another alternative would be to revive script-output, but using terminfo exclusively. I think my preference is to move tput, though I would also be interested in alternatives which allowed for a simplified output format (not involving terminal control sequences) for the very early stages of the boot process (pre-mountall.sh). It is important that we get this right, so I'd like to hear from others on this." Scott James Remnant was not totally comfortable with this saying, "For consistency we should move all of ncurses-bin into /bin ... I generally dislike moving single binaries just so they're available for you earlier in the boot process. I also slightly dislike moving this many binaries, they don't really belong there."  6. pmount and Breakage 2004/09/01 - 2004/09/02 (15 posts) Subject: "New group for local users (WILL BREAK UPGRADES without manual intervention!)" People: Martin Pitt, Fabio Massimo Di Nitto, Matt Zimmerman Martin Pitt, continuing his wave of good work on pmount made and important announcement to the list: I just uploaded a new version of base-config that creates an additional group 'plugdev' and puts the initially created user into it. This group's purpose is to control access to removeable devices like USB sticks and digital cameras. I will upload a new version of libgphoto2-2 soon which will allow access to USB digital cameras to plugdev members without any further configuration. More importantly, I will upload new versions of mount/umount and pmount/ pumount which can _only_ be executed by plugdev members. If you install new Ubuntu versions from scratch, you will not have to do anything. However, if you are upgrading from an older Warty, then the already existing user(s) will not be put into plugdev automatically, so you have to do this by hand: # addgroup --system plugdev # adduser plugdev (as root). If you don't do this, you will not be able to mount devices (as user) any more in the near future! Thanks for your attention and have a nice day! Fabio Massimo Di Nitto raised an objection/suggestion saying, "Can't we check if the group is there and take appropriate actions? This will break all upgrades from Debian to us. The admin should also be informed of these changes (only on upgrades) and perhaps (really a wishlist here) ask for the default user to add to that group." Martin Pitt replied saying, "I'm aware of that and I already discussed that with Matt. The current "solution" is to put this into the release notes since it does not directly break upgrades, but the usage afterward (I just exaggerated the subject a little to get some attention)" and "One possibility is to have the new mount/pmount packages display a high-priority debconf note if the group is not present. But I think somebody will shoot me if I did that. Alternatively, this change could be mailed to root@localhost, but I'm not sure whether the admin really reads this." Matt Zimmerman replied saying, "This is actually only an issue if we change the privileges of mount and umount. If we leave mount/umount alone, and only change pmount/pumount, then no behaviour changes on upgrade."  7. Desktop/Computer 2004/09/01 - 2004/09/02 (5 posts) Subject: "The Desktop/Computer Battle" People: Scott James Remnant, Sebastien Bacher, Jeff Waugh Scott James Remnant posted the following summary and suggestion about the placement of desktop icons and the like: Ok, everyone seems to have different opinions about this; I count three major ones: □ Put Home/Documents, Filesystem, Floppy/CD-ROM, Network, etc. on the Desktop (Me) □ Put Home/Documents and Computer on the desktop, other icons inside the desktop Computer icon (Jeff) □ Put nothing on the desktop and put the icons under the Computer menu instead (Mark) Each of these has different merits, however I don't think we're going to reach a good consensus just yet and more importantly... WE ARE RELEASING IN TWO WEEKS! This stuff requires serious user testing, actually sitting users of each of the different classes down with each of the setups and working out what they react well and badly to. I therefore think we should consider this modification a feature, just like usplash etc. We had to cut that for warty (sadly, I was getting quite excited about it) because it wouldn't receive enough testing -- likewise I think we need to cut this for warty because it won't receive enough testing. Sebastien Bacher agreed saying, "I totally agree with that. We have only 10 days before the 2.8.0 tarballs release and still a lot to do (as mentioned yesterday in my mail on the list), that's not a good time to make big changes." Jeff Waugh suggested that, " I've suggested this many times with both my desktop team lead hat and release manager hat on. Let's not go there. The toughest issue now is not what's going to be on the desktop (that's the easy bit), it's making sure that the very new Computer menu and its side-effects are a) not lame and b) bug-free."  8. mdadm Daemon 2004/09/01 - 2004/09/02 (5 posts) Subject: "mdadm daemon -> off by default?" People: Jeff Waugh, Matt Zimmerman Jeff Waugh was unsatisfied with the current status of the mdadm daemon and posted the following summary and suggestion: So, we have established that mdadm runs only to provide notification of RAID b0rkage, which is not necessary in the majority of cases where Ubuntu will be installed. But it is good to have mdadm in base by default. So, here are two potential solutions: 1. change the mdadm default to not start the daemon, users can fix this up if they want it running 2. check /proc/mdstat to see if any raid devices exist, and only run the daemon if they do The first option is an easy way out. The second could introduce bugs, but it is a pretty simple format to check. I'm tempted to do the second. Let me know if there's a better way, or if this is an unwise change to make now: https://bugzilla.no-name-yet.com/show_bug.cgi?id=966 Matt Zimmerman replied saying, "I'm OK with doing the second, but I am unsure that it will receive sufficient testing, considering the relative rarity of RAID configurations. If it doesn't work correctly, we (or the user) have to fix or bypass the test."  9. Bugzilla 2004/09/04 (2 posts) Subject: "Bugzilla update Saturday 6pm UTC" People: David Miller David Miller made the following mini-announcement: "Just a heads up, I'll be updating https://bugzilla.no-name-yet.com/ around 6pm UTC today (Saturday) to pick up fixes from upstream for problems we've run into and a few of the requested features that have been implemented upstream. The downtime will most likely be 30 seconds or less, but I'm warning you all so you know what's going on in the off chance that anything goes wrong." If you do run notice anything wrong, please find Dave in IRC.       We Hope You Enjoy Ubuntu Traffic  Ubuntu Traffic is created and produced by Canonical Ltd. All pages are copyright Canonical.