[10:50] <davmor2> cjwatson: the message about iana will that effect automated installs?
[10:51] <cjwatson> sorry, no idea what you're talking about :)
[10:52] <davmor2> on alternate there is now a popup message from dns iirc saying it needs to download the iana info which can't be shipped
[10:52] <davmor2> you need to say yes or no to the script running
[10:52] <davmor2> I'm just wondering if that will affect automated installs
[10:52] <cjwatson> soren: sounds excellent and it's not unlike the 'digress' framework that Debian use to do installer testing, at least in spirit (digress operates by serial console installation and expect, or similar). There are a number of things that one can't test with preseeding, the manual partitioner chief among them, so I'd definitely support the usefulness of this approach!
[10:53] <cjwatson> davmor2: yes, I would expect so; they would have to preseed it away
[10:54] <soren> cjwatson: Excellent. I'm glad you approve.
[10:54] <cjwatson> davmor2: don't suppose you already have a DEBCONF_DEBUG=developer log of this happening?
[10:54] <davmor2> cjwatson: okay ta
[12:02] <michaelforrest> cjwatson: I understand we want to add a stage to the installer where proprietary drivers are approved
[12:02] <michaelforrest> cjwatson: I was wondering if we could use the same stage to let users opt-in to get mp3s and Flash?
[12:03] <cjwatson> hmm, that's definitely strictly more difficult since we can't put that software on the CD
[12:04] <michaelforrest> I was thinking more that it would be a script that would automatically install those from the network on first login or something
[12:04] <cjwatson> that's kind of unpleasant :(
[12:04] <michaelforrest> it's more unpleasant to go to youtube and not be able to watch any videos
[12:04] <michaelforrest> or that warning you get when you want to install mp3 support...
[12:04] <cjwatson> we have to include that warning
[12:04] <cjwatson> or something with equivalent meaning
[12:05] <michaelforrest> sure -
[12:05] <michaelforrest> absolutely
[12:05] <michaelforrest> but if we could make it something during install
[12:05] <michaelforrest> I would like that a lot
[12:05] <michaelforrest> and so would 99% of users.
[12:05] <cjwatson> hmm, I don't know. I'm not immediately keen on it because it makes it less likely that we'll be able to get the whole thing done - jockey already exists and it's probably fairly straightforward to just drop it in
[12:05] <michaelforrest> (non-scientific estimate)
[12:06] <cjwatson> I'd like to make it a separate development task rather than piggybacking
[12:06] <michaelforrest> I don't know anything about Jockey
[12:06] <michaelforrest> (googling now)
[12:06] <cjwatson> jockey = System -> Administration -> Hardware Drivers
[12:07] <cjwatson> I understand the request - I can see how it makes some kind of sense to have all the legally nasty stuff in one place
[12:07] <michaelforrest> ok so technically they're separate things, but conceptually they're not too far apart
[12:07] <michaelforrest> yeah
[12:07] <cjwatson> just worried about the implementation
[12:07] <michaelforrest> sure
[12:08] <michaelforrest> I think it would be a worthwhile pain-in-the-ass though.
[12:08] <cjwatson> I think we should avoid first-login work because that has a non-intuitive kind of effect on boot performance
[12:08] <michaelforrest> ok
[12:08] <cjwatson> for lucid, I'd honestly rather offer it but say "only works if you have networking" or something
[12:09] <michaelforrest> yeah I was about to say
[12:09] <cjwatson> will probably be a pain in the arse for broadcom
[12:09] <michaelforrest> I'm terrified of the complexity involved in reliably establishing network..
[12:09] <cjwatson> (boot performance: reason for this is that we profile the first boot)
[12:10] <cjwatson> perhaps we could have a staged design - here's what the page looks like for hardware drivers only, here's what it looks like with proprietary software as well
[12:11] <cjwatson> oh and obviously the whole page vanishes if you say "free software only" at the bootloader
[12:11]  * ev is lost - how did we get from installing codecs as part of ubiquity to a second stage installation breaking boot profiling?
[12:11] <michaelforrest> what if I want to make the 'use all the proprietary stuff' the recommended option :)
[12:11] <cjwatson> ev: 12:04 <michaelforrest> I was thinking more that it would be a script that would automatically install those from the network on first login or something
[12:11] <ev> ahh
[12:12] <cjwatson> michaelforrest: I suspect your boss' boss might object
[12:12] <ev> sorry, I missed that
[12:12] <michaelforrest> I think we can bring him round
[12:12] <cjwatson> we keep proprietary software out of main for a reason
[12:12] <cjwatson> and we've committed to that at the highest level
[12:13] <cjwatson> this has pretty serious implications on how Ubuntu is perceived in the community
[12:13] <michaelforrest> yeah but in the world-at-large people want Flash, MP3s and Skype.
[12:13] <cjwatson> making it easy but not doing it by default is a different matter
[12:13] <michaelforrest> nevertheless, this is not an #ubuntu-installer discussion :)
[12:13] <cjwatson> we can only conquer the world at large if we keep our community
[12:13] <cjwatson> it's not an either-or kind of thing
[12:13] <michaelforrest> We will find a way.
[12:14] <cjwatson> we've settled on the path of making it straightforward to get at proprietary software when you need it, but not installing it by default
[12:14] <cjwatson> this is a compromise, which does tend to mean that neither side is totally happy ;-)
[12:15] <ev> so, lucid+1 *tentatively* add the ability, but not as the default selection, to install flash, mp3, skype, etc in the installer, much like we're going to do with jockey in lucid?
[12:15] <ev> perhaps integrating them into the same UI, but to be determined after design consideration
[12:15] <cjwatson> MP3 is likely to get easier in the near future, BTW
[12:15] <ev> oh?
[12:15] <cjwatson> the patents are expiring pretty soon
[12:15] <ev> hooray
[12:16] <cjwatson> they already have in some jurisdictions, I believe
[12:16] <michaelforrest> maybe we should make YouTube work by showing the h.264 videos that Apple made them sort out so it works on iPhones...
[12:16] <michaelforrest> how soon is pretty soon?
[12:16] <cjwatson> hmm, we discussed h.264 in the TB a while back, I can't remember the exact outcome
[12:17] <cjwatson> http://en.wikipedia.org/wiki/MP3#Licensing_and_patent_issues
[12:17] <ev> December 2012, according to wikipedia.
[12:17] <cjwatson> "may", so we would need to lawyer-check that
[12:17] <michaelforrest> still important to do something in the meantime, I think.
[12:18] <cjwatson> sure
[12:18] <ev> sure, I imagine we'll always have this issue of non-free stuff that people want and we can technically distribute
[12:18] <michaelforrest> indeed
[12:18] <michaelforrest> subject to the whims of fashion.
[12:18] <ev> haha
[12:18] <cjwatson> I would add that our historical understanding has been that there's a difference in liability between just distributing things on our servers and actively going out and offering them to people
[12:18] <cjwatson> it's all a bit vague, but this has to be run past legal
[12:19] <cjwatson> flash is probably ok
[12:19] <cjwatson> it's merely a licensing nightmare rather than a patent nightmare
[12:20] <cjwatson> if we really wanted to fix that we would fund gnash :P
[12:22] <ev> I wanted to like gnash, but I fear it will never catch up.  Adobe is hardly static in its development of Flash.
[12:22] <cjwatson> as long as things are proprietary we're always going to be at a fundamental disadvantage
[12:23] <cjwatson> the things that are currently fashionable I mean
[12:25] <ev> indeed, I'm not disagreeing, just being pessimistic about reaching that goal with flash anytime in the near future.
[12:25] <cjwatson> yeah
[13:30] <ev> shtylman: would you mind putting your scripts for nfs mounting ubiquity trunk inside a VM somewhere public?  I'm keen to see how you handled things like ubiquity/components/ubi-*
[13:40] <shtylman> ev: no probs...will do that tonight when I get home from work
[13:40] <ev> coolness
[14:33] <ev> michaelforrest: a new CD is available at <http://cdimage.ubuntu.com/daily-live/20091209/>.  It has the networking fix and seems to have the graphical changes you were after.
[14:34] <michaelforrest> thanks ev
[17:08] <CIA-15> grub-installer: cjwatson * r828 ubuntu/ (debian/changelog grub-installer): GRUB 2 now supports installation on SATA RAID and multipath.
[17:10] <michaelforrest> sooooooooo…. we do ship proprietary drivers on the CD?
[17:10] <cjwatson> some of them, can't ship all
[17:10] <michaelforrest> is there anything documented about the permission being requested as part of the install process?
[17:11] <cjwatson> could you rephrase that?
[17:11] <michaelforrest> um.. evan said something about a discussion about asking people if they wanted to use these drivers during installation
[17:12] <cjwatson> ev: oh, pretty much the same as what System -> Administration -> Hardware Drivers does, would you say?
[17:12] <michaelforrest> in my experience, I was confused that my wireless card worked on the live cd but not on the install
[17:12] <cjwatson> the kernel team will want us to ask because in general they can't effectively support proprietary drivers
[17:12] <michaelforrest> so is there anything written down about that anywhere?
[17:12] <michaelforrest> so I can learn more
[17:13] <michaelforrest> and stick a step into the design
[17:13] <cjwatson> I would copy-and-paste the text from Hardware Drivers to you but I can't because it isn't selectable ;-)
[17:14] <michaelforrest> I will have to use my touch-typing skills :P
[17:14] <cjwatson> "Proprietary drivers do not have public source code that Ubuntu developers are free to modify. Security updates and corrections depend solely on the responsiveness of the manufacturer. Ubuntu cannot fix or improve these drivers."
[17:14] <michaelforrest> hmm.
[17:14] <michaelforrest> "But if you want your wireless card to work then this is the only option" ??
[17:15] <cjwatson> how true that is varies between drivers
[17:15] <cjwatson> don't generalise too far from your own experience :)
[17:15] <michaelforrest> 'Ubuntu' is an entity that fixes drivers?
[17:15] <cjwatson> the Ubuntu kernel team does that kind of work
[17:15] <michaelforrest> strange phrasing that's all.
[17:15] <cjwatson> in the case of video drivers, the proprietary driver might work better for some use cases
[17:15] <cjwatson> I didn't write that text
[17:16] <cjwatson> better> or it might totally break
[17:16] <cjwatson> we just have no reasonable way to tell, across the board
[17:16] <michaelforrest> I am just trying to understand what a user is expected to do with that information!
[17:16] <michaelforrest> Is it relevant if there is not a choice?
[17:17] <cjwatson> in a number of environments, people might rather have a supported system that can only do wired networking, than have a dodgy wireless driver that isn't supported properly
[17:18] <cjwatson> it's in the nature of proprietary drivers that they sometimes fail to keep up with kernel changes, and sometimes the result of using one can be worse than having no driver at all
[17:18] <cjwatson> the difficulty here is that we're trying to present an inherently bad situation
[17:19] <cjwatson> but one that's relevant to a pretty large number of users
[17:19] <michaelforrest> I would have thought that open-source drivers would be equally prone to lagging behind the latest kernel changes… is that wrong?
[17:19] <cjwatson> not so
[17:19] <cjwatson> open-source drivers are maintained in the kernel tree itself, and when kernel maintainers make changes that require driver changes, they tend to upgrade the whole tree in step
[17:19] <cjwatson> out-of-tree drivers have to catch up for themselves
[17:20] <michaelforrest> ok
[17:21] <michaelforrest> I am thinking about this phenomenon http://snowleopard.wikidot.com/
[17:22] <michaelforrest> listings of third-party drivers
[17:22] <michaelforrest> I know.
[17:22] <michaelforrest> Can we say 'third party' instead of 'proprietary' instead do you think?
[17:22] <cjwatson> that's an application checker, isn't it?
[17:22] <cjwatson> at least primarily so
[17:22] <michaelforrest> oh
[17:22] <michaelforrest> I was looking for the one about hardware drivers
[17:22] <michaelforrest> didn't really read it
[17:23] <michaelforrest> but you get the idea
[17:23] <cjwatson> applications are a different kettle of fish, because the interface between the kernel and applications is much more rigidly defined
[17:23] <michaelforrest> yeah let's not talk about applications
[17:24] <cjwatson> the fact that we don't have the source code is relevant - it means that nobody other than the manufacturer is really capable of offering support. There are a couple of cases where multiple drivers are offered, and savvy users may well want to make a distinction on that basis
[17:25] <cjwatson> for example, in the case of one of the two Broadcom drivers, we're able to offer the source to the driver (and so it's a lot easier to keep that part in step, so it's less likely to cause random kernel crashes and the like) but we can't ship the firmware
[17:25] <michaelforrest> it feels like an upgrade-related issue more than an installation issue
[17:26] <michaelforrest> can ubuntu reliably drop back to open-source drivers if the proprietary ones don't work after upgrade? or will things most likely just go mental?
[17:26] <ev> sorry, was on the phone
[17:26] <ev> reading the scrollback now
[17:26] <cjwatson> not completely reliably and automatically
[17:26] <cjwatson> if you can manage to get to the Hardware Drivers application, you can deactivate the broken thing there
[17:27] <cjwatson> I don't really agree that it's just an upgrade issue though, as we often have to be in the position where the proprietary drivers we ship don't work on some hardware
[17:28] <michaelforrest> if there was a way to give the user enough information to make an informed choice it would all make a lot more sense
[17:29] <michaelforrest> I don't feel that saying 'this might not work.. not our problem' is adequate from a user-perspective.
[17:29] <cjwatson> would be easier if we had reliable access to the network :)
[17:29] <cjwatson> this entire problem does not exist in an ideal world
[17:29] <cjwatson> this is, unfortunately, entirely best-effort territory :(
[17:32] <cjwatson> "not our problem" is probably a bit too strong; the idea is more "degraded service"
[17:32] <michaelforrest> Inaccessible third-party drivers are such a fact of life in the Windows and Mac world I wonder why we'd even comment on it!
[17:32] <cjwatson> the Windows installer doesn't install inaccessible third-party drivers for you
[17:32] <michaelforrest> That is true.
[17:33] <cjwatson> your OEM might, but in that case they have QAed them
[17:33] <michaelforrest> And the Mac installer knows what it's gonna get.
[17:33] <michaelforrest> More-or-less.
[17:33] <cjwatson> also nobody expects bug reports on Windows to have any effect :)
[17:33] <cjwatson> (to a first approximation)
[17:33] <michaelforrest> "Ubuntu has found third-party drivers for your hardware - do you want to use them?"
[17:33] <michaelforrest> hmm.
[17:34] <michaelforrest> ("yeah I found em all right, I found 'em on a cd! cackacakcakcle")
[17:37] <cjwatson> the problem here is that there are reasons to use these drivers (sometimes excellent, e.g. no wireless without them) and there are reasons not to use these drivers (sometimes excellent, e.g. they'll crash your computer). If I knew how to present that kind of choice I'd be suggesting something already :(
[17:38] <cjwatson> I do think that we should not be shy about Ubuntu's selling points with respect to the things we ship ourselves that are open source
[18:34] <michaelforrest> cjwatson: looking at the daily build now. logo looks good on my mac's screen and external monitor.
[18:35] <michaelforrest> I'd like to switch the colours though - orange for the selected and a light grey for the unselected options
[18:35] <michaelforrest> can I give you some colour values?
[20:27] <jdoelger> Hi, is there any way to do an expert install without a network connection?
[21:56] <cjwatson> michaelforrest: sure
[21:56] <cjwatson> jdoelger: certainly, expert mode is entirely unrelated to whether you have a network connection or not - have you tried selecting it from the F6 menu at the CD boot loader?
[22:04] <jdoelger> cjwatson: when i used the alternate install disk, i just hit f6 and select expert mode -- when it gets to detecting network, it fails to recognize the card and won't let me continue with the installation
[22:05] <cjwatson> expert mode shouldn't make a difference to that - I'd guess it would have failed in normal mode too?
[22:05] <jdoelger> the desktop version sees it and uses the atl1c module
[22:05] <cjwatson> expert mode in the alternate installer just causes some more questions to be shown
[22:05] <cjwatson> ok, so that's a bug
[22:05] <cjwatson> almost certainly in the kernel, believe it or not
[22:06] <cjwatson> which release?
[22:06] <jdoelger> karmic
[22:06] <cjwatson> so I'm afraid the alternate installer will be broken for you, but if you file a bug on the 'linux' source package in Ubuntu and tell me the bug number, we'll get that fixed for lucid
[22:07] <cjwatson> http://paste.ubuntu.com/338319/ is the patch
[22:08] <jdoelger> heh, well that's fun.  from the look of that the installer is just missing the atl1c module?
[22:10] <jdoelger> is there a way I could copy that module to a flashdrive, then plug that in when the installer asks if i have anything with additional drivers on it?
[22:13] <cjwatson> you can certainly try - it's just the module out of the regular Ubuntu kernel
[22:14] <cjwatson> unless you fancy building udebs for yourself, though, it might be easiest to simply copy it into the filesystem from a shell prompt
[22:14] <cjwatson> (when the installer complains)
[22:15] <cjwatson> I'm surprised it won't let you continue, though
[22:15] <cjwatson> are you installing from the network, or from a CD?
[22:15] <jdoelger> so basically i can copy it to /lib/modules/*kernel*/kernel/drivers/net/atl1c then modprobe it?
[22:15] <jdoelger> from a cd, made into a liveusb
[22:15] <jdoelger> with unetbootin
[22:15] <cjwatson> right
[22:15] <cjwatson> it should be possible to just continue past that message
[22:16] <jdoelger> yeah the screen comes up red, and says it can't continue when i get to the 'select mirrors' part
[22:16] <cjwatson> worst case you get dropped to the main menu and then you have to bypass some steps
[22:16] <cjwatson> choosing a mirror shouldn't be mandatory in the installer image you have
[22:16] <cjwatson> it's got the base system on the CD, after all
[22:16] <jdoelger> that's what i thought!
[22:16] <jdoelger> lol
[22:16] <cjwatson> but you might have to bypass some of the main menu steps
[22:17] <jdoelger> maybe i just didn't try to jump ahead far enough
[22:22] <jdoelger> if i get the module into /lib/modules/*kernel*/kernel/drivers/net/atl1c modprobe should see it though, right?
[22:22] <cjwatson> you might have to run depmod -a as well, but basically yes
[22:23] <jdoelger> okay, good to know.
[22:28] <jdoelger> got it!
[22:30] <davmor2> cjwatson: I lose keyboard on new user in oem mode
[22:30] <jdoelger> thanks so much for your help.
[22:30] <cjwatson> jdoelger: great. let me know that bug number and I'll make sure it gets fixed properly.
[22:31] <cjwatson> davmor2: "lose" in what way?
[22:32] <davmor2> cjwatson: it works in oem, it works on end-user setup, when you get the end user gdm you have no keyboard
[22:32] <cjwatson> literally no response to keypresses?
[22:33] <davmor2> cjwatson: that's the one
[22:33] <cjwatson> I blame pitti, he touched that code latsst
[22:33] <cjwatson> er, last
[22:37] <davmor2> cjwatson: works if I reboot the system into the fresh user
[22:37] <davmor2> so it is only the transfer from oem user to end user
[22:37] <cjwatson> I definitely blame pitti
[22:37] <davmor2> sound like a plan ;)
[22:38] <cjwatson> either his ubi-reload-keyboard script is broken, or udevified X ain't working