[03:23] cjwatson, I hate to bother you.. but I think I know what happened [03:25] I believe my data was encrypted, I'm looking up how to fix this atm [09:34] cjwatson: explaining the situation to mpt now [09:34] thanks [09:34] is it a simple adjustment or is there contention? [09:35] well, we're trying to phrase it without being too verbose [09:36] right [09:36] cjwatson, I've changed "No files will be deleted." to "Documents, music, and other personal files will be kept." [09:36] I don't want to overwhelm people with words either [09:36] Beyond that, whether to show the upgrade option at all in 11.04 is up to you. [09:36] cjwatson: http://docs.google.com/View?id=dfkkjjcj_101gnkrpg5v#4_5_1_Automatic_partitioning_o_8475526086986065 could you look that over and let us know if there are any lingering issues? [09:36] mpt: OK - can we do anything about the "upgrade" word? [09:37] mm [09:37] so mpt has left that up to us, and I'm not entirely convinced changing it to reinstall it is the best path forward [09:38] I'm not sure about "reinstall" either - that's too far in the other direction [09:38] it's also used elsewhere there [09:38] cjwatson, I don't see how it could possibly then be understandably distinct from "Erase Ubuntu and reinstall" [09:38] "Erase everything and reinstall Ubuntu [maps to use_device]" for example [09:39] whoops, not trying to gang up [09:39] no that's fine, it's a valid point [09:39] I'm just trying to make it distinct from "upgrade" as used elsewhere [09:39] (in Ubuntu) [09:39] sure, I understand your concerns there [09:40] It is an upgrade, it just has differences from the internal upgrade process if you have server software installed. [09:40] (Albeit those are important differences if you do.) [09:40] if we describe it as an upgrade, users will turn up saying they upgraded, and the behaviour won't be as we expect for upgrades [09:40] I asked ev what software this would affect. Web servers, database servers, what else? [09:41] it's impossible to narrow it down to particular kinds of software [09:41] for example I've seen Linux games that you downloaded as a tarball and unpacked into /usr/local [09:41] Perhaps our best option is to make the description a bit verbose "Installed software will be kept where possible, system settings will not."? [09:41] we keep /usr/local [09:41] you designed that bit :-P [09:41] heh, gotcha [09:43] cjwatson: I believe mpt meant what are the common cases in which this would burn you? What software would you have installed that you'd be making conf file changes to. [09:43] right, well the thing that matters is whether dpkg-repack will pick it up or not [09:43] presumably because that's who we need to message this to [09:44] I think we should just assume dpkg-repack will not be invoked for the purposes of explaining all this [09:44] I know we can't have that in the explanation, but we need to think about it [09:44] any case where you create a new file in /etc (or change a configuration file that isn't a conffile?) will not be preserved [09:45] correct? [09:45] any time you change a system level setting to a package period, unless that package is unavailable and thus repacked with dpkg-repack [09:45] in which caseu I believe your above assessment is correct [09:46] aha, I missed that subtlety [09:46] so anything in the System menu that requires unlocking, basically [09:46] (lp:apt-clone if it's easier to follow along in code) [09:46] I think so [09:46] "system settings" seems a reasonable way to put that, then [09:46] just trying to think of the things I could possibly be missing here :) [09:47] ideally I'd like a new word that's a synonym for "upgrade" but that means that users will show up saying they did that rather than that they upgraded [09:47] (and that distinguishes it for more clueful users) [09:47] that's a really good point that I hadn't considered [09:48] and was demonstrated in that exchange from last night [09:48] I take mpt's point that it is an upgrade, in much the same sense as e.g. a Windows upgrade is [09:48] (mpt is radio silent because he's helping Ivanka) [09:49] although it's worth noting that video that somebody posted recently where they did a load of Windows upgrades and noted what kinds of configuration it kept [09:49] (did you see that?) [09:49] http://techland.time.com/2011/03/03/video-starts-upgrade-process-at-windows-1-0-ends-at-windows-7/ [09:50] I haven't actually watched it yet -- I always feel a bit weird about going on Youtube while in the office, but I had heard about it, yes [09:50] actually, better, http://rasteri.blogspot.com/2011/03/chain-of-fools-upgrading-through-every.html [09:55] yay monkey island [09:56] the rest of that section of the UX spec looks largely OK, although I'd point out that in the case where the same version of Ubuntu is already installed, we have the same problem that mpt raised above - "Reinstall Ubuntu {same version}" is not very distinct from "Erase Ubuntu and reinstall", unless you read the small print [09:56] failing to think of appropriate not-hopelessly-jargon synonyms for upgrade [09:58] system settings> BTW, I think it's OK if that doesn't cover all the bases, because people who've installed non-trivial software will likely read "system settings" and think "hm, that's vague and I'm not sure whether it will nuke my configuration, better keep a backup" (slightly optimistic, but I think it guides them in that direction) [10:11] agreed [10:11] mpt: when you get a chance, what are your thoughts on expanding the description as mentioned at 9:41? [10:12] hm [10:13] An unfortunate collision there is that as of 11.04 we're starting to update "System Settings" as the name for non-application settings in general, whether personal or system-wide [10:13] s/update/use/ [10:13] I guess "system-wide" settings would work [10:13] (and so the syllable count creeps ever upwards) [10:13] :) [10:13] lucky you don't speak French :-) [10:14] (but yeah, that works for me) [10:14] maybe it's better finessed in other languages. Perhaps the germans have a single word for all of this ;) [10:14] having failed to think of any suitable synonyms, I'm happy to shelve the "upgrade" bit if we fix everything else to be clear [10:14] okay [10:14] though we need to warn the rest of the development community [10:14] it's going to affect their bug triaging [10:16] indeed. Do you think a post to ubuntu-devel-announce will be enough? [10:17] there may be some documentation that needs to be updated, but I don't know where it is ... -devel-announce is a good start [10:18] I wonder why that user lost /home though - he said he thought he knew what the problem was and that his data was encrypted, so maybe we're failing to copy ecryptfs configuration across or something? [10:18] I thought we handled that... [10:18] so did I [10:18] we most certainly force them back into encrypt home if they selected it before [10:18] though I wonder [10:19] what if they pick a different password? [10:19] I have a note that the UID preservation is probably broken with the move to parallel steps [10:19] aha [10:19] perhaps that did it? [10:19] though for single-user systems it'll be 1000 either way [10:19] true [10:19] I mean, chances are [10:27] cjwatson: "Documents, music, and other personal files will be kept. Installed software will be kept where possible. System-wide settings will be cleared." [10:27] +1 [10:33] sounds good [10:43] http://paste.ubuntu.com/590149/ - how's that? [10:43] speaking of documentation, I really need to find some time to learn restructured text and sphinx [10:45] so we can generate pretty (and hopefully comprehensive) developer documentation for ubiquity like we do with d-i [10:45] for example: http://www.voidspace.org.uk/python/mock/ [10:59] 590149> looks good [11:07] cool, thanks [11:14] ubiquity: evand * r4645 trunk/debian/ (changelog ubiquity.templates): [11:14] ubiquity: Provide a better description for the upgrade/reinstall option that [11:14] ubiquity: intends to make the difference between it and a traditional Ubuntu [11:14] ubiquity: upgrade clear. [11:36] ubiquity: cjwatson * r4646 trunk/debian/ (changelog source/options): [11:36] ubiquity: Switch to xz compression for the source package, saving nearly a [11:36] ubiquity: megabyte. [11:39] hmm, does the removal of the live seed packages also remove the recommends they pulled in ? [11:39] ubiquity: evand * r4647 trunk/debian/changelog: Add LP bug reference. [11:41] ogra_: yes, it should prune everything back to the desktop seed, basically [11:42] it works based on differences between manifests [11:42] hrm, k [11:42] then i'll probably move the TI ppa handling back into jasper [11:44] having a package doesnt seem to make much sense here [12:04] cjwatson: aquarius just mentioned to me an idea of installing to /Ubuntu_$VER to make multiple versions easy, with the real intent being able to try an upgrade while still being able to go back [12:04] so I mentioned this was the intent of some of our work with btrfs snapshots [12:05] right [12:05] which got me thinking, given that you can seemingly mount snapshots without rolling back to them, could we muck about with the grub config and initramfs such that you could boot into one as well? [12:05] I'm not sure enough about the semantics of btrfs snapshots [12:05] it should be possible [12:05] yeah, me either [12:06] on the initramfs side it's just rootflags=subvol=@whatever [12:06] I get hazy in what happens to the data from that point [12:06] oh nice [12:06] that is, does it always start from the snapshot, or do any changes you make get carried into that snapshot path [12:07] presumably breaking the ability to roll back to it [12:07] grub.cfg would need you to construct entries rooted at /@whatever rather than /@, I think [12:07] right, that's the sort of thing I have no idea about [12:07] whether they're tags or branches, which ones are significant, etc. [12:08] I suspect that in part btrfs just provides some tools and it may be up to us to layer meaning on top, but not sure [12:08] across dist-upgrades, home directory data is going to start diverging too - think firefox profiles, for example [12:09] so that gets tricky for full rollbacks, might need to snapshot both in theory and offer various possibilities [12:11] I wonder if you could snapshot both and only present the user the one where they can muck about, but keep the other around for being able to roll back [12:11] emm, that wouldn't be ideal [12:11] yeah [12:11] tricky [12:12] worthy of a UDS discussion? [12:13] yeah, probably [12:15] I'll add it to the wiki then, thanks [12:16] (and will endeavor to learn more about this particular bit of machinery before then) [12:53] * cjwatson wonders how entertainingly http://paste.ubuntu.com/590197/ is going to fail [12:54] gah [12:55] will need to be paired with a Wubi change [13:14] ev: would you object to something like http://paste.ubuntu.com/590207/, to make ubiquity/nonfree_package preseedable the way the documentation says it is? [13:14] it would make it possible to preseed "install restricted drivers, but not restricted host software" [13:14] e.g. "Colin is doing Wubi tests and wants his wireless to work each time" [13:15] (OK, that's a crappy use case, but YKWIM :-) ) === kdoandal is now known as nevrax [14:00] sorry, was out at lunch [14:00] looking now [14:01] lol [14:01] looks okay to me [15:42] mterry: reviewing your branches now. Just trying to get your changes to work locally. [15:58] seems to not like client.SetSortMode [16:11] setting the hostname in a preseeded and networked installation of natty with the alternate image doesn't seem to work with netcfg/get_hostname whereas it seems to work with the desktop image for a networked install over nfs, might it be possible that something changed? [16:11] for some reason, the hostname seems to be set to "ubuntu" rather than the string provided to d-i [16:12] changed? bug 218965 [16:13] cjwatson: heh, we've been using desktop images so much more frequently that I haven't noticed so much the side effect with alternate images [16:14] cjwatson: mathiaz mentionned remastering the iso, does that mean I could potentially set a kernel parameter to set the hostname instead of the preseed? [16:15] the bug log remembers more than I do right now [16:16] cjwatson: no worries, thanks for the pointer [16:26] ev: could you build wubi r207? [16:26] on it now [16:27] * cjwatson fires up a scary lupin upload [16:31] cr3: I saw that hostname="ubuntu" thing about a week ago. network install, hostname supplied by dhcp, I thought it was because I had a dash, (test-1) but when I tried to track it down it went away. [16:32] CarlFK: hostname supplied by dhcp is a very different code path from what cr3 is doing [16:32] the bug I referenced is pretty specific [16:35] ah - known issue. never mind. [17:07] cjwatson: wubi is up [17:12] thanks [17:45] ev, :( about sphinx [17:45] still trying to get it going [17:45] is it working for you? [17:45] ev, I must have misused the python api. If I sort on command line it works [17:45] or was this a cowboy? [17:45] But I couldn't get python api to connect to my server at all, so I didn't get far enough to test tht [17:45] or the python api is broken [17:46] * mterry isn't even sure he's actually running a sphinx server vs local instance [17:46] ev, it was sorta-cowboy [17:46] ev, I looked at the python api code to make sure I got the arguments right... and it matched what I was doing on console [17:46] but I must have screwed something up [17:47] from what I can tell you seem to be matching the docs [17:50] * mterry tries to get my sphinx server up [17:55] ah ha [17:55] client._error helps [17:55] sort-by attribute 'population' not found [17:58] ev, I just retested locally and it worked in python [17:58] did you re-index? [18:00] doh. [18:00] sorry [18:00] doing that now [18:01] success! [18:01] thanks mterry [18:02] sweet [18:02] that branch really improves the quality of the results [18:06] and I've merged the other one now as well [18:06] yay [18:06] thanks a bunch for doing that [18:08] mterry: do you have other things in the pipe or should I pester IS to update from that branch? [18:19] ev, I have something in the pipe for localization, not sure when. I think end of week at worst [18:19] okay, I'll hold off then [18:19] best to batch these things up, methinks [18:55] I'll have to unearth the Windows VM from my NAS tonight to give this a try - http://paste.ubuntu.com/590378/ [18:56] hmm, actually that's going to need to be a bit more clever, given Windows' insane locking [18:56] bum [19:24] ubiquity: evand * r4648 trunk/ (3 files in 2 dirs): [19:24] ubiquity: Don't show the biggest_free option if we can't fit Ubuntu in the [19:24] ubiquity: free space (LP: #751145).