/srv/irclogs.ubuntu.com/2011/10/06/#ubuntu-mozillateam.txt

maxbAnyone around who remembers that thing that happened once before in natty with all but one of the Firefox search plugin disappearing because the .xml files had been mislaid in a natty update?11:56
maxbWell, it seems to have regressed in oneiric11:56
maxbhmm, but only on one of the two computers I have upgraded11:59
maxbfirefox -ProfileManager no longer works?12:05
knomehmm, there seems to be a bug in TB12:19
knomecan somebody confirm?12:20
knomestart writing a message, and type in a "attachment keyword" so you get the notification about the issue12:20
knomethen start writing an another message, and do the same, and TB should soon start throwing errors that a script is already in use12:20
=== m_conley_away is now known as m_conley
micahgmaxb: that happened on the upgrade to 7.0 in oneiric, should be fixed in 7.0.115:55
maxbmicahg: Am on 7.0.1, it's still broken15:55
micahgplease don't tell me that a week before release :(15:56
micahgmaxb: did you have any interrupted updates?15:56
maxbmuch to my confusion, I upgraded two natty machines today at the same time, and it's only broken on one of them15:56
maxbyes.... on the machine which it's working on :-)15:56
micahgah, upgrade from naty....15:56
micahgthat would make sense...15:57
micahgwell, no is shouldn't still...15:57
micahg*it, the new script should DTRT15:57
micahgmaxb: what's the status of /usr/lib/firefox-7.0/distribution on the broke machine15:58
maxbSo, on the working machine, I upgraded natty's 6.0.2 to oneiric's 7.0.1;  and on the broken machine I upgraded natty's 7.0.1 to oneiric's 7.0.115:58
maxbI don't have a /usr/lib/firefox-7.0 - I have a -7.0.1, though15:59
micahgyes, sorry, 7.0.115:59
micahgbrb15:59
maxbIt's an empty directory15:59
maxboh, and on the working machine it's a symlink15:59
maxbOK, so I've mv-ed aside the directory and created a symlink like on the working machine, and it's working fine now16:02
maxbUnfortunately, sounds like the migration isn't working properly :-/16:02
micahgright, that's the problem, I"ll need a bug to fix, let me see if I have one16:03
maxblaunchpad is not showing me anything obvious16:04
micahgmaxb: please file a bug, I"ll let the release manager know..thanks16:05
maxbfiling16:05
maxbbug 86931116:14
ubot2Launchpad bug 869311 in firefox "searchplugins installation damaged after natty->oneiric upgrade" [Undecided,New] https://launchpad.net/bugs/86931116:14
micahgmaxb: thanks, we have a fix that I"ll be uploading shortly16:16
debfxis it possible to confine firefox plugins without modifying the firefox apparmor profile?16:24
jdstranddebfx: you mean as in to avoid conffile changes?16:26
debfxjdstrand: yes, as it changes with every upstream version16:26
jdstranddebfx: yes, modify /etc/apparmor.d/local/usr.bin.firefox16:26
debfxjdstrand: the problem is that firefox has the rule "/usr/lib/firefox-7.0.1/** ixr,"16:27
debfxI'm not sure how to override that for /usr/lib/firefox-7.0.1/plugin-container16:27
jdstranddebfx: that's ok, just be more specific. I recommend something like this in local/usr.bin.firefox:16:28
jdstrand/usr/lib/firefox-7.0.1/plugin-container Cx -> plugincontainer,16:28
jdstrandprofile plugincontainer {16:28
jdstrand  ...16:28
jdstrand}16:28
jdstranddebfx: I do something like that for gedit16:29
jdstranddebfx: I'm also curious about what you come up with for plugin_container. when you have something working, can you file a bug on it and assign it to me?16:30
micahghi chrisccoulson, I'm about to upload your fix for the symlink issue since we just got a report one it16:30
debfxjdstrand: that hardcodes the upstream version16:31
* micahg just discovered yet another branch he's not subscribed to16:31
jdstranddebfx: fyi, you will actuall probably want to use globbing in local/usr.bin.firefox16:31
jdstranddebfx: heh, yes :)16:31
jdstrand/usr/lib/firefox-*/plugin-container Cx -> plugincontainer,16:31
debfxjdstrand: does that still work? how does apparmor determine which rule is more specific?16:32
jdstranddebfx: that should work. how it decides is actually pretty complicated and based on DFA and hybrid DFA theory. I'll direct you to #apparmor on OFTC and http://wiki.apparmor.net/index.php/Documentation if you are really interested16:33
chrisccoulsonmicahg, i already talked to pitti about that yesterday, we're going to do a SRU for that rather than squeeze it in to final16:39
micahgchrisccoulson: why?  people upgrade from the live media16:39
chrisccoulsonthey will still get the latest version after the upgrade16:40
chrisccoulsonpitti didn't raise that as an issue, anyway16:40
micahgyes, but they lose their search engines16:40
debfxjdstrand: yeah, sounds complicated :)16:41
debfxjdstrand: the problem with a general plugin-container profile is that it would apply to all firefox plugins16:42
jdstranddebfx: I like to think of it as "the most specific rule wins". it is not necessarily totally accurate, but it gets me through the day :)16:42
micahgchrisccoulson: where was the conversation (I'd like to look at the discussion if possible)16:43
jdstranddebfx: well, I'm still interested with what you come up with. I'd like to be able to lock it down, even if it is opt in16:43
jdstrandmaybe using Px for some things16:44
jdstrandit will be more complicated, but it is something I'd like to have16:44
jdstranddebfx: re most specific> I thinking if the basename is specified, that'll take precedence over a glob16:45
debfxjdstrand: this is the profile I currently use for nspluginwrapper: http://paste.ubuntu.com/703496/16:51
debfxit's very strict though16:51
jdstrandthanks16:52
debfxjdstrand: the apparmor parser complains about the plugin-container rule: "profile has merged rule with conflicting x modifiers"16:59
jdstranddebfx: can you bring that up in #ubuntu-hardened?17:00
debfxjdstrand: yep17:01
=== m_conley is now known as m_conley_away

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!