[09:59] right in the center next to the photo camera [10:08] keybuk: o hai. My VM setup: http://heh.fi/tmp/mountall-test-setup-20091009, debug log with my branch: http://pastebin.com/f175ac4af [10:09] Nothing special about them, except that it seems to work. :-P [10:50] ion: all merged, except I had a couple of alternate ways of doing it [10:50] rather than the trigger thing, I changed the count to never call mounted() more than once [10:50] Alright [11:56] \o. [11:56] .o/ [11:59] keybuk: I wonder how to handle e.g. the ‘apt-local /opt/apt-local vboxsf defaults 0 2’ mount in my VM fstab that gives me easy access to, say, a recent build of mountall deb? The vboxsf filesystem becomes available when the virtualbox-ose-guest-utils init script loads a module. Currently i just have it as ‘noauto’, so that the system at least boots. :-P [12:03] why doesn't it boot without noauto? [12:10] It gets stuck at the end, never sending the FHS event: http://pastebin.com/f775b90db (the last “mounted:” entry being “local 6/7 remote 0/0 virtual 12/12 swap 0/0”) [12:13] (I’m still running a build from my branch pre-merge. I’m about to do a new build, but that shouldn’t fix it.) [12:27] oh, hmm [12:27] ah, I see the bug [12:27] d'oh [12:28] ion: try -r85 [12:35] About to impact http://www.nasa.gov/55644main_NASATV_Windows.asx [12:36] keybuk: Great, it works. [12:38] it makes you wonder whether Obama has a "sorry, we blew up the moon" speech prepared [12:38] just in case [12:39] We launched a rocket toward the moon with an intention to detonate an explosive, but upon impact it accidentally the whole moon. [12:41] i think it will be revealed that blowing up the moon was, in fact, necessary, as an alien invasion was being staged there, and that that is, in fact, the reason why obama is getting the nobel peace prize [12:48] keybuk: Can we get events from Upstart about things being mounted by others? The vbox-guest-utils init script runs mount -a -t vboxsf after loading the module, but mountall is left to sleep forever thinking /opt/apt-local never got mounted. [12:48] s/Upstart/udev/, brainfart [12:48] oh, I was going to poll mountinfo! [12:54] ion: how do you watch it on Linux? LCROSS [12:58] vlc played both the stream i linked and http://playlist.yahoo.com/makeplaylist.dll?id=1368163 which had better quality but which suffered from congestion. [13:12] it seems to be the same thing.. that .asx link is just a xml file and it has a ref to http://playlist.yahoo.com/makeplaylist.dll?id=1369080&segment=149773 [13:48] keybuk: http://pastebin.com/m403ae9cd btw ;-) [13:56] keybuk: How about making nih_logger_printf() fflush the fd after each message? It would help with e.g. stracing mountall; you’d know which part of the code is running based on the debug write()s. Also, the latest log messages might only reach a log file after a long delay if the output has been redirected to a file, mountall is waiting for something and the buffer doesn’t quite reach its capacity. [13:57] fflush is expensive [13:57] I encountered both situations with debugging mountall. I simply added the flushing to nih_logger_printf temporarily. :-P [13:57] if (nih_log_priority >= NIH_LOG_DEBUG)? [13:57] <= [13:57] ion: tabstop-agnostic indentation - give me an emacs file for that ;) [13:58] Since my editor doesn’t do that either (AFAIK), i just indent with spaces in my own code. :-P [13:59] style < effort :p [13:59] and I use C-A-\ all the time [13:59] (reindent buffer) [14:00] TBH, i’d indent my code with spaces only anyway, i rather like a smaller amount of indentation. [14:04] :p === robbiew-afk is now known as robbiew [14:21] my JLS confirmation just arrived :) [14:21] In Japanese :( [14:57] fail [14:58] 敗 even [15:04] Keybuk: from what translating my literate friend has done, I need a copy of the email to get into the conference too. That's gonna be fun for some people. [15:04] Keybuk: especially those people who's spam filters have figured out the incredibly effective "foreign language => junk" heuristic [15:07] JLS? [15:07] ion: Japan Linux Symposium [15:09] “• Upstart 1.0 by Scott James Remnant [slides]” It’s a placeholder for a link. :-( [15:10] ion: where is that? [15:10] http://linuxplumbersconf.org/2009/program/ [15:11] yeah, I wonder when the videos will be available [15:13] ion: my laptop hard drive died [15:13] so I haven't rescued the slides off it yet [15:14] Ouch [15:18] Keybuk: mine showed up broken and yours didn't make it out [16:31] keybuk: As for “don’t queue filesystem check when device is ‘none’”, how about “if device doesn’t begin with a slash” instead? [16:41] you might have a legitimate reason for checking those [16:41] *shrug* [16:41] fsck.fuse may exist for example [16:51] keybuk: http://bazaar.launchpad.net/~ion/ubuntu/karmic/mountall/fsck-locking/revision/88 [16:51] no, I meant mnt->type there [16:52] Ah, ok, the debian/changelog entry mislead me. :-) [16:52] you can add another check though ;) [16:55] http://bazaar.launchpad.net/~ion/ubuntu/karmic/mountall/fsck-locking/revision/88 [16:55] you'll need to rebase that too ;) [16:58] pushed [16:59] Or did you already publish 0.2.0, i.e. should i increment version? [17:06] i did [17:30] Wouldn’t it be better for mount_event to be asychronous? Send the call, go back to main loop and proceed to run_mount/swapon when the call finishes. [17:45] maybe [17:45] but the idea is to just start essential services for that mount === notting_ is now known as notting === robbiew is now known as robbiew-afk