Conversation with #inferno at Sat Dec  1 21:20:17 2012 on powerman-asdf@irc.freenode.net (irc)
(22:14:35) dreadlorde left the room (quit: Ping timeout: 240 seconds).
(23:20:27) dreadlorde [dreadlorde@c-68-42-82-10.hsd1.mi.comcast.net] entered the room.
(00:42:27) Fish left the room (quit: Quit: WeeChat 0.3.8).
(04:26:03) powerman-asdf: is it possible to somehow read all files with same name in union directory?
(04:27:38) powerman-asdf: sys->dirread() show them all, but sys->open() opens file by name, not QID or something like that, and thus opens only first of same-named files
(04:28:39) powerman-asdf: I'm just thinking about issue with updating /man/*/INDEX when we `bind /someapp/man/2/ /man/2/`
(04:31:55) powerman-asdf: if it were possible to somehow read all INDEX files in /man/2/ dir, then each 3rd-party application can just generate it's own /someapp/man/2/INDEX with index for only it's own man pages, and then /dis/man should just read all files named INDEX instead of just one of them
(04:40:20) powerman-asdf: Dir.dtype + Dir.dev + Dir.qid.path expected to be unique id of any file, and we get them from dirread. But looks like there is no way to use this unique id to open/read that file. :(
(04:48:07) powerman-asdf: probably only way to do this is virtually "unroll" details of current namespace (from /prog/PID/ns) to find real paths on real devices (like #U) of all directories bind to /man/2/, and then open all INDEX files using full paths starting from #U…
(04:49:15) powerman-asdf: but I'm not sure is this reliable - i.e. is this algorithm always can succeed, or it's possible to form such a namespace which can't be "unrolled" to real and accessible device/path names
(05:14:49) leetspete: powerman-asdf: You could just spawn, forkns, drop layers of the union mount.
(05:34:42) anth_r: powerman-asdf: what you're asking for isn't really doable.
(05:35:13) anth_r: you could try to reconstruct the binds used to create your view and walk them manually, but that's error prone (and a lot of work).
(05:36:32) leetspete: I suspect somewhat undesirable besides.
(08:22:58) `jpi left the room (quit: Quit: WeeChat 0.3.2).
(10:11:59) raphaelsc left the room (quit: Ping timeout: 246 seconds).
(10:30:58) jbrhee [~jess@cpe-76-174-171-57.socal.res.rr.com] entered the room.
(12:56:47) jbrhee left the room (quit: Ping timeout: 265 seconds).
(13:51:19) `jpi [~jpi@unaffiliated/pseud0cod3r] entered the room.
(15:02:48) acmeuser [~acmeuser@budec-0020.koleje.cuni.cz] entered the room.
(15:18:30) acmeuser left the room.
(16:00:15) powerman-asdf: anth_r: hmm. leetspete's idea with spawn/fork/unmount everything sounds good. why you think it shouldn't work?
(16:01:17) powerman-asdf: in which cases unmounting one-by-one every mount/bind-point in current namespace may fail?
(17:37:53) acmeuser [~acmeuser@cpe-24-209-4-13.cinci.res.rr.com] entered the room.
(17:37:59) acmeuser left the room.
(19:48:13) Fish [~Fish@9fans.fr] entered the room.
(21:17:35) The account has disconnected and you are no longer in this chat. You will automatically rejoin the chat when the account reconnects.