Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2016-05-04

| Channels | #darcs index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:47 ilbot3 joined #darcs
01:47 Topic for #darcs is now Darcs/Pijul Sprint, Helsinki, 6th-8th May http://darcs.net/Sprints/2016-05 | http://darcs.net/ | logs: http://irclog.perlgeek.de/darcs/ | darcs 2.10.3 is out http://darcs.net/Releases/2.10
01:58 Riastradh joined #darcs
04:06 Big_G joined #darcs
07:43 isxek joined #darcs
07:47 Weltraumschaf joined #darcs
08:16 gal_bolle joined #darcs
09:30 Weltraumschaf joined #darcs
11:22 Meeh joined #darcs
12:26 fr33domlover hey sm! Some DarcsDen thoughts: I'm implementing some darcs related things (init a repo, view file tree,view history, push/pull over SSH) for a web app and one of th ways to do it is the modules exported by the darcs package. But I noticed a little issue: Darcs changes the current directory in every command. It seems darcsden doesn't mind that and still works fine, but I wonder: Is it safe in a web app (or any
12:27 fr33domlover other concurrent context) to do these frequent changes? Also what if 2 handlers serve 2 darcs repo views concurrently etc.?
12:27 fr33domlover Another way, although a partial solution, is the hashed-storage package... i'm going to try that
12:27 fr33domlover (and yet another one is to run darcs program and parse the XML output...)
12:29 fr33domlover I wish the darcs code didn't do these silly working dir changes... would make darcs better as a library
12:53 BitPuffin joined #darcs
13:29 mizu_no_oto joined #darcs
14:26 Big_G joined #darcs
15:40 mizu_no_oto joined #darcs
16:22 ronny_ left #darcs
16:25 amgarchIn9 joined #darcs
16:37 bmp fr33domlover: On the security front, it should be possible to chroot things just before invoking darcs.
16:37 bmp You'd need to put a Unix socket or something similar into the chrooted location so you could communicate with the darcs process, but it'd let you deal with this somewhat sanely in the meantime.
16:37 bmp Completely agree with you that that behavior is suboptimal, though
16:54 amgarchIn9 joined #darcs
17:11 Heffalump fr33domlover: that's a well known problem, I'm afraid. darcsden builds without the concurrent runtime to  work around it.
17:12 Heffalump it's actually an optimisation: it's much faster to access a relative path than construct the absolute one, pass it to the OS and have it take it apart again. But I don't know how material that is nowadays.
17:12 Heffalump re hashed-storage, that isn't really maintained as an independent library any more, it got folded into darcs. I think it may also rely on working dirs.
17:13 Heffalump Overall I think we should just stop darcs doing that.
17:13 fr33domlover I don't mind working on a fix, even if gradually, starting with the specific functions I need
17:13 fr33domlover I'm just not sure they'll want my patch :P
17:14 fr33domlover I certainly can't fix the entire code right now
17:14 fr33domlover Way too much work
17:14 Heffalump I think we'd accept it, but feel free to start a discussion on darcs-devel first to establish the principle.
17:25 fr33domlover Heffalump, in my web app I can't use that workaround because without -threaded I can't waitForProcess which I need in some cases... :P
17:35 Heffalump I'm not sure how workable a partial solution is though - you first need to get the 'right' folder passed to all the places that need it
17:39 fr33domlover I mean it can be done gradually. For example, "darcs init" seems to be a reasonably small and promising start
17:41 Heffalump well, maybe, but have you looked at the code?
18:00 sm fr33domlover: that is the reason darcsden is built without -threaded :)
18:02 sm what should be done about https://github.com/fpco/stackage/issues/1399 ?
18:04 sm tests requiring network access to run seems like a code smell, especially if not network related
18:04 fr33domlover Heffalump, yes, I have :)
18:05 Heffalump sm: heh, we were discussing this on darcs-devel and concluded we should enable them by default
18:05 Heffalump the tests are network related though
18:05 Heffalump I shall comment on the relevant thread..
18:05 Heffalump (darcs-internal thread, that is)
18:14 sm thanks Heffalump
18:15 Heffalump http://bugs.darcs.net/patch1378
19:18 der-landgraf joined #darcs
19:40 mizu_no_oto joined #darcs
20:32 mizu_no_oto joined #darcs
20:46 pointfree joined #darcs
20:50 7GHAAYAYL joined #darcs

| Channels | #darcs index | Today | | Search | Google Search | Plain-Text | summary