Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2015-07-15

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

All times shown according to UTC.

Time Nick Message
00:06 mizu_no_oto joined #darcs
00:59 byorgey joined #darcs
01:46 ilbot3 joined #darcs
01:46 Topic for #darcs is now http://darcs.net/ | logs: http://irclog.perlgeek.de/darcs/ | darcs 2.10.0 is out http://darcs.net/Releases/2.10
01:56 mizu_no_oto joined #darcs
02:09 mizu_no_oto joined #darcs
02:42 Pastaf joined #darcs
02:50 byorgey joined #darcs
03:12 mizu_no_oto joined #darcs
03:18 sprang joined #darcs
04:03 Riastradh joined #darcs
07:25 amgarching joined #darcs
07:31 amgarching joined #darcs
09:31 amgarching joined #darcs
09:44 alexei joined #darcs
11:07 mizu_no_oto joined #darcs
11:46 alexei joined #darcs
11:53 notdan sm: I have a question: why are you fetching changes with AJAX when viewing patches? You told me the reason some time ago but I kinda forgot :(
11:54 notdan also annotate doesn't work on your repo http://hub.darcs.net/simon/darcsden/annotate/src/DarcsDen/Pages/RepoPages.hs
12:01 maerwald notdan: not a good trace :/ hard to see where openFile (or similar) actually fails
12:39 byorgey joined #darcs
12:39 byorgey joined #darcs
13:05 Heffalump sm: I think that's just so the page as a whole loads quicker while the patch is rendering
13:05 Heffalump s/sm/notdan/
13:18 alexei joined #darcs
14:21 gh_ joined #darcs
14:22 gh_ hi
14:22 gh_ notdan, if you send patches to darcs about exporting more functions, please mention in the export that it is for darscden, otherwise in the future someone will come and remove the export seeing that it is not used in darcs (re: http://irclog.perlgeek.de/darcs/2015-07-14#i_10893421 )
14:33 gh_ and I hope we have a 2.10.2  after the september sprint
14:34 gh_ that would also contain all you need for darcsden (and hopefully bugfixes for http://darcs.net/Releases/2.10#known-issues )
14:40 alexei joined #darcs
14:51 notdan gh_: so you basically try to export as little as possible from every module?
14:51 gh_ notdan, yes , to avoid dead code
15:11 sm notdan: yes, one of the goals of darcs hub is to always serve a page reasonably quickly, but some darcs ops take much longer
15:12 sm g'day all
15:21 alexei joined #darcs
15:48 sm notdan: thanks for the report, I didn't have that one. Opened http://hub.darcs.net/simon/darcsden/issue/138
15:50 sm the darcsden open issues list is starting to look formidable.. ~90
15:50 sm it's the darcsden & darcs hub issues list, but still
16:53 Heffalump gh_: did you hear back about a venue for the sprint?
16:54 alexei joined #darcs
17:16 gh_ Heffalump, I wrote to IRILL but still no news yet. Probably because it's July.
17:20 gh_ trying again with a more official irill mail address
18:14 notdan man, hidden conflicts are driving me nuts :( I am doing rebase of 40+ patches and I have to manually check (basically record again) each
18:14 notdan and I think I screwed up somewhere midway
19:00 notdan Oh my god, i think i did it!
19:01 notdan I successfully (?) rebased the changes D:
19:01 Heffalump notdan: I'm around now
19:01 Heffalump good to hear about the rebase, sorry it was a pain..
19:02 Heffalump I'd forgotten you said you'd already resolved the conflicts - the "right" thing then is to merge the resolutions into the conflicting patches with amend/rebase, but I can't think of a trivial way to identify the right patches et
19:02 Heffalump c
19:04 notdan To be honest, I still don't understand the source of the problem. So you have a bunch of conflicting patches + a resolution patch, then there is a new patch which conflicts with a previous change. Well, so what, it shouldn't conflict with the resolution?
19:04 notdan Heffalump: that's actually one of the things I wanted to work on in darcs: some sort of UI that tells you what depends on what and what conflicts with what
19:04 Heffalump this would be easier to explain with a whiteboard :-) But I'll try..
19:05 Heffalump [around a bit intermittently for a few more minutes, biab]
19:05 notdan oh well if it's hard to explain on IRC then I can definitely wait, my head is full of stuff right now
19:06 notdan we can do a whiteboard session during the sprint!
19:07 notdan OK, now I need to make sure that everything works and ask simon to merge the changes
19:07 Heffalump it's basically that if patch B depends on patch A, and patch A is in a conflict, then B is implicitly in a conflict too
19:10 Heffalump sounds good
19:14 Heffalump did you still want a meeting today, btw?
19:16 notdan yeah, can we postpone it a bit tho?
19:17 notdan if not, I can just send you the questions by email
19:17 Heffalump sure, I'll be around for a while
19:22 notdan oops found a bug in repo comparison
19:32 byorgey joined #darcs
19:44 Riastradh joined #darcs
19:45 notdan sm: Hey, I've refactored the repository, not compare doesn't time out: http://hub.darcs.net/simon/darcsden/compare/co-dan/darcsden-local-take2
19:45 notdan Heffalump: alright, so I am ready for the mini-meeting
19:46 Heffalump ok, great
19:46 notdan So basically I've implemented the repository list thingie that we talked about last time
19:47 notdan the repository list is stored in ~/.darcs/darcsden_repos or in a windows equivalent
19:47 Heffalump cool
19:47 notdan the code is at http://hub.darcs.net/co-dan/darcsden-local-take2/
19:47 notdan You can use the command line tool to add repositories to that list: `den --add repo1 repo2`
19:48 notdan I haven't merged the CLI tool into the main repository yet -- which actually turned out to be a good thing :P
19:48 Heffalump :-)
19:48 Heffalump one idea for the cli tool - it should automatically add a repo if you launch the tool on that repo.
19:49 notdan however, the tools are not in the full sync yet
19:49 notdan 1) I used to have code in darcsden that would touch a file in the /tmp folder, and then den would check for that file before starting a server. if the file is present it would not start the webserver, it would just open the url
19:50 notdan that method however is quite dumb
19:50 notdan first of all, darcsden should remove that file on exit -- i havent implemented that because that would require having backend-specific signal handlers
19:50 Heffalump why are signal handlers backend-specific?
19:51 notdan secondly, that file would have to be either hardcoded or read from the configuration file. either way any other program can touch that file
19:51 notdan Heffalump: well, we only need to create that file if we are running darcsden local instance
19:53 Heffalump oh, I see
19:53 Heffalump how is the port the server runs on determined?
19:53 notdan 2) if `den` does not find the file, it starts a web server. however, the port specified in the configuration file might already be in use. in that case `den` tries a different port, recursively
19:53 Heffalump if it's in a config file anyway, then you could just ping that port instead
19:54 Heffalump ah, right
19:54 notdan Heffalump: usually the port is specified in the config file
19:54 notdan but anything can run on that port :)
19:54 notdan the problem with that method is the following: we need to run a web server and forkIO open a web browser
19:55 notdan and in order to open a web browser we need to know the port
19:55 Heffalump an alternative would be to just fail if the port is in use
19:55 Heffalump and to detect an existing server by checking an URL on the port
19:56 notdan but in order to know the port we have to check if a port is busy bla bla
19:56 notdan the way i detect it right now is by catching a resource exception
19:56 notdan but that is happening in a separate thread
19:56 notdan so there is a lack of sync in there
19:56 notdan Heffalump: but would that always work?
19:56 Heffalump you can use MVars or similar to synchronise
19:57 Heffalump errm, dunno. What could go wrong? :-)
19:57 Heffalump but perhaps just do something and see how it goes for now, in the interests of getting to the minimal mergeable product
19:57 notdan Heffalump: could it be that a port is busy (we cannot bind to it), but a server that is running on that port is not accepting connection?
19:57 notdan or for instance we just cannot allocate that port because of some privilege problem?
20:00 notdan Heffalump: regarding MVars, I thought about it but not sure how would it work. I don't think we can really check that the server did bind the port sucessfully; we can only check for failure via exceptions
20:00 notdan Yeah, I am definitely focused on just getting something minimal done
20:01 Heffalump I guess the temp file has a problem with portability
20:01 notdan I think I would just make `den` spew an error if the port is busy. This seems reasonable, I think hledger-web is doing that
20:03 Heffalump sounds good
20:07 notdan OK, so I will put on hold the part that switches the port dynamically, and abandone temp files in favour of checking the url directly
20:07 Heffalump ok
20:07 notdan another problem that trying to connect to the server might introduce some lag and slow things down a bit
20:07 notdan but we can worry about it later
20:09 Heffalump true, especially if the connection times out for some reason
20:11 notdan another that I had is about modification dates. i've realized that directory last modification date is a bad metric: it can stay the same even if there are new commits in the repository, and it can change even if there is no darcs-related activity
20:12 notdan after I get the local branch merged I will try to use the timestamp of the latest patch instead of directory modification date
20:13 Heffalump hmm, true - though making local changes to the repo is arguable still relevant activity
20:13 Heffalump but as you say the root directory modification date won't cover that
20:14 Heffalump looking for timestamp of the latest patch might be doable with the timestamp of _darcs/patches
20:14 notdan yeah and, IMHO, local changes to boringfiles shouldn't count as well
20:14 notdan Heffalump: ooh that's a very good idea
20:14 notdan at the very least it is definitely "good enough"
20:18 notdan Alright, hopefully by the end of the week we will get everything merged upstream and all the aforedmentioned changes implemented
20:19 notdan and next week I will be able to finish up a prototype of the patch record form
20:19 Heffalump sounds good
20:29 * Heffalump assumes we're done, but will be around for a bit more in case there is anything else
20:29 notdan yeah, sorry
20:29 notdan i dont have any further comments :]
20:30 notdan thanks for helping me with the rebase thingie, btw
20:30 notdan now I basically *have to* write a blog post about that
20:30 notdan like, legally
20:38 Heffalump :-)
20:38 Heffalump I look forward to that - I'm sure there's plenty to do to make it more usable
21:03 Pastaf joined #darcs
21:33 mizu_no__ joined #darcs
21:46 byorgey joined #darcs
21:46 byorgey joined #darcs
23:18 mizu_no__ joined #darcs
23:31 lf94_ joined #darcs
23:39 lf94 joined #darcs

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