Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2015-12-02

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

All times shown according to UTC.

Time Nick Message
01:07 dolio joined #darcs
01:23 mizu_no_oto joined #darcs
02:11 mizu_no_oto joined #darcs
02:33 Big_G joined #darcs
04:09 mizu_no_oto joined #darcs
04:44 mizu_no_oto joined #darcs
06:40 Heffalump maerwald: I've submitted http://bugs.darcs.net/patch1413 - gh is the one who decides on what goes in the 2.10 branch. And I'm not sure when a release is planned.
12:21 mizu_no_oto joined #darcs
14:02 maerwald Heffalump: btw. I feel it's barely possible to abuse the <from-id>..<to-id> feature, if you feed it crazy long ranges, it will just show nothing, probably because the patches are too disconnected
14:10 Big_G joined #darcs
14:19 maerwald does anyone know if it's safe to nest "withSettings"? I don't know much about implicit parameters to judge that
14:20 maerwald withSettings :: Settings -> ((?settings :: Settings) => a) -> a
14:20 maerwald withSettings s c = let ?settings = s in c
15:17 byorgey maerwald: I assume the inner one would shadow the outer one
15:18 byorgey so it should be "safe" (though I'm not 100% sure what you mean by "safe" in this context)
15:21 maerwald the alternative is to delay the initial "withSettings" call, which gets really messy, because I have to pass the settings explicitly until that point
15:22 maerwald and yeah, I mean safe in the way that I can be sure everything that follows the second "withSettings" will also use those settings
16:12 ggherdov` joined #darcs
16:12 theone joined #darcs
16:12 sea-gull joined #darcs
16:12 carter_cloud joined #darcs
16:12 Reyu joined #darcs
16:12 MasseR joined #darcs
16:12 navilan- joined #darcs
16:12 maerwald joined #darcs
16:12 mal`` joined #darcs
16:12 siel joined #darcs
16:12 carter joined #darcs
16:12 peb` joined #darcs
16:12 lelit joined #darcs
16:12 burp_ joined #darcs
16:12 lambdabot joined #darcs
16:12 haasn joined #darcs
16:12 Igloo joined #darcs
16:12 sm joined #darcs
16:12 Meeh joined #darcs
16:12 aristid joined #darcs
16:12 Heffalump joined #darcs
16:12 dleverton joined #darcs
16:12 vikraman joined #darcs
16:12 jackhill joined #darcs
16:15 ggherdov` joined #darcs
16:28 dolio joined #darcs
16:28 dixie_ joined #darcs
16:28 dino-_ joined #darcs
16:28 kaol joined #darcs
16:28 IcyFoxy joined #darcs
16:28 thorkilnaur joined #darcs
16:48 Riastradh joined #darcs
16:48 ggherdov` joined #darcs
16:48 theone joined #darcs
16:48 sea-gull joined #darcs
16:48 carter_cloud joined #darcs
16:48 Reyu joined #darcs
16:48 MasseR joined #darcs
16:48 navilan- joined #darcs
16:48 maerwald joined #darcs
16:48 jackhill joined #darcs
16:48 vikraman joined #darcs
16:48 dleverton joined #darcs
16:48 Heffalump joined #darcs
16:48 aristid joined #darcs
16:48 Meeh joined #darcs
16:48 sm joined #darcs
16:48 Igloo joined #darcs
16:48 haasn joined #darcs
16:48 lambdabot joined #darcs
16:48 burp_ joined #darcs
16:48 lelit joined #darcs
16:48 peb` joined #darcs
16:48 carter joined #darcs
16:48 siel joined #darcs
16:48 mal`` joined #darcs
16:51 ilbot3 joined #darcs
16:51 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
17:05 Riastradh joined #darcs
17:30 notdan joined #darcs
17:30 notdan joined #darcs
19:14 maerwald darcs record seems to hang on large patches... it's now at 100% cpu since 30minutes
19:17 maerwald for the same operation, git takes ~30s
19:23 Riastradh joined #darcs
19:39 maerwald almost 1 hour now
20:04 aristid joined #darcs
20:14 ggherdov` joined #darcs
20:19 carter_cloud joined #darcs
21:04 JamesJRH joined #darcs
21:05 JamesJRH sm: Hi.
21:07 JamesJRH sm: If I was to contribute patches to Darcsden, would you accept them if I license my changes under the AGPLv3?
21:08 maerwald the whole project should just be made AGPLv3
21:08 maerwald I don't see why we should allow that loophole in GPL-3
21:12 JamesJRH Indeed.
21:13 JamesJRH But it can only apply to new code – the existing code will always remain GPLv3.
21:14 maerwald that's not totally accurate. All old code will be AGPL-3 too in the project folder, but if people get an old tarball, then that will still have GPL-3
21:18 Heffalump maerwald: I wasn't worried about abuse, I was worried about it not being stable under repository operations
21:19 Heffalump maerwald: why do you need to call withSettings a second time? I suspect the existing code isn't properly factored if that's necessary, as the design was intended to just use a top-level withSettings call
21:19 maerwald there's no sensible way around it
21:20 maerwald but I'm not sure the patch as a whole makes sense currently
21:20 Heffalump JamesJRH: FYI the Haskell ecosystem tends to be mostly BSD3, so that's how my contributions to darcs and darcsden are licensed (to the extent that they standalone, which is a bit variable)
21:21 Heffalump maerwald: fair enough, would just like to understand why as I was responsible for withSettings in the fristirst place
21:21 maerwald because I need to call an operation in the Snap monad... that's not possible at the point where withSettings is called at first
21:22 maerwald the earliest point is the function where the handlers are defined
21:22 sm JamesJRH: hi.. um remind me why you'd want to do that ?
21:24 Heffalump maerwald: why does calling in the Snap monad imply calling withSettings? Is it because there's a utility function that both calls withSettings and runs the Snap monad?
21:25 maerwald Heffalump: because I modify a global setting that I can only construct before the handler code gets executed
21:25 maerwald the alternative would be a huge amount of code duplication
21:26 maerwald or introducing more implicit parameters, data structures and modules
21:26 Heffalump ok, I guess I'd need to see the code to understand properly
21:26 maerwald https://darcs.hasufell.de/hasufell/darcsden-ssl/patch/6a9bcfe6bac699f785dab2557a1b6978bd9bef4b
21:27 maerwald but that patch has a few other problems I don't know how to solve, so I'll probably discard it
21:27 Heffalump oh, I see, you're actually changing the settings in one particular scope?
21:28 maerwald the easiest way is to not make it possible to use http and https at the same time
21:28 maerwald as soon as you do that, you run into the problem of how to get the full URL for stuff, which is needed for both OAuth and the "darcs get" url
21:29 maerwald for the rest (links, js, css) relative paths work fine
21:30 Heffalump so you pass the "real" full URL just to the things that need it?
21:30 maerwald well, since it's part of the implicit parameter, I can access it from any handler
21:30 maerwald everything else gets really messy
21:31 maerwald the call stack is pretty deep for a few of those things
21:31 Heffalump right, but if you're calling withSettings earlier, then some code might or does execute with fullUrl = baseUrl
21:31 maerwald it isn't used before
21:32 Heffalump how about initialising it with an error, so it's easier to find out if that changes?
21:33 maerwald well, I'm rather in favor of not allowing http+https
21:33 maerwald things get even more complicated if you use a front proxy
21:33 maerwald then the whole thing just doesn't what you think it does
21:34 Heffalump you mean relative paths?
21:34 maerwald http-only is common if you use a https enabled front proxy... if you don't, just go with https only
21:34 maerwald Heffalump: the fullUrl' function doesn't know that there is a front proxy...
21:34 maerwald so you'll get http links
21:35 Heffalump I see. Wouldn't the front proxy have to rewrite things, or you would have to configure the application to know about the front proxy URLs?
21:35 maerwald I think that's a lot of work to support a corner case
21:35 maerwald it can be implemented in the future still
21:35 Heffalump I guess that rewriting is infeasible anyway when it comes to things like "darcs get foo" presented to the user
21:36 maerwald so I'd make the type: data Protocol = HTTP | HTTPS
21:36 maerwald not allowing both
21:36 Heffalump anyway, I'm not sure I fully understand what you mean, I'm just suggesting that you use an error for the "initial" value of fullUrl, instead of using baseUrl
21:36 maerwald if someone wants to implement support for using both, he can add a "ALL" constructor and start doing it properly
21:36 Heffalump the more strongly typed solution would be to add a second implicit parameter, but I guess that could be overkill
21:37 maerwald http should be discouraged anyway, unless darcsden is just the backend server behind an ssl enabled front proxy
21:38 maerwald github is https only too
21:38 maerwald and I don't mind that
21:54 maerwald sm: "The GNU Affero General Public License is designed specifically to ensure that, in such cases, the modified source code becomes available to the community.  It requires the operator of a network server to provide the source code of the modified version running there to the users of that server.  Therefore, public use of a modified version, on a publicly accessible server, gives the public access to the source code of the modified
21:55 maerwald otherwise you are effectively allowing people to fork darcsden, develop it in a proprietary manner and deploy it
21:55 maerwald so users cannot be sure what code is actually running there
23:24 sea-gull joined #darcs
23:38 maerwald so I managed to record the huge patch with a few tricks, but then darcs just chokes on "darcs push" (or darcs apply on the remote, not sure)
23:39 maerwald it seems like it's hanging, or it does some weird conversion from BS to String maybe?

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