Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2015-08-24

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

All times shown according to UTC.

Time Nick Message
02:12 mizu_no_oto joined #darcs
02:31 favonia joined #darcs
09:37 emax joined #darcs
10:37 mizu_no_oto joined #darcs
11:01 carter_cloud joined #darcs
13:01 timothyw joined #darcs
14:45 Igloo joined #darcs
14:50 favonia joined #darcs
15:06 gh_ joined #darcs
15:27 notdan sm: hi! what do you think about two patches at http://hub.darcs.net/simon/darcsden/compare/co-dan/darcsden-to-merge ?
15:35 sm notdan: sounds good.. let me try it on hub-dev
15:42 sm notdan: is there a user-visible change ? should we be seeing dep lines at http://hub-dev.darcs.net/simon/darcsden/compare/co-dan/darcsden-to-merge ?
15:43 notdan hm yes
15:43 notdan if you are logged in, I believe
15:43 sm I am.. see if you can find them, it's running your latest
15:46 notdan hm
15:46 notdan weird
15:56 notdan yeah the patch is bad
15:56 notdan sorry :(
15:57 sm I can obliterate it if you want, otherwise just send another
15:58 sm will the dep calculations create a heavy load in some cases ?
15:58 sm have you tested any complex cases ?
16:00 Riastradh joined #darcs
16:06 notdan oh i found the bug
16:06 notdan sm: hm, well it was already calculating the dependencies
16:06 notdan it just wasn't showing them
16:11 notdan if you could obliterate the patch while none is watching.. :P
16:17 sm notdan: oh actually I didn't push it yet, no need
16:17 notdan bugger, JS is *hard*
16:17 sm heh
16:17 sm it's a whole world of its own
16:19 favonia joined #darcs
16:21 notdan :)
16:37 notdan OK, pushed a proper patch this time :]
16:46 sm notdan: http://hub-dev.darcs.net/simon/darcsden/compare/co-dan/darcsden-to-merge looks the same
16:47 sm how about on your instance
16:47 sm make sure it works there first
16:47 sm I know sometimes (usually) there are web deployment issues on the production site too
16:48 notdan uh it did work! :(
16:48 notdan ok i apologize for this
16:49 sm ok no problem.. maybe you can check what I'm serving with web inspector ? I tried shift-reload, maybe I'm missing some file ?
16:49 notdan hm
16:49 notdan it does work for me tho
16:49 notdan http://hub-dev.darcs.net/co-dan/darcsden/compare/co-dan/darcsden-to-merge
16:49 sm ok! great!
16:49 sm I guess there's no dependencies at the url I tried
16:50 sm interesting!
16:50 notdan oh
16:50 sm a static image.. how does that even work :)
16:50 notdan yeah, that's true haha, those patches are independant
16:51 notdan sm: it's HTML5 canvas actually
16:51 sm ok
16:51 sm oh
16:51 sm would it be hard to highlight the lines on mouse hover ?
16:52 sm I wonder why it's not done on the changes screen.. too costly there ?
16:53 sm I guess it fails gracefully in browsers without HTML5 canvas
16:54 notdan highlight the dependencies?
16:54 notdan yeah i think it might be too costly. i can experiment and see
16:54 sm I meant highlight the lines, so you could more easily see where they lead, but actually that's a good idea.. highlight the deps when you mouse over a patch
16:54 sm with or without the lines
16:56 sm maybe that would be cheaper too.. do it lazily
16:58 notdan that would be cool
16:58 notdan and i guess easier once we get "production-ready" ghcjs
16:58 sm maybe
16:59 sm there's also elm
16:59 notdan yeah but with ghcjs you could share the same datatypes at least
16:59 sm yeah
17:00 notdan there is a really cool F# web framework - WebSharper - it allows you to select which parts of your codebase are you going to compile to JS and which to .NET byte code. and it provides seamless-y integration between two parts
17:00 sm nice
17:28 Heffalump notdan: does that work well in practice? I'm aware of it but have never tried it
17:29 Heffalump I think ghcjs produces huge "executables" which may be a downside
17:56 notdan yeah the size is an issue
17:56 notdan the fact it GHCJS has its own RTS
17:56 notdan which has both upsides and downsides, one of the latter being huge resulting files
17:59 favonia joined #darcs
18:18 mempko joined #darcs
18:55 gh_ darcs -- the beautiful dream https://pbs.twimg.com/media/CMFNV8fWoAAhr6w.jpg
19:00 sm gh_: sweet! :)
19:00 sm should be Davey and the Til-Darcs
19:02 gh_ Til as "today I learned"
19:09 mempko joined #darcs
20:02 notdan Heffalump: can you try `den' again?
20:28 Heffalump 500 :-)
20:28 Heffalump assuming you wanted me to pull from your repo first
20:28 notdan uh
20:28 notdan re-started
20:29 Heffalump it says I have everything already, is that expected? I haven't pulled since last week
20:29 notdan i cant figure out if it's a memory leak or my dirt-cheap vps doesnt have nearly enough memory
20:29 notdan uh i forgot to push
20:29 notdan sorry i am acting weird this week for some reason *shrug*
20:30 notdan ok, pushed
20:42 * Heffalump builds. What am I looking for when I run it?
20:42 notdan dataraces?
20:43 notdan I  forced dataraces on my machine, but I want to check if it works "in the wild"
20:43 Heffalump my machine definitely seems to count as wild :-)
20:46 Heffalump not looking promising :-(
20:46 Heffalump I have a den process that hasn't printed out anything or launched anything, apparently.
20:46 Heffalump (Linux)
21:02 Heffalump can you add some diagnostic print statements so I can try and see how far it gets/what is going on?
21:02 notdan yeah, on it
21:04 notdan pushed a wee patch
21:06 Heffalump ohh, I think I know what's wrong
21:07 Heffalump so, when I ran it, it was printing out 'pollServer localhost' or whatever repeatedly
21:07 Heffalump which made me remember that in my nix shells with GHC configured, name resolution is a bit broken for reasons unknown
21:08 Heffalump so then I tried to run it in another shell without GHC configured, and it just worked...
21:09 Heffalump so, I guess a lot of the problems are my fault for not realising it before, but also perhaps a timeout and reporting the last error message after the timeout would help
21:10 Heffalump I can probably handle the error message/timeout, I have a good test case to try it on :-)
21:10 Heffalump sorry about that, you probably ended up spending disproportionate effort on a race that was probably only theoretical for users without messed up name resolution
21:13 notdan no, it's useful!
21:15 notdan So I am not sure about pritning the message on the time out: when there is an actual data race, the server on localhost is not running. But polling it is very fast, so in that situation `den` would end up printing a lot of error messages before launching the browser
21:15 Heffalump only if the timeout times out, and then only the last message
21:16 Heffalump also, possibly it should be more discriminating about identifying "server isn't listening" (i.e. connection refused) and any other errors
21:19 notdan hm actually it should discriminate against real time out
21:19 Heffalump the time out would just be a limited number of iterations of the polling
21:19 Heffalump so e.g. retry 50 times, then print out the last error
21:20 notdan ah
21:20 notdan yea, this sounds good
21:53 favonia joined #darcs
22:07 mempko joined #darcs
23:48 favonia joined #darcs

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