Perl 6 - the future is here, just unevenly distributed

IRC log for #darcs, 2015-05-31

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

All times shown according to UTC.

Time Nick Message
01:30 c74d joined #darcs
09:16 alexei joined #darcs
11:55 Heffalump should darcs 2.12 support GHC 7.6? (I have a vague feeling I asked this before, but I can't find the conversation if so)
12:10 kaol Dropping the support would prevent a backport to Debian stable. As a case point. But if someone was set to run the newest darcs on Debian stable, they could get their own GHC to build it.
12:32 Heffalump kaol: yeah, I'm not sure how much that use case matters. We've certainly had people ask for it in the past. On the other hand, I think use of Debian is in (relative) decline, just judging anecdotally by the people I speak to
12:33 peb` looking at popcon I'd say you're mistaking
12:34 Heffalump fair enough
12:37 * Heffalump wonders about introducing Darcs.Prelude or Darcs.Util.Prelude, instead of spending ages maintaining our Prelude imports - I'm currently sorting out the (<$>) warnings
12:47 mizu_no_oto joined #darcs
13:22 Riastradh joined #darcs
13:51 mizu_no_oto joined #darcs
13:59 mizu_no_oto joined #darcs
14:32 sm sounds nice Heffalump
14:32 sm (Darcs.Prelude)
14:34 sm re tests, that's a good question. I think: make obvious to present and future devs the steps to follow and benefits of doing it
14:37 Heffalump http://hub.darcs.net/ganesh/darcs-patch1345/patch/20150531142627-81bb2#src/Darcs/Prelude.hs
14:38 Heffalump sm: so the two annoying things about them now is that you need couchdb and selenium running
14:38 Heffalump how much of a nuisance is that?
14:43 sm Heffalump: annoying enough to stop most contributors I'd say
14:43 sm (biab)
14:47 mizu_no_oto joined #darcs
15:10 Riastradh joined #darcs
15:21 sm Heffalump: like a typical lazy dev, I lost track of the tests since being unable to install them (with ghc 7.10). I had the vague impression there were probably just a few and not testing anything I'm likely to break at the moment, therefore low priority
15:21 sm if pressed to try again, I would have tried cabal install --enable tests && cabal test
15:22 sm I see there is some documentation in the README, and it's a little more complicated
15:23 sm currently all darcsden tests are "functional" rather than unit tests, right
15:23 sm personally, I might also be still a little scared that I'll screw up the config files or something and end up overwriting the production db again
15:28 sm also, I have the impression that our current commit rate and breakage rate is low and I haven't been feeling enough pain yet to invest more time in running tests
15:30 * sm <- typical lazy, short-sighted, unappreciative developer :)
16:47 sm nice discussion/intro for lenses: http://www.reddit.com/r/haskell/comments/37qnhq/motivation_for_lens/
17:30 Heffalump sm: fair enough. Turns out they are all working. I guess I'll try to get them to run selenium themselves and to mock out couchdb. Yes, they're functional tests.
17:31 Heffalump Would it be ok to add the selenium jar to the repo?
17:31 sm Heffalump: sure, if that would be helpful
17:34 sm I hear snap has a lot of tests, I wonder how they ensure people run those
17:36 Heffalump HTTP has plenty of tests, and they do work with 'cabal test' out of the box
17:36 sm nice.. are those webdriver/selenium based too ?
17:36 Heffalump no, it's a client library
17:37 sm oh true
17:37 Heffalump but it does run up an http server in the background (on a separate thread)
17:37 Heffalump but the server is also a Haskell one so it's easy enough to orchestrate
17:38 sm https://github.com/snapframework/snap-server 's how to run tests doc seems out of date
17:44 sm I see, snap-server has hunit tests, and functional ("blackbox") tests which spin up a server, like HTTP
17:45 sm (and test it via HTTP)
17:45 sm whereas for darcsden you built tests which actually go through the web browser and test what a user would see
17:49 * sm mentally thinks unit tests, system tests via HTTP API, system web tests via web UI [, production instance tests...]
17:50 Heffalump I'd probably have written some lower-level tests first if I'd written them to begin with.
17:51 Heffalump FYI I'm pretty certain that the test binary can't clobber production settings now
17:51 sm how come tests require couchdb ?
17:51 Heffalump because noone has written a mock backend
17:51 sm ah
18:00 sm oh bsrk made these tests eh
18:04 BitPuffin|osx joined #darcs
18:04 sm they seem pretty great, if you want to include the UI in tests
18:05 sm hackage-server also tests via HTTP. darcsden might be the only public haskell package using the browser
18:17 sm darcsden tests build with GHC 7.8 but don't run (listening on 8950, connecting on 4444 ?)
18:18 sm I'm not sure if they use the new config file(s)
18:20 mizu_no_oto joined #darcs
18:21 sm so yes, the simplest answer is the obvious, make them easy to run
18:40 mizu_no_oto joined #darcs
19:54 alexei joined #darcs
20:34 alexei joined #darcs
21:53 BitPuffin|osx joined #darcs
22:46 Heffalump sm: 4444 is for selenium
22:49 Heffalump I think it's useful to test with the browser given we're an actual interactive website, but I think I'd have taken a simpler approach for the first tests.

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