--- Log opened Mon Dec 14 00:00:35 2009
09:18 < koeien37> any progress on 0.4 ?
14:00 < koeien> h_buildbot: build
14:00 < h_buildbot> Build for all started. If one was running, no new one is started.
14:01 < stepcut> koeien: did you update sendfile from darcs in the build environments?
14:01 < koeien> stepcut: ah no. not from darcs.
14:01 < koeien> i should do that first.
14:02 < koeien> the thing sometimes hangs at the darcs pull step, like now
14:02 < stepcut> koeien: I have not testing sendfile on Windows, OS X, or FreeBSD yet... the changes I made should be OS agnostic, but it would be nice to check before uploading a new version to hackage..
14:03 < stepcut> koeien: I believe I patched sendfile to work on 6.8 and 6.12 now
14:03 < koeien> i only have linux :( the FreeBSD machine i had access to was a bit too limited on resources to test with
14:03 < koeien> yes, but there may be other problems with 6.12
14:03 < stepcut> yeah
14:03 < stepcut> oh ?
14:03 < koeien> e.g., syb-with-class
14:04 < koeien> should be fixed soon now 6.12 is out
14:04 < stepcut> ah, you need syb-with-class from darcs as well
14:04 < koeien> as long as it's fixed in darcs, i can upgrade
14:04 < stepcut> it is fixed.
14:04 < koeien> good
14:04 < stepcut> I have built happstack via a fairly recent 6.12
14:05 < koeien> that sounds promising
14:06 < koeien> first 6.8
14:06 < stepcut> yeah
14:08 < koeien> h_buildbot: build ghc-6.8.3
14:08 < h_buildbot> Build for ghc-6.8.3 started. If one was running, no new one is started.
14:08 < koeien> h_buildbot: build ghc-6.8.3
14:08 < h_buildbot> Build for ghc-6.8.3 started. If one was running, no new one is started.
14:08 < koeien> good. the build starts. sorry for the spam
14:09 < stepcut> I might disable the failing XML tests for 0.4 as well..
14:09 < stepcut> that way it will be all green!
14:09 < koeien> yeah
14:09 < koeien> i remember looking into that once, and it seemed to test something which wasn't implemented
14:09 < koeien> but it was somewhat opaque for me
14:09 < h_buildbot> Build for ghc-6.8.3 failed. Check http://buildbot.happstack.com/ for details.
14:09 < koeien> cabal: cannot configure happstack-0.4. It requires base ==4.* && >=3
14:09 < stepcut> koeien: I emailed Ian(?) or whoever commited the patches, and they said they might look at them sometime...
14:10 < h_buildbot> Build for ghc-6.12.1rc1 failed. Check http://buildbot.happstack.com/ for details.
14:10 < koeien> Igloo i thought (=Ian, right)
14:10 < stepcut> yeah
14:10 < stepcut> the tests might make sense if you expected newtypes to be treated as Transparent XML
14:10 < stepcut> where 'Transparent XML' is something specific to that library
14:11 < stepcut> similar to how tuples and lists are handled
14:11 < stepcut> one moment and I will look into the 6.8 failure
14:12 < koeien> hmm. i cannot find the problem with 6.8. it says it needs base>=3 and base==4.*
14:12 < koeien> i don't know where the base==4.* comes from
14:12 < stepcut> odd
14:13 < koeien> isn't there some cabal flag i can use?
14:14 < stepcut> one moment
14:15 < stepcut> in happstack/happstack.cabal
14:15 < stepcut> try changing this:
14:15 < stepcut>   if flag(base4)
14:15 < stepcut>     Build-Depends:     base >= 4 && < 5, syb
14:15 < stepcut>   else
14:15 < stepcut>     build-depends:     haskell-src-exts == 0.3.9,
14:15 < stepcut>                        hsx == 0.4.5
14:15 < stepcut>  
14:15 < stepcut> in the else section add
14:15 < stepcut> base < 4 ?
14:15 < koeien> argh, i forgot syb i think
14:15  * koeien is stupid
14:15 < koeien> oh no, that was for 6.12
14:16 < koeien> hmm, the flag base4 seems to be set...
14:16 < stepcut> ah
14:16 < koeien> (if i add base >= 37 there, i get a complaint about base >= 37 not being available)
14:17 < stepcut> are we looking into 6.8 or 6.12?
14:17 < koeien> 6.8
14:17 < koeien> this is quite strange
14:18 < koeien> let me nuke the repos
14:18 < stepcut> ok
14:19 < mightybyte> stepcut: If you get me a formlets patch that allows me to build happstack-extras, then I should be able to test my site with 0.4RC1
14:20 < koeien> cabal: cannot configure happstack-0.4. It requires hsx ==0.4.5 && >=0.5.5 &&
14:20 < koeien> <0.6
14:22 < stepcut> mightybyte: ok
14:22 < stepcut> koeien: odd
14:22 < koeien> stepcut: yeah.
14:22 < koeien> i'll do ghc-6.10 and .12 first to see whether that works
14:22 < stepcut> ok
14:23 < koeien> h_buildbot: build ghc-6.10.2
14:23 < h_buildbot> Build for ghc-6.10.2 started. If one was running, no new one is started.
14:23 < koeien> (this is with darcs sendfile.)
14:23 < stepcut> ok
14:24 < koeien> after that, i'll rm -rf the .cabal and .ghc dirs on the 6.8 chroot
14:25 < h_buildbot> Build for ghc-6.10.2 OK. Test suite build failed. Check http://buildbot.happstack.com/ for details.
14:25 < koeien> ah this is a genuine type error
14:25 < stepcut> ah
14:27 < koeien> syb-with-class darcs doesn't compile for me on ghc-6.12.1rc1
14:27 < stepcut> :-/
14:27 < stepcut> one moment
14:28 < koeien> i'll be glad to upgrade to ghc-6.12.1 if that resolves it
14:29 < stepcut> it worked for me with GHC 6.12 11-18-2009
14:29 < koeien> syb-with-class darcs head?
14:29 < stepcut> yeah
14:29 < koeien> i'll look into it later this week on my main machine, which is a little bit faster
14:30 < stepcut> ok
14:31 < koeien>                        hsx == 0.4.5
14:31 < koeien> this line, do you know why it's there for flag(base4) not set ?
14:32 < stepcut> no
14:32 < stepcut> darcs annotate maybe?
14:32 < koeien> earlier in the file we want >= 0.5.5, that's going to conflict
14:33 < stepcut> locked down hsx and haskell-src-exts versions so that it wouldn't break 6.8.3 build in happstack
14:34 < stepcut> perhaps newer versions of haskell-src-exts / hsx  don't build with ghc 6.8.2?
14:34 < koeien> that would seem likely
14:34 < koeien> we need the same trick for HStringTemplate it seems
14:35 < stepcut> we can move the hsx >= 0.5.5 && < 0.6 and friends into the if flag(base4) ... else part I guess?
14:35 < koeien> i did
14:36 < stepcut> did that fix it ?
14:36 < koeien> HStringTemplate<0.6 seems better for ghc-6.8.2 too
14:36 < koeien> building... :)
14:36 < stepcut> k
14:37 < koeien> it builds now
14:38 < stepcut> sweet!
14:38 < stepcut> do you have commit access?
14:38 < koeien> the test build fails with the same type error as in the ghc-6.10.2 build
14:38 < koeien> no, i don't
14:38 < koeien> i want to test the patch for 6.10 too
14:39 < stepcut> k
14:40 < stepcut> oh, I can fix that type error in the test, I know why it is failing
14:40 < koeien> yeah
14:40 < koeien> that seemed pretty easy
14:41 < koeien> h_buildbot: pause
14:41 < h_buildbot> Paused
14:41 < stepcut> what version of darcs are you using? I have seen darcs 2.3.0 seem to hang or run really slow a lot
14:41 < koeien> 2.2.0
14:41 < stepcut> hrm
14:42 < koeien> good. the patch doesn't seem to break the ghc-6.10 build
14:42 < koeien> i'll send it along on the mailing list
14:42 < stepcut> k
14:44 < stepcut> is the base4 flag set even for base5?
14:45 < koeien> i believe base5 doesn't exist
14:45 < stepcut> yeah, I meant, will base4 be set when base >= 5 ?
14:45 < koeien> i wouldn't know
14:45 < koeien> base >= 5 is far in the future?
14:46 < stepcut> me neither :)
14:46 < stepcut> no idea? As long as it isn't in GHC 6.12 I am ok..
14:46 < koeien> 6.12 ships base-4.* iirc
14:46 < stepcut> ok
14:46 < koeien> at least, mine does :)
14:46 < koeien> you have a more recent version
14:46 < koeien> but i doubt it changed between rc's
14:47 < stepcut> yeah
14:47 < koeien> my GHC 6.12 is from 2009-10-10
14:47 < koeien> so i should probably upgrade for a working syb-with-class
14:48 < koeien> i sent a patch to the mailing list
14:48 < stepcut> yeah
14:48 < koeien> h_buildbot: resume
14:48 < h_buildbot> Resumed.
14:48 < stepcut> plus it would be more sensible to build against 6.12.1 since that is what people are actually going to be using
14:48 < koeien> yes
14:49 < stepcut> I'll apply you patch as soon as I get it, and fix the test suite type error as soon as I get sendfile and happstack to build in my autobuilder
14:49 < koeien> good
14:49 < koeien> i'll install ghc-6.12.1 later this week
14:50 < stepcut> ok
14:50 < koeien> but 6.12.1 is not really ready for prime time without HP :)
14:50 < stepcut> yeah
14:50 < stepcut> but I don't want to have to come back and fix 0.4 to work with 6.12 when we are supposed to be working on 0.5
14:50 < koeien> but support for 0.4 would be great of course, since that'll be ready in a few weeks/months
14:50 < koeien> agreed
14:51 < koeien> do you have any idea for a policy regarding GHC version compatibility?
14:52 < stepcut> yeah -- whatever isn't too much work :)
14:52 < koeien> and it also depends a bit on the demand i guess
14:52 < stepcut> I would not be disappointed if 0.5 dropped support for 6.8
14:52 < koeien> 6.8 support is nice because it's in debian
14:52 < stepcut> yeah
14:53 < koeien> but debian squeeze is going to ship 6.12 in all likelihood
14:53 < stepcut> right
14:53 < stepcut> now that Debian is actually getting serious about Haskell packaging, that will help
14:53 < koeien> yes, the debian-haskell mailing list is pretty active
14:54 < burp> debian still needs 6.10 in testing -.-
14:54 < stepcut> yeah
14:54 < burp> still ancient 6.8 in testing... it's unbelievable
14:54 < koeien> that is strange actually
14:54 < koeien> but as long as they ship 6.12 with squeeze it's ok for me
14:54 < koeien> just hope that the HP is ready on time
14:55 < koeien> anyway, /me is going to get some coffee elsewhere. good night
14:58 < stepcut> thanks!
15:01 < stepcut> applied, btw.
15:01 < h_buildbot> Build for ghc-6.10.2 OK. Test suite build failed. Check http://buildbot.happstack.com/ for details.
15:02 < h_buildbot> Build for ghc-6.12.1rc1 failed. Check http://buildbot.happstack.com/ for details.
--- Log closed Tue Dec 15 00:00:36 2009