| Time |
S |
Nick |
Message |
| 00:01 |
|
|
secorp left #darcs |
| 00:01 |
|
|
owst joined #darcs |
| 00:10 |
|
|
gwern left #darcs |
| 00:12 |
|
|
gwern joined #darcs |
| 00:17 |
|
|
balor left #darcs |
| 00:30 |
|
dixie |
yes, it is in http://darcs.net repo |
| 00:30 |
|
dixie |
the main page is in doc/index.html |
| 00:30 |
|
dixie |
wiki has the other repo, http://wiki.darcs.net/; |
| 00:32 |
|
|
secorp joined #darcs |
| 01:27 |
|
|
raichoo joined #darcs |
| 02:57 |
|
|
owst left #darcs |
| 02:59 |
|
|
raichoo left #darcs |
| 03:23 |
|
|
secorp left #darcs |
| 03:29 |
|
|
secorp joined #darcs |
| 03:39 |
|
|
intripoon_ joined #darcs |
| 03:42 |
|
|
intripoon left #darcs |
| 05:25 |
|
|
sm left #darcs |
| 07:28 |
|
|
balor joined #darcs |
| 08:01 |
|
|
darcscommitbot left #darcs |
| 08:02 |
|
|
darcswikibot left #darcs |
| 08:02 |
|
|
darcscommitbot joined #darcs |
| 08:03 |
|
|
darcswikibot joined #darcs |
| 08:05 |
|
|
balor left #darcs |
| 09:15 |
|
|
ManateeLazyCat joined #darcs |
| 09:15 |
|
|
ManateeLazyCat left #darcs |
| 09:34 |
|
|
lelit joined #darcs |
| 09:35 |
|
|
lispy left #darcs |
| 09:35 |
|
|
lispy joined #darcs |
| 09:54 |
|
|
lispy left #darcs |
| 10:30 |
|
|
raichoo joined #darcs |
| 11:40 |
|
|
ManateeLazyCat joined #darcs |
| 11:56 |
|
|
ManateeLazyCat left #darcs |
| 11:56 |
|
|
ManateeLazyCat joined #darcs |
| 12:19 |
|
|
iago joined #darcs |
| 12:22 |
|
iago |
hi |
| 12:23 |
|
dixie |
hi |
| 12:24 |
|
Heffalump |
hi |
| 12:25 |
|
Heffalump |
I think if True || FlexibleContexts was because HSE thought FlexibleContexts was needed for some code construct in a case where GHC didn't think that |
| 12:28 |
|
|
owst joined #darcs |
| 12:37 |
|
iago |
Heffalump, could be |
| 12:38 |
|
iago |
I remember I had to add FlexibleContexts to the list of extensions because HSE complains about it |
| 12:45 |
|
|
ManateeL` joined #darcs |
| 12:45 |
|
iago |
Heffalump, did you do something to allow parsing of identifiers with symbol # ? |
| 12:46 |
|
|
ManateeLazyCat left #darcs |
| 12:47 |
|
|
ManateeL` left #darcs |
| 12:52 |
|
|
balor joined #darcs |
| 12:56 |
|
Heffalump |
isn't that just the MagicHash extension? I don't remember having to specifically worry about it. |
| 12:59 |
|
iago |
yes, it is |
| 12:59 |
|
iago |
ok, perhaps is another extension you don't have to specify in the cabal file |
| 13:00 |
|
Heffalump |
it's explicitly specified in a couple of source files, I guess those are the only ones that need it |
| 13:01 |
|
iago |
despite that I'm parsing GHC and I have problems with magic-hash, even with that extension enabled |
| 13:02 |
|
iago |
I also experience more problems with Darcs when I parsing with Annotated |
| 13:02 |
|
iago |
(Language.Haskell.Exts.Annotated) |
| 13:02 |
|
iago |
seems that it doesn't like the use of "rec" as an identifier |
| 13:07 |
|
|
raichoo left #darcs |
| 13:08 |
|
|
shenshei left #darcs |
| 13:08 |
|
iago |
Heffalump, it is mainly used for witnesses, unsafeCoerce# |
| 13:35 |
|
|
balor left #darcs |
| 13:38 |
|
iago |
Heffalump, excuse me again, how did you to handle CPP (since enable CPP extension seems to do nothing) ? I have run cpphs -I but I wonder if there is another way |
| 13:44 |
|
|
raichoo joined #darcs |
| 13:55 |
|
|
raichoo left #darcs |
| 13:58 |
|
Heffalump |
iago: did you look at how hs-export-check invokes HSE? I remember it was a bit tricky |
| 14:03 |
|
iago |
I'm taking a look |
| 14:03 |
|
iago |
how check is supposed to work ? |
| 14:04 |
|
mornfall |
Woo. HEAD passed tests on 6.12.3 Vista, after a long while. |
| 14:04 |
|
mornfall |
Now the question is, what's up with the OSX boxes. |
| 14:04 |
|
mornfall |
Well, box. |
| 14:05 |
|
mornfall |
Seems that hspwd is pretty cranky. |
| 14:05 |
|
mornfall |
Oh. I see. |
| 14:08 |
|
mornfall |
Heffalump: I have sent in a patch (and there are couple more from 2 days ago or so) that should help with the buildbot failures. |
| 14:08 |
|
mornfall |
Heffalump: (The win32 is still relevant, although it is only needed for when some tests fail, to avoid a darcs-test crash...) |
| 14:09 |
|
mornfall |
Heffalump: The remaining 2 should fix the OSX slaves and get us nearer a working GHC 7 one, too. |
| 14:11 |
|
Heffalump |
mornfall: cool |
| 14:12 |
|
Heffalump |
on my win7 laptop, when I run the test suite, I get random complaints from Windows itself about corrupt files. Have you seen anything like that? |
| 14:13 |
|
mornfall |
Hm. Not really, actually. But I haven't tried getting stuff up on win7. |
| 14:13 |
|
Heffalump |
it's kind of scary - how could darcs be doing anything that causes the filesystem to complain |
| 14:17 |
|
mornfall |
If anything, Win7 seems even slower than Vista... :\ |
| 14:19 |
|
mornfall |
Multitasking fail, really. |
| 14:19 |
|
mornfall |
(With HP installer running, even creating a folder gets you into "not responding" situation... kinda icky.) |
| 14:21 |
|
mornfall |
4 minutes to unpack a 400k zip file isn't very optimistic either. :( |
| 14:22 |
|
mornfall |
Well, I can get some food while it ... does stuff. |
| 14:23 |
|
|
sm joined #darcs |
| 14:33 |
|
dixie |
hmm, why sometimes "darcs log -v --match 'hash ...'" output the primite patches enclosed in {} and sometimes not. |
| 14:34 |
|
dixie |
is there anywhere described the format/syntax ? or have look into source code :) |
| 14:40 |
|
Heffalump |
dixie: it may depend on whether it's a darcs 1 or a darcs 2 repo |
| 14:40 |
|
Heffalump |
mornfall: I already wrote patch549 for 2.5.1, I guess I didn't merge it to HEAD/screened yet |
| 14:40 |
|
dixie |
Heffalump: it can be a reason, I observed difference between darcs and ghc repo. |
| 14:41 |
|
mornfall |
Ohwell. :) |
| 14:49 |
|
|
balor joined #darcs |
| 14:53 |
|
|
darcsbb left #darcs |
| 14:53 |
|
|
darcsbb joined #darcs |
| 14:54 |
|
Heffalump |
dixie: it's basically an artefact of code changes between v1 and v2 patches; the on-disk representation of the patches is stuck like that, but we should try to make the output from commands more consistent. |
| 15:01 |
|
mornfall |
Ok, Windows 7 is (almost) rolling now. |
| 15:01 |
|
mornfall |
Just need to reinstall darcs with --global so the slave sees it. |
| 15:02 |
|
Heffalump |
cool |
| 15:02 |
|
Heffalump |
have you tried running the tests yet? |
| 15:02 |
|
mornfall |
No. :) |
| 15:02 |
|
Heffalump |
it'll be interesting to see if my problems reproduce elsewhere |
| 15:02 |
|
mornfall |
But I can try as soon as I have a checkout. |
| 15:09 |
|
|
balor left #darcs |
| 15:22 |
|
|
lelit left #darcs |
| 15:27 |
|
Heffalump |
darcs.net is quite heavily loaded with what seem to be loads of parallel pulls of darcs-screened |
| 15:28 |
|
Heffalump |
mornfall: is that because you're setting up buildbot or something? |
| 15:28 |
|
mornfall |
Don't think so. |
| 15:28 |
|
mornfall |
I don't pull screened so far, in buildbot. |
| 15:40 |
|
darcsbb |
build #96 of 6.10.4 Vista is complete: Success [build successful] Build details are at http://buildbot.darcs.net/buil[…]20Vista/builds/96 |
| 15:47 |
|
|
Fistyhotbaby joined #darcs |
| 15:48 |
|
mornfall |
! |
| 15:48 |
|
mornfall |
:) |
| 15:48 |
|
Fistyhotbaby |
hey |
| 15:48 |
|
Fistyhotbaby |
mornfall |
| 15:48 |
|
mornfall |
Umm. Yes? |
| 15:49 |
|
Fistyhotbaby |
what is darcs? |
| 15:50 |
|
mornfall |
See topic. |
| 15:50 |
|
Fistyhotbaby |
;) |
| 15:50 |
|
Fistyhotbaby |
ok |
| 15:52 |
|
|
Fistyhotbaby left #darcs |
| 15:59 |
|
darcsbb |
build #2 of 6.12.3 Win7 is complete: Success [build successful] Build details are at http://buildbot.darcs.net/buil[…]3%20Win7/builds/2 |
| 16:01 |
|
mornfall |
Owned. |
| 16:01 |
|
Heffalump |
excellent |
| 16:08 |
|
* dcoutts |
is running cabal install http://darcs.net/releases/darcs-2.5.1.tar.gz |
| 16:08 |
|
dcoutts |
so far so good |
| 16:08 |
|
dcoutts |
testing new URL target feature in cabal-install |
| 16:09 |
|
* dcoutts |
notes that this may help for getting users to try out betas that are not on hackage |
| 16:09 |
|
Heffalump |
ooh, yes |
| 16:10 |
|
Heffalump |
it'd be great to be able to drop darcs-beta |
| 16:10 |
|
dcoutts |
right |
| 16:10 |
|
dcoutts |
that was partly the inspiration |
| 16:10 |
|
dcoutts |
plus local file/dir targets |
| 16:10 |
|
|
gwern left #darcs |
| 16:10 |
|
dcoutts |
like cabal install ../lib/thing ../lib/the-other |
| 16:10 |
|
Heffalump |
though mornfall did actually make darcs-fastconvert depend on darcs-beta. But I guess we should just get him not to do that :-) |
| 16:10 |
|
dcoutts |
or cabal install ./foo-1.0.tar.gz |
| 16:13 |
|
dcoutts |
Heffalump: yes, that's probably sensible |
| 16:13 |
|
Heffalump |
from the conversation a while back it didn't work out at all well with cabal anyway |
| 16:13 |
|
dcoutts |
he put the default to be to use darcs-beta |
| 16:14 |
|
mornfall |
Yeah, without realizing it, of course. |
| 16:14 |
|
dcoutts |
which isn't what he really wanted |
| 16:14 |
|
dcoutts |
mornfall: right |
| 16:14 |
|
dcoutts |
it's not really specified what the priority of these soft preferences is |
| 16:14 |
|
mornfall |
Now, did I upload 0.2.1? Doesn't seem to be the case... |
| 16:15 |
|
dcoutts |
we don't quite have enough experience with the solvers to work out if we can actually sensibly distinguish levels of preference |
| 16:15 |
|
mornfall |
Let's see. |
| 16:15 |
|
mornfall |
Ok, the beta SNAFU should be fixed now. :) |
| 16:16 |
|
dcoutts |
instead you'll be able to tell people to do things like: cabal install http://blah/darcs-x.y.tar.gz http://blah/darcs-fastconvert-x.y.tar.gz |
| 16:16 |
|
dcoutts |
it'll download both, resolve deps, pull down any other deps from hackage and install |
| 16:16 |
|
mornfall |
Yeah, sounds reasonable. |
| 16:17 |
|
* Heffalump |
realises he still needs to figure out how to send patches from Windows and not get a result like http://bugs.darcs.net/patch538 |
| 16:17 |
|
mornfall |
Ah indeed, that was a bit cranky. |
| 16:18 |
|
mornfall |
I had to grab it from screened. |
| 16:18 |
|
mornfall |
Which makes me wonder why is the patch still needs-review. |
| 16:18 |
|
mornfall |
Oh, prolly because darcswatch couldn't parse the bundle. |
| 16:20 |
|
Heffalump |
I think darscwatch may be asleep anyway, it didn't notice something else I pushed to screened the other day |
| 16:24 |
|
* Heffalump |
has "fun" trying to figure out what's wrong with this MIME encoding |
| 16:26 |
|
|
raichoo joined #darcs |
| 16:26 |
|
mornfall |
Heffalump: Could be extra carriage returns? |
| 16:26 |
|
Heffalump |
that would be the most likely candidate, yeah |
| 16:27 |
|
mornfall |
It should be possible to test by sending --to self. |
| 16:27 |
|
mornfall |
If nothing else. |
| 16:27 |
|
mornfall |
(Not sure we could come up with an automated test for this, though... I don't think windows comes with a standalone mime decoder or anything...) |
| 16:28 |
|
Heffalump |
I'm just comparing the emails I get by sending to me on windows and linux |
| 16:28 |
|
|
gwern joined #darcs |
| 16:28 |
|
|
gwern left #darcs |
| 16:28 |
|
|
gwern joined #darcs |
| 16:29 |
|
|
kolmodin joined #darcs |
| 16:31 |
|
Heffalump |
can't we just test that the result is byte-for-byte what we expect? |
| 16:31 |
|
mornfall |
We probably can. |
| 16:31 |
|
mornfall |
(Modulo dates and stuff, but shouldn't be hard to fix up. Or we can use a canned repository.) |
| 16:32 |
|
Heffalump |
yeah |
| 16:37 |
|
Heffalump |
hmm, the two emails are completely different |
| 16:46 |
|
darcsbb |
build #105 of 6.12.3 Vista is complete: Failure [failed test] Build details are at http://buildbot.darcs.net/buil[…]0Vista/builds/105 blamelist: Ganesh Sittampalam <ganesh earth.li>, Ganesh Sittampalam <ganeshearth.li>, Ganesh Sittampalam <ganeshearth.li>, Ganesh Sittampalam <ganeshearth.li>, Ganesh Sittampalam <ganeshearth.li>, Eric Kow <koweydarcs.net>, Eric Kow <koweydarcs.net>, Ganesh Sittampalam <ganeshearth.li>, Ganesh Sittampalam |
| 16:49 |
|
mornfall |
Kablaam. "Bad file number" whatever that means... |
| 16:51 |
|
mornfall |
Heffalump: Btw. I didn't get any complaints from my windows 7 box so far (I ran the suite manually once, and buildbot ran it twice). |
| 16:56 |
|
Heffalump |
interesting, that's coming from hspwd |
| 16:56 |
|
Heffalump |
oh, I don't think that has the >/dv/null patch |
| 16:57 |
|
Heffalump |
at least, the quoted ghc command doesn't have it, but perhaps it doesn't display that |
| 16:57 |
|
mornfall |
It does. But I am not sure it could be related. |
| 16:57 |
|
mornfall |
Well, I have no idea how /dev/null works on MSYS. |
| 16:58 |
|
Heffalump |
I'm sure more tests than that one use pwd. |
| 16:58 |
|
mornfall |
It's definitely random. |
| 16:58 |
|
mornfall |
It failed just once. |
| 16:58 |
|
mornfall |
Even if the test ran 3 times. |
| 16:59 |
|
Heffalump |
got to love Windows |
| 16:59 |
|
Heffalump |
seems my win7 is rather more unreliable than yours, because I also get random permission denied errors presumably caused by timing issues |
| 17:01 |
|
|
sm_ joined #darcs |
| 17:03 |
|
mornfall |
It's probably also a lot slower (which may mean the races don't happen due to general slowness). |
| 17:04 |
|
|
sm left #darcs |
| 17:04 |
|
|
sm_ is now known as sm |
| 17:06 |
|
Heffalump |
you think yours is slower? Mine is only a laptop, albeit a fairly new one. |
| 17:11 |
|
Igloo |
It might be interesting to try with things like virus checkers turned off, but I don't know if there's anything useful you can do with the results of the experiment |
| 17:16 |
|
|
jutaro joined #darcs |
| 17:16 |
|
mornfall |
Well, I have a clean win7 install, running in qemu-kvm. |
| 17:16 |
|
mornfall |
It seems to be IO-bound, too. |
| 17:16 |
|
mornfall |
Heffalump: Anyway, if yours is as slow as mine, I sure wouldn't want to use it. |
| 17:21 |
|
Igloo |
I meant for the random failures, not for performance |
| 17:24 |
|
|
shenshei joined #darcs |
| 17:41 |
|
iago |
does cabal test run quickcheck tests? |
| 17:42 |
|
mornfall |
Not by default. |
| 17:42 |
|
mornfall |
But I think you can ask it to. |
| 17:44 |
|
iago |
cabal test --help does not say anything useful with that respect |
| 17:50 |
|
|
iago left #darcs |
| 17:56 |
|
|
iago joined #darcs |
| 18:06 |
|
mornfall |
Ok, you can't directly. You need to run darcs-test manually for that. |
| 18:07 |
|
mornfall |
But it wouldn't be hard to extend Setup.hs to allow running unit tests (see runTests in Setup.lhs). |
| 18:19 |
|
iago |
with darcs-test I can run ALL tests? |
| 18:19 |
|
mornfall |
What'd ya mean ALL? |
| 18:19 |
|
Heffalump |
yes |
| 18:19 |
|
iago |
all means all |
| 18:19 |
|
mornfall |
You can run unit & functional tests that we have. |
| 18:20 |
|
mornfall |
And that are enabled etc. |
| 18:20 |
|
iago |
and, excuse for a non-darcs related question but |
| 18:20 |
|
iago |
$ hpc report hpctixdir/sum.tix |
| 18:20 |
|
iago |
69% expressions used (41706/60263) |
| 18:20 |
|
iago |
can I assume this is the % of executed code? |
| 18:21 |
|
iago |
someone knows? |
| 18:21 |
|
Heffalump |
I guess it means what it says, but that seems a reasonable way to judge percentage coverage |
| 18:21 |
|
Heffalump |
rather than by line count, which would be rather less precise |
| 18:24 |
|
iago |
I just have doubts about what is an expression to be executed, and how they are count |
| 18:24 |
|
iago |
if B then X else Y, when only then-branch is taken... |
| 18:24 |
|
iago |
the whole expression is considered to be executed? |
| 18:24 |
|
iago |
.. |
| 18:24 |
|
iago |
etc |
| 18:27 |
|
Heffalump |
dunno, I imagine the hpc docs cover this. There was also a paper when it first came out. |
| 18:28 |
|
iago |
I will look for the paper |
| 18:28 |
|
iago |
I have read the doc that comes from GHC (which is the pointed in the HPC page) |
| 18:28 |
|
iago |
just says this Source coverage is the extent to which every part of the program was used, measured at three different levels: declarations (both top-level and local), alternatives (among several equations or case branches) and expressions (at every level). |
| 18:36 |
|
Heffalump |
I'm sure you could figure out the precise answer with some experimentation then. |
| 18:36 |
|
iago |
yeah... maybe the best option |
| 18:37 |
|
Heffalump |
I would guess that if the then branch of if B then X else Y is taken, the whole expression will be covered but only one alternative will be covered |
| 18:37 |
|
Heffalump |
since it measures both expressions and alternatives |
| 18:37 |
|
Heffalump |
are you plannign to get hpc running again for darcs? |
| 18:37 |
|
|
jutaro left #darcs |
| 18:40 |
|
iago |
for analysis purposes I'm working with 2.5, but as part of testing work I expect to fix hpc.README to work with screened |
| 18:40 |
|
Heffalump |
cool. |
| 18:41 |
|
iago |
I don't know why but with 2.5 I get a sum.tix file... with screened I get lot of *.tix files |
| 18:41 |
|
iago |
now I'm trying to sum them with hpc sum, though it is taken lot of time |
| 18:49 |
|
|
owst left #darcs |
| 19:06 |
|
|
gwern left #darcs |
| 19:07 |
|
|
gwern joined #darcs |
| 19:07 |
|
|
gwern left #darcs |
| 19:07 |
|
|
gwern joined #darcs |
| 19:20 |
|
mornfall |
iago: If 2.5 has the old shell harness, I had code in there that did the summing along the way. |
| 19:20 |
|
mornfall |
iago: The hpc summer is extremely slow (I think it's not linear in input size, so doing it many times over fewer files is a lot faster). |
| 19:21 |
|
mornfall |
Also, hpc by default wastes insane amount of disk space (it'll eat gigabytes for the darcs testsuite, even though the data would fit in maybe 100k...) |
| 19:21 |
|
mornfall |
The problem is that for every darcs invocation, it creates a new file and there's no way to tell it to just increment counters in an existing file. |
| 19:23 |
|
iago |
if you have a script to sum them then it would be very helpful |
| 19:23 |
|
iago |
with screened I get more than 6k .tix files |
| 19:23 |
|
mornfall |
I only have what's in Distribution/ShellHarness |
| 19:23 |
|
iago |
I ctrl-C hpc sum after an hour... |
| 19:23 |
|
mornfall |
You could probably port it to darcs-test. |
| 19:30 |
|
iago |
mornfall, your summer sums tix 2 by 2 ? |
| 19:31 |
|
mornfall |
I don't remember, let me look. |
| 19:31 |
|
mornfall |
iago: It actually sums after each test. |
| 19:31 |
|
iago |
ok I see |
| 19:31 |
|
iago |
it sums all |
| 19:31 |
|
mornfall |
So it's a couple of files every time. |
| 19:31 |
|
iago |
but it is run after each test |
| 19:31 |
|
mornfall |
Between 1 and maybe ten or twenty. |
| 19:32 |
|
mornfall |
Yes. |
| 19:32 |
|
mornfall |
Gotta go, anyway. |
| 19:33 |
|
iago |
ok thanks |
| 19:40 |
|
|
jeltsch joined #darcs |
| 19:43 |
|
iago |
how can I run unit tests in darcs 2.5 ? |
| 19:43 |
|
iago |
I see "cabal test" only run shell-tests |
| 19:44 |
|
darcsbb |
build #106 of 6.12.3 Vista is complete: Exception [exception darcs] Build details are at http://buildbot.darcs.net/buil[…]0Vista/builds/106 blamelist: Alex Suraci <i.amtoogeneric.com>, Alex Suraci <i.amtoogeneric.com>, Eric Kow <koweydarcs.net>, Ganesh Sittampalam <ganeshearth.li> |
| 19:49 |
|
Heffalump |
iago: there's already a darcs-test binary to run |
| 19:49 |
|
Heffalump |
it just doesn't do the shell tests |
| 19:49 |
|
Heffalump |
maybe it's called 'unit' actually |
| 19:50 |
|
iago |
Heffalump, are you talking about screened? |
| 19:51 |
|
Heffalump |
no |
| 19:51 |
|
Heffalump |
2.5 |
| 19:51 |
|
iago |
oh ok I see |
| 19:51 |
|
iago |
unit |
| 19:51 |
|
iago |
uhmm no |
| 19:51 |
|
iago |
unit is just a directory |
| 19:52 |
|
Heffalump |
you might need to build with -ftest |
| 19:55 |
|
iago |
are you interested in the coverage of shell tests? |
| 19:55 |
|
Heffalump |
yes |
| 19:55 |
|
iago |
actually I think the important thing is unit testing |
| 19:55 |
|
Heffalump |
perhaps in theory, but those only cover the patch modules |
| 19:55 |
|
Heffalump |
so the rest is important too |
| 19:56 |
|
|
lispy joined #darcs |
| 19:57 |
|
iago |
Heffalump, what are the sources that generate darcs-test ? |
| 19:57 |
|
sm |
if you darcs tag again with an existing tag name, does it move the old one by any chance ? |
| 20:04 |
|
|
pumpkin joined #darcs |
| 20:06 |
|
|
copumpkin left #darcs |
| 20:15 |
|
Heffalump |
iago: src/test.hs I think |
| 20:15 |
|
Heffalump |
sm: it just hides it |
| 20:15 |
|
Heffalump |
it's like making a patch with the same name |
| 20:16 |
|
Heffalump |
makes it harder to get at in the UI, but it's still there |
| 20:17 |
|
sm |
Heffalump: thanks |
| 20:19 |
|
|
balor joined #darcs |
| 20:28 |
|
Heffalump |
mornfall: you know shellish doesn't build with GHC 7? |
| 20:30 |
|
Heffalump |
and hashed-storage has some probably mtl-related issue |
| 20:40 |
|
|
pumpkin left #darcs |
| 20:40 |
|
|
copumpkin joined #darcs |
| 20:41 |
|
sm |
alexsuraci: hoping to raise darcsden's profile a little: http://thread.gmane.org/gmane.[…]askell.cafe/86330 |
| 20:41 |
|
alexsuraci |
heh, nice |
| 21:03 |
|
dixie |
alexsuraci: i have added some patch viewing into that toyish slow web repo viewer :) but it is quite dirty mess of code... http://tinyurl.com/6956wp5 |
| 21:05 |
|
dixie |
I hope it doesn't cause a harm that the web design is very similar to darcsden (at least some effects) :) |
| 21:05 |
|
alexsuraci |
heh, no problem |
| 21:05 |
|
alexsuraci |
pretty neat |
| 21:06 |
|
|
preflex left #darcs |
| 21:09 |
|
dixie |
hope I will figure out some cache schema since it is very slow. the VM where it run is resource-less too. But I guess there are not a too much repos scattered on private serves. With repos on patchtag/darcsden it doesn't make sense. |
| 21:10 |
|
* dixie |
bbl. |
| 21:11 |
|
|
preflex joined #darcs |
| 21:17 |
|
|
JaffaCake joined #darcs |
| 21:19 |
|
|
copumpkin left #darcs |
| 21:19 |
|
|
JaffaCake1 left #darcs |
| 21:27 |
|
Heffalump |
mornfall: is there a repo for the hashed-storage 0.5 branch? |
| 21:30 |
|
mornfall |
Heffalump: <head>/0.5 (http://repos.mornfall.net/hashed-storage/0.5 IIRC) |
| 21:30 |
|
mornfall |
I'll look at shellish. |
| 21:30 |
|
Heffalump |
mornfall: oh yes. I did actually look there but missed it :-) |
| 21:30 |
|
mornfall |
(cmdlib 0.3 should build with GHC 7) |
| 21:30 |
|
Heffalump |
yeah, I pushed that patch to darcs |
| 21:30 |
|
Heffalump |
your patch for the updated dep that is |
| 21:30 |
|
Heffalump |
actually, hashed-storage doesn't seem to build with any mtl on the latest GHC 7.0 snapshot |
| 21:31 |
|
* Heffalump |
is confused |
| 21:36 |
|
Heffalump |
ah. instance (Monad m, MonadError e m) => TreeRO (TreeMonad m) where |
| 21:37 |
|
Heffalump |
I bet the unbound on the 'e' on the RHS is problematic |
| 21:38 |
|
Igloo |
See http://hackage.haskell.org/trac/ghc/ticket/4952 |
| 21:39 |
|
Heffalump |
ah, so it'll be fixed? |
| 21:40 |
|
Heffalump |
cool, thanks. How did you find it? |
| 21:40 |
|
Igloo |
Hackage regression testing |
| 21:40 |
|
Heffalump |
excellent :-) |
| 21:40 |
|
Heffalump |
looking at the code, I don't really blame GHC for complaining though. |
| 21:41 |
|
Heffalump |
mornfall: what is the MonadError constraint actually for? I can't spot any direct uses of it. |
| 21:46 |
|
|
jeltsch left #darcs |
| 21:56 |
|
mornfall |
Heffalump: Could be a leftover. |
| 21:59 |
|
|
jutaro joined #darcs |
| 22:03 |
|
Heffalump |
I'll see what happens when I get rid of it then :-) |
| 22:04 |
|
Heffalump |
mornfall: the 0.5 branch doesn't have 0.5.4 |
| 22:05 |
|
Heffalump |
seems like the constraint is totally unnecessary :-) |
| 22:15 |
|
* Heffalump |
sends a patch in case it's useful |
| 22:18 |
|
iago |
there is some plan for darcs 3 ? |
| 22:18 |
|
Heffalump |
I have one in my head, if that helps :-) |
| 22:18 |
|
Heffalump |
at least for the patches |
| 22:18 |
|
iago |
Some "special" plan I mean |
| 22:19 |
|
iago |
theoretically Camp would be a source of inspiration for darcs 3 or something, ? |
| 22:19 |
|
iago |
Heffalump, re-write all code? :P |
| 22:20 |
|
Heffalump |
probably, though not certainly. |
| 22:20 |
|
iago |
I ask because I realize you will release 2.8 "in short" |
| 22:20 |
|
iago |
arg |
| 22:20 |
|
iago |
briefly |
| 22:20 |
|
Heffalump |
2.8 is due this summer. I think there's actually quite a bit of work to do for all the things we'd like to get in to it |
| 22:21 |
|
Igloo |
In case it isn't clear, 2.8 can be followed by 2.10, it doesn't mean 3.0 is next |
| 22:21 |
|
Heffalump |
right |
| 22:21 |
|
iago |
well, yeah, of course |
| 22:21 |
|
* Heffalump |
wonders if now is a good time to explain graphy things to Igloo |
| 22:21 |
|
iago |
perhaps that is what will happen |
| 22:22 |
|
Igloo |
Not really, sorry |
| 22:26 |
|
|
jutaro left #darcs |
| 22:35 |
|
mornfall |
Heffalump: Sorry, I have pushed to the public repo. |
| 22:36 |
|
Heffalump |
ok, that does conflict with what I sent. Are you happy to apply that? If so I'll send a new version. |
| 22:37 |
|
mornfall |
Let me have a look. |
| 22:37 |
|
mornfall |
Oooo, new send format. :D |
| 22:37 |
|
mornfall |
Yeah, I don't have a problem with that. |
| 22:42 |
|
|
copumpkin joined #darcs |
| 22:54 |
|
|
owst joined #darcs |
| 23:05 |
|
iago |
there is some way to select just patches from a given author for unrecord ? |
| 23:05 |
|
mornfall |
--match 'author ...' |
| 23:08 |
|
mornfall |
Heffalump: I have pushed your cleanup. (Do we also need a release?) |
| 23:11 |
|
iago |
thanks |
| 23:12 |
|
|
raichoo left #darcs |
| 23:37 |
|
|
twb joined #darcs |
| 23:41 |
|
|
balor left #darcs |
| 23:44 |
|
iago |
mornfall, (about hashed-storage) there is some (easy) way to work with a pure (non-IO) Tree in memory and then persist it to disk? |
| 23:45 |
|
mornfall |
iago: Not that I can think of. The types don't work out. |
| 23:46 |
|
iago |
would not be possible to have some Tree m1 -> Tree m2 conversion ? |
| 23:47 |
|
iago |
I'm using a copy of h-s Tree data type as a RepoModel for patch testing |
| 23:47 |
|
iago |
I would like to rewrite it to just use Tree Identity |
| 23:47 |
|
mornfall |
Well, the question is why you want to persist the Tree? |
| 23:47 |
|
iago |
but despite avoid code duplication (which is not a lot), the interesting case is to use h-s to persist it and make tests involving application of patches |
| 23:48 |
|
iago |
for example, I could test that a sequence of patches and its commutation produces the same tree |
| 23:48 |
|
mornfall |
iago: You don't need to persist for that, though. |
| 23:48 |
|
iago |
:? |
| 23:49 |
|
mornfall |
You can compare Trees in memory. |
| 23:49 |
|
iago |
Darcs.Patch.Apply code is able to handle any kind of Tree |
| 23:49 |
|
mornfall |
The problem with "converting" the monads is that the Tree will have items of the type (m a) in it, and you'd need to provide m1 a -> m2 a for this to work. |
| 23:49 |
|
iago |
even Tree Identity ? |
| 23:50 |
|
mornfall |
So you can't go from IO to anything anyway. |
| 23:50 |
|
mornfall |
Or, well. |
| 23:50 |
|
mornfall |
Maybe you can. |
| 23:50 |
|
mornfall |
Hm. |
| 23:50 |
|
mornfall |
iago: Well, I don't know, but the straightforward thing would be to try using Tree Identity for everything and just not involve disk at all. |
| 23:51 |
|
iago |
well, you are right anyway |
| 23:53 |
|
iago |
now I'm wondering why I did not think in that before :) |
