Check your feather email | spec.pugscode.org | paste: sial.org/pbot/perl6 | pugs.blogs.com | www.treehugger.com/files/th_images/paradigm.jpg
Set by audreyt on 29 August 2006.
TreyHarris anyone here who has a moment to look at a few lines of MMD logic to see if it's sane? 00:01
00:04 weinig is now known as weinig|bbl
markstos TreyHarris: Still there? I'll look at MMD 00:11
You might check examples/dispatch_quiz.pl though-- it shows some cases that I think are bugs. 00:12
pasteling "TreyHarris" at 216.254.17.198 pasted "markstos: have a look" (16 lines, 412B) at sial.org/pbot/19457 00:13
markstos will do. give me 5.
TreyHarris thx
00:15 buetow joined
markstos TreyHarris. I think it's sane, but pugs is buggy WRT () vs (*%h) 00:15
That's one of the cases in the dispatch_quiz.pl
I would be more assertive about creating related tests, but the spec is not detailed about how MMD happens with optional and slurpy fields. 00:16
I asked p6l to clarify recently
and you may recall I also added some related language to mmd-draft.txt
To get code working, I combined the () and (*%h) cases into one method ... 00:17
with an "if %h { ... }" clause.
TreyHarris right. this is for a draft i'm writing, so it doesn't have to work :-)
yet. 00:18
thanks much
markstos Let me check the dispatch quiz result.
Those are actually the first a second tests in the quiz. Pugs does different things for subs and methods in that case! 00:19
TreyHarris *grin* 00:20
markstos The subs case looks correct, it sends (a => 'b') to (*%h).
the multi method case sends ( a=>'b') to (), which seems ... wrong.
TreyHarris yes 00:24
markstos I'll go ahead and poke around to make sure it is covered by a test.
00:25 hikozaemon joined
markstos Yes, it looks like I already made a failing test for it in t/oo/methods/multi.t 00:26
(And 3 other cases from the quiz!)
00:27 nekokak joined 00:40 frederico joined
nothingmuch wow, adding roles made it even slower =) 01:16
TreyHarris whoo-hooh! push for O(Inf)! 01:17
nothingmuch ;-)
push for correctness actually
i am explicitly not memoizing anything
nothingmuch.woobling.org/MO/t/roles.t 01:18
TreyHarris O(Inf) is just, after all, O(c), which is equivalent to O(1)....
nothingmuch heh
porting this 1:1 to haskell should speed it up a bit
i also want to wirte Moose::Aspect::Memoize 01:19
which implicitly declares attrs over methods
and does auto caching
with lazy => 1
and auto expires caches using triggers
which should bring this impl closer to the haskell version in terms of O(n) complexity
01:27 ayrnieu left
nothingmuch audreyt: when you backlog: nothingmuch.woobling.org/MO/t/roles.t 01:29
01:43 syntaxfree joined
markstos ?eval class Foo { has $.a; method a { delete $.a } } Foo.new.a; 02:12
02:12 evalbot_13026 is now known as evalbot_13028
evalbot_13028 () 02:12
markstos Shouldn't it throw an error to try to delete something that's not a hash or array element ?
markstos checks the spec
I don't see this behavior spec'ed. I'll review the tests, and test for the kind of msg Perl 5 would throw: 02:16
delete argument is not a HASH or ARRAY element or slice
02:19 revdiablo joined, b00t joined
svnbot6 r13029 | markstos++ | [t/builtins/arrays_and_hashes/delete.t] 02:25
r13029 | markstos++ | New failing test for delete: mis-use of delete should return:
r13029 | markstos++ | Argument is not a Hash or Array element or slice
SamB @timefor audreyt 02:31
lambdabot Unknown command, try @list
SamB @localtime audreyt
lambdabot Local time for audreyt is Tue Sep 5 04:31:39 2006
markstos @localtime markstos 02:32
lambdabot Local time for markstos is Mon Sep 4 22:28:16 2006
markstos That's the wrong time for Audreyt. It's reporting the time on feather, which is located in France I believe. 02:33
Audrey's localtime is Tues, 10:28am, I think.
bsb hi markstos, I just responded to your p6l qn 02:49
with non-authoritative guesses :) 02:50
02:52 weinig|bbl is now known as weinig
markstos bsb: thanks 02:52
02:52 reZo joined
TreyHarris @localtime TreyHarris 02:53
lambdabot Local time for TreyHarris is Mon Sep 4 19:53:36 2006
02:55 Aankh|Clone joined, weinig is now known as weinig|zZz
TreyHarris anyone know if there's a way to tell a mac's dvd drive to spin down? i TSTP'd the program that was using it, but it still spins.... 02:55
02:59 revdiablo joined
syntaxfree my Mini does that sometime. 03:18
03:19 Bit-Man joined 03:39 markstos left
TreyHarris how can I require an arbitrary file not in @*INC? 03:44
ah, evalfile 03:45
03:46 frankg joined 03:48 frankg joined 03:58 Khisanth joined 04:08 frankg joined
masak @localtime masak 04:08
lambdabot Local time for masak is Tue Sep 5 06:08:26
masak yep
04:23 aufrank joined 04:58 agentzh joined
agentzh markstos: in case you don't know, there's a even shorter url for feather.perl6.nl, that is, perlcabal.org, which points to exactly the same IP. 04:59
lambdabot Title: Perl6 Community Development Server
agentzh (bbiab) 05:00
05:00 agentzh left
audreyt yeah, I registered it and used it as a, sortof ironic, sister domain to perlcabal.com 05:09
lambdabot Title: 404 Not Found
audreyt (don't be fooled by the 404 :)) 05:10
05:13 justatheory joined
araujo coding a bit on the Types Graph 05:19
audreyt araujo++
araujo :-)
hi there audreyt
audreyt heya 05:20
audreyt at $job
araujo is _supposed_ to be UP within 4 hours :-) 05:21
audreyt wow :) 05:22
05:25 EdLin joined
gaal hey 05:32
audreyt: I'd like to fix the use of existentialCoerce in Parser 1428 (that is, change Param.p_unpacking back to Maybe PureSig) 05:37
but as it is there's a type error
I hate it when Code.hs becomes stale. /me splines for the beziers or ghc 6.6 05:41
TreyHarris erm. 05:54
?eval sub foo($a) { say "hi $a" }; foo("mom")
05:55 evalbot_13028 is now known as evalbot_13029
evalbot_13029 OUTPUT[hi mom ] Bool::True 05:55
TreyHarris ok, wasn't sure it was allowed to leave out the space between the subname and the sig
gaal you are, but the style used thoroughly in existing code is to use the space there. 05:57
this makes me wonder though if the space is optional here: sub foo:($a) { ... } 05:58
(a Sig in a sub decl really is a Sig, dropping the colon is "just" an abbreviation) 05:59
05:59 [particle] joined
TreyHarris well, method self:meth can be defined for .= meth operations, but I don't think that causes any ambiguity 06:00
araujo off to sleep 06:01
06:07 marmic joined
TreyHarris is there a Perl 5 impl of sub MAIN yet? 06:07
gaal I haven't heard of one 06:08
jdv79 MAIN? 06:28
06:32 kasei joined
kasei does anyone know if new ghc dists are being compiled for debian stable? 06:33
06:42 traecer joined 06:44 frederico joined
TreyHarris gaal: how do I refer to the current type symbolically? "my $?CLASS $x" doesn't work, nor would I expect it to... 06:58
gaal jdv79: it's a mechanism for a default entry point to executables, that takes care of command line options automatically
TreyHarris: in signatures? 06:59
TreyHarris signatures or storage defs
gaal you need to already have bound a tyvar, so this:
:(Mammal ::T $x, ::T y) 07:00
will bind ::T to say Moose at call time
er, that's $y in there
I don't know how to bind a tyvar in other scopes so that your example works
TreyHarris hmmm. i should be able to write class Moose { method clone ( --> <mumble>) { ... } } without making <mumble> eq "Moose" 07:03
I kinda think 'Self' (capital S) should work 07:05
or maybe SELF
gaal better ask the oo guys. from my unwashed viewpoint, I don't see that Moose should know how to clone arbitrary types correctly, though sure, superficially it should be able to. 07:06
07:11 kanru joined
TreyHarris hmm. looks like ::?Class may be what I'm looking for 07:14
07:19 iblechbot joined
svnbot6 r13030 | gaal++ | * fix 'make ghci' on colinux and possibly Windows. I hope this doesn't 07:33
r13030 | gaal++ | break it elsewhere.
r13030 | gaal++ | * noticed that assert_ghc in Module::Install::Pugs litters ghc_flags
r13030 | gaal++ | with threading options coming from perl5, so I added a crutch to
r13030 | gaal++ | filter those out via the environment (set PUGS_NO_THREADS). I don't
r13030 | gaal++ | know if this should be moved to config.yml, or be made to use some
r13030 | gaal++ | heuristic to just do the right thing automatically (if there is a
r13030 | gaal++ | right thing here: can ghci work correctly when linked against a
r13030 | gaal++ | threaded perl5?).
gaal wow! pugs in ghci shares readline buffers with its ghci! that's so cool 07:37
svnbot6 r13031 | gaal++ | * fix whitespace bug in Signature parsing, reported by salty-horse++: 07:42
r13031 | gaal++ | :($x = 42 is rw)
r13031 | gaal++ | now parses again.
lumi ls 07:44
gaal . .. 07:45
lumi Erp. Wrong window
Wrong computer
Wrong os, even
gaal Did you get the universe right? 07:46
lumi Yes, it's a start, right?
07:48 justatheory joined
gaal what does the Param object look like in unspecified-invocant Sigs? :( : $moose) 07:49
is there a varname? $?SELF would be wrong 07:50
07:55 neoesque joined 08:10 drrho joined
TreyHarris 'self'? it's what you use when you don't specify a varname... 08:12
gaal it's a callconv question, I guess. you can't do this after all: foo( self => $obj ) 08:16
hmm, maybe you can?
but that means there can be no parameters called self.
08:26 ruoso joined
ruoso www.perlmonks.org/?node_id=571140 08:28
lambdabot Title: Mary had a Little Camel
ruoso one of the most fun posts I read in the last weeks
svnbot6 r13032 | trey++ | [t/oo/attributes/recursive.t] 08:33
r13032 | trey++ | Added new tests for recursive classes--
r13032 | trey++ | classes with attributes of the same type
r13032 | trey++ | as the class (or the ::?CLASS symbolic name).
08:35 elmex joined 08:39 kane-xs joined
TreyHarris oops 08:43
karma classes
perlbot: karma classes
perlbot Karma for classes: -2
TreyHarris classes++
perlbot: karma classes
perlbot classes doesn't have any karma
TreyHarris better :-)
08:44 markstos joined
drrho is away: cinema 08:45
09:00 chris2 joined 09:13 traecer left
gaal audreyt: I don't understand dist/build: why are some things under src/ there and some not? 09:36
I'm getting errors linking Pugszm6zi2zi12_PugsziEmbedziPerl5_perl5zueval_closure
the symbol is in dist/build/Pugs/Embed/Perl5.o 09:37
why not dist/build/src etc.?
the actual error is after 'make ghci', running 'main' in the ghci prompt: 09:39
<interactive>: src/Pugs/Run/Perl5_stub.o: unknown symbol `Pugszm6zi2zi12_PugsziRunziPerl5_zdfpugszueval_closure'
I had it working for a blissful hour :(
09:46 meppl joined
svnbot6 r13033 | trey++ | [util/version_h.pl] 09:46
r13033 | trey++ | Script expects cwd to be base of the working
r13033 | trey++ | copy. Made that so when using svn. I don't
r13033 | trey++ | know how to do it with svk, so hopefully
r13033 | trey++ | someone else can come behind and fix that.
r13033 | trey++ | It's no more broken for svk than it was
r13033 | trey++ | before...
r13033 | trey++ | [util/smartlinks.pl]
r13033 | trey++ | Made file executable
r13034 | trey++ | [t/oo/attributes/class.t] 09:52
r13034 | trey++ | Tests for Class.^attr syntax;
r13034 | trey++ | better smartlinks
09:54 meppl joined 10:03 _meppl joined
svnbot6 r13035 | trey++ | [t/blocks/return.t] 10:08
r13035 | trey++ | Fixed old slurpy syntax
10:11 _meppl is now known as meppl 10:13 int-e joined
svnbot6 r13036 | trey++ | [t/blocks/return_function.t] 10:20
r13036 | trey++ | Fixed smartlink; added an actual test for
r13036 | trey++ | a line that was tried but whose value was
r13036 | trey++ | thrown away.
10:27 Grrrr joined 11:00 buetow joined 11:03 TimToady joined 11:26 iblechbot joined 11:41 mako132_ joined 11:42 justatheory joined 12:07 iblechbot joined 12:21 Limbic_Region joined, hermax_ joined
audreyt gaal: oy. the idea is that "make ghci" should ignore embedding 12:36
and _not_ link in Perl5_stub 12:37
i.e. you never get p5embed in ghci
can you simply remove the _stub.o in src/?
"make ghci" uses src/ to put the .o files
normal "make" uses dist/build/
totally separate spaces
(I lost my old laptop's adaptor at $job, and new MacBook is still running ddrescue, so I'm stuck on this iBook G4 without much productivity tonight :/) 12:38
hopefully ddrescue will finish on tomorrow morning
markstos++ # subtle nuge at pugs.blogs.com for full MOP dispatch support 12:39
nudge, even
and with that... over and out for tonight :) &
Limbic_Region yeah - that one liner is quite impressive when taking in context of the amount of p5 code it is replacing 12:40
sleep well audreyt
dunno if it is a good thing or not, but the binary snapshots appear to have ceased being built as of the 1st 12:41
12:45 drbean joined
svnbot6 r13037 | agentz++ | [util/smartlinks.pl] 12:57
r13037 | agentz++ | - various English fixes from Sun Xin++ for the Pod.
SamB ?eval 1 12:59
12:59 evalbot_13029 is now known as evalbot_13037
evalbot_13037 1 12:59
SamB @tell audreyt please restart evalbot-loop, kthxbye! 13:03
lambdabot Consider it noted.
13:04 bpphillips joined 13:08 svnbot6 joined, evalbot_r13037 joined 13:12 rafl joined 13:18 cjeris joined
svnbot6 r13038 | agentz++ | [t/syntax/unicode.t] 13:20
r13038 | agentz++ | - broke too long lines
r13038 | agentz++ | - used eval_is to work around a parsefail.
13:23 vel joined 13:25 agentzh joined
agentzh audreyt: there's a typo in S04: "When used as statement modifers". 13:25
audreyt: it should have been "modifiers" 13:26
svnbot6 r13039 | agentz++ | [t/statements/modifiers/for.t] 13:29
r13039 | agentz++ | - adjusted the order of the tests a bit.
SamB agentzh: looks like she went to bed 13:33
agentzh SamB: oh, she sleeps quite early now. :)
svnbot6 r13040 | agentz++ | [util/smartlinks.pl]
r13040 | agentz++ | - now we would not try any further if the first match
r13040 | agentz++ | of a certain smartlink had been found.
agentzh it's just 21:32 in her timezone. :) 13:34
SamB heh
Limbic_Region well, I don't think she sleeps on a regular schedule 13:35
agentzh hehe
Limbic_Region and if she has been up for a prolonged period of time and tomorrow is a $work day - she may just be ensuring she gets some rest
and judging from her over and out, I don't think she has gone to sleep - just off channel so she doesn't get distracted
SamB oh 13:38
hmm
she did say & 13:39
that does sort of imply some kind of continued activity
theorbtwo People on this channel tend to always say &.
Limbic_Region normally it is /me & #<verb that will take us away from the channel for an extended period of time> 13:44
as in /me & # sleep
lumi #perl6 madlibs 13:45
Limbic_Region s/libs// 13:46
;-)
13:46 xerox joined 13:51 ofer0 joined
Limbic_Region anyone else having trouble building an optimized pugs? 13:51
or rather - the pugs portion completes ok - it is the post build activities such as P::C::R 13:52
pasteling "Limbic_Region" at 129.33.119.12 pasted "Current Pugs build output" (147 lines, 12.4K) at sial.org/pbot/19463 13:53
svnbot6 r13041 | agentz++ | [t/TASKS]
r13041 | agentz++ | - added one task of updating legacy smartlink notations
r13041 | agentz++ | in the test suite.
Limbic_Region I have a pugs.exe but apparently it is somewhat b0rk
svnbot6 r13042 | agentz++ | [t/README] 13:59
r13042 | agentz++ | - mentioned the URL perlcabal.org/syn/ there
lambdabot Title: Official Perl 6 Documentation
14:05 frederico joined 14:10 amnesiac joined
gaal audreyt: yay, rm src/Pugs/Run/Perl5_stub.o worked! 14:19
svnbot6 r13043 | agentz++ | [util/smartlinks.pl]
r13043 | agentz++ | - added support for generic links used in the test suite.
r13043 | agentz++ | Notations like
r13043 | agentz++ | L<blah.blah.blah>
r13043 | agentz++ | and
r13043 | agentz++ | L<"foo.baz.bah">
r13043 | agentz++ | are both recognized.
gaal however: what I don't understand is something about src/ *under dist*. 14:20
under some circumstances, files get put there *not* under src
that is, dist/build/Pugs/* instead of dist/build/src/Pugs/* 14:21
14:21 josh joined
gaal right now for example I have both dist/build/Pugs/Embed/Perl5.o and dist/build/src/Pugs/Embed/Perl5.o 14:22
svnbot6 r13044 | agentz++ | [util/get-irc-logs.pl] 14:29
r13044 | agentz++ | - fixed the --channel option
r13044 | agentz++ | - changed the --out-dir option to ordinary command line
r13044 | agentz++ | argument.
14:32 elmex joined
Limbic_Region gaal - did you see my nopaste and does it make any sense to you? 14:54
gaal didn't see, looking now
hm looks like a ghc thing. 14:55
does manual yml creation work sometimes?
.\pugs.exe -CParse-YAML ext\Test\lib\Test.pm > blib6\lib\Test.pm.yml
if that doesn't work, try 14:59
.\pugs +RTS -A10m -RTS -CParse-YAML ext\Test\lib\Test.pm > blib6\lib\Test.pm.yml
also try -H200m 15:00
15:01 ludan joined
Limbic_Region gaal - sorry, was afk 15:05
15:05 weinig joined
gaal no worries 15:05
Limbic_Region no, the first one doesn't work
though it did yesterday from home 15:06
Limbic_Region is using GHC 6.6rc btw
second one fails with same error
gaal how big does the pugs get before dying?
Limbic_Region pugs: getMBlocks: VirtualAlloc MEM_COMMIT failed: The parameter is incorrect.
not sure how to answer that
the pugs process ? 15:07
it returns almost immediately so not much chance to check the process table
gaal yes. use Process Explorer?
oh
that's.. interesting
does this only happen with Test.yml? if you try to compile (say) the prelude again, does it also happen? 15:08
Limbic_Region yes, much too fast for me to read it
and I would try and compile the prelude again how?
gaal just put it as the arg for .\pugs.exe -CParse-YAML 15:09
instead of Test.pm
src/perl6/Prelude.pm
Limbic_Region k, just a sec
same error gaal 15:13
15:13 yhsuan_ joined
gaal so from within make, it worked ok (your nopaste showed the prelude turned out fine) but subsequently it fails. 15:14
oh!
15:14 yhsuan_ is now known as yhsuan
gaal try del blib6\lib\Prelude.pm.yml 15:14
and then gen some .pm file
Limbic_Region just a sec
gaal or maybe, even before that 15:15
how big is that file?
the .yml
Limbic_Region 727,460
gaal hm no, that's kinda reasonable
ok, try removing it and genning something. 15:16
15:16 avarab joined
Limbic_Region same error 15:17
gaal well, then I don't understand how the first gen succeeded
nmake again? (no clean)
Limbic_Region tried that
no diff 15:18
that's one of the things I tried in between your "try this"
gaal hm. ok. walk away from the computer please for a few minutes, then try genning again. keep the computer idle while you're away.
Limbic_Region did that too 15:19
between the time I pasted the error and when I asked you if you had seen it
[particle] did you kick it?
Limbic_Region that I haven't done
gaal do you have something that can show cpu temperatures? SpeedFan or something?
Limbic_Region no 15:20
gaal try installing one of those. also, try rebooting (no kidding).
Limbic_Region is about to realclean, svn up, reboot, nmake
gaal - I can't install one of those (work machine)
but the reboot is on my radar as indicated above
gaal don't realclean yet
Limbic_Region ok - just the reboot
gaal actually
do a hard shutdown, wait five, start up 15:21
Limbic_Region ok - see you in a few
gaal (ghc predicted a cpu death on my computer once.) 15:22
xerox ?remember gaal (ghc predicted a cpu death on my computer once.)
hahahaha.
15:22 Eimi joined
gaal it did! 15:22
xerox How? 15:23
gaal it kept giving random errors at different stages of compilation. 15:24
hm. how do I get a 'fail' to propagate upwards? here's my situation: 15:25
xerox I see..
You'd better ask on #haskell, I've gotta dash in a moment.
gaal nyeh
see you
gaal stares at his code some more 15:26
[particle] ooh, a staring contest.
[particle] watches intently
15:32 hexmode joined
nothingmuch m 15:34
gaal oose
15:38 c6rbon joined
pasteling "[particle]" at 144.81.84.141 pasted "make unoptimized fails w/ghc 6.4.1 on win32 for rHEAD" (512 lines, 25.6K) at sial.org/pbot/19465 15:41
[particle] could be processor heat death :P 15:43
svnbot6 r13045 | gaal++ | * Signature parser: more whitespace fixes, and handle method signatures
r13045 | gaal++ | with default invocant:
r13045 | gaal++ | :(: $x)
r13045 | gaal++ | The Param consturucted for this is perhaps a little shoddy still.
r13045 | gaal++ | (i.e. what's its intended varname?)
r13045 | gaal++ | * traceM, a golf course in the desert.
gaal wow, glimpse++: 15:49
(cd ghc-sources ; glimpseindex -o .) 15:50
glimpse "^!* ::"
er, alias hf glimpse "^!* ::"
now 'hf lookAhead' is like, instantaneous
xerox Hey. What is hf? 15:51
gaal a shell alias :)
xerox Tags?
gaal scroll back three lines
xerox I don't know what glimpse is. 15:52
gaal @google glimpse search
lambdabot webglimpse.net/
Title: Webglimpse and Glimpse: advanced site search software for Unix : index websites ...
Daveman waves to xerox 15:53
xerox Yuck.
gaal actually this is a better link: webglimpse.net/gdocs/glimpsehelp.html
lambdabot Title: Glimpse manual page(1)
gaal what are you yucking about? 15:54
15:54 c6rbon joined
xerox Thanks. 15:55
15:56 onsen joined 15:57 c6rbon joined 16:01 lichtkind joined 16:04 prism joined 16:08 lisppaste3 joined 16:10 Psyche^ joined
gaal Parser:1406 looks almost like it could be golfed into a State monad... 16:15
16:26 kanru joined, Psyche^ is now known as Patterner 16:29 Limbic_Region joined
Limbic_Region gaal - left it powered off the entirety of lunch, same exact error 16:30
rebuilding from scratch now
[particle] l_r: i'm having trouble building with 6.4.1 16:31
"make unoptimized fails w/ghc 6.4.1 on win32 for rHEAD" (512 lines, 25.6K) at sial.org/pbot/19465 16:32
lambdabot Title: Paste #19465 from "[particle]" at 144.81.84.141
16:33 sili-babypuncher left
Limbic_Region looking particle though I will only be guessing if I offer any advice 16:35
[particle] yeah, i figured it was unrelated to your error
Limbic_Region particle - you could try disabling perl5 embedding 16:36
since that is where it is failing
set PUGS_EMBED="!perl5" 16:37
[particle] i guess it's defaulting to that embedding
i haven't set any special env vars
Limbic_Region yes, p5 embedding is enabled by default now
you have to specifically tell it not to if you don't want it
this btw is not a fix, just a workaround (presuming it works)
[particle] do i need makefile.pl again? 16:38
Limbic_Region the issue apparently is with some conflicting header files
particle, I never risk it - I always realclean;svn up;makefile;etc
[particle] perhaps it's perl-version dependent?
i'm on 5.8.6
Limbic_Region AS Perl?
[particle] yep
Limbic_Region I believe it has been certified to be working with AS and Vanilla Perl
not sure about versions 16:39
[particle] i'll kick the certifier :)
Limbic_Region in any case make unoptimized is easier spelt make fast
it appears to be an issue with conflicting header files
[particle] i'll retrain my fingers :)
Limbic_Region the certifier likely didn't try with ghc 6.4.1
[particle] i'm hoping to go direct to 6.6 16:40
16:40 SubStack joined
[particle] i don't want to install yet another ghc version 16:40
Limbic_Region well, you shouldn't have to
but do try disabling perl5 embedding 16:41
16:41 Odin- joined
[particle] i'm recompiling now 16:41
Limbic_Region fwiw, the conflict appears to be between usr/local/perl/lib/CORE/win32.h and /usr/local/ghc/ghc-6.4.1/include/mingw/stdint.h
16:42 avarab is now known as avar
xerox You surely know this one: how is it called the env var of a shell script that contains the number of arguments passed? 16:49
16:50 fglock joined
Limbic_Region xerox - what kind of shell? 16:51
sorry, was AFK
xerox bash
hmm
sh in fact.
Limbic_Region shouldn't matter much
let me check - been a while since I did any shell programming
xerox For me too :( 16:52
16:52 mauke_ joined
xerox I ought to do a script that lets people add packages to hackage, in order to use them with cabal-install. 16:52
Limbic_Region $* is all of them right?
so wouldn't it just be $# ?
xerox That's right! Thank you. 16:53
Limbic_Region no worries - had to stop and think about it
16:53 araujo joined
xerox I'll need to do some sed/awk/perl(?) hackery for this thing, gawk... it's ages since I did these kind of things. 16:53
16:54 c6rbon_ joined
Limbic_Region well, if this is to be portable - be sure not to assume versions of any of them 16:54
gawk != awk and what not
xerox haha, I meant it as an exclamation, gawk1 16:56
gawk!
Limbic_Region heh 16:57
well, I used to do an extensive amount of shell scripting with sed and awk so if you need help let me know 17:01
Limbic_Region likes to feel useful every now and then 17:05
17:07 mauke_ is now known as mauke
xerox Limbic_Region: still around? 17:09
Limbic_Region in and out
but if you have questions I will stick around
xerox Yes! thank you. 17:10
Limbic_Region shoot
xerox I need to do these steps:
./cabal-put foo-1.0.tar.gz
1. check the filename format is correct: PACKAGE-VERSION.tar.gz
2. check that it unpacks to (case-sensitively) the same PACKAGE-VERSION/ dir
3. check it exists (case-sensitively) PACKAGE-VERSION/PACKAGE.cabal 17:11
4. check .cabal contains 'name: PACKAGE' and 'version: VERSION', 'name' and 'version' case-insensistively (one can use Name or name, and so on), while PACKAGE case-sensitively
Limbic_Region is there a 4?
let me know when you have provided the last step 17:12
xerox 5. copy out .cabal to foo/PACKAGE/PACKAGE-VERSION.cabal and .tar.gz to the same dir. (foo is the base-packages-dir)
Done.
How would you do that? 17:13
Limbic_Region ok - so we are assuming bash/sh/ksh type shell
and sed and/or awk but no perl?
xerox Let me see
This is perl, v5.8.4 built for i386-linux-thread-multi
I can use this perl too
Limbic_Region oh, well then there is no need for a shell script at all 17:14
xerox Except I can't maintain the perl thing
But well, yeah.
Limbic_Region ok - well I will just do shell with sed then
xerox Oh well
If the perl one is easyer
*easier, go with it
You're too kind :)
Limbic_Region sed is only needed for 1 step
xerox Okay!
Limbic_Region let me throw something together
xerox I probably can manage some perl. 17:15
Limbic_Region I will do both
xerox hugs Limbic_Region
Limbic_Region you check which one you want
xerox I'm so stuck... too much higher level programming makes me unable to do this... yuck.
Heh, no pugs on monk 17:16
Otherwise we could've used perl6 ;-)
17:27 salty-horse joined 17:32 Eimi joined
justatheory seen audreyt 17:39
gaal "option Nothing $ fmap Just ruleThisOrOther" -- is there a golfier way? 17:40
xerox ?type option 17:41
lambdabot Not in scope: `option'
xerox What is it?
gaal parsec
Limbic_Region perlbot nopaste
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
pasteling "Limbic_Region" at 129.33.119.12 pasted "For xerox: This will likely need to be debugged." (41 lines, 1.2K) at sial.org/pbot/19467 17:42
gaal xerox: option x p = p <|> return x
and <|> is mplus IIRC
Limbic_Region xerox - there is a shell script version that is pretty ugly if you ask me
I will write a perl version which should be much cleaner
xerox Limbic_Region: great!!!
Limbic_Region++
I'll be back in short time. 17:43
gaal I can of course write a combinator - but I feel like there's a native way to exploit the Maybe here...
Limbic_Region I already know the shell script is missing one command - the actual cabal-put
xerox gaal: why do you need to option Nothing . fmap Just?
gaal I *want* the parser to return a Maybe
Limbic_Region as well as step 5 - must not have copied that 17:44
gaal it's not Nothing . fmap just
xerox It's option Nothing . fmap Just :)
gaal it's option (Nothing) (fmap Just someRule)
xerox Yeah, the point is someRule I mean.
Just make someRule fail with Nothing, i.e. return a Maybe a?
Limbic_Region: right, the actual cabal-put *is* step 5 right? or what do you mean? 17:45
Limbic_Region xerox - let me finish the perl version then I will rectify any problems with the shell version 17:46
xerox nods
gaal xerox: it compiles, but does not work :/ 17:48
xerox Aw.
Hmmmm.
gaal that is, having the rule 'return Just x' instead of 'return x'
because apparently option catches the failure? 17:49
xerox (option x p) tries to apply parser p. If p fails without consuming input, it returns the value x, otherwise the value returned by p.
I think the `without consuming input' part is the key. 17:50
gaal but my rule has a 'try' combinator around it... 17:51
it doesn't consume input when failing.
xerox Right, I was going to suggest that.
gaal I think I'll just write a maybeRule function that does thi 17:52
s
:pl maybeRule r = option r $ fmap Just r 17:53
@pl maybeRule r = option r $ fmap Just r
xerox Strange. Cale is good at Parsec, if he comes about ask him about that.
lambdabot maybeRule = ap option (fmap Just)
gaal okay, thanks.
I gotta moose for a bit...
17:53 weinig is now known as weinig|bbl
xerox Moose freely. 17:53
gaal named parameters are in. 17:54
salty-horse gaal, hi
gaal hey salty-horse. the bug's fixed
salty-horse yay. watching
gaal as of this morning
salty-horse can you help me figure out why my changes to pretty print Param.Default had no effect? (the ones that i msg-ed you when you were away) 17:55
gaal i replied on the channel, please look in the irc.pugscode.org 17:56
gotta go for a bit
xerox There's another irc network ? 17:57
gaal logs
xerox Ah-ha.
salty-horse thanks
Limbic_Region xerox - "copy out .cabal to foo/PACKAGE/PACKAGE-VERSION.cabal and .tar.gz to the same dir." where foo is base-packages-dir - how will I know what that is? 17:59
is it passed in as a command line argument?
xerox Limbic_Region: a variable would be sweet 18:00
It's now a subdir in my $HOME but it'll be something different 18:01
Limbic_Region what I mean is, something has to tell the script what it is
I can query the environment or it can be passed in as an arg
please disregard the shell script - I made a lot of incorrect assumptions when I wrote that
I read more carefully for the perl 18:02
xerox OK
Limbic_Region: can it be a "global" variable?
It is just a fixed directory, users aren't supposed to change it
Limbic_Region I don't care what kind of variable it is 18:03
I care how my script finds out about it in the first place
xerox Just a variable in the script, it's a "fixed" path.
Limbic_Region almost finished
xerox One writes it in the source.
pasteling "Limbic_Region" at 129.33.119.12 pasted "for xerox - does a better job" (45 lines, 1.3K) at sial.org/pbot/19469 18:05
Limbic_Region ok, I have to go - bbiab
xerox Thank you! 18:06
svnbot6 r13046 | gaal++ | * Handle named parameters in Signatures. 18:11
r13046 | gaal++ | :( :$justify ) # which is shorthand for...
r13046 | gaal++ | :( :justify($justify) ) # but you can also say
r13046 | gaal++ | :( :justify($justification) ) # the long name is used inside body;
r13046 | gaal++ | # the short name by callers
xerox brb 18:15
gaal rehi 18:16
wow, 13 minutes to commit. that's... slow 18:17
so, modulo bugs, I think the new sig parser handles everything the old one does, and more. 18:21
18:22 xerox joined
gaal I wonder whether at this point it would be more fun to write the pugsy introspection interface, or to transition the the parser to use this in actual sub declarations. 18:22
xerox Back.
gaal or maybe a pretty printer, though perhaps writing the pretty printer in Perl would be a good way of testing an introspection interface! 18:25
xerox Limbic_Region: anyway, there's no such thing as cabal-put. The script *is* cabal-put.
fglock compiling Test.pm from scratch with v6 dropped from 13.5s to 9.1s (about 30%) 18:32
svnbot6 r13047 | fglock++ | v6 - just removed a debug comment
18:41 renormalist joined 18:42 DaGo joined 18:53 prefiks joined 18:56 prism joined 19:01 ruz_ joined 19:10 Shonhen joined 19:11 Shonhen left 19:27 fglock joined
TreyHarris my make smoke is hung in ext/libwww-perl/t/live.t 19:38
does test status get written to a file somewhere as it goes along, so I can try to figure out what test it's hung on? tracing the process is proving unhelpful 19:40
nothingmuch 2/w 12 19:43
19:44 bernhard joined
TreyHarris nothingmuch: hmmm... on my numlock, that would translate to "k0w jk", which is no more enlightening than what you actually typed ;-) 19:45
nothingmuch that was just a typo ;-)
19:45 polettix joined
nothingmuch i was switching to window 12 in irssi 19:45
mauke /w is an abbreviation for /window 19:46
TreyHarris meh, i just killed live.t. i'll come back behind once i've got the smoke.html and figure out where it hung :-) 19:48
19:53 larsen joined
svnbot6 r13048 | Sage++ | The translator will now accept multiple modifiers after one -, so -VR and similar are valid. In addition, -O is now used for heavy object-oriented translation (but -Oo is still useable. 19:56
r13048 | Sage++ | In addition, I found a stupid little error in ASTParser that I fixed (just some bad indentation, not sure how that got there).
19:58 polettix joined 20:03 soisoisoi joined
gaal this paper is fun to read: research.microsoft.com/~simonpj/Papers/inlining 20:04
lambdabot Title: Simon Peyton Jones: papers, tinyurl.com/kevk6 20:05
gaal [ Secrets of the GHC inliner ]
Limbic_Region gaal - not sure if you saw my report earlier but leaving the machine turned off for the entirety of lunch had no effect, neither did recompiling from realclean 20:08
gaal looks like a GHC bug to me
Limbic_Region except that it worked fine yesterday
gaal but I'm not sure.
Limbic_Region I will know tonight if it is related to the pugs source as what I have at home is using the same version of ghc - so I verify that it still works, svn up, and see if it starts failing 20:09
gaal the bug has festered, perhaps
Limbic_Region not that it couldn't still be a ghc bug - but if so, then it is one that just recently got tickled
and AFAIK, I am the only one using ghc 6.6rc on Win32
gaal okies.
gaal heads for bed
lumi Night gaal 20:10
gaal eigens &
nothingmuch mooooooooooooose 20:19
clkao moose 20:24
20:25 Zemyla joined 20:29 deplifer joined 20:30 deplifer left 20:32 weinig|bbl is now known as weinig 20:34 bpphillips left, c6rbon_ is now known as c6rbon 20:41 DaGo joined 21:01 penk joined 21:05 radian joined 21:10 crem_ joined 21:14 Limbic_Region joined, weinig is now known as weinig_, weinig_ is now known as weinig 21:17 fglock left 21:18 salty-horse left
xerox Question. 21:33
To check for directory existence is it:
dir "arr" if ! $dir;
Or:
die "arr" if ! -d $dir;
?
wolverian just open/glob it and check the return value. 21:34
there's a race condition in checking its existence first.
er.
oh, this is #perl6 :)
xerox Hm?
wolverian well, the advice still applies.
Limbic_Region xerox - in perl, use ! -d $dir
xerox hehe okay
wolverian die unless -d $dir; ...someone removes the directory here...; glob $dir... 21:35
Limbic_Region race conditions are not applicable here
xerox Just checking for existance and moving on.
Limbic_Region neither is portability and a half a dozen other things you might think are good ideas 21:36
;-)
xerox paolo@monk:~$ bin/cabal-put fps-0.7.tar.gz
paolo@monk:~$ ls projects/packages/fps/
fps-0.7.cabal fps-0.7.tar.gz
Oh yeah!
wolverian stop hacking and do it the right way :)
21:36 weinig is now known as weinig|bbl
xerox prods himself... and Limbic_Region for the help hehe 21:36
wolverian: do you also know the idiomatic way to cwd to a directory and run a command there? 21:41
wolverian: I'm pretty sure it's system() for the latter part, but I don't know about the former.
wolverian { temp $*CWD = ...; ... } # :) 21:42
xerox Nice!!
wolverian no, I'm not sure if that works. it's scary.
xerox :(
wolverian I remember discussion about it.
xerox Something that works.. ?
wolverian in perl6 or perl5? 21:43
xerox perl5.
wolverian I don't think there's a neat idiom. 21:44
use Cwd qw/getcwd/; my $old = getcwd; if (chdir $target) { system command => -flags } else { die "foo" }
er, and chdir back to $old :)
xerox Yuck.
wolverian quite.
21:49 jlf joined 21:54 prefiks left 21:57 cjeris left
TreyHarris xerox: if you can stomach turning commandline processing over to the shell, you can just do 'system "cd $target && $command"'... when the system exits, you'll be back in your old cwd 22:37
xerox TreyHarris: that's what I am doing now in fact
Thanks anyway :)
TreyHarris of course, this Is Not Portable... but incantations involving system() never are
xerox Yeah
TreyHarris what wolverian describes of temporizing $*CWD is already implementable by the Synopses. I dunno if it's implementable under pugs yet 22:38
actually, i know it's not. Proxy objects aren't implemented. 22:39
23:01 [particle] joined, weinig|bbl is now known as weinig 23:18 jlf joined 23:23 reZo joined 23:24 jferrero joined 23:45 weinig is now known as weinig|food, SubStack joined
svnbot6 r13049 | renormalist++ | cperl-mode.el: 23:58
r13049 | renormalist++ | - Merged changes from original cperl-mode v5.20
r13049 | renormalist++ | - perl 5.9 state variable declarations
r13049 | renormalist++ | - experiments to fix y/tr bug with twigils, but no success yet