Summer of Perl | 6.2.12 released! | paste: sial.org/pbot/perl6 | pugs.blogs.com | pugscode.org | pugs.kwiki.org
Set by ingy on 4 July 2006.
nothingmuch bye guys 00:09
we're leaving
00:10 drbean_ joined
clkao due to software problem we are stucked 00:17
audreyt mm
obra clkao: what software?
gaal bye softwarestuck mountainmoose! 00:19
00:22 Bit-Man joined
clkao printer driver 00:24
obra oh. 00:26
come by the office. I'll print whatever you want.
you can grab and go
clkao it's 50% printed 00:29
svnbot6 r11226 | audreyt++ | * try to fix building with GHC 6.4.1
obra ok
clkao: how can we get svn mailer.py to ignore property-only commits?
kolibrie audreyt: what is the magic setting to have minibufexpl.vim show the filenames at the bottom? 00:31
clkao obra: not sure, might worth ask on #svn 00:34
clkao &
audreyt kolibrie: I no longer remember 00:35
obra clkao: no worries. enjoy hiking 00:36
pasteling "TimToady" at 64.0.7.8 pasted "build failure" (112 lines, 8.2K) at sial.org/pbot/18251 00:38
TimToady I think this is probably the same problem other people have been having with ghc...
this is with a realclean r11226 00:42
I particularly wonder about the lines:
Unregistering pugs-HsSyck-0.1...
ghc-pkg: /usr/local/lib/ghc-6.4.1/package.conf: you don't have permission to modify this file
gaal TimToady: I cheated and chmod a+w /usr/local/lib/ghc-6.4.1/package.conf but I don't think that's Right 00:43
kolibrie audreyt: very sad, can I have your whole .vimrc then?
gaal she's in a braineat^H^H^Hstorming session
kolibrie gaal: you can extricate it from her when she's done 00:44
gaal she'll backlog I'm sure :) 00:45
kolibrie no doubt
TimToady well, still the same result as my pbot log...no Data.Yaml.Syck 00:49
audreyt fixing that 00:51
01:01 xdg joined
TimToady btw, use v6-() also fails the -w test... 01:20
audreyt yeah. 01:22
v6-perl then?
01:23 mdiep joined
obra Don't pure perl implementations usually use pp? 01:24
obra interjects useless commentary 01:25
marksto1 obra: I think so. DBI uses that convention.
01:25 marksto1 is now known as markstos_
ajs v6-perl or v6-perl5 or v6-pp. I think they all look purty 01:25
markstos_ v6-pp++ "v6-perl" looks redundent and v6-perl6 is a bit confusing, although accurate. 01:26
obra audreyt: have you already explored making v6 a "Best" like shim?
TimToady according to p6 that field is the "authority".
obra use v6 pugs 01:27
audreyt obra: yes. 01:29
obra ok
kolibrie gaal: audreyt: 'let g:miniBufExplSplitBelow=1' puts the MiniBufExplorer at the bottom, but there are still extra (useless) lines audreyt's setup did not have 01:32
svnbot6 r11227 | audreyt++ | * remove --global for real
kolibrie sleep & 01:33
svnbot6 r11228 | audreyt++ | * ...ditto
01:35 cmarcelo joined
audreyt TimToady: the build should now work... 01:35
(I think)
still waiting for "perl Makefile.PL" to finish at szbalint's machinre 01:36
cmarcelo dear perl hackers: i'm having trouble with line 60 of util/build_pugs.pl, it's a call to system(), which call ../../Setup passing a few arguments, but seems one of the arguments is being executed somehow ("/bin/sh: --with-hc-pkg: command not found") during the build. it's like system is considering only some arguments and taking the rest as another command. any idea? [ghc 6.5 and perl v5.8.8] 01:41
audreyt cmarcelo: r11288? 01:42
have you rerun Makefile.PL? 01:43
cmarcelo audreyt: 11226.. i'm up'ing to 228.. [yes, "re-ran" =P] 01:46
same error.. and after make clean, still the same and a new one, after dep building: $ Setup: executing external program failed: "util/ghc-pkg-wrapper" --global list >tmp11183 01:49
audreyt ok 01:50
svnbot6 r11229 | audreyt++ | * even more attempt at taming Cabal from rafl++'s change.
audreyt try again?
cmarcelo just a sec 01:51
audreyt 11230
svnbot6 r11230 | audreyt++ | * err, debug 01:53
01:53 xdg joined
TreyHarris hmm. given the choice between "my %x = map { {$_ => 1} }, 'a'..'c'" and "my %x = map { $_, 1 }, 'a'..'c'", I don't think I like either of 'em (though I'm willing to believe that's just my perl5 brain irrationally fighting change). is there a better idiom? 01:54
audreyt my $x = any('a'..'c')
TreyHarris audreyt: :-) of course, but I was actually trying to construct a hash in order to play with hash-in-regex semantics 01:56
so actually that would have been { $_, "" } 01:57
audreyt well, in p5 we have the idiom 01:59
my %x; %x{'a'..'c'} = ();
and use .exists
TreyHarris ahh. yes.
so undef is treated identically to "" inside a regex? 02:00
no warning either?
audreyt not currently 02:01
that may change
TreyHarris hm 02:02
?eval my %x{'a'..'c'} = ();
evalbot_11168 [undef, undef, undef]
TreyHarris hah! recent change. my pugs says [1, 1, 1] :-)
audreyt :)
in any case
that is a good demo of how p6's generalization helps surface syntax
you can say "my @x[1,2,3] = ()" 02:03
while in p5 you have to say "my @x; @x[1,2,3]= ()"
TreyHarris interesting: 02:04
?eval my %z{'a'..'c'} = (17,17,17); %z<a b c> = ();
evalbot_11168 [17, 17, 17]
TreyHarris is that right?
audreyt no, it's wrong :) 02:08
write a test?
TreyHarris :( unfortunately, i have no machine I own fast enough to rebuild pugs in anything resembling useful time. it takes about 29 hours on my poor 256MB G4. 02:09
audreyt er no 02:10
just write a test :)
gaal what's the difference between Bit and Bool? Both are mentioned in S06
audreyt Bool has helpers True and False
...and that's it
otherwise they are the same thing
I wonder if "bit" is just boxed to "Bool" 02:11
TreyHarris audreyt: ok, but if the test works on the latest binary build and breaks now, i won't be held responsible ;-)
audreyt sure.
gaal audreyt: so storagewise they're both haskell bools in Pugs yes?
audreyt also, incremental builds take less time
gaal: yes.
though you can invent
data Bit = On | Off
if you feel masochistic
gaal my Bit $hurts 02:12
audreyt cute
gaal Thy fingers make early flowers / of all things. / thy hair mostly the hours love: 02:13
svnbot6 r11231 | fglock++ | v6 - regex parser: 'multi', signature, attributes
gaal a smoothness which / sings,saying / (though love be a day) / do not fear,we will go amaying.
02:14 scook0 joined 02:20 cmarcelo joined 02:59 ilogger2 joined
vsoni audreyt: how are u 03:01
TimToady hmm.... 03:04
dist/build/libHSPugs-6.2.12.a(Parrot.o): In function `sMgK_slow':
ghc26105.hc:(.text+0x3cd7): undefined reference to `Parrot_find_global'
audreyt parrot no longer havea find_global? 03:05
oy.
TimToady looks like it only has hungarian find_globals from the nm listing... 03:06
audreyt find_global_p ?
TimToady dunno what the bare one changed to.
03:08 xinming joined
gaal how come the .h still has it? 03:08
just an oversight?
audreyt no
#parrot refactored that into global.h 03:09
thinking about how to go about that 03:10
trying to just include parrot.h
and see if it works
03:27 kanru joined 03:39 sanug joined
TreyHarris hrm: "svn: Failed to add directory 'pugs/perl5/Pugs-Compiler-Rule': object of the same name already exists" 03:47
TreyHarris does not know svn
TimToady if you don't care about what's currently in the directory, just blow it away and "svn up" again. 03:48
otherwise rename it.
TreyHarris TimToady: danke 03:49
so what's this svk i heard mutterings about at yapc? is it a totally new vcs, or is it a different frontend for SVN? 03:50
TimToady the latter 03:51
taps in at the C api layer, if I recall.
TreyHarris ah-hah, looks like my cron job to update the pugs src was always croaking on that error, which is why i was so often confused when my source didn't match what people were talking about here. *sigh* 03:52
TimToady on this channel it seems to inspire both tremendous loyalty and intense swearing, often from the same people.
TreyHarris :-) 03:53
audreyt TreyHarris: fixed 03:56
ingy hi audreyt
TreyHarris audreyt: already? 03:57
?eval my %z{'a'..'c'} = (17,17,17); %z<a b c> = ();
audreyt TimToady: try now?
evalbot_11168 [17, 17, 17]
audreyt TreyHarris: the recompile takes time even on feather :)
svnbot6 r11233 | audreyt++ | * TreyHarris++ reported that slice assignment is broken:
r11233 | audreyt++ | @a[1,2,3] = 1,2,3;
r11233 | audreyt++ | @a[1,2,3] = 4;
r11233 | audreyt++ | used to leave the values of @a[2] and @a[3] intact.
r11234 | audreyt++ | * Pugs.Embed.Parrot: Include <parrot.h> to chase the recent
audreyt ingy: hi!
svnbot6 r11234 | audreyt++ | change on find_globals being refactored into global.h
audreyt TimToady: it worksfor me with 11234 and trunk parrot
TimToady audreyt: actually, I think v6-pugs is fine, if you take pugs as a naming authority.
making... 03:58
03:59 lollan joined
ingy use v6-|*|_|6$; 04:00
audreyt TimToady: yeah. 04:01
|n6yl33t use v6-|*|_|6$;
TreyHarris ingy: and a syntax error under perl5 is even better than an errant warning? :-) 04:02
ingy we'll special case it
gaal use v6-lol; 04:04
04:04 lollan joined
TimToady use v6-1e1; 04:05
TreyHarris how can i look at what's just changed before i update? like cvs? "svn diff -r HEAD pugs"? 04:06
TimToady I often use svn diff -r PREV on things 04:07
PerlJam TreyHarris: svn st
TimToady svn status is good too.
PerlJam TreyHarris: svn st -u
TimToady hmm, still didn't find Parrot_find_global... 04:09
prolly shoulda done a realclean first... 04:10
well, gotta commute a bit before I try that... &
ingy use v0b0110; 04:11
TreyHarris hrm. shouldn't i be able to just run a .t file directly with pugs? it's failing on the first :todo and never tries any of the tests beyond 04:17
ivanfrey audreyt: Well I'm back and I find myself at this point:env PERL_HARNESS=./pugs prove t/01-sanity/*.t
t/01-sanity/01-tap.................dyld: /usr/bin/pugs can't open library: /System/Library/Perl/lib/5.8/libperl.dylib (No such file or directory, errno = 2)
t/01-sanity/01-tap.................dubious
Test returned status 0 (wstat 5, 0x5) and so on and so forth.
04:21 guaumiau joined
audreyt ivanfrey: is libperl.dylib somewhere in /system/LLibrary? 04:22
find /System/Library/Perl | grep libperl
04:23 stevan joined
ivanfrey This one: /System/Library/Perl/5.8.1/darwin-thread-multi-2level/CORE/libperl.dylib 04:23
audreyt ok. can you perhaps do a symlink? 04:24
(I know it's wrong, but give it a try)
TreyHarris audreyt: where do i get a prove that works with perl6? 04:27
audreyt it's the same prove 04:28
just a different PERL_HARNESS
TreyHarris oh. I just tried running "PERL_HARNESS=/usr/local/bin/pugs prove array.t" and got lots of errors that certainly looked like perl5 was trying to execute perl6 code 04:29
audreyt well, do you actually have /usr/local/bin/pugs?
or is /usr/bin/pugs ?
TreyHarris either way, same error. i have both, /usr/bin/pugs symlinks to /usr/local/bin/pugs 04:30
i'm using the binary pkg you pointed me at
under osx 10.4
ivanfrey audreyt: can't open /usr/lib/libncurses.5.4.dylib 04:31
audreyt ivanfrey: do you have a ncurses /usr/lib/ 04:32
04:33 stephanepayrard_ joined
audreyt /usr/lib/*ncurses* 04:33
TimToady use v6-alpha; 04:35
TreyHarris I have an ancient version of Test::Harness, I'll try upgrading that
ivanfrey audreyt: I have a libncurses.5.dylib and a libncurses.dylib
audreyt ivanfrey: more symlink? 04:36
vsoni audreyt: hi
audreyt vsoni: hi!
vsoni audreyt: did you get my mail
ivanfrey audreyt: should I symlink libncurses.5.dylib or the other one? 04:37
TimToady do they have the same size? 04:38
audreyt ivanfrey: should be the same; but try .5.
yes.
vsoni: src/Pugs/Embed/Parrot.hsc 04:39
in the pugs tree
and also the horrible workaround for imcc
brokenness
is in src/pge/parrotembed.c
I hope your work can make src/pge/parrotembed.c go away :)
vsoni thanks.......would you like to have better interface to imcc
audreyt: We can talk about it somet time 04:40
audreyt I'm using imcc_init and imcc_compile_pir
that's the only ones I use actually
well also Parrot_compreg
but that's separate
vsoni ok
Hey about the :immediate discussion 04:41
audreyt well also leo also made me do
PackFile_new
PackFile_Segment_new_seg
PF_create_default_segs
etc etc
as the boilerplace init sequence
I never understood that
so I'd welcome them be folded away somehow
into one single _inic perhaps
_init
vsoni how about IMCC_push_parser_state 04:42
audreyt the :immediate -- I don't see why an assembly language should have BEGIN block
vsoni do you use that
audreyt allison ruled that it's a good thing; I'm still wondering if she'll explain why
but at this moment I don't care anymore :) I don't use it, so I pretend it's not there.
ivanfrey audreyt: t/01-sanity/01-tap.................dyld: /usr/bin/pugs version mismatch for library: /usr/lib/libncurses.5.4.dylib (compatibility version of user: 5.4.0 greater than library's version: 5.0.0)
vsoni audreyt: How about immediate was an Interface in IMCC.... that calling programs could implement
audreyt vsoni: well, and IMCC will transfer the partially compiled .pbc to the embedder? 04:43
vsoni: or the textual .pir?
basically I can't see a use case
ivanfrey: hm. 04:44
get perlcabal.org/~audreyt/tmp/libncurses.5.4.dylib ?
(hopefully it's Universal)
vsoni audreyt: It will depend on the how the interface is defined....when IMCC: encounters :immediate....The interface that the calling program defined gets called.....So in that interface you could create a new IMCC instance
TimToady hmm, still getting:
dist/build/libHSPugs-6.2.12.a(Parrot.o): In function `sMgK_slow':
ghc1790.hc:(.text+0x3cd7): undefined reference to `Parrot_find_global'
that's after realclean
04:44 sanug left
audreyt TimToady: hrm. 04:45
leo: do you know what's happening here? :)
vsoni Audreyt: did you get the latest Parrot code....I think Chip made some changes today 04:46
audreyt TimToady: I'm seeing
00165fe0 T _Parrot_find_global
from nm
oh wait, I'm 2 revs behind.
aha. chip removed the API. 04:47
very well.
04:47 phpError joined
vsoni I am not sure if he removed the API 04:47
audreyt he did. 04:48
vsoni ok
audreyt and introduced several variants
vsoni Break in Binary Compatibility :)
audreyt it's now Parrot_find_global_s
same thing
just different name!
vsoni ok
audreyt oy. what's the #ifdef. 04:49
ivanfrey audreyt: oh, Camino tried to open libncurses.5.4.dylib in MS Excel.
audreyt ivanfrey: wow.
oh well. I'll just test PARROT_VERSION > 0.4.5. 04:52
ivanfrey audreyt: I've got a problem. Why are the errors referring to /usr/bin/pugs when I wanted to run ./pugs in the Perl6-Pugs-6..2.12 directory. 04:53
TimToady it has PARROT_REVISION.
in config.h
audreyt yeah, but that will break with 0.4.5-devel up until an hour ago 04:54
but that's fine.
TimToady it was r13170 that changed it, i think. 04:55
audreyt TimToady: retest?
TimToady commit?
audreyt already!
svnbot6 r11235 | audreyt++ | * chase post-0.4.5 Parrot API change: Parrot_find_global 04:57
r11235 | audreyt++ | now becomes Parrot_find_globals
ivanfrey audreyt: maybe I should delete the parrot and pugs builds I installed earlier from Udobe's site.
audreyt ivanfrey: that is likely 04:58
maybe rm /usr/bin/pugs first?
05:00 mauke_ joined
TimToady looks good. 05:00
couldn't make miniperl6, but that's 'cuz I don't have Padwalker, apparently. 05:01
ivanfrey audreyt: I'm going to use the script in /usr/local/bin. 05:02
audreyt ivanfrey: k.
ivanfrey env PERL_HARNESS=./pugs prove t/01-sanity/*.t 05:03
t/01-sanity/01-tap.................Can't exec /usr/bin/pugs at t/01-sanity/01-tap.t line 1.
t/01-sanity/01-tap.................dubious
Test returned status 2 (wstat 512, 0x200)
TimToady audreyt: what do you thing of v6-alpha? 05:04
*think
ivanfrey audreyt: How do we run ./pugs in all incantations of pugs in the test scripts? Is there an environment variable for this. 05:05
audreyt TimToady: when 6.0.0 is here we call it -omega?
ivanfrey audreyt: maybe a symlink will work. 05:06
audreyt (I'm fine with it. still I think v6-perl, esp without #! line, is more descrptive, but -alpha is a neat social hack.)
TimToady as in that number we can't calculate? :)
audreyt -aleph
well, it's good, as we can semi-freeze the spec some time in the future 05:07
and call that level -beta
it's really confusing version slot with authority slot
05:07 frederico joined
audreyt but otoh, it means that before release, nobody can really claim authority to 6.0.0 05:07
which is a neat social hack.
TimToady that's what I thought...
and -perl claims a little too much, I think.
audreyt with the surface syntax still changing, yeah. 05:08
svnbot6 r11236 | audreyt++ | * findSymRef is made from MonadSTM to Eval, thanks to gaal++
r11236 | audreyt++ | for noticing it.
audreyt ok, -alpha it is :)
TimToady or able, baker, charlie... alpha, bravo, etc...
TreyHarris eek. now i'm really confused. I thought maybe some old version of Test::Harness was hanging around somewhere on my filesystem confusing things, so I created an empty test with nothing but a CHECK { warn "version of Test::Harness is $Test::Harness::VERSION." }. when i try to run prove on it, I get "version of Test::Harness is ."
TimToady -abel, that is... 05:09
TreyHarris oh, duh, we're in test::harness. *smacks head* 05:10
TimToady or use the hurricane list for this year...
audreyt so instead of specifying the variant of implementation, it's specifying the "stage of language" 05:12
I like that.
since the implementation in my mind belongs to #! anyway.
TimToady since the idea is to cut across implementations
audreyt eexactly.
so if you have a script that has no impl prefence, don't put a #!
some point in the future maybe #!env perl6 05:13
but not now :)
TimToady we're still trying to encourage convergence at this point. :)
audreyt oh, the idea is that various implementation will symlink to cc :) 05:14
er I mean runghc
*runhaskell
ivanfrey audreyt: env PERL_HARNESS=./pugs prove t/01-sanity/*.t
t/01-sanity/01-tap.................dyld: /usr/bin/pugs can't open library: GMP.framework/Versions/A/GMP (No such file or directory, errno = 2)
t/01-sanity/01-tap.................dubious
Test returned status 0 (wstat 5, 0x5)
audreyt ivanfrey: ok, you are at the end of the tunnel!
05:15 mauke_ is now known as mauke
TreyHarris hmmm. I just found *my* problem... in my version of prove, it's HARNESS_PERL, not PERL_HARNESS. 05:15
audreyt ivanfrey: perlcabal.org/~audreyt/tmp/GMP.framework.tar.gz
TimToady okay, we're schmokin'.
audreyt ivanfrey: cd /Library/Frameworks
and extract it there.
TreyHarris ivanfrey: where did you get PERL_HARNESS from? 05:16
audreyt it's HARNESS_PERL and never PERL_HARNESS I think... 05:17
gaal ;2~.
audreyt ;3-!
TimToady ;5-? 05:18
ivanfrey TreyHarris: I thought it was Test::Harness
gaal wow, ssh escape codes escape windows
ivanfrey TreyHarris: I used cpan to install it. 05:19
TimToady it's an OS-in-the-middle attack...
audreyt TimToady: in P6AST work today, I think that 05:20
TreyHarris ivanfrey: you've been running "PERL_HARNESS=./pugs prove..."
audreyt use v6-alpha
is short for
use Perl-6-urn:alpha;
makes sense?
gaal use v6-א_0;
TimToady urn?
TreyHarris ivanfrey: it should be HARNESS_PERL. i was about to type the same line when you mentioned it on here, and so I reversed them, too
audreyt the spec has it that authority has to be a URI
which means it has to have a scheme
and the default scheme, for things that cannot be located, is urn: 05:21
as opposed to everybody else like http:, which are URLs
TimToady I was kinda hoping to locate perl...
ivanfrey TreyHarris: Well audreyt advised me to run env PERL_HARNESS=..... as opposed to make test 05:23
TreyHarris and so I spent the past hour reinstalling Test::Harness because I was typing PERL_HARNESS instead of HARNESS_PERL. but now I have the spelling right, the tests run just fine, no syntax errors or erroneous testing of TODOs. :-)
TimToady it's all very Greek: alphas, and urns...
ivanfrey TreyHarris: I apologize for the confusion. I'm confused myself. 05:24
TimToady gaal: I'm afraid perl5 doesn't deal with א_0 very well. We're hoping to fix that in perl6...
TreyHarris ivanfrey: no problem. but given this, i think it's pretty clear that you have a problem with your Perl (5) installation, whether or not you have a problem with pugs 05:25
TimToady on the other hand it doesn't seem to have much trouble with -Inf. go figure...
TreyHarris ivanfrey: since the line you were running was never actually sucessfully calling pugs 05:26
ivanfrey audreyt | TreyHarris: Well it appears to now.
audreyt TimToady: I committed S11 05:28
sanity check?
ivanfrey But I certainly have made a dog's breakfast of things. It appears that some setup of pugs occurred when /sw/bin/perl was still in my path.
audreyt | TreyHarris: t/01-sanity/06-use.................pugs: *** Can't locate lib.pm in @*INC (@*INC contains: /sw/lib/perl6/darwin-thread-multi-2level /sw/lib/perl6 /sw/lib/perl6/site_perl/darwin-thread-multi-2level /sw/lib/perl6/site_perl /sw/lib/perl6/auto/pugs/perl6/lib /sw/lib/perl6/site_perl/auto/pugs/perl6/lib .). 05:30
at t/01-sanity/06-use.t line 1, column 1
t/01-sanity/06-use.................dubious
Test returned status 1 (wstat 256, 0x100)
TimToady sane. 05:31
ivanfrey audreyt: I guess I better start over? 05:34
audreyt cool.
try
geoffb rezzes in for a moment
audreyt env HARNESS_PERL=./pugs -Iblib6/lib t/01-sanity
you didn't install, so the blib is needed (for hopfully obvious reasons) 05:35
geoffb just wanted to congratulate y'all on the v6.pm getting through the sanity suite.
Damn fine progress, that is!
audreyt geoffb: thanks :) yay :)
stevan just release new Moose
with Module and Package encodings
so it also agrees with the 6.28.0 AST's MOP encodings!
(or rather, "so we can import en mass from Moose without reservations") 05:36
gaal will commit the new AST sketch in a bit.
geoffb mmmm ... Perlish goodness
audreyt then the task is to switch both the v6.pm runtiem and Haskell runtime to use it
one both are using it, then we call it 6.28.0.
it's like, the plan, exactly one year ago, finally starts to come together :) 05:37
geoffb wonderful, I was wondering when that was going to happen
heh
TreyHarris TimToady, audreyt: i hope you're future-proofing perl 6 by allowing some value of "use v7...." that will cause Perl 6 to load v7.pm ;-)
audreyt lol
05:37 nothingmuch joined
gaal committed 05:38
audreyt gaal++ gaal++ gaal++ gaal++
TimToady: it's actually a good point 05:39
when perl6 sees use v6
use v7
instead of curling up and die
it should first try to load v7.pm :)
TimToady sure, why not?
audreyt fix S11? :)
audreyt needs to shower+sleep 05:40
alternately...
TimToady though arguably it should really just load Perl-v7.pm
audreyt yes.
Perl.pm with multiversioning selector
ivanfrey audreyt: env: -Iblib6/lib: No such file or directory
TimToady or maybe we just install a hook that runs if the first word in a Perl file is completely unrecognized. 05:41
audreyt ivanfrey: sorry
geoffb seen rafl
jabbot geoffb: rafl was seen 8 hours 24 minutes 29 seconds ago
audreyt env HARNESS_PERL=./pugs prove -Iblib6/lib t/01-sanity/*.t
TimToady: yeah. since as we know "there will be no perl 7" etc :) 05:42
gaal whereof one cannot complie thereof one must not run
audreyt or rather "perl 7 will be a perl 6 module"
svnbot6 r11237 | gaal++ | * WIP towards an overhaul of Pugs.AST.Internals.
r11237 | gaal++ | It includes a new and more detailed AST which captures much more
r11237 | gaal++ | Perl 6 semantics; captures; MOP based on Stevan++'s Moose.pm; and
r11237 | gaal++ | cleans up some accrued cruft.
audreyt but hey, perl 6 is now just a perl 5 module.
gaal++
lol
gaal one must, so to speak, shed away the antlers after one has paraded around yapc with them # XXX WRONG 05:43
geoffb Anyone know if rafl (or someone else) is still maintaining a Debian package for pugs?
ivanfrey audreyt: Take a look at this: All tests successful, 2 subtests skipped. 05:44
Files=19, Tests=63, 50 wallclock secs (34.37 cusr + 1.69 csys = 36.06 CPU)
dduncan so, with headings like "use v6-alpha;" ... does this mean that we can take identical Perl 6 scripts and run them under multiple runtimes, like saying "pugs <script>" or "perl <script>" ... just like running normal Perl 5 scripts?
so only the program name changes on the command line, and everything else is the same?
ivanfrey audreyt: Ain't that sweet?
TimToady dduncan: that's the point of scrapping the #! line, which can only indicate a single interpreter. we wanted to run the tests under various interpreters. 05:46
xinming If `use Perl-v7` is true... someone might even do `use Python-2` or `use Basic` :-P
ivanfrey audreyt: So I guess it's sudo make install followed by util/package-osx.pl
audreyt ivanfrey: yeah :) 05:47
dduncan I would just like it if this means I can simply write a perl 6 version of stuff only, and not bother with a separate perl 5 version, assuming I'm willing to live with alpha quality support structures for awhile
audreyt dduncan: yes, that would mean that. 05:48
dduncan good ...
TimToady well, it's still closer to halloween than christmas...
dduncan then with the current rapid pace of development, I may very well skip a pure perl 5 version of Rosetta since a perl 6 version should be all that is needed, even in a plain perl 5 environment ... good news
audreyt dduncan: yes. that is I think good news to all CPAN authors :)
(as maintaining -6 and -5 takes lots of energy.) 05:49
dduncan but you know what else that means ...
geoffb can't wait until v6.pm is far enough long to convert all the fun stuff from his code library .... 05:50
dduncan it means that lots of stuff in ext/ can now conceptually just be put on their own on CPAN now
rather than fattening up the Pugs bundle
unless you like their presence in Pugs for testing Pugs in smokes etc
audreyt yeah, but multiversioning on CPAN needs work. 05:51
i.e. it needs to implement the full "use" spec. 05:52
which robkinyon is working on, I believe
dduncan good point, actually ... the CPAN indexer will need to know how to parse the Perl 6 modules ... perhaps a reason to wait on that to-CPAN move then
geoffb Hmmm, if ext/ stuff was moved to CPAN, having a bundle of it for use as test fodder would be key
audreyt geoffb: yes.
dduncan I think that separating the test suite soon would be a good idea, maybe 05:53
TimToady speaking of which, finally finished my smoke, 44 errors.
audreyt but CPAN indexer multiversioning ("The Stas Spec") and client multiversiong (JIB) and coexistence of modules (the hack I did in YAPC::NA) needs to be specced and implemented.
dduncan if all the t/ stuff is a separate downloadable bundle, it can be used with just v6 or just Perl6-Pugs, etc
ivanfrey audreyt: I don't have a util/package-osx.pl 05:54
audreyt before that if we just throw random v6 modules on CPAN, the indexer, client, and "require" will all just be very confused.
TimToady interestingly, for the first time, I passed Perl-Grammar/t/basic.t.
audreyt ivanfrey: it's in the svn checkout
TimToady: evidently the parrot cage cleaners are doing some good.
ivanfrey: download it from svn.openfoundry.org/pugs/util/package-osx.pl 05:55
TimToady hope so.
audreyt ivanfrey: but I havn't run it myself -- unobe is the one you want to consult
and I need to shower+sleep anyway :) bbiab &
05:55 buetow joined
dduncan I don't suppose that the now rapidly updated version.pm is something that can help with the CPAN multiversioning issues ... or only.pm, or if the fix needs to be done separately from those 05:56
ivanfrey pugs works! 06:00
audreyt yay! 06:07
dduncan: version.pm is one piece of puzzle 06:10
only.pm too
but yeah, need to graffle them into a roadmap of sorts.
there's always tomorrow :)
audreyt waves &
geoffb derezzes as well 06:11
dduncan yes ... but still, this v6 development seems to move some kind of production ready 6.0.0 a lot closer to the present in my eyes ... Perl 5 may be the first production ready runtime
ivanfrey audreyt: I didn't bother with the package-osx.pl script because my implementation of pugs appears to be installed in /sw and not /usr/local
06:12 kanru joined
dduncan though not likely the fastest ... Parrot would likely take longer to be production ready, but end up being the fastest one when it is 06:12
those are my predictions 06:13
audreyt dduncan: that may very well be true.
Haskell is the one to prototype semantics and unknown interactions in; once it's well understood we code it as CPAN modules
and have another layer of sugar modules like Moose that makes it idiomatic
and have finally the v6.pm layer that targets the new idioms 06:14
TimToady and makes sure all the source filters run in the right order. :)
audreyt so the lower-layer modules are goign to be production-ready long before v6.pm covers all of perl6
there's no source filters anymore :)
"precompilers"
but yes.
dduncan: though it's also possible to retool v5 runtime so that v6-related features runs fast. 06:15
but that will probably not happen with vanilla 5.8. 06:16
dduncan yes, of course
audreyt but parrot is needed for platform-independent bytecode, JIT, ease of deployment, etc.
...though there is a fallback plan, namely javascript... :)
obra ease of deployment? 06:17
audreyt obra: .pbc files
and AOP compilation
er, AOT
(ahead-of-time)
though admittedly that's broken with parrot right now
but it's part of the spec :)
dduncan right, so here's a bold prediction of mine ... a subset of Perl 6 that will cover what people need for at least 90% of their work will be at least beta quality by Christmas of THIS year ... over Perl 5 ... and if that isn't bold enough, some people will use it in production by then ... including me
audreyt dduncan: well, minus the v6.pm sugar layer, I'd say stevan and I and other lambdamoose are already deploying it in production 06:18
_with_ the sugar layer however... yeah, Christmas _this_ year seems possible.
dduncan yes, I mean _with_
audreyt a graffler is coming to this channel... 06:19
06:19 justatheory joined, shachaf joined
audreyt dduncan: yeah. it would help though 06:20
dduncan that all said ... I am assuming here that, if someone is going to update a legacy Perl 5 module to use a new Perl 6 module under the hood, that doing so would be fairly easy ... like one can invoke Perl 6 in Perl 5 now ... maybe assuming for simplicity that only one author-version of said perl 6 module exists
06:21 Bleadof joined
dduncan after all, one intended purpose of Rosetta is to have existing ORMs use it under the hood 06:21
audreyt dduncan: sure, but the calling convention is a bit different
to call a Perl 6 method from Perl 5 land
currently you have to pass your positionals and nameds separately
dduncan ideally, the users of the legacy modules won't have to change anything at all
audreyt but we can always have a idiom layer around Data::Bind
dduncan so such updated ORMs can be drop-in replacements
audreyt that, like Params::Validate, tries to intuit what the caller means.
stevan was suggesting something like that 06:22
Class::MOP : Moose = Data::Bind : Pugs::Runtime::Call
or something like that
Buffalo.pm maybe ;)
06:23 sanug joined
audreyt oh wait, Buffalo.pm is actually a monger group. 06:23
hi sanug!
sanug [01:22am] sanug: audreyt: I will repeat from before... I looked at the announcement you sent on the v6pm now runs t
[01:22am] sanug: audreyt: layers are from the user perspective - CPAN - Class::MOP - Data::Bind
[01:22am] sanug: audreyt: Or CPAN Moose Class::MOP Data::Bind
[01:23am] sanug: audreyt: So we put a user of module using first Moose - which layers on top of Class::MOP which uses Data:Bind...
audreyt sanug: so, from user point of view, CPAN.pm is still toplevel
it installs some modules
some of which are written in perl 6
for those modules, v6.pm is the topmost "sugar" level
dduncan speaking of all the undersides of Perl 6 being implemented as Perl 5 modules ... is there something existing to make strong variable typing easy? 06:24
audreyt dduncan: see Moose TypeConstraints
dduncan okay
audreyt sanug: v6.pm parses perl 6 code and generates perl 5 code that uses "idiom" level modules
so they are the next level
dduncan effective strong typing is a prerequisite for my database stuff
audreyt the idiom level does not use Perl 6 rules
and just use normal perl5 declarative sugar to make things easier 06:25
sanug So I think that colors would do this best - Perl 6 gets one color and when we get to Perl 5 get another color to show the switch
audreyt Moose.pm would be an example for that 06:26
well, yeah, but it's all Perl 5 up and including v6.pm
the user module, say Rosetta.pm
installed from CPAN
is written in perl 6
handled in p5 land by v6.pm
and from v6.pm onward it's all perl5
sanug OK - so perl6 code gets the Perl 6 color
and then all below are perl5 06:27
audreyt yup
and Rosetta would be a nice example.
sanug I would like to make special diagram for the magic of v6.pm
TimToady s/Cfront/Perlfront/
audreyt that would be nice
yes :)
sanug It is important that all grok the wow of that module
audreyt below the idiom level it's the "runtime" level 06:28
Class::MOP, Pugs::Compiler::Rule, Pugs::Runtime::* etc all go there
sanug Explain idiom in more detail so I can express tat in the graphic 06:29
audreyt ok.
sanug s/tat/that/
audreyt look at the synopsis of search.cpan.org/dist/Moose/lib/Moose.pm
lambdabot Title: "Moose - Moose, it's the new Camel - search.cpan.org"
dduncan slight change of topic, but I was wondering how .id works with non-trivial types
eg, what does the .id of a Pair look like?
I know that to users it shouldn't matter, but to people implementing composite types, it does
audreyt dduncan: one possibility - could be just itself. 06:30
sanug: note the new pseudo-keywords: "has", "isa", etc
sanug: those are just regular perl 5 functions
but it hides the underlying Class::MOP calls
the "has" actually stands for $class->meta->_process_attribute(...) ect etc 06:31
sanug audreyt: How does the installing of the attributes work?
audreyt sanug: it calls into Class::MOP
sanug: basically, for each perl 6 class, there is one "meta object" behind it 06:32
accessible by the "meta" method
dduncan one key thing I'm wondering about .id for immutable types is ... are they supposed to generate some neutral value like an integer, two of which can then be compared independently of the type definition, or will they contain references to the actual object all the time and that the object's class still needs to declare a === method which is invoked as needed? 06:33
audreyt so you can add an attribute in perl 6 by saying
Dog.meta.add_attribute(...)
dduncan if it is the latter, I imagine that implementation will be simpler, at a possible cost of performance if the same comparison is done a lot
audreyt and with Class::MOP it's
Dog->meta->add_attribute(...)
but it's not declarative
and hence non-idiomatic
so Moose.pm introduces the idiom of
has 'x' => (is => 'rw', isa => 'Int'); 06:34
which translates to a bunch of underlying calls to meta objects.
dduncan: the latter
dduncan okay, that answers my question
audreyt sanug: and the win of the idiom layer is that Perl 5 CPAN modules
that do not desire to use perl 6 syntax
can neverthelss use those idioms to improve their code 06:35
_and_ the perl 5 code emitted by v6.pm will look clean and idiomatic too
geoffb
.oO( ... which is *extremely* nice ... )
 06:36
sanug When i spoke to Stevan Little on Saturday - he showed me a diagram - that had all of the layers inside the meta object protocol
audreyt yes.
sanug Do you recall where that is?
It was in graffle as well - no colors
audreyt sanug: somewhere in svn.openfoundry.org/pugs/perl5/Perl...odel/docs/
lambdabot Title: "Revision 11237: /perl5/Perl6-MetaModel/docs" 06:37
audreyt there are 4 graffles there
sanug: so. the runtime layer are, at least ideally, still all pure perl5 06:38
06:38 janus` joined
audreyt but sometimes the perl5 runtime doesn't quite let us do what we want 06:38
so we delve deeper
and use the "core" layer 06:39
janus` hi
audreyt or "kluge" layer ;)
or just neutrally, the "XS" layer
which includes autobox.pm, Devel::Caller, etc.
also, PadWalker.
sanug So please explain the magic of v6 again, that you exploited..
audreyt those augment the perl5 runtime to provide neccessary semantics 06:40
sanug the name v6 doies something magical
s/doies/does/
audreyt and I'd like to work with p5p people to make sure that they are documented, supported, and mabye even made into core.
sanug: ok. the magick is simple and explained in search.cpan.org/dist/v6-pugs/lib/v6.pm
TimToady notes that it's 2:40am where Audrey is having her little slumber party...
lambdabot Title: "v6-pugs - an experimental Perl 6 implementation - search.cpan.org"
audreyt true... 06:41
sanug: basically, when perl5 sees "use v6" it's an error
"use v6 -pugs" is an error
"use v6 -alpha" too
but when it sees "use v6-alpha"
then it magickally loads v6.pm with -alpha as the import list.
v6.pm then uses Module::Compile to precompile the module that invoked it -- say it was Rosetta.pm 06:42
parse it using Pugs::Grammar::Perl6
emit it with Pugs::Emitter::Perl6::Perl5
and save the emitted Perl 5 code into Rosetta.pmc
so that the next time around, "use Rosetta" will simply pick up the .pmc.
end of explanation. see Module::Compile docs for more info.
xerox blinkes at chosen names ;) 06:43
sanug Ok so I think this would be not 1 but several slides
So that the magic is shown one hand wave at a time.
and then the result is that perl5 is again there when you are done.
audreyt yes. we collapse one hand wave form at a time
yup.
sanug Then Perl 6 disappears into Perl 5. 06:44
pmc
audreyt yup.
sanug And the Objects - can they be collapsed too?
audreyt yes. 06:45
sanug Stevan talked about optimizations to pare down the 15x - 30x performance hit of the dynamic side.
audreyt oh, he finished that and released it today as the new Class::MOP.
it's now as fast as handwritten code.
sanug Would like to show those optimizations.
audreyt but you have to tell v6.pm to "optimize" 06:46
sanug So that has to occur early...
audreyt which is to say, you promise to close all the classes except those are declared open
yeah, it has to happen before the .pmc is written.
sanug So you can still have your cake and eat it too.
audreyt yes. 06:47
so, a Moose object consturcted from v6.pm
is indistinguishable from a Moose object constructed using Mosoe idioms
the idioms layer is where v6 and regular p5 meet
in short, they become normal Perl5 objects. 06:48
and Perl 6 methods and subs become normal Perl5 subs
etc.
sanug I will refer back to this tomorrow when i resume work.
audreyt cool. 06:49
ditto.
thanks for the graffling work :)
audreyt waves goodnight &
sanug I see value in public discussion.
06:49 sanug left 06:51 Aankhen`` joined 06:56 putter joined
putter ok, one attempted robbery later let's see if I can update mmd draft. its very late, so i may just punt till tomorrow. 06:57
xerox b/win 21
Ops.
svnbot6 r11238 | audreyt++ | * mmd-draft.txt - putter++ says he's going to edit it, 07:00
r11238 | audreyt++ | so maybe I'll check in my local clarifications to the
r11238 | audreyt++ | spoiler-seppuku paragaph in first... :-)
07:12 rashakil_ joined 07:13 scook0 joined 07:19 drrho joined 07:31 sanug_ joined 07:48 kane_ joined 07:54 putter joined
putter audreyt: sigh :) 07:54
TreyHarris guess i didn't read this right. anyone know how to suppress :todo tests? prove hits the first :todo and then fails the remaining tests without trying them 07:58
07:59 elmex joined
putter audreyt: I'm afraid my oh so very rough new draft changes story perspective to the meaning of a single signature. so the whole paragraph you modified is gone. sorry. i'll clean up tomorrow. middle of my night now. 08:02
oh, and for clarity, I was the target, rather than the instigator, of the aforementioned attempted robbery. just in case anyone was uncertain. ;) walking home from 'thon (well, looping neighborhood in thought). nothing like it has happened in many years. sigh. 08:03
TreyHarris: there is currently no mechanism but skip to, err, skip tests. Todo is just a post-test, how should test failure/success be treated, annotation. 08:04
TreyHarris putter: ack, I thought you were making some joke that went over my head. i'm sorry to hear that! 08:05
putter: huh, but with my prove at least, it prevents any later test from running. how do i make it continue later tests?
putter the inability to skip deadly tests is a pain when doing a new pugs backend, or complete pugs alternative. the tests grew up with pugs, using evals, and commenting out, to deal with problematic ones. but we dont have a more general mechanism to easily say "skip/eval this test when running under X". 08:06
TreyHarris so if I add a test, I need to do so before the first :todo? 08:07
putter I assume you are running the new p5 non-pugs version of p6?
Or is the observation that normal prove is now broken. There was fear of this in yesterday(?) removing the #!/usr/bin/pugs line from test files. which the t/README file said were helpful to Test::Harness and prove. Have we just hit this problem. 08:09
TreyHarris eh? no. i'm running "HARNESS_PERL=/usr/bin/pugs prove -v t/data_types/array.t"
putter try adding #!/usr/bin/pugs to the top of your array.t, and see if it works again. 08:10
08:11 penk joined
svnbot6 r11239 | audreyt++ | * Squash warning for Pugs::Grammar::Perl6's invalid syntax error 08:11
pasteling "TreyHarris" at 207.171.180.101 pasted "prove failure after first :todo" (24 lines, 1.1K) at sial.org/pbot/18254 08:12
TreyHarris putter: that's what I'm getting ^^^^ 08:13
08:16 ajs_ joined
TreyHarris (that's after inserting the shebang line) 08:16
svnbot6 r11240 | audreyt++ | * Pugs::Emitter::Perl6::Perl5 - more warnings squash. 08:20
r11241 | audreyt++ | * v6.pm - change "v6-pugs" to "v6-alpha" in distname and docs.
r11242 | putter++ | mmd-draft.txt: Describes new modified algorithm. Also altered the approach taken by the spec (now more local, "what it means for my signature here", and less global/algorithmic). Might need separate "reference implementation" section. Or maybe not, if the rest ends up clear enough. Currently very rough - cleanup in morning. Back to zeroth-draft status.
08:20 elmex_ joined
svnbot6 r11243 | audreyt++ | * This be v6-alpha, version 0.006. 08:21
putter TreyHarris: I'm sorry, I really need to call it a night at this point. With everything I said I would do now done. I will look at it in the morning (t+sigh 3 hours. maybe I'll push it back a bit). It is something which needs to be explored, and I appreciate your pursuing it. thanks...
good night all &
08:22 froh-doh joined
unobe audreyt: i installed ghc-6.5, did svn up, and now Data/YAML/Syck.hsc is barfing: Syck.hsc:180:8: Not in scope: `Buf.unsafeUseAsCStringLen' 08:30
ever run into that? 08:31
xinming hmm, Is current pugs can be compiled with ghc 6.4.2? 08:34
08:34 pmurias joined 08:35 pmurias left
unobe xinming: the 6.2.12 release tarball i compiled with ghc 6.4.2, but hadn't yet tried it on svn 08:37
xinming /usr/bin/ar: creating dist/build/libHSPugs-6.2.12.a 08:38
*** Wasn't able to find 'Parrot_stub.o', aborting..
unobe xinming: i've switched to ghc-6.4.2 and am trying to build it 08:45
xinming unobe++
08:55 Aankhen`` joined
xinming unobe: can you build it? 09:05
09:09 cognominal joined
unobe nope, but it's not the problems you're running into 09:09
parsec isn't being found even tho' configure finds it
still working on it
mental note: don't forget to re-symlink ghc-pkg when going back to ghc-6.4.2 09:19
xinming: are you compiling with PUGS_EMBED="perl5 parrot"?
xinming unobe: yes...
unobe oh, okay. i was told a bit ago that it's unstable, so i've been building only with perl5. 09:20
i'll re-build parrot and try with PUGS_EMBED="perl5 parrot" tho' to see if I can get what you're getting 09:21
xinming unobe: I just have ghc-6.4.2 installed. and I didn't tried ghc-6.5.x series... 09:23
09:29 renormalist joined
unobe xinming: it's building fine so far. is the error near the end of the build? 09:34
xinming Skipping Pugs.CodeGen ( src/Pugs/CodeGen.hs, dist/build/Pugs/CodeGen.o )
Skipping Pugs.Help ( src/Pugs/Help.hs, dist/build/Pugs/Help.o )
Skipping Pugs ( src/Pugs.hs, dist/build/Pugs.o )
/usr/bin/ar: creating dist/build/libHSPugs-6.2.12.a
*** Wasn't able to find 'Parrot_stub.o', aborting...
make: *** [pugs] Error 9
unobe ok 09:35
right at the end
when was the last time you updated parrot? 09:36
xinming: did you see this? xrl.us/oe3v 09:41
lambdabot Title: "#perl6 2006-07-06,Thu"
unobe maybe that has something to do with it...
xinming: built fine for me. 10:05
have you tried updating your parrot to the latest in subversion?
xinming unobe: Yes, It's the latest... 10:06
both parrot and pugs are latest...
strange... :-/
10:07 Debolaz joined
xinming unobe: what's your perl version? 10:10
don't know if it's perl's problem, I'm using gentoo, and perl version is 5.8.7
10:14 joples joined
unobe oh, i'm on os x 5.8.6 10:24
10:40 lollan joined 10:42 ludan joined
ludan hi 10:42
10:55 theorbtwo joined
rafl geoffb: I am still doing this. The next upload will happen in some days, as soon as the latest hs-plugins package is in unstable. 11:12
geoffb: Which will probably be tomorrow. 11:13
11:27 elmex joined 11:37 Odin-LAP joined 11:43 chris2 joined 11:47 kanru joined 12:03 bpphillips joined, bpphillips left, bpphillips joined 12:33 shachaf joined 12:38 datasieve joined 12:42 iblechbot joined 12:48 Limbic_Region joined
Limbic_Region audreyt ping 13:02
13:03 elmex joined 13:04 xdg joined 13:06 thepler joined 13:09 lollan joined 13:23 xdg joined, vel joined
Limbic_Region seen audreyt 13:23
jabbot Limbic_Region: audreyt was seen 6 hours 34 minutes 24 seconds ago
meppl guten morgen 13:33
13:41 hexmode joined 13:44 shachaf joined 13:45 frederico joined 13:55 rzoz joined 14:03 buetow joined 14:05 justatheory joined 14:22 elmex joined 14:23 elmex joined, prefiks joined
geoffb rafl, cool, thanks. there are times when I just want to apt-get and see how things are going, rather than try to get source trees and build tools up to date after some months ... 14:32
14:38 snowstalker joined
szbalint hm. When building pugs, make runs fine, but when trying to make install: 15:03
Cannot import `Data.ByteString': it was found in multiple packages: pugs-fps-0.7 fps-0.7 (imported from src/Pugs/AST/Internals/Instances.hs)
I think putter had this problem last night, but I can't find how he solved it... 15:04
rafl geoffb: Of course.. I just couldn't update pugs for a while because of various issues with the build system, which now seem to be fixed.
szbalint: You can hide your local fps.
szbalint: ghc-pkg hide fps-0.7
audreyt ghc-pkg unregister --user fps 15:05
geoffb rafl, ah, I get it. 15:06
Limbic_Region audreyt - when/if you feel like working the Win32 build failure - I am a willing guinea pig
rafl geoffb: At least parrot is pretty much up to date in Debian these days.
szbalint thanks audreyt. rafl: that didn't work, it said: ghc-pkg: cannot find package fps-0.7
audreyt Limbic_Region: I think it may be already fixed.
szbalint seems to build now though. 15:07
audreyt szbalint: did you add the -user?
Limbic_Region as of when?
geoffb rafl, cool -- will HEAD pugs build against Debian parrot?
Limbic_Region I tried it a few hours ago and still b0rk
audreyt oh ok. same error?
Limbic_Region I think so,
particle_ hrmm, the 'ghc-pkg unregister fps' incantation gets me to 'could not find module Data.Yaml.Syck'
audreyt ok. give me the complete nopaste again?
Limbic_Region let me nmake realclean (which is also still b0rk), svn up, and try again
will nopaste in a few
audreyt particle_: have you rerun Makefile.PL ? 15:08
particle_ i'll follow l_r's lead and update working copy
audreyt k.
rafl: seems "make install" needs to be updated to carry over the packageconf
geoffb rafl, OT Debian question: Why sometimes do packages get pulled from testing, while existing in both stable and unstable? This happened to PDL for a while a few months ago. 15:09
szbalint audreyt: yes, its working now, the comamnd you gave fixed it.
pasteling "Limbic_Region" at 129.33.119.12 pasted "'nmake realclean' failure on Win32 (not entire output)" (197 lines, 15.8K) at sial.org/pbot/18260
Limbic_Region ok - that's the nmake realclean failure - will post the nmake failure RSN 15:10
audreyt - fwiw, this is with gcc-lib NOT in pugs root 15:11
since gcc-lib is not part of PATH, not sure how one would go about fixing find_path anyway 15:12
xinming audreyt: pugs doesn't compile here, with ghc 6.4.2, svn pugs, and svn parrot 15:14
audreyt xinming: compile without parrotembed and it should work. 15:16
xinming: it looks like maybe you had an incrementally work 15:17
s/work/build/
one with PUGS_EMBED=parrot and one without
and they clashed
rm -rf dist/build ; perl Makefile.PL ; make
should work
xinming hmm, I didn't touch the %ENV<PUGS_EMBED>, and also, pugs is newly checked-out... 15:19
pasteling "Limbic_Region" at 129.33.119.12 pasted "nmake failures on Win32 (entire paste of perl Makefile.PL and nmake)" (515 lines, 26.4K) at sial.org/pbot/18261 15:20
Limbic_Region audreyt - there you go
if you would like, I can confirm that copying the gcc-lib back into pugs root works around the failure?
audreyt k 15:21
xinming: well, but without PUGS_EMBED=parrot, pugs will not attempt to load Parrot_stub.
o
xinming audreyt: can't pugs be built with parrot embed? 15:24
audreyt xinming: it can, but the problem is that the pugs_embed flag has to be consistant 15:25
if you run makefile.pl once with embed
then make
then change some file via svn up
then run makefile.pl again without embed
then make
then you end up with the error you are seeing
gaal audreyt: coffeeeeeeee 15:27
Limbic_Region audreyt - out of curiosity, do you know why Pugs suddenly started to need to know where gcc-lib is when it never needed to in the past and would the fact that I have mingw/msys installed be screwing with things? 15:28
Limbic_Region was thinking something along the lines of probling $ENV{PATH} for ld.exe - if found but not with GHC delete 15:29
this of course is subject to breakage of the unexpected and hard-to-track variety 15:30
particle_ i seem to be compiling fine now after up and realclean 15:31
ghc-6.4.1 on win32
Limbic_Region is using ghc-6.4.2
can you nopaste your PATH somewhere for me?
particle_ after compile finishes, sure 15:33
btw i'm not using msys
hrmm, heap exhausted 15:34
pasteling "particle_" at 144.81.84.166 pasted "win32 path for l_r" (11 lines, 850B) at sial.org/pbot/18262
Limbic_Region yep - me too (heap that is) 15:46
particle - is there an ld.exe in your path anywhere? 15:47
particle_ no 15:48
i just restarted nmake after the heap problem, worked fine 15:49
finished to completion, that is
now running 'nmake test'
interestingly, 'nmake test' semms to be recompiling pugs :( 15:50
ah, i see why
d:\usr\local\ghc\ghc-6.4.1\bin\ar.exe: creating dist\build\libHSPugs-6.2.12.a 15:51
ghc.exe: file `s' does not exist
Limbic_Region is still catching up with you (second nmake after heap blowup)
though I needed to copy gcc-lib from GHC into pugs root 15:52
15:52 ruoso joined
rafl geoffb: I don't quite get the problem. What distribution are you running? 15:59
geoffb: Yes, HEAD pugs builds against debian parrot. 16:00
szbalint wth
rafl audreyt: Why? I thought the haskell lib isn't installed anyway at the moment?
szbalint audreyt: I think perl5 broke or my build is totally whacky 16:01
Limbic_Region audreyt - aside from having to run nmake twice (ran out of heap), compiling on Win32 works fine if I copy \gat\ghc\gcc-lib to \gat\pugs
szbalint (I was about to test your changes)
16:02 vel joined
szbalint exception handling passes, but array.t, class.t, etc does not. 16:03
*** No such method in class Array: "&array" at array.t line 60, column 1-30
audreyt wow, for real? 16:05
retesting.
rafl: but people still see errors for some reason 16:06
rafl: and it should be made installable too
rafl audreyt: Any idea where those problems come from? I don't see how they could be related to make install. 16:09
audreyt rafl: hm. maybe it was "make register" 16:10
which does copy+register still
rafl audreyt: Right. paths need to be changed from the build dir to the install dir. I'll look into that either later today (where I unfortunately won't have net access) or tomorrow. 16:14
16:14 amnesiac_ joined 16:15 amnesiac_ is now known as amnesiac
Juerd Limbic_Region: nack, I'm on vacation 16:15
audreyt rafl: cool, thanks! 16:16
szbalint: duplicated. fixing 16:17
Limbic_Region Juerd - ok, just wanted to let you know in addition to the hardcoding of channel/network - the ability to use the plus symbol is also gone too
Limbic_Region can't karma anyone
audreyt szbalint: fixed 16:18
it's an offbyone
szbalint audreyt: okay, thanks!
Juerd Limbic_Region: That sucks. Sorry. 16:19
My internet connecton has over 3000 ms lag, so I can't resonably vim anything.
Especially not such complex spaghetti code
SamB ed(1)!
Juerd SamB: Lack of overview 16:20
I've already used ed today a few times for config files.
SamB just kidding, mostly
Juerd It's great. But useless for CGI:IRC
Limbic_Region Juerd - no worries, just letting you know
audreyt - I am headed off to lunch but if you think of anything /msg me and I will try when I get back
Juerd I feel sorry for upgrading CGI:IRC withuot remembering the custom changes
audreyt Limbic_Region: ok. 16:22
Limbic_Region: also, I mised your wife's comment yesterday; that was sweet of her :) 16:23
16:27 guaumiau joined 16:35 pmurias_ joined
pmurias_ seen chromatic 16:35
jabbot pmurias_: chromatic was seen 147 days 17 hours 22 minutes 5 seconds ago
pmurias_ a long time ago...
szbalint Feb 9th. Indeed. 16:37
16:39 jkanter joined
audreyt szbalint: fixed 16:41
svnbot6 r11244 | audreyt++ | * Repair Perl5 embed; was an off-by-one error. 16:45
szbalint audreyt++
:)
16:46 shadowhawk_pl joined, elmex joined
Limbic_Region returns from lunch and decides that his achey body is a sign that he is or is soon to be under the weather 16:46
16:49 putter joined
putter Hmm. So an airline passenger can simultaneously be over the weather and under it. 16:51
No doubt old news, but, I have a build. Yay.
audreyt yay.
16:52 pmurias joined
audreyt putter: I like the new story of mmd-draft 16:52
putter :) 16:53
audreyt the last paragraph needs work
putter oh yes
audreyt otherwise I'd say it exceeds the average clarity level of Synopses
pmurias does anyone now anything of chromatics parrot in perl5 embbeding?
audreyt pmurias: no. write him?
putter ok, that made me pause frozen for a good 3 secs. yipes. re exceeds. ;) 16:54
particle_ pmurias: yes, he started it in chicago last week
pmurias audreyt: wrote to him eariler today 16:55
particle_ it's not yet been checked into the parrot repo
putter audreyt: but ok, I'll try to remember not to over wordsmith it/
audreyt cool :)
pmurias particle: do you know what technology is he using, or how much progress did he make? 16:57
particle_ i believe he's using p5nci
pmurias i tried doing the same thing before with Inline::C, that's why i'm intrested 16:58
particle_ i think he's got strings going back and forth 16:59
pmurias i got pmc and numbers
audreyt gaal: search.cpan.org/dist/Proc-ParallelLoop/
lambdabot Title: "Byron Darrah / Proc-ParallelLoop - search.cpan.org"
putter some items from my "open conceptual issues" todo list include at the moment "how does the new constraint distance metric interact with type compatibility paths which contain implicit conversions", and "think of motivating story which spinclad buys, for why two different role param types need to be comparable".
geoffb rafl, re: the OT Debian question: I tend to install new Debian boxes every few months, and I always use testing (to get hardware support, etc., without bleeding). However, I sometimes find a common package like PDL is just plain missing from testing; it exists in oldstable, stable, and unstable, but not testing -- and there's a note in the developer package info page that it was pulled by some automated system. I was wondering why this silliness happens, 17:00
since I had the ear of a DD. :-)
It's almost like an RC bug was found, but instead of reverting to a previous version, the package is removed from testing entirely! 17:01
putter P-PL - neat. 17:02
Limbic_Region audreyt - if you can explain to me why Pugs needs to specifically call something in gcc-lib (assuming it is ld.exe) where this never seemed to be an issue before, I can work on a sane fix 17:03
17:04 shachaf joined
rafl geoffb: packages.debian.org/cgi-bin/search_...ywords=pdl 17:08
lambdabot Title: "Debian -- Debian Package Search Results"
audreyt Limbic_Region: I cannot, I don't know :) 17:09
rafl geoffb: pdl obviously is in testing.
audreyt Limbic_Region: I suspect it's hardcoded in GHC-win32 somehow.
geoffb rafl, that was an example I fell over a few months ago
rafl, sorry for the confusion
Hmm, I wonder if I can find a recent one that did that
rafl geoffb: Maybe it has been removed from testing in the past because it was RC buggy and the maintainer nor an other developer fixed it for quite a while.
geoffb: That's done to keep the rc candidate bug-free as the next release is quite near :-) 17:10
geoffb rafl, why would it not just be replaced with the version from stable?
hmmm
painful, though.
Ah, here's a more recent example: packages.qa.debian.org/n/nvidia-gra...ivers.html 17:11
lambdabot Title: "Overview of nvidia-graphics-drivers source package"
rafl geoffb: Because a) it's not said that the given RC bug which caused the removal doesn't exist in stable and b) It's not guaranteed that the older package from stable would work in a testing environment.
geoffb rafl, OK, that's fair enough. 17:12
17:12 penk joined
rafl geoffb: However, I generally don't recommend using testing. For me it caused much more trouble than using sid. 17:12
Limbic_Region audreyt - let me try again. gcc runs fine as is, and pugs did as well until very recently, do you know what changed to make it break (and if not, should I do a binary revert search until I find the checkin that caused the problem)?
geoffb rafl, really? I've heard so many horror stories of "dear god, don't dist-upgrade today, Perl / X / etc. is in pieces on the floor" 17:13
Limbic_Region btw - how do I revert with svn anyway?
geoffb That's why I've never run one of my main boxes on sid
rafl geoffb: Of course sid breaks much more often, but those things are pretty easy to fix most of the time, whereas testing b0rkage might be more troublesome.
geoffb rafl, nod 17:14
rafl geoffb: If things are too fucked to sort them out manually or when I just don't have the time to do so I just downgrade the package and hold it until it's either fixed or I find the time to fix it.
geoffb rafl, is there a defined place to look for daily sid horrors before doing my regular dist-upgrade?
audreyt Limbic_Region: yes, it's rafl's change that makes it look at ../../util/runcompiler instead of ghc 17:15
not sure which rev
(and I'll bbiab)
Limbic_Region k
rafl geoffb: I know a wiki of a german comunity and I think there's something on wiki.d.o as well.
Limbic_Region starts pestering rafl then 17:16
geoffb rafl, hmmm, OK, I'll go look at wiki.d.o
Limbic_Region rafl - what exactly is util/runcompiler and why is it being used instead of GHC? 17:17
geoffb rafl, looks like wiki.debian.org/Status/Unstable is what you were referring to. Thanks!
lambdabot Title: "Status/Unstable - Debian Wiki"
Limbic_Region actually - I bet I can think of a silly workaround 17:19
invoke runcompiler from the same directory as ghc.exe instead of from util/ 17:20
geoffb rafl, back to on topic questions: Will you be packaging v6.pm for Debian? 17:21
Limbic_Region waits for rafl before doing anything whacky
svnbot6 r11245 | gaal++ | * set svn:ignore flag on autodownloaded files 17:27
particle_ is '...' listop, or term? 17:28
(same for ??? and !!!)
S03 mentions '...' in listops (precedence table,) but i'm not sure that's right 17:29
17:32 Debolaz2 joined
putter re genders: on mozart-oz list: "One of my programs starts a lot of OS processes [...] However, at one point I get [error] when he tries to select the variable. " I saw a fun paper go by which compared how views of... keys and bridges (?), were distorted^H influenced by two languages making alternate choices on their genders. 17:33
The "he" is very startling for an english reader.
german email addr 17:34
audreyt particle_: my copy of S03 doesn't say that 17:37
I think it should be term.
instead of listop
otoh, listop would work equally 17:38
putter is there some wacky way to give them error messages?
audreyt nvm, I think it's listop :)
17:39 bernhard joined, TimToady joined
particle_ dev.perl.org's S03 has list ops as C<' print push any all '...' etc.> 17:40
okay, i'll add it to parrot/perl6's opp then 17:41
17:41 fglock joined
audreyt there is also !!! and ??? 17:43
fglock: hi!
fglock hi!
audreyt thinking about doing T-shirts with "Perl 6 is just a CPAN module" 17:44
fglock nice! with a big "v6.pm"?
audreyt yup.
got a "probably wise" comment from matt sergent
particle_ Perl6: JACM
audreyt # use.perl.org/comments.pl?sid=32005&...;cid=48530
lambdabot Title: "make perl faster?" 17:45
audreyt er, "probably very wise"
putter spinclad: how about something this: "any type can of course be viewed as a predicate on objects. but roles, by virtue of (one of?) their defining property of being folded into classes, and only existing by virtue of such use, look more like predicates on _classes_, and only indirectly like predicates on non-class objects. so 17:46
xinming audreyt: error still exists...
I just now failed again...
fglock what's a "Blue Peter"? 17:47
putter ... and that story doesn't work. :/ else :(OrthognonalRole1) couldnt be trumped by :(OrthognonalRole1 OrthogonalRole2). sigh.
audreyt xinming: if you comment out line 172 from utils/build_pugs.pl
and hit "make" again
does that work?
xinming ok, I'll try
pass.... :-) 17:48
audreyt putter: "if one side is a strict but smaller subset of another side, that side loses. otherwise, discard Role cards, and both side loses if they are not subset of each other" 17:49
xinming: you sure your ENV has no PUGS_EMBED?
perl -e "print $ENV{PUGS_EMBED}"
I need to run
bbiab
xinming I'm using ghc-6.4.2 and I don't know if ghc is installed in /opt matter..
audreyt no... that shoulfd not matter at all... 17:50
it's that -DPUGS_HAS_PARROT is somehow in your Makefile
trace Makefile.PL and see why it got there if neccessary
lunch &
17:52 nothingmuch_ joined
spinclad putter, audreyt: have some thoughts which for exiguity won't fit this channel, will discuss over high-bandwidth channels when i arrive 17:58
basically simpler counter-story and algorithm wrt roles; but i need to read your latest thoughts on my way out the door.
putter looking forward to it 18:03
so the card algo just mentioned seems equiv to "params with only roles are a worse fit than any param mentioning a non-Any class; params with only roles are only comparable to each other if a improper subset relation exists, in which case the largest list wins". hmm. 18:05
backing up, we've talked about the desirability of being able to switch between is and does without altering dispatch. is there a similar desire re roles and subtypes? (hmm, though from that, I guess, by transitivity, re classes and subtypes, which seems to go against the old hierarchical distance concept). 18:09
spinclad "Chosen|spoiler = alive.": "Contender | spoiler = alive." 18:13
putter ah, tnx 18:14
fglock what's the command to create/update META.yml automatically?
spinclad s/sepecu/sepukku/ 18:15
kolibrie make dist ? 18:16
putter re implicit conversion and cards, perhaps the card story fall back to "if this zero implicit conversion case fails, then we compare implicit conversion distances. if different, smaller wins. if same, then.... something."
fglock kolibrie: it doesn't work in Pugs-Compiler-Rule/
kolibrie :(
fglock bbiab &
putter fglock: dont know, sorry 18:17
ingy morning! 18:18
spinclad morning ingy!
spinclad heads out &
ingy hi spinclad++ 18:19
putter for anyone who is wondering what is going on, the objective is to find a plausible multi method/sub dispatch algorithm. it has three parts. an ignoring of any variants which are not compatible with the call. that we understand. a left to right way to combine the results of "which variant is best at just this position?" applications of some distance metric. that we are starting to get a grip on. and finally, 18:21
a distance metric. that we are still wandering about, trying to find something with nice properties. 18:22
gaal audreyt: with variant records... data MutObject = MkObject {} | MkForeign{} | MkPrototype{} should be made more qualified
putter indeed, still trying to picture what nice properties are.
gaal MkObject??? | MkObjectForeign | MkObjectPrototype is a little long-winded
audreyt ObjectInstance{...} | ObjectForeign{...} | ObjectPrototype 18:23
don't use Mk on variants
only use Mk for data/newtype with one con
gaal ah, oka
y
audreyt also, shorten to Obj perhaps
gaal yeah
audreyt things have wellknown 3-letter shorthands should
be shortened 18:24
gaal sure, like Sig
(not a record type, but still)
audreyt k 18:25
kolibrie gaal: what do you have your ghc_heap_size set to? What is recommended compared to physical memory, etc.?
putter it's sort of a three prong free for all. ("here is a algorithm, lets see what properties it has and how to explain it", "here is a nice software engineering property, lets see what might implement it, and how to abstract it", "here is a description of what it means to do dispatch, lets look for an algo and explore properties").
gaal kolibrie: for low mem machines, make sure it's a little less than available ram just before build time 18:26
kolibrie gaal: okay
gaal for ram-affluent machines, last time I checed 500m was a reasonable value
but then that was when making a Hs prelude (wow, a while ago!) 18:27
putter though with a stabilizing thread of audrey maintaining a reference impl, and doing minimum required tweaks as new insights go by.
gaal I managed to build pugs optimied with ~200m during the .il hackathon but you may need a little more nowadays
kolibrie I'm swapping like crazy right now. Can I kill, reset my heap size, then make again? 18:28
gaal sure, but perl Makefile.PL before make
(config.yml changes always need a makefile rebuild) 18:29
kolibrie oh, right, otherwise it won't see the change
gaal (which makes me wonder if we should make Makefile depend on it...)
kolibrie perhaps
seems to be much happier now 18:36
gaal cool 18:39
18:59 Aankh|Clone joined
TreyHarris if someone could take a look at sial.org/pbot/18254 ... I'm trying to figure out why prove drops dead after the first :todo test and doesn't even attempt the rest 19:03
lambdabot Title: "Paste #18254 from "TreyHarris" at 207.171.180.101"
19:03 larsen joined 19:10 mauke_ joined
putter TreyHarris: it works for me with r11243 (modulo adding -Iblib6/lib (or ../blib6/lib if in t dir), since I didnt precompile Test). 19:13
and modulo using a non-installed pugs. 19:14
are you using precompiled or compiled at this point?
19:17 shachaf joined
putter in the pugs build dir, r11243, this HARNESS_PERL="`pwd`/pugs -Iblib6/lib" prove -v t/data_types/array.t works for me. 19:18
as for why it might be failing in the way yours is... I'm not sure. sorry. 19:20
svnbot6 r11246 | gaal++ | * Pugs::Doc::Hack::Style - Haskell coding guidelines, starting with 19:24
r11246 | gaal++ | a convention for rationalized, punfree data type naming.
19:25 mauke_ is now known as mauke
putter lol # re punfree 19:29
thank you gaal :) 19:30
gaal putter: :-) 19:31
19:31 Aankh|Clone is now known as Aankhen``
particle_ moose! 19:31
gaal moose 19:37
ajs_ moo
svnbot6 r11247 | gaal++ | * Style.pod - minor tweaks 19:42
gaal kpm.monksofcool.org/images/innermoose.jpg
putter :) 19:44
hmm. someone really has to do an animation of the o'reilly language animals, or other symbols where needed, walk into frame and get assimilated into perl creature in center. perl1, perl2, perl3, perl4, perl5, perl6. 19:47
all sorts of room for inside jokes in the assimilation process.
a python looks on in disbelief, shakes its head and slithers off? 19:51
19:52 rzoz_ joined
ajs_ Heh, I'd see it as the python slithering by... spots a small camel... consumes it, and starts to leave... twitches and morphs into a moose with a camel's hump, rubies for eyes and a python's tail... ;-) 19:54
19:55 spinclad_ joined
ajs_ But speaking of python, has anyone here used twisted? I work with one of the primary folks involved, and it seems like a nice networking framework that would work even better in p6.... 19:56
ajs_ thinks thoughts of a large p5/p6 project and cringes in fear and anticipation 19:57
integral is twisted different from POE? 19:58
ajs_ Never toyed with POE
19:58 cognominal joined
putter ajs_: lol # morph 19:59
ajs_ I realy hate waiting for 10 minutes of tests to run just so I can type "perldoc The::Freaking::Module".... sigh 20:01
putter ajs_: I've never used twisted for anything real, but my impression is a number of core design decisions had a flavor of "we really want to do x, but we can't with python, so we do y instead". so I suspect one with twisted motivations would do things differently in p6. 20:02
ajs_ putter: interesting. I'll have to ask
particle_ ajs_: that's why there's google The::Freaking::Module 20:03
putter hmm. as soon as p6 is working a little better, I expect to see, well, at least if p6 doesnt get _too_ static, I expect to see a transliterating implementation of python on p6. which raises a couple of questions. 20:04
CPAN and python folks - what issues arise in trying to reduce the impedance mismatch?
PerlJam putter: I didn't get that impression of twisted at all.
audreyt putter: have you read Sam's piethon summary?
gaal hmm, "data Routine SingleRoutine | MultiRoutine" or RoutineSingle | RoutineMulti ? 20:05
the latter is more consistent but the former more English
audreyt gaal: RoutineSingle I think.
Test::Simple
instead of Simple::Test
gaal okay, give consistency a chance
all we are saying...
particle_ ups the pun count 20:06
putter wonsers if he is confusing twisted with... oy... what is python's big vcs cum everthing?
s/wonsers/wonders/ hmm, wonsers should be a word.
audreyt tailor? 20:07
putter audreyt: re have read, don't know. maybe not?
gaal The Punner Counter clicks! --More-- You feel irradiated
PerlJam twisted is roughly equivalent to stem (assuming I understand stem properly)
audreyt putter: intertwingly.net/slides/2005/oscon/
lambdabot Title: "OSCON: OSCon 2005"
putter PerlJam: I could well be mistaken.
audreyt the message seemed to be "Yes parrot may be faster than GCC, but only if you don't use PMCs; if you do, it becomes slower than python." 20:08
but there's been a lot of work in the past year, so take those slides with sodium
putter oh, the other question was just how to expose p6 capabilities present in python in a nice pythonish appearance.
s/present/not present/
audreyt but Sam talks about impedence mismatch too, a bit
and the Parrot/HLL design aims to resolve that as well.
see the namespace PDD 20:09
PerlJam audreyt: Sam also didn't have TGE :)
particle_ suggests the sodium be accompanied by chlorine, for health reasons
audreyt PerlJam: piethon was about speed :)
and TGE doesn't get you speed :)
(or lack of. it's orthogonal)
PerlJam well, that's true
putter audreyt's sodium hits putters dilute thoughts, and boom 20:10
gaal looks for a good name for 'Map Ident Val'
audreyt MIV
gaal used in Stmt/s_pragmas, Param/p_slots, Code/c_traits 20:11
and ObjSlots is TVar that
20:11 shachaf_ joined
gaal these are somewhat different uses 20:11
audreyt ValMap?
Vappings?
gaal putter: will you be moosing over?
audreyt Malues? 20:12
audreyt can't think of something good
well
particle_ as opposed to Mariable?
audreyt in Lua it's called Table
putter MapIndentVal until a better idea arises? ;)
audreyt let's call it Table and confuse dduncan some more
putter err, Ident
spinclad_ losing my marriables
20:12 dduncan joined
audreyt I think Table is good 20:12
gaal proposes Desk 20:13
putter ok, I'll keep my eye out for them.
maybe they will be attracted via similarity.
gaal Nights of the Malue Desk
audreyt Malloween
ajs_ To the question of exposing p6 (or any other language) to python through parrot, see my recent p6l comments. I think it's useful to start thinking in terms of the language-speicifc abstraction layers that will be needed, and in the case of restrictive languages like python, those will be LARGE layers. 20:14
spinclad_ eve of the martian mallows
putter gaal: yes. re office. I didn't want to join the circus until the folks who needed quiet had most of their day.
particle_ mmm... bone mallow
gaal putter: they never showed up today, so come
I have to leave not-so-far-in-the-future
putter gaal: ok, be over shortly 20:15
gaal wallops in puin
particle_ notices #perl6 is optimized for pun
audreyt yes, which is why we target multiple puntime 20:16
putter ajs_: different issue. two versions of python. directly on parrot. and on perl6. I personally expect the later to be more interesting, but that's not a universal view by any means.
ajs_ particle_: what do you meme?
audreyt it's nice to see the channel rejoyce in pun 20:17
putter p6 is basically mutable enough itself (hopefully), that it can be twisted into pythonness. err, no pun intended.
oy
gaal now that AST.Internals is not sucking the pun out of everything it is free to flow rampunt.
spinclad_ but how will the pythoness feel about that
?
ajs_ putter: no, I think there the same. Seeing python through p6/parrot is roughly the same problem (though potentially easier to solve) as seeing p6 through python/parrot. Both languages will need abstraction layers for each target langauge.
audreyt instead of pie-thon we need a punthon 20:18
ajs_ If v6.pm isn't running by yesterday, I'll take a pun in the face!
audreyt ajs++ # you win 20:19
gaal plays P&P
putter rests shaking head in hand, groans turning to laughter
audreyt well, if TimToady is here this will get much worse... 20:20
putter ajs_: no parrot need be involved. python on p6 on p5. python on p6 on haskell. python on p6 on js. whatever.
ajs_ spinclad_: pythoness doesn't have to feel. The language defines a set of semantics, and the (for sake of exaple) lang.parrot.perl6 module uses them as provided. Now, it might be ugly in some cases... but I think that's a second-order probolem
particle_ thank the maker for small favors :)
audreyt yeah, we can live in relative piece until the supunman returns 20:21
I mean, peace. (that's a typo, not a typun)
ajs_ If you're talking about using a python grammar in p6 on top of $runtime, then that's a different animal, true.
putter heads out for the hackathon. it looks suspiciously like fleeing. 20:22
particle_ defleas putter
TimToady not sure whether python will be our main squeeze...
particle_ *how embarassing*
audreyt oh no the supunman returns
ajs_ putter: I guess your ship sailed in, since you're time here was fleeting.... 20:23
TimToady I just think of myself as a pundit...
ajs_ "<TimToady> not sure whether python will be our main squeeze..." That's going in a quotes file....
spinclad_ putter: you won't escape so easily. out of the fleeingpan into the flier... 20:24
PerlJam TimToady: is that because you often punt language issues into the future?
TimToady I often shove off issues, but not usually in a punt.
particle_ this discussion might be described as punctuated equilibrium 20:25
20:25 rzoz_ is now known as rzoz
TimToady Holey acu-pun-cture, Batman! 20:26
PerlJam particle_: or punishment.
audreyt or just dyspunctional.
kolibrie hmm, my build died with: *** Wasn't able to find 'Parrot_stub.o', aborting...
audreyt kolibrie: hmm, you are the second one to report that 20:27
kolibrie: did you build with PUGS_EMBED?
kolibrie audreyt: yes
perl5 and parrot
particle_ builds with PUNS_EMBED
TimToady sure it wasn't PUNS_EMBEDD
audreyt trunk parrot?
TimToady darn
ajs_ heh, beat me to it!
kolibrie audreyt: yes
from this morning
gaal drunken parrot 20:28
audreyt kolibrie: try pugs r11248
TimToady it's fjording for the pines...
audreyt and see if it still run
kolibrie audreyt: will do
svnbot6 r11248 | audreyt++ | * try to instrument Parrot_stub.o error
spinclad_ fjording the river in a borrowed punt 20:29
audreyt is punning for the fnords
TimToady the punic wars fit audrey to a t
kolibrie audreyt: just make, no realclean or anything?
audreyt kolibrie: aye
spinclad_ audreyt: no, sorry, can't see em 20:30
kolibrie makes
PerlJam perl6 is scrumdiddlypuntious.
particle_ this is war!
particle_ brings out the big puns
audreyt hmm, "punning for the fnords" is a googlewhack. 20:31
PerlJam particle_: make puns, not way.
er, war
TimToady don't need big puns in a punic war...
ajs_ Good greif... I think I've been punt!
audreyt indeed, you just have to port perl6 to punix
particle_ punning is half the battle
TimToady punic("Can't win this one...") 20:32
gaal punts
20:32 bpphillips left
ajs_ audreyt: Isn't that what punie is for? 20:32
particle_ this reminds me of a drinking game... where's nothingmuch?
audreyt ajs_: yeah, it's a pun project
TimToady nothingmuch already *won*, apparently... 20:33
particle_ patrick ports perl6 to parrot
spinclad_ took to the hills yesterday in a borrowed car
ajs_ audreyt: I bet, after all, it's pretty impuntant
TreyHarris hrm, adding -Iblib6/lib doesn't help, I still get 'Undeclared variable: "$?CALLER::CALLER::SUBNAME" at blib/lib/Test.pm line 268, column 8-276' after I hit the first :todo test
20:33 Daveman joined
TimToady that boy has a 6th sense about impending, um, whatevers... 20:33
audreyt ajs_: sure, that's for the punrot-punters mailing list to decide 20:34
TimToady *impunding
ajs_ audreyt: mercy!
particle_ puncle!
audreyt TreyHarris: wait. that Test.pm is old
TreyHarris: no Test.pm should say CALLER::CALLER::SUBNAME
TimToady *rotevers
audreyt TreyHarris: that is curious. it appears your ext/Test/ checkout is really old 20:35
or somethiing.
maybe try replacing it with search.cpan.org/src/AUTRIJUS/Perl6-...ib/Test.pm
or just rm -rf blib/lib
and try make again
gaal audreyt: I don't see any Declarator enum in p6ast. did you add it w/o svk ci or push? 20:36
TimToady [We now return you to the usual mindless drivel of this channel.]
gaal #perl6: we drive you to drivel 20:37
audreyt my our has state constant
gaal: that's it
20:37 aufrank joined
spinclad_ putter: s/sepukku/seppuku/ # my bad 20:37
TimToady can I quote you?
PerlJam seppunku?
spinclad_ sure, but where
audreyt ?eval q[you]
evalbot_11168 "you"
ajs_ Is "is rw" part of the signature of a p6 sub? 20:38
audreyt yes.
TimToady only if you put it.
gaal "My state has our constant ???"
ajs_ As is "sub foo() is rw {}"
audreyt otherwise perl6 punt it.
ajs_: conceptually a Sub holds two signatures; one for the incoming and one for the return 20:39
gaal: that needs to be reflected in AST :)
kolibrie audreyt: looks like r11248 worked. Any expressions I should test before I run the smoke?
gaal audreyt: does the enum derive Ord like SScope?
audreyt ajs_: the "is rw" is saying that it returns something that conforms to :(Any is rw)
gaal: sure, why not. also, add a return-sgnature to Code
gaal vecause that's a little sick
ajs_ audreyt: Ok, so if I "multi sub foo() {}" and "multi sub foo() is rw {}" then "foo()=1" will do the right thing.... 20:40
gaal ah, good point about that
audreyt kolibrie: ./pugs -e 'say 1 ~~ /(.)/'
TreyHarris audreyt: yes, that's my problem. since make takes forever, i've been trying to use a binary build to write tests in a source tree i compiled a long time ago, but have since updated
audreyt ajs_: return value multidispatch, like named arg dispatch, is... well... unspecced.
TimToady the syntax is specced. :)
kolibrie audreyt: 1
audreyt the semantics not :) 20:41
ajs_ audreyt: hrm... ok, should I assume an implementation for my S29:splice update?
audreyt but it's not required for 6.0.0 :)
ajs_: don't assume that until Pugs 6.2831.0
spinclad_ a signature already has room for two signatures: **In -> **Out or something
particle_ trey: make unoptimized
ajs_ audreyt: Ok... then splice gets... interesting.
audreyt spinclad_: --> 20:42
it's actually chaining two Sig objects together
not sure about currying
spinclad_ so is --> :: Sig -> Sig -> Sig ? 20:43
20:43 monkey_ joined
TimToady Perl is turning into Haskell. Someone shoot me. 20:44
ajs_ TimToady: -- bang
audreyt spinclad_: I think it's Sig->Sig->Constraint
TimToady missed, haha! 20:45
szbalint I wonder if it means that I should learn Haskell or that I should be very afraid. Or both. :)
audreyt szbalint: both.
ajs_ TimToady: Of course, I missed... it was only a comment! The next one's loaded (and probably a continuation)!
spinclad_ definitely both. 20:46
TimToady learn -> Haskell -> be -> very -> afraid
spinclad_ even TimToady starts to speak it
audreyt @djinn learn -> haskell -> be -> very -> afraid
lambdabot -- f cannot be realized.
audreyt lol
TimToady I just like the shiny arrows.
I suspect they have a religious significance. 20:47
ajs_ TimToady: they represent apotheosis, I'm sure....
TimToady metatheosis, surely, or para... 20:48
could even argue ortho, I suppose.
spinclad_ they express the directionality of cognition
ajs_ TimToady: possibly apuntheosis....
TimToady who? ā†¬ what? ā†« huh? 20:49
audreyt it is the sign of Apuncalypse
TimToady no, that v6.6.6
ajs_ We'll need a Punopsis to understand it! 20:50
spinclad_ our one secret weapon is redirection, and stealth
TreyHarris stop, this is punishing.
audreyt but first: repunt for your sins! 20:51
TimToady how many angels can dance on the HEAD of a pun?
gaal audreyt: again and again? 20:52
TimToady "We never learn" --a punticle for liebowitz
audreyt goes back to work on her confession and punance. 20:53
TimToady just works on his confusion and pennance. 20:55
ajs_ Ok, next stupid question: can you pass a slurpy by name? e.g. "sub foo($offset=0,*@slurp)" ... "foo(:@slurp)" 20:57
PerlJam ajs_: wouldn't that kind of defeat the idea of slurpiness?
audreyt ajs_: no.
szbalint Isn't *%hash the slurpy for named params? 20:58
monkey_ :( for Devel::Caller failing to pass tests
ajs_ Hrm... I recall one of the things that Larry said about splice very early on was that in p6, we wouldn't have that silly "if you want to provide a replacement list, you have to pass all of the optional parameters" thing
I don't see a way to accomplish that, given the current slurpy mechanism... 20:59
svnbot6 r11249 | gaal++ | * WIP AST code: naming convention updates
audreyt monkey_: 0.10
monkey_: maybe your cpan still carry 0.09?
TimToady splice @foo, 10 <== @repl 21:00
audreyt ajs_: positional args (and named) can come in multiple batches
monkey_ monkey_: it's 0.10 that is failing :(
audreyt monkey_: nopaste the error?
prove -lv t
monkey_ k, one sec 21:01
ajs_ TimToady: that's interesting. It kind of breaks the P5 mindset, but it does work... hmmm
PerlJam ajs_: lots of things in perl6 "break the perl5 mindset" :-) 21:02
ajs_ PerlJam: treu
szbalint This reminds me
audreyt "break all; break test; break install"
ajs_ or true, if you're not french ;)
szbalint what is the current term for <== and ==> ?
PerlJam szbalint: good question. Still "pipe" AFAIK
audreyt I call them "batches"
TimToady it's in a quantum superposition between pipe and feed
audreyt because "pipe" confuses me
ajs_ dev.perl.org/perl6/doc/design/syn/S...tml#___top 21:03
lambdabot Title: "Synopsis 3: Summary of Perl 6 Operators -"
21:03 ebvv joined
ajs_ "PIPING OPERATORS" is the section 21:03
audreyt <== Atom <== RSS <== RDF
so a function is an aggregator
PerlJam "feed" makes me think of cattle
TimToady good
audreyt and *** is a portal
pasteling "monkey_" at 87.81.40.22 pasted "Devel::Caller failing tests" (26 lines, 1K) at sial.org/pbot/18274
"TreyHarris" at 207.171.180.101 pasted "compile error" (15 lines, 1.2K) at sial.org/pbot/18275
TimToady might change *** to @@
or just stick with named arrays. 21:04
PerlJam *** (or @@) is a worm-hole, <== is just regular space flight.
audreyt TimToady: what's wrong with @@_ :)
monkey_: wow, it crashed. nopaste a detailed run (perl -Mblib t/Devel-Caller.t)
and your perl -V
TimToady nothing, if you have an array of that name declared somewhere.
monkey_ k, sorry still a noob at tests 21:05
audreyt the idea like $_ it's implicit
monkey_ <3
TreyHarris audreyt: make unoptimized failed: "Interface file inconsistency: home-package module `Data.Yaml.Syck' is mentioned, but does not appear in the dependencies of the interface". i pasted the whole error output above
audreyt TreyHarris: yeah. is this release?
Prelude.pm line 78 doesn't contain a ; for me 21:06
if you have mixed unoptimised/optimised make
TimToady I like ==> * for sending stuff down the black hole. Just not sure what to call the corresponding white hole.
audreyt then it's possible that interface files gets confused
(brb)
monkey_: you can force an install and see if v6.pm's tests breaks 21:07
pasteling "monkey_" at 87.81.40.22 pasted "Devel::Caller failing tests (part 2 ;)" (106 lines, 5K) at sial.org/pbot/18276
monkey_ oh good stuff
21:08 ebvv left
monkey_ perl -Mblib t/Devel-Caller.t segfaults at the end :) 21:08
TimToady but I really don't like *** very much.
I'd rather say that ==> * feeds to the next * or ** it finds, while * <== feeds to the previous * or **, I think. 21:10
on the other hand, when do you actually execute the blunt end of the * <== in that case? 21:12
svnbot6 r11250 | Sage++ | split(/;/m, $something) now translates to split(/;/, $something) to avoid nasty issues with the immediacy of /m.
audreyt monkey_: please paste it to rt.cpan.org/Public/Bug/Report.html?...vel-Caller 21:15
svnbot6 r11251 | Sage++ | Updating completed translations to refelect work to date.
audreyt and force install, and see if v6.pm tests pass.
TreyHarris audreyt: I guess I'm just making assumptions about how svn works based on my experience with cvs. I ran "svn update src/perl6/Prelude.pm", and yet, it still hase a ; on line 78. ("multi sub evalfile (Str $filename; Str :$lang = 'Perl6')". what should i run if i really, truly want Prelude.pm to be updated? delete it first and then check it out again?
21:15 macroron joined
audreyt oh I see 21:16
sorry, it means your ./pugs is old.
TreyHarris audreyt: you're right. i thought i did a make clean. guess i forgot. sorry :-) 21:17
audreyt like, before 6.2.12
a fresh checkout perhaps?
TreyHarris huh, a make clean is in my history... 21:18
yeah, i guess i'd better
monkey_ audreyt: cool, k
TreyHarris can i just 'rm -rf *' and 'svn update'? or is checkout magical somehow? 21:19
szbalint uh-oh.
audreyt: perl5 tests still fail :(
With a different error now though 21:20
not ok 1 - elems
# Failed test (array.t line 62, column 1-53)
# Expected: '4'
# Actual: 'undef'
some tests from array.t and method.t fail, the rest passes. 21:22
audreyt szbalint: just that one fails? 21:25
TreyHarris: you can do that yes 21:26
szbalint no, but I didn't want to paste it all.
audreyt szbalint: ok. and the .elems one is blocking you? 21:27
szbalint no, that was just the first one I found to fail 21:28
but well I don't know what exactly is blocking me atm, trying to figure it out now
audreyt ok, please do
gaal gets ready to catch a train 21:30
szbalint okay, try traps the exception, $! is set, the problem is that $! is mangled 21:31
instead of "main::org::apache::derby::impl::jdbc::EmbedSQLException=HASH(0x97cd64c)", it is "user error (main::org::apache::derby::impl::jdbc::EmbedSQLException=HASH(0x97cd64c))" 21:32
at least for what I need.
audreyt szbalint: that's easy. fixing 21:36
21:37 froh-doh joined
szbalint thanks. 21:37
audreyt np :) 21:39
gaal is off to train station & 21:44
svnbot6 r11252 | audreyt++ | * szbalint++ requrested that IO errors should not be wrapped
r11252 | audreyt++ | under "user error".
rafl geoffb: Of course! 21:46
21:50 bpphillips joined
TreyHarris what is this Task::Smoke referred to in so many places? CPAN doesn't know about it, and Test::Smoke doesn't appear to be it, either, so i guess it's not a typo... 21:51
audreyt search.cpan.org/dist/Task-Smoke/ 21:55
lambdabot Title: "Gaal Yahas / Task-Smoke - search.cpan.org"
TreyHarris huh, how strange. maybe it's just a problem with CPANPLUS.pm 21:56
"[ERROR] 'Task::Smoke' does not contain an author part; [ERROR] Cannot find 'Task::Smoke' in the module tree; No such module: Task::Smoke; No modules found to operate on! Nothing done" 21:57
bpphillips \'\
monkey_ audreyt: rt.cpan.org/Public/Bug/Display.html?id=19011 :/ 21:58
lambdabot Title: "#19011: can&#39;t install latest version on debian linux sarge"
monkey_ i'm not alone
TreyHarris cpan says: "Strange distribution name [Task::Smoke]No objects found of any type for argument Task::Smoke". and I've tried it at several different sites, with perl installs built by me and by others. would someone for whom this has worked (via a CPAN install) tell me what version of CPAN.pm you have? 22:00
(yes, I can and will download and install it from the dist, but I'd like to track this problem down) 22:01
22:04 bpphillips left
audreyt TreyHarris: I think it's just your cpan mirror havn't been udpated. 22:05
TreyHarris huh, i swore i saw smoke tests happening at YAPC. lol. guess i imagined it :-) 22:06
22:09 DaGo joined
audreyt it was known as Bundle::Pugs::SmokeKit 22:09
gaal changed its name as it's not only pugs specific now
TreyHarris ah, that makes sense now. 22:10
22:15 macroron joined
ivanfrey audreyt | TreyHarris: I'd like to run the smoketest and upload the results, but I think I should clean up my perl5 installation first. 22:15
22:16 xinming joined
audreyt ivanfrey: yeah. 22:19
ivanfrey audreyt: possibly odobe might be more knowledgeable about this. 22:25
22:26 fhork joined
unobe ivanfrey: i don't think i can help you much. it seems like you installed perl via fink (if your perl is under /sw), and i don't have any experience with fink 22:48
ivanfrey unobe: I can leave the /sw out of my path and rebuild pugs. It's just that I'd like to help you guys debug pugs not necessarily by fixing bugs because that's a bit over my head but by installing and running tests on my 10.3.9 machine. I can even very easily uninstall the Fink installed perl5. 22:52
unobe: But I have a few questions. Does Pugs require Perl 5.8.6? Isn't the default Perl in 10.3.9, 5.8.1. Does this mean I can't run Pugs in 10.3.9 with the default Perl installation? 22:57
23:02 ivanfrey joined 23:06 putter joined
putter 1/2 kg peanum m&m's, 3 dollars. a long walk, free. new perspectives on mmd, err, 2 cents. 23:06
audreyt that's a nice one 23:07
kolibrie hmm, trying to build pugs at home now, I get the fps error szbalint had several hours ago
but the command ghc-pkg unregister --user fps does not seem to work 23:08
ghc-pkg: cannot find package fps
kolibrie puts children to bed & 23:10
putter try it without the --user? (there are a couple of different package.conf around)
:)
audreyt: :)
ok, let's see
story 1 - in which putter appeals for a ;; mark. 23:11
"less important" != "not important"
audreyt Very Good Idea. 23:12
surely, ;; is final?
i.e. no ; after it
putter yes
audreyt nor ;; after it
putter yes
:) 23:13
audreyt semantically the same
just better surface?
putter no
audreyt (and to confuse OCaml users?)
putter having erased next line of argument, now regrets same ;)
audreyt ok. how is it different than the "final ;" previously?
putter no different than final-; . 23:14
but final-; (aka last-;), is distinctly different than non-final-;.
and that given (A,B), to make A more important than B, you have to say (A;B;) instead of (A;B) is just wrong. 23:15
23:15 vsoni joined
putter ;; is implicit in end of arguments if _;;_ hasnt already been seen 23:16
ie, the end of the parameters which are involved in choice.
non-incidental to choice
silence == not buying yet? :) 23:18
audreyt it's all surface
so I fully agree
putter ah, ok. 23:19
audreyt now ; only appears between params
(A;) is bogus
so is (;A)
putter yes
audreyt k, vehement agreement reached.
I don't have to update the .hs :)
putter :)
unobe ivanfrey: to the first question: i haven't tried with anything less. To the second (without a question mark): I don't know. To the third: I don't know. 23:20
putter lol
yet...
unobe ivanfrey: what you could do is download the source code of perl from www.perl.org/get.html and try building it 23:21
lambdabot Title: "Get Perl - perl.org"
putter story 2 - in which cards are coerced
ivanfrey unobe: the question is how can I tell if my Perl 5 installation is really borked. I mean it doesn't appear to be, but then all these strange things happened. 23:24
unobe: like when I was make installing pugs it was placing a whole bunch of files under /sw.
vsoni audreyt: I need some help 23:25
unobe ivanfrey: well, you don't consider all these strange things as a sign of borkage, i don't know what to tell ya
putter distance again becomes hierarchical. the most significant component being the number of implicit coercions. the card story is what you do with two distances of equal coercion count. modulo, for non-zero coercions, you run the card story on the coerced-to types.
audreyt ivanfrey: well, yes, because we reuse perl5's sitelib and simply s/perl5/perl6/ as our sitelib
unobe s/you don't/if you don't/
audreyt vsoni: sure?
vsoni audreyt: I would like to write some tests before I start fixing IMCC to make it reenterant.........Any ideas how I could write a test 23:26
s/write a test/write some tests/ 23:27
unobe ivanfrey: maybe that's unfair of me to say.
ivanfrey audreyt: and the build process found the site lib by search the PATH variable, is that right?
audreyt ivanfrey: by asking the perl you run Makefile.PL with 23:28
vsoni: well... use the parrot_embed interface?
unobe ivanfrey: what you could do is just try building a new version of perl
vsoni and than I invoke the interpreter multiple times 23:29
unobe ivanfrey: well, maybe not should...but that's what i would do
putter oh, and if more than one coersion class (in a type conversion path) exists, you have to try all of them, and have an unambiguous win.
vsoni s/interpreter/imcc/
putter ok? 23:30
audreyt that seems natural. 23:31
ivanfrey unobe: Hmmmmm, if I keep fink's perl and upgrade that one and put that one back in the path and just leave Apple's perl alone, wouldn't that work.
putter ok, at that point I think we have subsumed spinclad's tree of bundles. 23:32
ivanfrey unobe:Fink by the way sets up a parallel unix universe under /sw
unobe ivanfrey: i'm don 23:33
'tknow
putter and the only capability that we have discussed that here doesnt exist is that two unrelated roles are now incomparable.
vsoni audreyt: any documentation on embedding
audreyt vsoni: parrot/docs/embed.pod 23:34
also check out the mod_parrot tree
unobe ivanfrey: remember, i don't use fink, so i'm not sure of what else (if anything) it does to the system
ivanfrey unobe: it uses standard deb packages, but I can choose to either compile from source or install the equivalent binary package.
audreyt mod_parrot and pugs are currently the only two publicly-available parrot host applications I'm aware of
vsoni: also note embed.pod is probably incomplete and/or misleading 23:35
putter story 3 - in which, with immediate problems believed solved, no doubt naively, putter looks further afield
audreyt very telling, that story
putter by suggesting there be two flavors of implicit coercioin
unobe ivanfrey: what version of perl are you using (under fink)?
did you already say? 23:36
vsoni audreyt: thanks
ivanfrey unobe: 5.8.6
putter high cost and low cost. "we can get there" vs "just as good as"
audreyt vsoni: np :) good luck. t/* is probably your best bed
putter: I think TimToady wants "we can get there" to be explicit
concrete examples?
putter counts as a distance level of coersion, and doesnt
ivanfrey unobe: but fink also has a 5.8.8 package available. 23:37
audreyt oh.
hm.
"Just as good as" is perhaps simply "does"
in which case it naturally 0-distances out
unobe i don't think that will make a difference. This is the perl i'm using:
putter I have another rep, which is a different class, but we're good friends and dont mind switching back and forth.
unobe This is perl, v5.8.6 built for darwin-thread-multi-2level
(with 2 registered patches, see perl -V for more detail)
kolibrie putter: ghc-pkg: /usr/local/lib/ghc-6.4/package.conf: you don't have permission to modify this file
do I want to sudo that? 23:38
audreyt putter: why don't we do this f2f again? :)
putter err, not another rep. a...
audreyt kolibrie: yes you do
putter f2f?
kolibrie thanks
audreyt face to face
ivanfrey unobe: the only snag here is if some perl programs specify /usr/bin/perl then that's perl 5.8.l
unobe: you're running 10.4 23:39
unobe: append ? to last statement
putter because the outer door was locked, I didn't have a cell number, and though standing in the street tossing a peanut m&m at the window might not be well received. ;)
unobe ivanfrey: yes. /usr/bin/perl is 5.8.6 here.
audreyt heh
putter: I can come downstairs and get you in 23:40
putter f2f tomorrow sounds like a plan though
audreyt sure, that can do too
putter oh, or that
audreyt just come over :)
putter will do
in a bit. will give you a 5min warning. ;) 23:41
unobe ivanfrey: i'd try removing /sw from the path when building pugs to see how it goes 23:42
kolibrie ok, if I run as root, suddenly ghc-pkg cannot find package fps, nor if I relax the permissions of /usr/local/lib/ghc-6.4/package.conf and run as myself
ivanfrey unobe: perl -v
This is perl, v5.8.1-RC3 built for darwin-thread-multi-2level
(with 1 registered patch, see perl -V for more detail)
23:42 Daveman left
unobe ivanfrey: try removing /sw from your path when building pugs 23:43
kolibrie if I put the permissions back, ghc-pkg sees the file, but complains about not having permissions 23:44
audreyt kolibrie: ghc-pkg list
and seek fps among them
if it's in your user conf 23:45
then
ghc-pkg unregister --user fps
should've removed it
putter: the doors are open
pasteling "kolibrie" at 66.160.89.114 pasted "output from ghc-pkg list" (9 lines, 469B) at sial.org/pbot/18284 23:46
kolibrie fps seems to be missing
but make complained of having two of them 23:47
23:47 Daveman joined
audreyt kolibrie: if you type "make" now it still complains? 23:47
are there two GHCs on your system?
kolibrie I hope not, let me look 23:48
audreyt kolibrie: there's a package.conf somewhere that has fps in it
putter audreyt: oh, great. thanks. 23:49
ivanfrey unobe: I removed it halfway through the last build, so I should retry rebuilding it. Is there a make uninstall command?
audreyt (bbiab) 23:50
kolibrie audreyt++ # /me finds /usr/bin/ghc and /usr/local/bin/ghc-6.4
unobe ivanfrey: the first "it" being /sw in $PATH, the second "it" being pugs, right?
ivanfrey: prolly `make realclean` 23:51
ivanfrey unobe: oops sorry, yes
unobe: I might download the latest snapshot from the subversion repository. Is there anything preventing me from using Xcode? 23:53
putter steven: thank you for your kind words on the spike regards moose. it disappeared from collective consciousness so fast, I've never been quite clear on just how much or how little impact it had. my thanks.
23:54 lampus_ joined
putter & 23:56
unobe ivanfrey: i've never used Xcode. can't help ya there. 23:57
ivanfrey unobe: I better go eat. You just use an svn client and download the source somewhere in your home directory. 23:58
unobe: forgot the ? again. 23:59
unobe ivanfrey: yeah, i do svn co in my ~/dev dir

Powered by Cro and the RakuĀ® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!