6.2.5 released! | Your opinions wanted: perlmonks.org/?node_id=459750 | pugscode.org <Overview Journal Logs> | pugs.kwiki.org
Set by autrijus on 23 May 2005.
clkao it's killing 00:01
svnbot6 r3839, autrijus++ | * unbreak again
autrijus I need to get xsinit done
otherwise the perl5 part doesn't have dynaloader
clkao hang on
wrong macro
autrijus and how much fun we can have without dynaloader? not much.
clkao mine is Perl_eval_pv of course. 00:02
autrijus !
why?
clkao what why what?
you didn't have proper macro included
autrijus so in proto
proto.h:565:PERL_CALLCONV SV* Perl_eval_pv(pTHX_ const char* p, I32 croak_on_error);
what's this PERL_CALLCONV? 00:03
clkao doesn't mater 00:04
autrijus ok, fixed 00:05
clkao how?
Perl_
there should be proper macro
you need to include embed.h
does all the magic. otherwise something might not work 00:06
japhy hey, rules question for anyone who can answer
svnbot6 r3840, autrijus++ | * change to Perl_eval_pv
japhy m:foo{arg}/ rules... / is perfectly valid syntax
autrijus ok, included it
japhy it passes the closure {arg} to the user-defined modifier 'foo'
clkao $ ab -clkao- [~/work/pugs] ./pugs -e 'say eval_perl5("1")' 00:07
/libexec/ld-elf.so.1: Shared object "libperl.so" not found, required by "pugs"
japhy therefore, you'd have to write -> m:w {...}
instead of m:w{...} correct?
because the lack of whitespace makes the {...} denote a closure, as an argument to :w, right?
clkao you are not using ExtUtils::Embed
autrijus clkao: no I'm not... because the linking options to perl is not grokked by ghc 00:08
need the same sanitising thing jhorwitz did in Makefile.PL
for parrot-config
clkao hate
autrijus I did say that.
clkao you do want -Wl,-R/usr/local/lib/perl5/5.8.5/mach/CORE -Wl,-E
mrborisguy japhy: that sounds reasonable 00:09
autrijus I'm getting segfaults. fun
clkao: can you link libperl.so linked somehow?
./inc/Module/Install/Pugs.pm:< $ghc_flags .= " -I$Config{archlib}/CORE -L$Config{archlib}/CORE -i$Config{archlib}/CORE -lperl"
that line is the magic 00:10
japhy boris - ok.
then that's a typo in syn05 ;)
clkao which is not understood by ghc:
-Wl,-R/usr/local/lib/perl5/5.8.5/mach/CORE -Wl,-E -L/usr/local/lib /usr/local/lib/perl5/5.8.5/mach/auto/DynaLoader/DynaLoader.a -L/usr/local/lib/perl5/5.8.5/mach/CORE -lperl -lm -lcrypt -lutil -lc
mrborisguy i sure wish somebody who knew something would pipe up about it though!
japhy syn05 also doesn't clarify syntax for character classes. at least, it's not thorough
mrborisguy 'cause japhy, i really don't know... i think i've skimmed syn05 maybe twice at most 00:11
japhy boris - very soon I'll have a listing of all rule tokens on p6l for learned people to pore over
mrborisguy that'd be good
where are you going to post it? 00:12
japhy japhy.perlmonk.org/rules.txt, and then inform perl6-language of it
maybe add a perl6 directory there. 00:13
that'd make sense.
mrborisguy a'ight, i'll be sure to watch for it
japhy I think I'm done looking over apo5 and syn5, so it'll probably be posted in an hour or so
autrijus clkao: set libperl.so to your LD_LIBRARY_PATH
clkao: or ln it to /usr/local/lib
(For now)
clkao i know P 00:14
autrijus: now invoking anything returns 'Alarm clock' and then exit 00:23
svnbot6 r3841, clkao++ | Give proper ldopts to perl5 embed from ExtUtils::Embed.
japhy ok, email has been sent to p6l. 00:25
time to screw around for the rest of the night
g'night all
mrborisguy g'night
svnbot6 r3842, clkao++ | ccopts breaks the build. 00:34
clkao dear autrijus, eval_perl65 always returns 1 00:37
00:38
autrijus clkao: your change broke my build 00:41
/tmp/ghc36081.hc:6:20: EXTERN.h: No such file or directory
/tmp/ghc36081.hc:7:18: perl.h: No such file or directory
the -Wl,-R is not enough to make ghc see -L
clkao gah. still need ccopts 00:42
autrijus you need to filter it
see Makefile.PL for how
oh you already saw
clkao no i iddn't.. i just do it my way 00:43
ha
btw
autrijus heh
clkao eval_perl5 ("print 1") doesn't work
autrijus I'm fixing the eval prototpye and xsinit
try "warn 1"
it worksforme
clkao i'm confused. it worked again with ccopts
autrijus it worksforme too 00:44
what did you see?
mrborisguy are you guys trying to make it so perl5 code can be included?
autrijus some phantom thing?
clkao no nothing
mrborisguy will that mean i can include perl5 modules?
clkao eval_perl5('sfaksfl;'); 00:45
this returns 1 :)
autrijus mrborisguy: yes
mrborisguy: use DBI; # loads the p5 DBI if p6's DBI is not in path
that's the idea. 00:46
I think it's worth it.
mrborisguy oh, i think it's incredibly worth it!
p5 modules are the only reason i'm writing my current program in p5
autrijus you speak for all of us CPAN authors :)
clkao autrijus: ok, here's what i want 00:47
autrijus gets to port PAR's myldr/* to Haskell! fun! 00:48
clkao autrijus: can you make embed.perl5's eval_perl5 not calling eval_pv ectly 00:50
but call to something like perl5_eval_pv, which is some c file hooked to the build system
autrijus sure thing
clkao then i can do s interesting stuff
autrijus sure
clkao (of course perl5_eval_pv is a stu calling evaL-pv for now
autrijus let's call it pugs_perl5_eval() for now? 00:51
clkao sure.
make it a c file
src/Pugs/Embed/Perl5.o(.text+0x12c): In function `r2HZ_info':
: undefined reference to `boot_DynaLoader'
svnbot6 r3843, autrijus++ | * init dynaloader; add ccopts back.
clkao hate
autrijus clkao: you need ccopts
that will add back Dynaloader.a
which is neccessary
i.e. /usr/local/lib/perl5/5.8.6/mach/auto/DynaLoader/DynaLoader.a
clkao you are not including embed.h
i have it.
it was a 'g' merge 00:52
autrijus ok, add embed.h and commit then
clkao i rerun makefile.pl
autrijus ok. you are correct, we need a src/perl65/
re
src/perl5/
just like how syck is dealt
all C wrappers go there
clkao ok. just make ithooke to the build system so i can play with 00:54
{-# INCLUDE <embed.h> #-} 00:55
this is not comment no?
p 00:56
autrijus I'm fixing build system 00:57
no it's not comment.
clkao bah. it doesn't work. macro not resoled 00:58
commit is bloody slow 01:02
svnbot6 r3844, clkao++ | Make newXS work.
clkao great, now i have Alarm clock again 01:06
autrijus I'm fixing build system ;) 01:07
ijoyce_laptop where do the parrot folks irc at? 01:08
autrijus irc.perl.org
ijoyce_laptop thanks 01:09
autrijus $ ./pugs -e 'eval_perl5 "use DBI; print DBI->VERSION"'
1.48
please tell me you're happy
ijoyce_laptop fan-tastic 01:10
autrijus ok. now onto marshalling so that eval_perl5 can return something :)
btw, is it considered sane that under embed perl5, when pugs se
see
use DBI;
it search for perl5 when perl6 DBI.pm is not found?
i.e. we don't have another 01:11
use_p5 DBI;
clkao i think for now making it explicit is better
ijoyce_laptop use p5DBI;
use p5:DBI;
autrijus require_perl5('DBI');
require_perl5('DBI.pm');
that actually makes sense. I mean the .pm-less form 01:12
I hate the path munging anyway
so jusr module name
obra just magically falling back to perl5 seems very nice for migration 01:13
clkao now put eval_pv in c
autrijus yeah, but it has this burden
svnbot6 r3845, autrijus++ | * src/perl5/ for perl5 wrappers
r3846, autrijus++ | * dynaloader build infrastructure landed
autrijus of having CGI.pm6 shadowing of CGI.pm5
obra actually, i wonder if there's a way to use the perl6 module path naming to make it clean
autrijus and people will expect compat
obra: good idea!
ijoyce_laptop falling back without notice could cause fun debugging sessions
autrijus use DBI-(Any)-Perl5;
use DBI--Perl5;
is that the idea? 01:14
obra nods
autrijus use DBI-1.40-Perl5;
that does make the "Perl5" thing magical tho
whereas it's supposed to be uri
ijoyce_laptop bridge DBI;
obra it's supposed to be a URI? really
?
autrijus I think. check the syns?
obra wait. perl6 won't let me put a coderef in @INC?
autrijus er, I don't know 01:15
obra in the middle of a kwiki upgrade. it'll take a while
autrijus I'm not sure I care
obra so. if you can put a coderef in @INC, then you could make -Perl5 magic in a sane way
autrijus clkao: you can put eval_pv in c :)
clkao: I'll write the neccessary FFI in Perl.hs
clkao: just write perl5_eval
whatever way you like it to be
in src/perl5/perl5.c
clkao i'm still building
autrijus (and .h)
obra does that make sense? 01:16
autrijus obra: fwiw that doesn't quite make sense to me
obra: we are in a separate compilatino system.
it's no longer scripting :)
need to handle pugscc
clkao cc
autrijus let me do this typespace magic first. sec 01:17
I believe I can get it to work.
clkao++ # reduced trial-and-error by 100%
obra nods
clkao what trial and error? 01:18
you mean all this briding
autrijus clkao: getting the right embedding scheme figured out
clkao ?
autrijus right
clkao :)
autrijus I mean PAR's use case doesn't involve any marshal 01:19
it's purely perl5-to-perl5
and MzScheme's C part is too light
so I'm clueless about how to get two way bridging to work
clkao: here's a thought. 01:20
all perl5 values are SV right?
how about I make SV a native Pugs type?
clkao if that helps for ffi callbacks
autrijus I think it makes ~total sense
because then I can do introspection
clkao now 01:21
autrijus on haskell side
clkao foreign import ccall "perl5.h perl5_eval_pv"
perl5_eval_pv :: CString -> Word32 -> IO ()
what should this return?
autrijus () is void
you can change its return type by changing ()
I'm thinking having it retugning a Perl5SV
is the best bet
or just P5SV
and then I can make it a first-class Val 01:22
clkao ok... where to declare?
obra autrijus: URI identifying the author (or
authorizing authority).
clkao or you do that
kelan hmm. irc'ing with gaim
autrijus I'll do that now
then I'll make the SV stringify and numifyable
that instantly solve marshalling for simple cases.
then we move to SvHV and SvRV.
obra . o O { use DBI-1.40-perl5:/usr/bin/perl } 01:23
autrijus use DBI-1.40-perl5;
obra some sort of degenerate internal URI scheme seems not insane
autrijus I think that works best.
svnbot6 r3847, autrijus++ | * require_perl5
obra sure
I actually like that
autrijus then that is that.
obra will there be a perl5 interpreter compiled in or just on the fly bound?
clkao let's sort the internals out first 01:24
i have a simple wrapper 01:25
autrijus clkao: PerlSV type is in
obra: compiled in
obra: we can't quite do on the fly bound without a NCI
obra: clkao's layer is in pure C
so future NCI can target that
so it's not haskellish :) 01:26
obra nods 01:27
Oh god. the new kwiki antispam is so much better
clkao autrijus: ./pugs -e 'eval_perl5 ("print 1")' 01:31
svnbot6 r3848, autrijus++ | * PerlSV type for clkao.
r3849, autrijus++ | * add back boot_DynaLoader
clkao this still doesn't work for me
autrijus what's the werr? alarm? 01:32
clkao nothing
autrijus $ ./pugs -e 'say require_perl5 "DBI"'
pugs: cannot cast from PerlSV 0x084464f4 to [Char] 01:33
clkao ./pugs -e 'say eval_perl5 ("print 2")'
autrijus yay! I'm only one step till done!
clkao this printgs 1
autrijus clkao: try trunk
let me finish simple marshalling
clkao ?? i am using trunk 01:34
autrijus committing
r3850
clkao it's the print isn't got run 01:35
btw i have unspecced/p5/basic.t
laptop gets soooo hot 01:38
svnbot6 r3850, autrijus++ | * PerlSV is now a native Pugs type.
autrijus cool
try using warn
clkao same
autrijus then you're borked.
probably initperl5 didn't get run 01:39
trace it at perl5.c level?
clkao fsck
autrijus it works fine here btw.
you can write perl5.c without running it I'm sure, too
clkao pugs: cannot cast from PerlSV () to [Char]
ha
coool
autrijus I've done the casting 01:41
clkao no it's not called 01:42
what should i do?
autrijus committing 01:43
sproingie swears up and down 01:44
i swear, windows BOOTS faster than it responds to a damn delete keypress in explorer 01:45
clkao autrijus: i put a fprintf in perl5_init.. not called
autrijus clkao: look at pugsconfig 01:46
src/pugs_config.h
make sure perl5 is defined
clkao #undef PUGS_EMBED_PERL5
!*%#%
autrijus ...
!$#!$ 01:47
you did run makefile.pl one without the env right? ;)
fix the config bah
clkao it doesn't regen
?
i did run it with the env
and i even make clean
autrijus $ ./pugs -e 'say eval_perl5 "use DBI; DBI->VERSION"'
1.48
clkao still the same :( 01:48
autrijus hm? rm src/Pugs/Embed/Perl5.{o,hi} 01:49
then make again
svnbot6 r3851, clkao++ | Basic tests for eval_perl5.
clkao does the test pass for you now?
autrijus 1..3
ok 1 - number
ok 2 - number
Can't locate object method "perl6" via package "now" (perhaps you forgot to load "now"?) at (eval 4) line 1.
clkao hahaha 01:50
autrijus heh bad test
clkao fix that
more quotes
autrijus all pass!
done
amazing how fast we did this
now moving to handle the "use" syntax and namespace numbing
clkao :))
alarm clock for me again 01:52
right after perl5_init 01:53
autrijus hrm. 01:54
svnbot6 r3852, autrijus++ | * SV stringification!
r3853, autrijus++ | * bad test
autrijus I can't duplicate it at all
clkao it's after perl5_init finsihes 01:55
50929 pugs PSIG SIGALRM caught handler=0x82dd588 mask=0x0 code=0x0 50929 pugs PSIG SIGALRM caught handler=0x82dd588 mask=0x0 code=0x0 01:56
maybe it's becaues ghc is threaded but my perl5 isn't 02:01
autrijus ah. that can be it 02:03
you can remove -thread and try
ok, "use" parsed 02:04
class space initialised
clkao WORKS
initialzied as?
you need a magic object for class method 02:05
autrijus I did
and yes I'm doing that
eric256 after reading the "Why aren't you using Perl6 yet?" post it seems that most people either don't understand pugs or are perfectly happy to let other people write the next progamming language for them....odd and slightly disheartening. until i joined pugs development i though perlmonks was pretty cutting edge and moving forward...now they seem like cavemen who run from the burning stick. ;) 02:08
autrijus eric256: heh, no, it's perfectly reasonable to expect something usable in production before using it 02:10
ok, now we need AUTOMETH. 02:11
eric256 i suppose before using it in production. just seems like lots of people expect perl6 to materialize out of then air...i don't realy have the time for it, and all there reasons apply to me as well. but perl6 has enough new amazing features that i'd like to do whatever possible to help get it here now! lol. dinner either way. 02:12
later
clkao why isn't freebsd building threaded perl by default?
autrijus clkao: no idea
eric256 autrijus++ for pugs! 02:14
svnbot6 r3854, autrijus++ | * parse for `use DBI-1.40-Perl5`
r3855, clkao++ | Disable threads when embedding non-threaded perl5.
clkao autrijus: so i guess what you will do is make sv convertable to obj that can invoke method.
eric256 just thought i would throw that in there in case it wasn't assumed already ;) 02:15
clkao and then you forge an object and link the invocation to perl5_call.. then i can take over (or you are not tired yet and want to see if c is adorable again :)
for now i am going to sleep
autrijus enjoy 02:17
I've already got the obj autoload done.
I can now invoke arbitary p5sv's methods.
clkao coooool
autrijus the other way is gonna be harder.
clkao if you can make some easier way to create pugs objects from c .. 02:18
anything better than eval
autrijus trying to think if that makes sense actually 02:19
I can export a seriest of casters
int -> pugsval
char* -> pugsval
will that help?
clkao hang on. 02:20
i just give you sv and it turns to pugsval
autrijus to avoid class clash I'll use scalar space for now 02:25
i.e. $DBI
$*DBI actually
clkao i take it's an SV pv ? 02:26
commit!
autrijus committing 02:28
* Full PerlSV object reification in Pugs space.
r3856.
$ ./pugs -e 'use DBI-1.41-Perl5; say $DBI.VERSION' 02:29
*** Unimplemented unaryOp: Scalar::Perl5::AUTOLOAD at -e line 1, column 25-37
fun not?
now how do I implement that ;)
need your callconv! 02:30
clkao make it call perl5_call
autrijus perl5_call (SV *sv, char *methname, SV**args)
clkao make a stub entry to return "ping clkao"
autrijus can we have that?
svnbot6 r3856, autrijus++ | * Full PerlSV object reification in Pugs space.
clkao wtf SV**args?
autrijus $DBI.foo($DBI, $DBI)
or rather, just 02:31
clkao AV *args?
autrijus perl5_call(char *subname, SV**args)
why AV? AV is a SV you know
perl5_call(char *subname, int argc, SV**args)
that make sense?
that's easiest to me
can you manufacture that? just push args to stack 02:32
and call_pv subname
clkao sure
autrijus easy as pie
k, please do that, I'll do haskell side 02:33
clkao but i need to sleep now
make a stub!
autrijus aww.
clkao and a failed test
autrijus nah. I'll just implement that ;)
clkao bwah
autrijus is having too much fun to stop
clkao don't take away my fun!
autrijus ok ok ;)
clkao actually you want a flag for CALL_METHOD or CALL_PV
call_pv ('VERSION', 'DBI') won't properly resolv 02:34
autrijus I know. I'll use call_method
with G_SCALAR
<- actually read EaEP 02:35
done 02:37
give it a sanity review if you can 02:38
r3857
clkao i will do a quick review
autrijus foreign import ccall "perl5.h perl5_call" perl5_call :: CString -> CInt -> Ptr PerlSV -> IO PerlSV 02:39
FFI++ # too easy!
clkao refcount on POPi ?
er POPs i mean
autrijus dunno how to do that properly
_inc? 02:40
please just fix away
clkao can wait till a failed test :)
return sv not char* 02:41
autrijus k.
ah right.
please fix away 02:42
<- is already in haskell side
clkao -> sleep 02:43
svnbot6 r3857, autrijus++ | * add perl5 callconv
autrijus ok. callconv passed. now linking to autoload 02:46
will it work?
clkao people must be curious 02:48
svnbot6 r3858, clkao++ | call returns SV *, not char *.
autrijus it compiles... 02:49
(ship it!)
clkao ship it!
i mean, commit it XD
autrijus committing 02:51
ah. callconv need SPAGAIN 02:52
fixing
eric256 what are you guys working on?
autrijus eric256: use DBI from perl6
we're very close.
eric256 all perl5 modules? or are you just focusing on DBI? i would guess getting it to work would get others to work 02:53
svnbot6 r3859, autrijus++ | * AUTOLOAD now links to callconv
autrijus all perl5 modules.
$ ./pugs -e 'use DBI-0.01-Perl5; say $DBI.VERSION' 02:57
1.48
WE DID IT!
jql impressive
autrijus $ ./pugs -e 'use CGI-0.01-Perl5; say $CGI.header'
Content-Type: text/html
clkao :) 02:58
mrborisguy YEAH!!!! 02:59
svnbot6 r3860, autrijus++ | * callconv fixed.
mrborisguy rejoices
autrijus moving to typespace!
eric256 starts a new build 03:01
obra autrijus noce 03:05
nice
autrijus :) 03:06
Khisanth you have libperl linked into pugs? 03:07
autrijus yes. 03:08
this time with dynaloader.
clkao fix my test and you rule
clkao really tries to sleep 03:09
autrijus typespace populated. filling in metaobj 03:11
svnbot6 r3861, clkao++ | Tests for p5 method calls.
autrijus $ ./pugs -e 'use CGI-0.01-Perl5; say CGI.header'
Content-Type: text/html
svnbot6 r3862, autrijus++ | * type space populated! 03:20
Khisanth hrm 03:31
so pugs just acquired all of CPAN
jql theives 03:32
Khisanth hostile takeover :)
autrijus ./pugs -e 'use Digest-0-perl5; my $md5 = Digest.MD5; $md5.add("Pugs!"); say $md5.hexdigest'
f08bcb32d6eea3d3c9ec2f924ce33067
mrborisguy i'm loving it! 03:33
jql you can't have my modules
:P
Khisanth what is with the version numbers though?
autrijus Khisanth: minimal version number thing
I'm making it optional 03:34
jql waits for pugs to compile
autrijus works
mrborisguy starts new build 03:36
crysflame win move 5
autrijus yo. :)
crysflame i thought it was just quiet
svnbot6 r3863, autrijus++ | * string and integer parameter passing.
mrborisguy am i supposed to reconfigure every time?
or i guess, rerun Makefile.PL every time? 03:37
jql feels this sudden urge to get a faster computer
japhy ignore that urge. it's bad and wrong. 03:38
mrborisguy i can hardly wait... i wish i were running something faster too at the moment!
jql yeah. pugs.parser just compiled, and the urge is diminished
Khisanth heh compiling pugs does tend to cause an urge to get a faster computer 03:39
mrborisguy i just may have to rewrite the entire program i just wrote in p6 now
autrijus mrborisguy: no, just "make" is enough 03:40
jql hell, with CPAN starting to work on P6, my desire to rewrite modules just went down
jql chuckles
autrijus that's as expected :)
japhy what will the Perl 6 interpreter be written in, ultimately? 03:41
is the Haskell just a prototype thing?
mugwump forth
jql prolog would be a natural
japhy befunge! 03:42
kelan malbolge
mrborisguy autrijus: thanks
jql Perl6#
japhy intercal.
Khisanth japhy: Perl 6 of course ;)
mrborisguy isn't perl6 going to be written in python?
autrijus japhy: the perl 6 interpreter will no longer exist 03:43
japhy: there will be a perl 6 _compiler_ though :)
Khisanth mrborisguy: only if they decide to kill the whole project :p
jql C:\Perl6-Pugs-6.2.5>pugs -e "say 'Hello World!'"
autrijus that takes a perl 6 program and emit some kind of assembly.
jql Hello World!
sweet
mrborisguy Khisanth: haha
crysflame hiya, japhy
autrijus japhy: that "some kind" may be parrot assembly, or C, or C--, or LLVM, or god knows what.
japhy: and the ultimate goal is for that compiler itself written in perl 6 03:44
so itself can also run on parrot, or C, or C--, or LLVM, or god knows what.
japhy hi crys
autri - ok, by "interpreter" I meant "parser and compiler"
my point is, you're using Haskell right now to interpret Perl 6 because it seems to be a natural fit for the job 03:45
autrijus japhy: pugs can also emit parrot assembly from perl6 code :)
japhy that's cool.
autrijus ./pugs -BParrot examples/mandel.p6 # for example
japhy I'm just curious how this isn't pulling yourself up by your bootstraps
autrijus japhy: so the interpreter part is pluggable 03:46
japhy: the next logical step is to make the parser to use rules
japhy: the next logibal step is to port the codegen to perl6.
that's all.
japhy so you'd use pugs to temporarily support the work until you've used it to write the actual perl 6 compiler?
autrijus that is one route.
it is also possible to simply machine-translate pugs to PIR or to Perl6. 03:47
japhy ok
autrijus at least the codegen/parsing part.
mrborisguy it seems really odd... but if you think about it, the C compiler is written in C too
autrijus but one way or another it makes bootstrapping much simpler.
japhy yeah.
autrijus since at any time we have an interpreter to fallback to.
so eg. BEGIN{...} can work for the first time
otherwise it's very hard.
have to essentially recompile for each BEGIN 03:48
japhy interesting
one day I'll wrap my brain around this whole thing.
autrijus :)
japhy I'll just get back to working on that Perl 6 rule parser
mrborisguy did you ever post your txt file? 03:49
i couldn't find it, japhy
japhy yeah.
jql it's like trying to understand quines
japhy japhy.perlmonk.org/perl6/rules.txt
mrborisguy japhy.perlmonk.org/rules.txt?
japhy no, I added perl6/ to it
mrborisguy got it
japhy I figure there'll be more than one perl6-related file I ever write ;) 03:50
already got a bit of feedback from J S Duff
mrborisguy it looks good 03:51
i'll have to really sift through it to give you any useful feedback
yeah, i'm sure you'll write more
stevan autrijus: so do we have perl5 in pugs now? 03:52
stevan backlogs in amazement
autrijus stevan: yes. 03:53
mrborisguy autrijus: i must admit, i'm rather amazed at how fast you guys did that! 03:54
stevan very very very very cool
autrijus use Digest--perl5;
my $md5 = Digest.MD5;
$md5.add('Pugs!');
mrborisguy dd you do it all today?
autrijus say $md5.hexdigest; # f08bcb32d6eea3d3c9ec2f924ce33067
mrborisguy: yes, most of it.
in 3 hours or so 03:55
mrborisguy that's just unbelievable
crysflame that's autrijus
autrijus (and clkao.)
stevan its the SVK :)
jql Digest.MD5 can't be right, can it? 03:56
japhy Digest::MD5.new, no?
or am I loopy?
jql that's how it should be, yeah
stevan isnt Digest the factory module
autrijus perldoc Digest
japhy WOW! I KNOW PERL 6!
Khisanth or we need to read up on perl6? :P
autrijus "perldoc Digest" for enlightenment :) 03:57
japhy pshaw.
jql I will enlighten myself shortly
autrijus there's lots of rough edges
svnbot6 r3864, autrijus++ | * config tweks for PUGS_EMBED=perl5
autrijus but I'll fix them when I wake up.
japhy I never knew they could be used like that.
autrijus the bulk is all there.
japhy err, s/they/it/
jql interesting. I've only ever read the Digest::MD5/SHA1 pod, never the Digest pod 03:58
didn't realize it was gonna play that way
and Digest::MD5 says Digest::MD5->new :P 03:59
japhy yeah, seriously. I always just used the Digest::X module I needed
since I was never in need of a factory
autrijus I posted to PM 04:00
# perlmonks.org/?node_id=460233
jql checks to see if nmake test is done
Khisanth I can usually hear if it is done :)
japhy cool beans, autrijus.
jql urge to upgrade computer... rising
I'm in southern cali. fan is on, TV is on, can't hear hard-drive over that. :) 04:01
Khisanth flips jql's upgrade urge overflow bit
jql: I don't hear the hard drive, I hear the fan 04:02
this is a laptop
jql ahh
stevan autrijus: as always, I am utterly amazed 04:03
stevan edges ever closer to using Pugs in production :P 04:04
autrijus :)
stevan I am too sleepy to write tests now, but I am setting aside some tuits for tomorrow 04:05
jql urge to link software with pugs... rising
stevan so very very cool
jql anyone written a gui with pugs, yet?
crysflame if i were the borg, i would leave borg artifacts everywhere and wait for cultures to link to them
at which point the culture would become a borg extension culture 04:06
mrborisguy *** cannot cast from VUndef to ()
at ./dates.p6 line 6, column 11-21
f0rth gui with hs-plugins and wxHaskell?
stevan wanders happily off to bed, to dream of dancing pugs :)
jql <dalek>assimilate</dalek>
mrborisguy line 6 is: my $md5 = Digest.MD5;
in the exact example autrijus gave 04:07
crysflame maybe $md5= Digest--MD5.new;
autrijus mrborisguy: you didn't build with embedding perl5 04:08
mrborisguy: you need to do:
env PUGS_EMBED=perl5 perl Makefile.PL
rm src/Pugs/pugs_config.h src/Pugs/Embed/Perl5.o
and make again
mrborisguy oh, okay 04:09
i'll try that
thanks
autrijus np :) 04:10
now I sleep. &
autrijus waves
eric256 to embed perl do you need to be building perl5 from source? 05:10
gaal i don't think so eric, but you probably need the same compiler as the one your perl5 was built with. or close enough. 05:11
mugwump aiui you need dev. headers and libperl
eric256 well i have a ActiveState release soo that is probably not going to happen :)
gaal eric, the other day i just happened to build my own perl on win32. it really wasn't all that bad, if you're up to d/ling about 250mb. 05:12
:) 05:13
eric256 is there a paste bot i can put my huge error on?
gaal actually the prospect filled me with dread, since i hate the windows command line so much, but now that i have a working cpan it's such a relief. 05:14
perlbot nopaste
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
eric256 yea i don't realy feel like screwing with my perl5 install since it is all working and happy
pasteling "eric256" at 69.234.109.88 pasted "Build Problems. ActiveState + Pugs ;) + EMBED_PUGS=perl5" (169 lines, 11.1K) at sial.org/pbot/10395 05:16
eric256 nmake and i arn't friends so i just included the whole nmake unoptimized out put.....i'm hoping someone can help me figure out why it failed ;)
gaal hmm, it's working for me 05:20
eric256 with ActiveState?
gaal no, my shiny own perl 05:21
eric256 yea i'm pretty sure its related to activestate...what compiler did you use with your perl? cpan on windows would be nice
no more icky ppm
gaal oh, you can use cpan with activeperl 05:22
if you do have a compiler.
(or even without one, if you don't need to compile XS)
eric256 yes. if i had a compiler i wouldn't be using activeperl though 05:23
gaal i used the free MS toolchain
eric256 toolchain?
free and MS in one sentence? you must me the diseas and not the company ;)
gaal okay, it goes like this: Microsoft have a free version of their compiler plus build tools
Khisanth well free as in beer anyway 05:24
gaal it's mostly command line
and not the very newest.
but it works!
eric256 hmmm. seems like going over to the dark side. ;)
gaal (and also, seems very fast.)
Khisanth hmm
gaal however, you need to gather bits and pieces a bit to make it useful 05:25
Khisanth does ghc build with mingw?
eric256 where would i find this? i've tried compiling perl with VC 6 a couple times with no luck.
gaal it's build *with* msys, Khisanth.
msys ~~ mingw
eric256 mingw?
gaal eric, okay, i gotta run to work unfortunately but the short of it:
- grab perl-5.8.6.tar.gz 05:26
Khisanth www.mingw.org :)
gaal - read README.win32
- d/l three kits from microsoft.com [1]
- set up paths
- cd perl-5.8.6/win32
- vi Makefile
ingy anyone have knowledge about guid generation? 05:27
gaal - nmake etc.
that's it; really
oh, there's a couple of things you have to build manually but it's in the README
Khisanth what happened to Configure?
ingy is there a perl module to generate generic guids?
eric256 is quite sure gaal left out all the hard things. lol. thanks
gaal ingy, on p5 i think there's one on cpan
Khisanth what is a guid? 05:29
gaal eric, just remember, one of the three kits has many subfeatures you can choose from; two of them are necessary for building perl so don't leave them out. this is mentioned in the README.
oops, too late.
ingy I suppose just taking an md5 of a hires time and a couple other unique vals would work
or sha1
gaal for what, ingy?
obra ingy: Data::UUID
ingy mmm
gaal ingy, you can call that from pugs now :)
ingy cool! 05:30
thanks y'all
Khisanth renames pugs to frankenstein's monster and autrijus to dr. frankenstein
gaal the frankensteins were fast and graceful? 05:31
Khisanth it depends on the version
ingy :) 05:32
obra: that's whatI was looking for
wow it's all XS and stuff 05:33
obra though there's a pure perl version too
ingy how does that work under pugs? 05:34
ah
svnbot6 r3865, autrijus++ | * new perl5_can API to test whether a method can be found 05:46
r3865, autrijus++ | in perl5 land, or fallback to perl6.
r3866, autrijus++ | * bareword classes now stand for their metaobjects, not the
r3866, autrijus++ | simple type literal. this makes class-level methods possible.
r3867, autrijus++ | * only look up AUTOLOAD when there is possibly a hit for AUTOLOAD.
r3867, autrijus++ | * therefore `DBI.connect` and `Digest::MD5.new` no longer refers to
r3867, autrijus++ | the builtin prim `connect` and `new`, but rather to their Perl 5
r3867, autrijus++ | space methods.
gaal oops, i was wrong earlier when i told eric PUGS_EMBED works for me on win32 - i get the same error 05:59
looks like it may be a ghc issue :( 06:03
as its definition is wrapped in a defined(_MSC_VER) in HsBase.h 06:04
#if !defined(...)
oy, there's alsmost certainly a bug there, if i'm looking at the sources for the version i'm using.. but the cvs HEAD looks fixed 06:06
i'll try this later when i have tuits for it, for now it seems as though p5 embedding on win32 might require a newer ghc than 6.4 06:07
gaal -> $work & 06:12
06:32 chady_ is now known as chady 06:33 castaway_ is now known as castaway
castaway mornin 06:35
x86 morn 06:39
07:10 wolv is now known as wolverian
Aankhen`` Closure interpolation in an s/// doesn't work? 07:47
nothingmuch write a test? 07:49
Aankhen`` OK.
Where should it go? 07:50
t/rules/s_perl5_closure.t?
nothingmuch i think probably under quoting
Aankhen`` Ah.
Khisanth shouldn't that be perl6?
nothingmuch wait, should s:p5/// behave like normal s///?
Aankhen`` I guess so...
nothingmuch if it wants to be compatible, then shouldn't the quoting work like it did before?
or should the translator escape this stuff 07:51
Aankhen`` I think the :P5 only changes the interpretation of the pattern.
# The new :perl5 modifier allows Perl 5 regex syntax to be used instead:
m:perl5/(?mi)^[a-z]{1,2}(?=\s)/
(It does not go so far as to allow you to put your modifiers at the end.)
nothingmuch okay, fair enough, since the substitution is just like quoted text 07:52
i would put it in one of the quoting tests
Aankhen`` OK, I'm going to commit my changes to CGI.pm, then I'll write up a test for the closure thing.
nothingmuch "If file does not graphic or documentation then backup it" 07:57
readable comments... *sigh*
castaway nice comment
nothingmuch getHeaderNema 07:59
it takes just as much effort to fix that where it's defined instead of use it this way
gah!
Aankhen`` Wow, svnbot6 is slow at picking up commits today... 08:07
svnbot6 r3868, Aankhen++ | * rearranged parameters for &CGI::header to be more like P5 &CGI::header; made signature more specific.
r3868, Aankhen++ | * added support for extra headers in &CGI::header.
r3868, Aankhen++ | * fleshed out &CGI::redirect to support all parameters supported by P5 &CGI::redirect.
r3868, Aankhen++ | * updated header.t.
r3868, Aankhen++ | * added redirect.t to test &CGI::redirect.
r3869, Aankhen++ | * fixed a few tests broken by my previous commit.
Aankhen`` shuts up.
OK, examples/memory_game/ doesn't seem to be using header() or redirect(). 08:08
nothingmuch >> Where do the quoting tests go?
nothingmuch t/operators/quoting.t 08:09
but that might not be 100% correct
Aankhen`` Ah.
nothingmuch it might be better placed in t/syntax/string_parsing.t
or t/rules/s_perl5
oh, i dunno
Aankhen`` Hrm.
I think this is mainly testing interpolation in regular strings. 08:10
I'll add some in the rules section as well which specifically test manipulating and interpolating $0, $1, etc.
nothingmuch look in those files to find tests similar to what you wrote
and see if it fits there
i didn't understand that sentac
nce
Aankhen`` Which one? 08:11
nothingmuch "I'll add some ..."
Aankhen`` Ah.
I will add some tests to t/rules/s_perl5 (or whichever file is appropriate) which specifically test stuff like s:P5/foo(bar)/{ ucfirst(lc($0)); }/ and so on. 08:12
But first... lunch. ;-)
BBIAB.
nothingmuch bon appetite 08:13
Aankhen`` Weird... "is export returns Str" won't work; you must use "returns Str is export". 08:54
vcv- I'm sure autrijus had it fixed before you hit enter 08:57
clkao *yawn* 08:58
now let's see if my p5 tests pass
Aankhen`` Heh.
Is the latest trend to use try { ... } in place of eval_is()? 08:59
svnbot6 r3870, Aankhen++ | * &CGI::header: use method call interpolation to uppercase letters.
r3870, Aankhen++ | * re-added "returns Str" and "is export" traits for &CGI::escapeHTML and &CGI::unescapeHTML.
clkao (VType (mkType "Class"),(mkType "Type"),(mkType "Class")) 09:01
debug message!
Aankhen`` I couldn't figure out why that appears.
nothingmuch Aankhen`` eval_is is for string eval 09:06
as for try - i think lives_ok is a better choice
unless the return value is needed
Aankhen`` Ah. 09:07
Hrm.
OK.
nothingmuch wow! 09:08
nothingmuch wrote his first line of C++ code that actually worked like he wanted on the first try
i think i'm starting to make progress
Aankhen`` Congrats!
;-)
castaway grats nm! 09:09
nothingmuch and this involved weird ifdefs and extern char * foo
i hate lousy design... it's so permanent
Aankhen`` lives_ok takes a code block? 09:10
nothingmuch afaik yes
but i haven't used it
Aankhen`` OK.
castaway lives_ok sounds fun :) 09:15
svnbot6 r3871, Aankhen++ | * actually added ext/CGI/t/redirect.t
vcv- nothingmuch: wait until you write a 1000line library and have it work on the first try :p 09:16
nothingmuch vcv-: i hope to have less C++ experience than that 09:18
vcv- smart move 09:20
Juerd fun.sdinet.de/pics/raid.jpg 09:28
Aankhen`` GODDANGIT 09:32
pasteling "Aankhen``" at 61.246.35.93 pasted "Pugs going crazy" (22 lines, 323B) at sial.org/pbot/10402 09:34
Aankhen`` It was working perfectly fine... then I added tests for French quotes and it started giving errors. 09:35
Bleh. 09:50
Nevermind. -_-
svnbot6 r3872, Aankhen++ | * added t/builtins/lists/enum.t 10:02
clkao autrijus: ping me if you are back 10:49
svnbot6 r3873, iblech++ | * EOL at EOF added to times.t. 11:05
r3873, iblech++ | * Usual svn properties added to src/perl[56] and t/pugsbugs/namespaces.t.
r3873, iblech++ | * Changed times.t to treat a diff of 0.0001 as 0.
r3873, iblech++ | * Added a diag() to sleep.t and removed n_ge and n_le in favor of &infix:«>=»
r3873, iblech++ | and &infix:«<=».
r3874, iblech++ | hyper.t -- EVALed two tests and added two new ones. 11:32
r3875, iblech++ | * unTODO a succeeding test of oo/class/basic.t. 11:48
r3875, iblech++ | * Comment out the debugging message when creating objects using "new".
r3876, autrijus++ | * Fix eval_perl5 building on Mac OS X 12:24
Juerd is compiling a kernel 13:05
for feather
svnbot6 r3877, scook0++ | * Minor English tweaks to PA02
r3877, scook0++ | * Fixed a comment that was confusing Haddock
Juerd Installation so far went well
I used this machine to test my new installation script, automating the procedure at juerd.nl/debianraid
juerd.nl/files/raidthingy.pl 13:06
scook0 Juerd: so the name is chosen? I was having so much fun learning 'onion' in different languages... :)
Juerd It was fun, but wildly off topic :)
scook0 aye
Juerd I think that was the largest off topic subthread p6l has known so far :)
(Oh, re raidthingy.pl, it's very messy - I blame my not having coded for a long time) 13:07
nothingmuch Juerd: know how to combine raid 1 on root with LVM? 13:08
as for a clean installer, when we have tuits stevan or I will release something we wrote for that 13:09
Juerd I have no LVM experience yet 13:10
I try to stay away from anything that makes a drive unreadable without tools
scook0: Feel free to start a subthread of multilingual feathers :) 13:11
feather boots! 13:14
and works!
woo.
The first raidthingy-installed box is a fact.
nothingmuch Juerd: if you're interested my C++ brain is fried, so I'll work on releng for Verby 13:15
mostly documentation
Juerd releng? 13:17
verby?
what are you talking about?
nothingmuch verby is that installer framework i was talking about
and releng is release engineering
autrijus uses it here occasionally, so I thought it was well known
wolverian Juerd: why are you compiling the kernel? the standard debian kernel didn't fit? 13:22
theorbtw1 (Compiling kernels)++
Juerd Oh. 13:27
wolverian: Standard linux kernels, not limited to debian, suck. 13:28
wolverian: They are bloated and inefficient.
And having the very core of a system be inefficient is stupid :)
wolverian Juerd: hmm. okay. do you have any tips/urls for me to look at if I want to recompile my kernel? I do know _how_ to recompile, but I don't know what I should be optimising.
Juerd Besides that, a moduleless kernel is much more secure, because almost every rootkit out there makes use of kernel modules
wolverian: Just include everything you need, and exclude everything you don't. :) 13:29
wolverian Juerd: right. :) reminds me of my FreeBSD days.
Juerd wolverian: See step 15 at juerd.nl/debianraid for rough instructions
theorbtw1 wolverian: Make sure your CPU and memory size is set correctly, and if you need SMP or not.
dvtoo it's not that hard to load modules with /dev/kmem ;)
Juerd dvtoo: True.
dvtoo if someone has local access to your box, you've already lost
theorbtw1 dvtoo: He's giving out ssh accounts; that's the point of the box. 13:30
13:30 theorbtw1 is now known as theorbtwo
wolverian Juerd: does grsecurity have any downsides currently? 13:30
dvtoo oh, sorry, he's just kind of screwed then ;)
Juerd wolverian: Restrictions often make users complain. 13:33
autrijus jabbot: seen clkao 13:37
jabbot autrijus: clkao was seen 1 hours 35 minutes 30 seconds ago
autrijus rehi all :) 13:38
clkao *yawn* 13:39
Limbic_Region anyone seen chip? 13:40
salutations all btw
autrijus clkao: how's life? perlembed working for ya? 13:41
theorbtwo Is there a reason that PUGS_EMBED="perl5" isn't the default?
Juerd one of these: gcc vim-perl zsh ghc-cvs ssh
depends on X!
argh. :)
theorbtwo: You may not have libperl installed 13:42
clkao bwah
autrijus: i want the vcode map to sv
autrijus theorbtwo: no good reason except when the C compilers for GHC and Perl are not the same it'll break -- also what Juerd said -- but you can apply a similar probe like parrot's -- in that case might as well enable parrot by default when it's probed -- can you help hacking it in? 13:43
clkao: you mean calling into p6?
clkao: or exporting p5 funcs in p6? 13:44
theorbtwo tries it, and gets a bunch of syntax errors.
clkao see my p5/method.t ?
nothingmuch autrijus: what implications does perlembed have on perl from GHC?
is the bridge very pugs specific?
autrijus nothingmuch: no, the bridge is complete generic
see src/perl5/*
nothingmuch because i would *love* to use perl and haskell together
this is an excuse to start coding that Pesto project 13:45
autrijus and Pugs.Embed.Perl5
both are not specific to pugs
I mean they don't use any datatypes in AST
nothingmuch how close is Inline::GHC?
gaal|work shoudl PUGS_EMBED be perl5 or parrot? :)
autrijus I was asking for opinions regarding whether to work on perlembed or ghcembed
gaal|work: you can put both in -- separate by some isPunct
nothingmuch autrijus: i'll try to add perl5 to embed later today 13:46
autrijus nothingmuch: I was leaning to ghcembed aka Inline::GHC, but clkao showed up and proved that perlembed is trivial
nothingmuch: cool
nothingmuch: so I'm afraid I'm going with perlembed for now
nothingmuch but it's lag baaomer, so no promises on tuits
autrijus ghcembed can wait ;)
that's fine
nothingmuch it's not really lag baaomer
theorbtwo Oh, at the least, we should make the messages at Makefile.PL time be more obvious -- since the p5 one says "containing", and the parrot one says "equal to"... IIRC.
nothingmuch it's the work party for thereof
theorbtwo wonders what ghcembed is.
autrijus theorbtwo: sure. fix away!
theorbtwo: Inline::GHC
via hs-plugins
gaal|work perlbot nopaste 13:47
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
pasteling "gaal" at 212.179.43.225 pasted "make errors / linux" (32 lines, 2.2K) at sial.org/pbot/10405
gaal|work oh, does perl5_embed require hs-plugins? 13:48
autrijus nope 13:49
hm, seems ghc has different 64bitness than perl5
gaal|work hmm. the crypt error is prolly some missing ssl lib / dev package, but i don' tknow about the others.
autrijus are you compiling both with the same gcc?
gaal|work ghc-6.4 btw.
err, 6.5 snapshot actually 13:50
clkao autrijus: so, if you can bind vcode into a cv
or some special bless sv
gaal|work my perl5 is not 64bit. how do i check my ghc? 13:51
theorbtwo I get similar errors, and I'm on an ia32 chip, so shouldn't be any 64-bittyness. 13:53
(Unless you mean filesizes?)
gaal|work my system gcc is 3.3.6-5; does 'ghc-cvs' on debian come with its own gcc like it does on msys?
nothingmuch i think this is the largefile support complaining
gaal|work dpkg -L ghc-cvs | grep gcc # it doesn't
well, i do have largefie in my p5. 13:54
svnbot6 r3878, iblech++ | Parse [+]«. 13:55
r3879, iblech++ | Main.hs -- die() should output correct utf-8.
r3880, iblech++ | Pugs.Prim -- readline() should read utf-8.
r3880, iblech++ | (This was the real reason why svnbot6 outputted funny chars instead of the
r3880, iblech++ | correct ones!)
Juerd After unpacking 43.3MB of additional disk space will be used.
and that's just emacs
WHOA.
autrijus iblech++ # svnbot6 13:56
nothingmuch as they say, emacs is a wonderful OS, it only needs a good editor
Juerd When emacs is installed, I have all requested software there
autrijus and mozilla is a wonderful OS, it only needs a good browser 13:57
Juerd I like firefox
autrijus ...which is called firefox, so it's no longer funny
Juerd heh
gaal|work googling suggests -DGNU_SOURCE etc., but we have that. 14:01
autrijus clkao: ok, got it. will make it work 14:07
Limbic_Region [plus plus] autrijus, clkao 14:11
that 3 hour hack yesterday gave me a lot of advocacy ammunition this morning 14:12
theorbtwo For l~r: autrijus++ clkao++
Juerd users are created
Juerd preps for the voyage to the data centre 14:13
But first, let's see if pugs compiles with the currently installed tools
autrijus Juerd: with embed_perl5 :) 14:14
EMBED_PUGS=perl5 that is -- also check src/Pugs/pugs_config.h to see it's really set to 1
Limbic_Region++ # advocacy
Juerd real 1m12.981s 14:16
unoptimized
autrijus: Sorry, no time right now for extensive tests :)
autrijus k :)
Juerd autrijus: In a few hours you'll be able to test this yourself
autrijus yup
Limbic_Region the hyper derferencing tests still need a . added, will try to do that tonight (assuming I actually get the cable installed this time) 14:19
autrijus bbiab. 14:27
q[merlyn] so 6.2.5 still not working on OSX is known? 14:28
or does someone want a bug report?
theorbtwo Uff -- if I do PUGS_EMBED="perl5" perl Makefile.PL; make (errors), then perl Makefile.PL; make -- it still errors (differently). 14:29
I don't have time to track it down now, though.
q[merlyn]: if this is tiger, search in the README for "tiger" -- there is a workaround.
Juerd I'm off
theorbtwo If it's not that problem, then AFAIK it's not a known problem. 14:30
Juerd Bye!
(feather will now travel to Schuberg Philis (www.schubergphilis.nl/)
)
q[merlyn] ahh. gcc 3.x not 4.
darn it. :)
q[merlyn] types make all test install 14:31
Limbic_Region q[merlyn] - out of curiosity, are you writing any p6 code or just verifying that it is building and trying out examples 14:32
PerlJam is continually shocked at just how much perl6 pugs understands. 14:33
Limbic_Region yeah - you and me both 14:34
I find some piece of p5 code I have written and think - no way this will work
and other than working around bugs - it does
I am however waiting on a few features for my latest project 14:35
specifically hyper dereferencing
theorbtwo Hm, Makefile.PL should possibly notice what version of GCC it is using. 14:38
clkao autrijus: and you need to give me some example how to hold and use an pugs val in C
theorbtwo (In the same way that it now notices what ghc it is using.) 14:39
Or possibly just print a warning whenever $^O says we've got an OSX machine.
(What is $^O on OSX?)
q[merlyn] I'm mostly tracking it for now 14:40
I want to be more involved, but timing is bad
Limbic_Region was just an inquiring mind 14:41
nothingmuch theorbtwo: 'darwin' 14:47
q[merlyn] OK, selecting gcc 3.3 didn't help 14:49
still dies in samespot
right after "Compiling Main" during "Linking..."
theorbtwo Thanks. 14:50
q[merlyn] ahh. it's picking up /opt/local/lib - how do I tell it to not do that? 14:51
Limbic_Region q[merlyn] - perhaps nopaste the error
q[merlyn] is there a nopaste here? 14:52
integral sial.org/pbot/
Limbic_Region perlbot nopaste
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel> 14:53
Limbic_Region there is a perlbot q[merlyn] - it knows lots of stuff
pasteling "q[merlyn]" at 200.179.244.157 pasted "Linking ... /usr/bin/ld: warni" (11 lines, 1.2K) at sial.org/pbot/10407
q[merlyn] nothing in my env contains /opt/local/lib
so I don't know where that's set
Oh. it's ghc 14:54
ghc is built in /opt/local
integral you don't use the pre-build binaries? They're /usr/local
q[merlyn] so I have a libdl in /opt/local/lib and in /usr/lib 14:55
I try not to install prebuilt binaries
not when source is available
even then, it'd probably conflict
Limbic_Region perlmonks.org/index.pl?node_id=460344 # plea for more better FMs 14:56
integral hmm, ghc -v5 prints a lot of config info
q[merlyn] Hmm. my "port" command appears to be broken too 14:58
integral hmm, you have a libdl in your /opt/local/lib ?
q[merlyn] this is not good
Yeah, I built ghc from darwinports because fink didn't have it.
q[merlyn] reinstalls darwinports just in case 14:59
stupid "make isntall" doesn't do a "make" 15:00
q[merlyn] retries first with "make" again 15:01
(distros where make install breaks because "make" hasn't been done)--
mandatory makefile line: "install: all" 15:02
theorbtwo Um, you did type "install" correctly, right?
q[merlyn] Yes
and it bombed because something wasn't made that "make all" makes
Limbic_Region YES - an advocacy victory
perlmonks.org/index.pl?node_id=460345
"Well that's amazing. I guess I have to check out Pugs now." 15:03
gaal|work what platform is diotalevi on? 15:06
q[merlyn] Hmm. ports suggests that ghc 6.4 is stable, but 6.5 is devel 15:07
is that true?
Limbic_Region gaal - I have asked several times and I don't remember 15:08
would you like me to ask again?
q[merlyn] ahh. ghc-devel doesn't depend on dlopen, so that will probably use the /usr/lib version 15:09
Limbic_Region gaal|work - see my reply at perlmonks.org/index.pl?node_id=460347 15:10
theorbtwo That's around right, merlyn. Uses the same scheme as the linux kernel, more or less. 15:11
Limbic_Region wanders off 15:12
q[merlyn] Oh, ghc-devel is trying to fetch something from cse.ogi.edu, and failing
I guess I'm back to 6.4 15:13
pdcawley "Luke and Damian provided the sanity" <-- What's wrong with this sentence? 15:14
q[merlyn] Hmm. out of time to do more downloads etc. 15:17
q[merlyn] wanders off
PerlJam pdcawley: Luke and Damian are on the wrong side of the sanity 15:18
Limbic_Region gaal|work - I'm using NetBSD 2.0.2 (the latest) on an SMP Sparc. "sparc-netbsd-thread-multi" 15:24
that was diotalevi's reply
gaal|work sorry, was away
interesting platform
Limbic_Region no worries - I actually got it in a post at the Monastery so I won't have to keep bugging him about it 15:25
iblech ping 16:15
Limbic_Region likes "use:p5 Digest;" as a way of designated p5 module use 16:22
obra if you're going to go that route, why not create a proxy 'perl5' module. 16:23
use perl5 DBI;
Limbic_Region isn't going any route 16:24
clkao *yawn* 16:26
Juerd LADIES AND GENTLEMEN 17:15
feather.perl6.nl is online! 17:16
gaal yay!
Juerd If you have requested an account, you may /msg me for your temporary password.
(They are preset, and I expect you to change it immediately)
PerlJam you didn't like carrot?
Juerd I like it, but I like feather better
Boogie Hey, it was my suggestion. :) As _I_ remember. ;) 17:19
knewt Juerd: not requested an account yet, can i do that now? 17:20
gaal logged in an password changed.
crazy npasswd :) 17:21
Juerd knewt: By sending e-mail, yes
gaal Thanks, Juerd!
PerlJam Juerd: What are the specs on that machine again? 17:23
Juerd p4 2.4 1024 80r1
sata disks
wolverian are we going to each build our own pugs? 17:24
PerlJam wolverian: I'm sure you could setup an automated build to share with everyone if you like.
wolverian oh, right. :) 17:25
gaal i'll just be posting smoke results there (nicely linked)
Limbic_Region Juerd plus plus btw for pushing to get the community server
PerlJam Juerd++ for L~R :)
Juerd pushing to get? That's one way to put it :)
wolverian hmm. no svn, only svn-mirror? :) 17:26
gaal Is L~R's keyboard broken?
Limbic_Region no gaal
Juerd wolverian: Well, I hope people will announce where their pugsies and parrots are
Limbic_Region my client is
Juerd wolverian: So others can use them
PerlJam gaal: no, but his client eats +
gaal lol
Juerd wolverian: There will be a wiki eventually
gaal you need a ++ bot 17:27
Limbic_Region the only way I can be here from work is to use cgiirc
wolverian Juerd: sure, but building and testing pugs is fairly resource intensive, isn't it? it'd be better to have only one pugs. then again, I don't care that much, and it's your box :)
gaal hmm, doesn't that read "pastebot" in haskell?
PerlJam Juerd: why eventually?
Juerd wolverian: You are right.
PerlJam: Because I wanted the box online asap, and haven't arranged any documentation apart from the motd yet. 17:28
PerlJam wolverian: Given that people will be using this server to do various things, one of those things might be to build a "special" version of pugs (i.e. testing a particular feature)
knewt ah, there's a pastebot on this channel. that's good to know
wolverian PerlJam: yes, of course. that shouldn't be disallowed, probably.
Juerd wolverian: It may be best to simply symlink /usr/local/bin/pugs to a single pugs
wolverian: Will you be building that one?
wolverian: Or writing a script to build it periodically?
etc.
wolverian Juerd: I don't know if I have enough tuits for that. never done anything like this before. 17:29
Juerd Ah, well, this is simply enough. I'll do it myself
PerlJam wolverian: good a time as any to learn! (screw work and family, you need to play with pugs! ;-)
wolverian PerlJam: true enough. :)
Juerd To skip the learning thing: 17:30
It's a matter of using cron (crontab -e) to set a timer that svn ups and builds.
wolverian in any case, as Juerd said, it's trivial. simply add a cron job to svn up, make all test install. if you want to get fancy, you'd redirect errors to /etc/motd or so, to let everyone know what is working and what is not
Juerd nah, on error it simply doesn't make install.
Limbic_Region cklao ping
PerlJam Speaking of cklao .... Juerd, are you going to install svk for everyone? 17:31
(or delegate appropriately)
Limbic_Region IIRC, cklao has written Benchmark.pm but hasn't released it yet 17:32
Juerd PerlJam: I've installed whatever debian thinks svk is, system wide.
Limbic_Region really would like to have it to benchmark various solutions to his recent project
PerlJam Juerd++
I may ask for an account yet.
Juerd You may 17:33
By e-mail.
PerlJam (currently I just build on my box at work (lucky me no restrictions there))
though it is a tad annoying when the entire campus drops off the net for 15 minutes at a time (like it did to me last night in the middle of composing an email)
wolverian sid's svk is up to date. 17:34
clkao people 17:39
please learn to spell so my irssi client will hilite
Limbic_Region: what can i do for you?
i ugess you will want a shared mirror on the box
and just tell people to depot map to that one
Limbic_Region clkao - I don't understand what you mean by "learn to spell so my irssi client will hilite" 17:40
clkao let me know if i can be of any help.
clkao &
you typed cklao
wolverian Limbic_Region: typing his nick correctly so that he will see the line in his awaylog
Limbic_Region oh
clkao anyway &
Limbic_Region wants Benchmark.pm clkao is all 17:41
clkao ok ok i will try later tonight
Limbic_Region perlmonks.org/index.pl?node_id=458728 # It is S-L-O-W
but since there are many ways to skin a cat, I want to see which ones are fastest
great - thanks clkao 17:42
gaal feather users, please create a ~/README 17:46
wtf? the "struct dirent" still happens with ghc snapshot :( 17:54
uh, the __hscore_d_name problem that is.
revdiablo Limbic_Region: thanks for lurking on that Meditation I posted, and replying to folks. I barely had the energy to read most of the replies, much less reply back to them. :) 17:56
q[merlyn] weird. installing ghc from ports is building Perl 5.8 from scratch. 17:57
japhy the perl6 summary's links don't work :( 17:58
q[merlyn] so now I'll have two copies of Perl 5.8 on my disk. :(
gaal dislikes link shorteners
q[merlyn] purl?
Limbic_Region revdiablo - I didn't do an exhaustive set of replies
just the ones I felt warranted a proper nudge
shapr Purle: Permanent Uniform Resource Locator, a service of OCLC in Virginia
gaal merlyn, i have five of 'em :(
revdiablo Limbic_Region: yeah, but you replied some. I think it's helpful just to keep some activity up 17:59
q[merlyn] wow
shapr um, Purl
Limbic_Region wanders off to a meeting
q[merlyn] hard to imagine why it would not recognize I have one already
gaal actually, three p5.6.1 + 2 5.8.x
q[merlyn] wow 18:00
"Perl - so good you'll want to install it 2, er, 3, er, 4, I mean 5 times!~"
gaal actually, i just compiled one the other day, and i *thought* it was the 3rd instance -- but i just discovered ghc comes with its own perl 18:01
mauke 5.6.1, 5.8.0, 5.8.1, 5.8.3, 5.8.4, 5.8.5, 5.8.6
I win
gaal and its own gcc :)
q[merlyn] heh
gaal mauke wins :)
revdiablo mauke always wins
mauke oh, and perl1
gaal but i have access to a machine with....
perl4 :
oy
ok ok ok
you WIN.
q[merlyn] ugh! 18:02
p4!
I remember that
and p3... and p2... and p1.
gaal i bet you do.
you remember p3?
q[merlyn] Hell. "I wrote the book" on p4. :)
mauke I have access to a machine with p4, p5.002, p5.005 and p5.005_02
gaal was there a p3?
mauke I think it's interesting that p5.8.6 passes p1's test suite 18:03
gaal lol
q[merlyn] of course there was a p3
gaal with warnings, i bet, about @ ?
mauke no warnings 18:04
the tests don't "use warnings" :-)
gaal sounds like a short test suite
ah :) 18:05
Juerd feather is offline, temporarily, because its connectivity is somehow borked
the fiber ring doesn't like this box.
mauke perl1 doesn't even have -w
18:05 orevdiabl is now known as revdiablo
gaal eeek! eerk! 18:06
Found = in conditional, should be == at /usr/share/perl/5.8.4/File/Path.pm line 266.
Bareword "getcwd" not allowed while "strict subs" in use at /usr/share/perl/5.8.4/File/Path.pm line 266.
lose, lose 18:07
ninereasons hi shapr 18:08
shapr 9r 9r
ninereasons I added a quote to your perlbash... er. Haskell quote page yesterday :) 18:09
shapr Oh really?
Oh, that's a good quote.
It's not really my page, I just added 80% of the quotes.
ninereasons :-)
shapr All bad guys listen to Reason
That is an excellent quote. 18:10
ninereasons yes, I thought so.
shapr I don't often get in the way of depleted uranium slivers moving at a noticeable fraction of c.
I'm actually recruiting TMR authors for TMR3 right now, was wondering if any pugs people would like to write an article about their experiencese. 18:12
clkao autrijus: *yawn* any news for me 18:13
autrijus: so, foreign export call invoke.. blablah ? 18:18
q[merlyn] continues to build ghc again, this time without dlopen 18:25
howver this time with its own private copy of Perl5.8.6
shapr It would be fun to get Pugs booting on the bare metal via House. 18:26
q[merlyn] & spending another day in rainy Rio De Janiero 18:32
wolverian gaal: why a ~/README? as in, what do I put in it? 18:52
what I'm doing? :)
gaal yeah, in lieu of a .plan, i guess 18:58
Limbic_Region is trying not to get his hopes up but can't contain himself - the prospect of actually having net access from home is too much 19:27
clkao summon autrijus 19:29
obra hi clkao!
Limbic_Region it's 3:30AM - he likely is having a cat nap
so you just need to speak a little louder 19:30
SUMMON autrijus
ninereasons iblech, your sub step() in mklivecd.pl is a really nice idea 19:37
wilx Somebody should add the http:// prefix to URIs in topic so that I can just click :) 19:43
Limbic_Region 6.2.5 released! | Your opinions wanted: perlmonks.org/?node_id=459750 | pugscode.org <Overview Journal Logs> | pugs.kwiki.org 20:00
Limbic_Region wilx - done 20:00
any reason you didn't do it yourself?
japhy character classes are a source of grief... 20:08
svnbot6 r3881, Aankhen++ | * basic port of Perl 5 CGI::Util.
ninereasons I also didn't realize that mortals could do that, Limbic_Region 20:09
Limbic_Region ninereasons - TIAS
perlbot TIAS 20:10
perlbot Try It And See: the best way to learn if something works.
ninereasons Limbic_Region, good advice :)
Limbic_Region well - one should always apply a good dose of sanity first 20:11
Aankhen`` Does %hash.keys() return a sorted list? 20:18
Limbic_Region TIAS
unless of course what you meant to say s/Does/Should/ 20:19
Aankhen`` Same difference. :-P
stevan Aankhen``++ # for taking over the CGI module and adding on to it 20:20
Aankhen`` Thanks. :-) I didn't really take it over though -- I'm just adding whatever I can, and hoping I don't collide with anyone else's changes. 20:21
stevan Aankhen``: oh no,.. its your now :P
Aankhen`` 's shoulders sag under the responsibility. 20:22
stevan feels a little like Tom sawyer :P
clkao i feel at home reading haskell tutorial
obra dooom
osfameron clkao: the haskell for c programmers?
obra waits for clkao to start hacking darcs 20:23
clkao no, yaht
i'm lisp native, btw
PerlJam buries clkao under a pile of parentheses 20:24
gugod_ clkao: hello)))))
clkao emacs helps
ya, so, it's like lisp without nails plus evil inference 20:25
theorbtwo nails? 20:27
mauke (clipped (finger (nails)))
obra "it came before sails"
Aankhen`` Strange. 20:28
I have: $string = q/bar"bar/; $string ~~ s:P5:g/"/&quot;/; 20:29
And I get: *** Can't modify constant item: VUndef
clkao i'm only 25% through
but dinner time
wilx Hmm.
Lim...bah, he left. 20:30
Aankhen`` Ah, nevermind. 20:31
I fixed it.
"is rw" is a dangerous tool... ;-)
japhy heh. 20:32
perl 6 rules are supposed to be one-pass-compiled, right?
as opposed to Perl 5 regexes which require two passes
theorbtwo Yes.
japhy das ist gud. 20:33
theorbtwo (Though I don't think p5 REs strictly require two-pass parsing, they are presently parsed that way.)
theorbtwo wonders what to do.
japhy they do, because backrefs and octal escapes are identical
and because, the way it's set up, the compiler needs to know the regex's size ahead of time 20:34
theorbtwo Ah. The first sounds like a good reason.
(Octal escapes require a o in perl6.) The second isn't a real requirement, it's just the way it's currently done. 20:35
japhy yeah, I know about the second
octal escapes are \o... now?
didn't see that in ASE05 :(
theorbtwo I think they are...
PerlJam japhy: I believe the goals are one-pass with minimal lookahead and no backtracking if we can help it. or something like that.
japhy it's \0 in the docs I've seen, orb 20:36
Aankhen`` Octal escapes are \0123...
Yeh.
And you can use \0[123] for disambiguation.
japhy perljam - there's only very little look-ahead needed.
PerlJam perl6 won't have \1 \2, etc back refs
Aankhen`` ponders whether to shift url_encode and url_decode to CGI::Util.
japhy the cases I can think of off the top of my head are '@array' vs '@array :=' and '<[abc]>' vs. '<[a-c]>' 20:37
Aankhen`` '^' vs. '^^'? 20:38
japhy I need to know ahead of time if a variable is being "used" or "bound"
aanken - no, that's a separate issue, the way I parse
Aankhen`` OK.
japhy and I need to know if a character in a character class is standalone or part of a range
oh, and the range is .. not -
oops
svnbot6 r3882, Aankhen++ | * added tests for CGI::Util.
r3882, Aankhen++ | * fixed svn:mime-type property for ext/CGI/lib/CGI/Util.pm
r3882, Aankhen++ | * fixed &CGI::Util::simple_escape
Aankhen`` Darn... porting &expires will require developing a familiarity with the time functions. 20:40
theorbtwo I could have sworn we decided that numbers were 0o123, and the string escape was \o123... 20:42
theorbtwo sighs.
Making a leading 0 octal is horrible huffmanisation -- it takes legal syntax for numbers, and steals it away for something very rarely used. 20:43
Aankhen`` Anyone feel like implementing gmtime()?
PerlJam theorbtwo: I believe that $Larry ruled on that on p6l and that the answer was different than what's currently in the AESs
theorbtwo sighs -- somebody should start indexing these things. 20:44
PerlJam Aankhen``: Actually, I felt like it earlier today, but then the feeling passed :)
theorbtwo Otherwise nobody can tell what's going on without reading every message that goes through p6l.
Aankhen`` mumbles about doggone fickle programmers.
theorbtwo mumbles about doggone fickle language designers. 20:46
Hm, "theorbtwo".chars == "Aankhen``".chars
Aankhen`` Indeed.
PerlJam theorbtwo: groups-beta.google.com/group/perl.p...a408b399d0 20:47
theorbtwo Ah, good. 20:50
theorbtwo wonders if L~R's internet is working at home. 20:58
wolverian wonders where Larry has disappeared 21:03
PerlJam He's probably working to pay the rent and put food on the table etc. 21:04
wolverian hm. yeah. sorry, I sounded stupid.
japhy what's with $Larry and @Larry?
wolverian (I'm in a situation where I don't have to worry about such things, if that's any excuse.)
japhy wolv - you're homeless?
wolverian no, just unemployed. :) 21:05
japhy heheh
PerlJam japhy: $Larry is *the* Larry, @Larry is the entire cabal (Luke, Chip/Dan, Pm, chromatic, Damian, etc.) 21:06
integral Not that there *is* a cabal though... 21:07
japhy ah, gotchya.
push @Larry, 'japhy';
japhy disappears
I really should've gotten involved two years ago. *sigh* 21:08
PerlJam wonders if chip participates in the weekly conference calls like Dan did (i.e. not at all :) 21:09
I wonder if chip even knows about the weekly calls.
theorbtwo I think Dan considered his job more or less dissepperate from that of the rest of the cabal. 21:10
That is, in a way, a real shame, but it's also quite interesting.
One of the reasons that parrot has been so slow in being really visibly useful is because it has a very ill-defined mission.
Pugs has always been written in a way that everything is supposed to be there to get perl6 working as quickly as reasonably possible. 21:11
wolverian pugs has a lot easier job, yes. 21:12
theorbtwo Parrot has always been written to be generally useful... (((and run perl6))) as almost an asside.
integral a good virtual machine is somewhat different though, it does need more design than Pugs
PerlJam theorbtwo: Are you on #parrot?
PerlJam looks
theorbtwo Not at the moment.
wolverian theorbtwo: asside :) (aside)
PerlJam theorbtwo: but you have been in the past?
theorbtwo Yes.
Most of the time, it seemed very very quiet.
PerlJam theorbtwo: freenode #parrot or magnet #parrot? 21:13
wolverian hm, is magnet #parrot open to all? :) 21:14
theorbtwo irc.perl.org parrot.
There's a freenode #parrot?
wolverian (that is magnet.)
PerlJam theorbtwo: were you on earlier this week when chip was talking about how he views his mission? 21:15
Aankhen`` G'night. 21:16
theorbtwo Nope, missed that.
21:19 Arathorn is now known as Aragone
theorbtwo Care to give me a hint? 21:19
PerlJam theorbtwo: unfortunately I don't have it logged (I don't think I do anyway ... hold on ...) 21:20
integral wonders if pugs/parrot would be faster than p5 for processing his logs 21:21
PerlJam oh good, chip does know about the weekly cabal calls.
theorbtwo: anyway, I wish I could paraphrase better but words are failing me at the moment. 21:22
AtnNn since diamonds dont do globbing anymore, what does? 21:23
wolverian glob(), I think.
PerlJam theorbtwo: suffice to say that chip views his job not as "finishing parrot" as much as "continuing with the general plan even if it means bad things"
where "bad things" could be scrapping loads of already done work (chip would not do this lightly of course) 21:24
the impression I got was that he feels slightly captive by earlier design decisions. 21:25
integral is leo the only person currently coding on parrot?
theorbtwo Ugh.
PerlJam integral: define "coding on parrot" a little more :-)
integral err, well writing C code, that gets checked into svn, and gets parrot closer to being "finished" 21:26
PerlJam If you mean "writing C code to implement parrot", then he's certainly still the number one committer.
I don't think he's the only one though
integral it seems a shame that such an exciting idea isn't really being hacked on 21:27
Juerd connectivity problems disappeard magically 21:28
feather's up again
have fun!
integral Juerd: have you emailed passwords to those who've asked for accounts?
Juerd no
irc people can /msg me
other people will have to wait longer
integral decides to wait since he's got exams 8-) 21:29
clkao sadly im learning pugs guts very quickly 21:40
PerlJam in the last 100 revisions to parrot there have been 10 committers. including pmichuad and autrijus 21:42
clkao oh no! autrijus touches C code
that brings segfault!
eric256 i think the pugs website needs stats show like that....x commits in last day, x commites in last day, average commites / committer ;) 21:43
PerlJam 40 revisions by leo, 20 by bernhard, 14 by autrijus, 9 by pmichaud, 9 by particle, the rest have < 4
theorbtwo That sounds like a better spread then pugs. 21:45
svnbot6 r3883, autrijus++ | * change -DPUGS_EMBED_PERL5 to -DPUGS_HAVE_PERL5 to agree
r3883, autrijus++ | with HSPLUGINS and PARROT; also remove the obsolete and
r3883, autrijus++ | troublesome code in config_h.pl.
clkao eek!
integral particle? 21:46
theorbtwo Of the last 100 revisions:
38 autrijus, 17 iblech, 9 Aankhen, 7 clkao, rest <= 5. 21:47
svn log -r3783:3883 |grep '^r[0-9]'|cut -d'|' -f2|sort|uniq -c|sort -r
clkao what, i'm on the list? you suck! more commits!
PerlJam heh ... total revisions to the parrot tree: leo:2632, dan:967, scog:476, mrjoltcola:450, sfink:332,simon:289,coke:259, etc.
theorbtwo: gosh your command line looks remarkably similar to mine :-) 21:48
clkao theorbtwo: btw, you can use svk log -l 100 21:49
theorbtwo I don't use svk, clkao.
clkao i'm so old to look up HEAD and calculate minus 100 in mind.
PerlJam clkao: Just use head!
clkao what about HEAD-100 ? 21:50
PerlJam svn log | grep '^r[0-9]' | head -100l | ...
clkao ya, network is also cheap these days.
autrijus greetings! 21:51
Juerd feather has global pugs now 21:52
rebuilt every 15 minutes
autrijus clkao: so you've read PA02?
clkao no not yet. i've skimming some code... i'm very sleepy today
autrijus that's fine. what segfaulting C code did I introduce?
clkao autrijus: so, export foriegn ccall should work?
was j/k :) 21:53
autrijus clkao: "foreign export ccall"
ahh.
integral Juerd: how fast can it rebuild pugs (un)optimizede?
autrijus but yeah it should work
clkao so
export eval's apply function
autrijus how do you construct Exp from C space? 21:54
writing simple strings?
Juerd integral: 1:12 minutes, if nobody else does anything
autrijus and I can read in back using haskell Read?
that I can do
clkao and need some code to hide vcode and other Type in sv
Juerd integral: without parrot and perl5
integral: unoptimized
autrijus Juerd++ # feather!
is feather dyndns or staticip? 21:55
integral that's very nice :-) 21:56
svnbot6 r3884, autrijus++ | * handle perl5-land AUTOLOAD.
autrijus ok, try www.perlcabal.org/ ? :)
perlcabal.org and no.perlcabal.org itself are CNAMEed to to feather as well. 21:57
so we get perlcabal.org/~autrijus/ URLs :D
clkao autrijus: so, little exercise. say we have a function in later.c. run_later(), takes vcode and args, returning an opaque object. and run_me() taking what run_later() returns and invoke that vcode. how would you do that? 21:58
Juerd autrijus: see motd :)
autrijus: Although that's no longer true, still use the hostname only.
autrijus Juerd: what is my password again? :)
Juerd autrijus: again? 21:59
autrijus Juerd: if it's not set yet, please set it to autrijus.org/tmp/id_dsa.pub 22:00
Juerd You can do that yourself
autrijus but I can't remove my password :)
I mean, like, completely
but that's fine
thanks!
autrijus reads motd 22:01
sure, I'm now using CNAMEs. strange how registerfly lets you use CNAME even for the domain itself
clkao autrijus: tell me about the above before i fall asleep :P 22:02
Juerd autrijus: set it to something complex that you won't remember.
autrijus clkao: run_later will obviously just contain a struct to the vcode and args
clkao: and run_me will feed them to apply
Juerd: sure 22:03
clkao ya, i mean how are they represented in c land?
if you can have that implemented i can make the final connect :)
autrijus clkao: see 22:04
typedef void *HsStablePtr;
that's how all haskell stableptrs look at C land
so I'll just use that.
with
newStablePtr :: a -> IO (StablePtr a)
I can make any haskell type into a stableptr
for you to put into C structs 22:05
when they are called back, I can deref with
deRefStablePtr :: StablePtr a -> IO a
that's all. easy as pie
theorbtwo Hmm...
perlbot nopaste?
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
autrijus clkao: also I'm already beginning to think about doing this without embed libperl 22:06
clkao autrijus: how?
pasteling "theorbtwo" at 84.245.186.159 pasted "Errors when compiling with perl5 support -- debian" (44 lines, 3.8K) at sial.org/pbot/10413
autrijus clkao: I think almost all of the api can be emulated using socketpair 22:07
Juerd Okay - had enough mcfood, enough hot chocolate
clkao yuck!
autrijus clkao: if you can load a SV by id
almost everything else is trivial
Juerd And I've been in these datacentra (both) more than enough for one day
autrijus and I think that is solvable with Devel::Pointer or Pointer.pm or some such
Juerd feather is up, connectivity is alright (should you manage to gain root access through some exploit, PLEASE DO NOT TOUCH THE AUTONEGOTIATION)
autrijus theorbtwo: yeah, gaal reported same thing 22:08
theorbtwo I saw the bottom half before, but not the top half.
autrijus hm, Inappropriate ioctl for device at util/config_h.pl line 106
clkao autrijus: you don't want to do that
autrijus clkao: why not? 22:09
having to embed perl is a pain, as theorbtwo observed ;)
clkao autrijus: trust me. this will be more painful
autrijus I mean, we already bridge external PGE that way
ok... anyway we'll get the embed mode figured out first
clkao oh, you know you want to use Acme::Ref
autrijus er not really. 22:10
clkao nevermind XD
autrijus no, ingy's solution is best
it is even pure perl
Pointer.pm
clkao so, if you feel bored, implement the above later stuff (with proper gc/refcnt). i can get the rest tomorrow
autrijus I can handle haskell gc fine 22:11
ingy oh no
Pointer.pm!
autrijus I don't really understand refcnt but I can fake
hi ingy!
ingy it was a *joke* ;)
clkao trusts no ingry
autrijus ingy: it is *invaluable*
ingy /kick clkao
autrijus ingy: it lets me write RPC perlembed code!
obra ponders ingrish 22:12
ingy you are mad
mad i tell you
clkao autrijus: make the run_later stub in a sample .c (maybe just in perl5).. so i can carry on
autrijus I thought that's not news
clkao: 'kay
ingy :)
clkao *sigh*
ingy hi clkao
autrijus clkao: I think pugs guts will be quite transparent to your eyes... real PA02 sometime :)
ingy clkao: still hate me?
clkao mainly because of kwiki 22:13
autrijus s/real/read/
clkao autrijus: it's quite. i decided to make blutwurst from the pugs
autrijus yeah. kwiki is so borderline. if it's worse, we won't feel compelled to suffer; if it's better, there won't be so much headache :)
clkao: cool
obra ingy's new despammy stuff makes life better 22:14
I despammed a week's worth of spam in 5 minutes
clkao anyway, get this done so i can implmement bin/svk in perl6 and at least one editor in perl6.
autrijus yay!
clkao++
I'll get this done :)
clkao autrijus++ # though you need no karma anymore :)
theorbtwo autrijus, I think it's simply missing a header. Ignore the 64-bit stuff it's whining about -- /usr/lib/perl/5.8/CORE/reentr.h:611: error: field `_crypt_struct' has incomplete type 22:15
autrijus obra: hm, url to said despamming stuff?
theorbtwo: that may be. I wonder if GHC is shadowing the .h somehow
ingy clkao: what is wrong with kwiki? 22:16
clkao it's ahteful
obra autrijus: um. no url. Ingy hasn't released it yet, i think
clkao internal server error when it's totally spammed
autrijus or if it's because if the munging in line 174 inc/Module/Install/Pugs.pm
clkao ::revisions plugin doesn't work with svk
ingy clkao: you are a hatful of hollow
obra basically, there's a mod to recent pages that takes you straight to page_info on the admin version of the kwiki 22:17
clkao maybe someone needs to write a wiki in perl6.
obra then there's a mod to PageInfo to reverse the listing.
autrijus nice.
ingy clkao: what obra say ^^
clkao autrijus: so you will setup shared mirror depot for feather?
obra ingy: that still doesn't stop the server error on spam
autrijus clkao: thru local exploit? ;)
obra it makes it easier to recover, though
clkao autrijus: what local exploit?
autrijus (meaning, there's no sudo on feather) 22:18
how am I to setup shared depot thing?
clkao oh
it's easy
i assume you are in same group?
fsfs handles that already
autrijus so everyone link my .svk
is that the idea?
clkao get a mirror in /pugs/, which points to ~autrijus/pugs-=depot
and people map their default depot or /pugs/ to that repositroy 22:19
don't have to do SVKROOT=/home/autrijus/.svk, it's a bit annoying
autrijus sounds yummy
ok. then we put it to motd
that will save lots of svn bandwidth 22:20
svk++ # technology!
clkao *nod*
and fscking disk space
did i tell you 22:21
autrijus no you did not
clkao at $work, the repository is 1.2G. svn co is 2.4G. svk co is 700M.
so a mirror of the full history (36k revs) + checkout is fucking smaller than svn co
autrijus are you using xdelta repo yet?
clkao no
autrijus I'm pondering a strategy of migrating foundry to that 22:22
clkao why
autrijus I especially want server side support for --limit
clkao what for?
you dont want to hear me rant about it
theorbtwo It will iff that svk repo is kept very up-to-date. Otherwise it'll just be annoying.
clkao IT DOESN"T WORK
oh, it's fixed for 1.2.1
autrijus er.
clkao theorbtwo: ? every commit will update it.
autrijus that renders the rant less effective
clkao theorbtwo: any time you do svk up -s
autrijus theorbtwo: everyone can update it with svk pull
theorbtwo Ah, of course. 22:23
Nevermind then.
autrijus :)
clkao we use shared svk depot at work, proven to be very helpful and efficient.
never had out-of-date problem.
autrijus wunderbar
clkao so it's proven technology
s/proven/enterprise/ 22:24
bwaha
knewt clkao: btw, i'm seeing an "unexpected success" from the svk test suite
autrijus clkao: so did you submit to euroscon?
clkao knewt: oh? which one?
autrijus: yes, with help from obra :)
autrijus do I get to see your submission?
ooh. 22:25
clkao autrijus: but i think i'll go anyway.
autrijus: nopaste.snit.ch:8002/2713
autrijus greetings mrborisguy
knewt clkao: 70symlink.t, [[ ok 50 - got svn:special # TODO revert overwritten symlink from file ]]
clkao knewt: interesting. 22:26
knewt clkao: however, the following TODO subtest fails [[ not ok 51 - t/checkout/symlink/non.lnk.cp is symlink # TODO revert overwritten symlink from file ]]
clkao ah, i must have fixed soemthing accidentally
knewt: would you like to be able to commit to svk?
autrijus (accidental fix)++ # my favourite kind of fix 22:27
autrijus looks at #svk backlog and grins
clkao autrijus: do you handle p5 returned arrahref/hashref yet?
autrijus 03:00 < clkao> bad news. i like haskell
clkao: planning to now
clkao autrijus++ 22:28
autrijus clkao: and I'll do auto deRV!
clkao COOL
knewt clkao: heh, yeah, sure. run stuff by you first before actually committing anything?
autrijus eval_perl5('@INC').[0]
eval_perl5('\@INC').[0]
will be the same thing.
knewt autrijus: so can you embed both perl5 /and/ parrot at the same time?
autrijus no more hateful cryptoreferencing
knewt: yes, using PUGS_EMBED=perl5,parrot
or parrot,perl5 if you prefer
q[acme] giggles
autrijus or outer punctuation chars 22:29
knewt autrijus: ah, ok. i tried doing PUGS_EMBED="perl5 parrot"
clkao q[acme]!
autrijus knewt: that should also do
knewt hmm. didn't seem to when i tried it. of course, that was several days back now *g*
q[acme] clkao: found your new project? ;-)
clkao q[acme]: no, sadly i'm just going to stick with svk, but slowly port things to p6 piece by piece 22:30
mrborisguy as added props for p6... it takes a lot less time to compile than 5.8.6
clkao q[acme]: how are you party boy
autrijus knewt: worksforme 22:31
knewt: I can eval PIR defined in SVs 22:32
it's not quite ponie, but it's still fun :)
theorbtwo So long as there is a \b between them, it's all good. 22:33
q[acme] parties 22:37
drbean The svn,svk tests for version in the Makefile are hanging the build on my machine 23:48
svk outputs a line about the base directory being not a checkout directory 23:49
this is printed when I run the pugs_version.pl script by itself 23:50
the line is not printed when run by make and the build hangs 23:51
Limbic_Region does a little dance as he is on the net from home again 23:58
theorbtwo Congrats, Limbic~Region!
Welcome to the wonderful world of connectivity!
Limbic_Region well, not quite
only got digital cable - not HD
and haven't gotten my VoIP phone working yet
but getting there

Powered by Cro and the Raku® Programming Language.

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

Thank you!