pugscode.org/ | nopaste: sial.org/pbot/perl6 | ?eval [~] <m oo se> | We do Haskell, too | > reverse . show $ foldl1 (*) [1..4] | irclog: irc.pugscode.org/
Set by diakopter on 11 July 2007.
00:14 thoughtpolice left 00:17 theorbtwo left 00:21 Lunchy joined 00:22 REPLeffect_ left 00:23 nekokak_ left, wilx left 00:25 nekokak joined, theorbtwo joined 00:29 mfowle joined
mfowle what parrot based perl6 implementations are there? 00:29
00:39 baest left 00:40 mfowle is now known as myren 00:59 vsmatck joined 01:02 nipotaway is now known as nipotan 01:04 lyokato joined 01:10 Lunchy left 01:15 Lunchy joined 01:16 Lunchy left 01:20 Lunchy joined 01:34 daxim joined 01:39 myren_ joined 01:48 daxim_ left 01:53 myren left, weinig left 01:55 [particle1 joined 02:11 [particle] left 02:18 myren_ left 02:25 sunnavy_ joined 02:49 amnesiac joined 02:52 kanru joined 02:53 Jamtech joined 03:03 weinig joined 03:06 weinig left 03:07 weinig joined 04:11 zbrown joined 04:12 zbrown left 04:23 sunnavy left 04:24 sunnavy_ is now known as sunnavy 04:33 melissa left 04:49 wilx joined 05:09 dduncan left 05:19 lyokato left 05:20 charsbar_ left 05:22 dduncan joined 05:25 lyokato joined, jisom joined 05:27 charsbr joined 05:30 REPLeffect_ joined, REPLeffect_ is now known as REPLeffect 05:38 Aankhen`` joined 05:49 weinig left 05:50 weinig joined 05:52 Ashizawa joined 06:02 drupek12 left 06:06 Jamtech left 06:12 john___ joined 06:17 john___ left 06:20 iblechbot joined 06:21 baest joined 06:22 johntaro joined 06:43 cognominal_ left 06:59 REPLeffect left, REPLeffect_ joined 07:06 jisom_ joined 07:08 elmex joined 07:21 penk joined 07:23 jisom left 07:26 qmole left 07:35 penk left 07:36 hakobe is now known as hkb_lab 07:42 penk joined 07:44 penk left 07:47 franck___ joined 07:51 drupek12 joined 07:58 REPLeffect_ left 08:00 polettix left 08:04 weinig left 08:09 Coke left, Coke joined, Coke left 08:14 Coke_ joined 08:15 weinig joined 08:18 qmole joined 08:22 hiarc left, masak joined 08:29 dduncan left 08:35 iblechbot left 08:36 hiarc joined 08:43 agentzh joined
agentzh g'localtime, folks :) 08:43
masak g'localtome, agentzh :) 08:44
agentzh heh
seen the CPAN release of kindaperl6 :)
masak yes
feels very official in a way 08:45
agentzh heh
masak ...except for the big red "UNAUTHORIZED" texts here and there
agentzh i'd try to find some time to tweak PCR in this month :)
masak tweak in what way? 08:46
agentzh really want to implement that regex tracer as a Firefox extension or an AJAX page.
masak sounds cool
agentzh i've been mostly working on a Firefox extension these days. it's called SearchAll.
it's on addons.mozilla.org now.
masak looks
agentzh maybe it's the first FF extension written in Perl 5. 08:47
agentzh means the UI part.
masak oh! what's the trick?
agentzh Template::Declare and XUL::App
:)
masak aha
agentzh Jifty's tricks :)
masak sounds very worthwhile
agentzh :) 08:48
masak next time I have an idea for a plugin, I'll try that path
agentzh svn.openfoundry.org/searchall/trunk/
lambdabot Title: Revision 117: /trunk
masak ah, thx
agentzh there's an FF extension framework in the same repos :)
XUL::App :)
masak excellent
agentzh i'll try to push it onto CPAN when it's no longer hacky ;) 08:49
masak agentzh: what makes it hacky?
agentzh a regex tracer in XUL will allow me to hack on PCR in a full time fashion in $work ;)
masak nice
agentzh masak: XUL::App is using a lot of linux shells to make things work ;) 08:50
masak not to mention all the cross-pollinating between js and perl5
agentzh i mean shell commands
masak agentzh: is that avoidable?
agentzh masak: sure!
masak good.
agentzh masak: just a lot of 'cp', 'mkdir', 'zip' and etc etc etc ;)
masak has searchall now
yes, should be avoidable
agentzh wow
*nod* 08:51
masak I'm frankly intrigued by the technology that makes this work 08:52
mainly because I've been trying to make extensions and failed
agentzh heh
masak and because I feel more experienced at p5 than at js
agentzh mozilla's docs are not that good.
masak problem is they describe a moving target 08:53
agentzh nowadays, js is already very perl-ish ;)
true.
masak yes, I like js
but that doesn't mean I master it
agentzh a good amount of the online documentation is out of date :)
masak that's what I found too
agentzh XUL/js is now part of my $work now :) 08:54
and my $boss is amazed to see i still use a lot of Perl in $work ;)
i'm also working on another perl project named ajaxperl. 08:55
its goal is to merge HTML::DOM and JavaScripdt::SpiderMonkey and run Ajax sites in perl ;)
masak cool 08:56
masak had $boss say "we need to integrate perl into our application" yesterday
big moment
especially since it's a Java application
agentzh my friend is writing a perl translator that can convert w3c's official dom test suite to perl test scripts. 08:57
wow
integrating perl into java apps :D
masak yeah, don't know how feasible that is 08:58
agentzh heh
masak but surely making an exec call from Java, and doing things with STDIN/STDOUT is at least possible 08:59
agentzh some people here have embedded activeperl interpreter into their C++ MFC apps for production and give positive feedback :)
heh, IPC is not that efficient (or elegant) :) 09:00
masak 'fraid not
but maybe that's all there is ATM 09:01
in ten years or so, we might have Perl 6 on the JVM
agentzh well, at least there's some atoms of Perl :)
i agree :)
masak but today, there's only those tenuous links between the worlds... 09:02
masak praises the current trend of cross-platform-ness
that's one of the things that makes the perl 6 effort fun 09:03
09:03 IllvilJa left
agentzh indeed :) 09:04
agentzh looks forward to using NPQ in production.
s/NPQ/NQP/
masak oh, is that the one we'll be using in production 09:05
masak is still confused by all the new Perl 6es 09:06
agentzh too.
supper &
09:06 charsbr left
masak bon appetit 09:06
09:06 charsbr joined 09:12 snearch joined 09:20 snearch left 09:29 drupek12 left 09:32 IllvilJa joined 09:35 drupek12 joined 09:43 lichtkind joined 10:03 mj41 joined 10:11 fglock joined 10:30 daxim left 10:31 nipotan is now known as nipotaway 10:34 jisom_ left, ruoso left 10:36 Ashizawa left 10:37 fglock left 10:40 cognominal_ joined 10:47 daxim joined 10:58 ruoso joined 11:05 fglock joined 11:13 johntaro left 11:23 lyokato left 11:44 kanru left 11:47 lyokato joined, lyokato left 11:56 Coke_ left 11:59 Coke joined 12:22 cmarcelo joined 12:28 prefiks joined 12:32 hkb_lab is now known as hakobe 12:49 stevan_ left, stevan_ joined 12:53 baest left 12:59 bloonix left 13:05 rindolf joined 13:14 [particle1 is now known as [particle] 13:27 rindolf left 13:37 chris2 joined 13:44 [particle] left 13:45 TJCRI joined, [particle] joined 13:49 jeffreykegler left 13:56 drbean left 14:20 thoughtpolice joined 14:28 pmurias joined 14:30 drbean joined 14:33 amnesiac_ joined
pmurias fglock: is it ok to have $!attr compile to $self->{_value}{'attr'}? 14:34
14:39 Aankhen`` left, Aankhen`` joined 14:47 amnesiac left
fglock pmurias: i think so (checking) 14:50
TimToady pmurias: if a base class or derived class also has $!attr, it needs to be stored in a different place
fglock hmm 14:51
maybe {_value}{"_${class_id}attr"} except that we don't have ids yet 14:54
pmurias md5 is used for that puropose now in ExtractBlock
ExtractRuleBlock 14:55
$class_id = $COMPILER::source_md5 ~ CLASSNAME 14:56
fglock: or do you have a better idea than using md5? 14:57
fglock you could just stringify the reference, it's random enough 14:58
moritz_ but isn't that backend dependant again?
fglock yes 14:59
moritz_ for example if you write a java backend - i don't think you can get pointer adresses there to stringify
fglock or just use a simple counter
actually, i'm reading about jvm right now (and others) - trying to get more knowledge of threads 15:01
ruoso suggested we could get our hashes handled by BerkeleyDB, for STM and thread-safety 15:03
15:03 cognominal_ left 15:04 cognominal_ joined
TimToady might make more sense to put the id in with the "_value" than in with the "attr" 15:05
pmurias fglock: the Hash class implemented by BerkeleyDB? 15:06
fglock the low-level hash
but this seems too heavyweight, as each object requires several hashes 15:07
pmurias it sounds way too heavy
fglock TimToady: makes sense
TimToady dunno, I can argue it both ways 15:08
which usually means neither way is right :) 15:09
fglock should probably just hide it in a closure, but this would make object creation harder 15:11
15:13 amnesiac_ is now known as amnesiac 15:15 blindfish joined 15:18 iblechbot joined 15:30 rindolf joined
ruoso fglock, I think it would be saner to say that the memory allocation would be handled by bdb 15:39
I mean, 15:40
the slots used by the values would be stored in bdb
it happens that, as a variable is a value for the namespace,
it'll be stored there also
15:42 pnu left 15:43 pnu joined
pmurias ruoso: would storing in bdb imply IO? 15:44
ruoso pmurias, not necessarly 15:45
pmurias ruoso: does it have an benefits besides locking? 15:48
15:54 chris2 left
ruoso pmurias, besides doing it in a very extremely performant way... and besides being thread safe... 15:55
15:55 justatheory joined
ruoso maybe not... 15:55
;_
;)
pmurias ruoso: how much slower would it be for normal code 15:57
?
ruoso pmurias, I never made such a comparision 15:58
but several bdb operations can fit in 2 or 3 microseconds 15:59
because it's backed on file
16:00 barney joined
ruoso but it uses a lot of the filesystem cache 16:00
and have buffers and cache
so it doesn't sync everytime... 16:01
in fact, it syncs very often
unless you asks for it, of course
pmurias ruoso: using a pluggable representation might make sense
ruoso sure sure
I don't think using bdb as a final solution
it's just a nice implementation of memory allocation and pointers 16:02
I mean
or else, all the memory management needs to be done
[particle] how does dbm::deep compare
?
pmurias is ignorant of threading 16:03
ruoso [particle], I'm talking about bdb in C, not in Perl
the problem is
ruoso just realizes he's assuming the result of a conversation he had with fglock yesterday
how to implement threads? 16:04
considering there's so much on p6 to benefit from it?
the problem with an interpreted language, is that everything is global
I mean
the interpreter is a single instance
where you can access all variables
so... how to implement thread-safety without absolute pessimistic locking? 16:05
there's when the memory mangement idea came out
if the interpreter implements the memory management by itself, it can know the owner of each memory page 16:06
and implement page level locking
which is much more optimistic than object-level-locking
and that's when I reminded that BDB does exactly thatr
and that it's thread-safe
so...
each thread would open a bdb database
where it would allocate the memory 16:07
then we can implement any of share-everything, share-nothing and even copy-on-write
just easily
16:10 jhorwitz joined
ruoso eventually, a "memory database storage" could be implemented in BDB, or just the code that manages the databases (the "access methods") and the locking could be taken 16:12
or simply, someone would write something that would take care of that 16:13
and replace bdb entirely
16:14 ac joined
ruoso does anyone thinks it makes sense? 16:14
16:16 ferreira joined
ruoso just realizes that other memory management libraries may also be a good idea 16:19
16:25 dlo joined
dlo Hello? 16:26
Aankhen`` No, sorry. :-(
^_^
dlo :) 16:27
ac So are any of you worried about Haskell?
Aankhen`` ac: Er, "worried"?
avar Aankhen``: IMPLEMENT!
dlo I'm thinking about putting 3 months into helping perl6 get out the door. However, I'm worried about the ramp up time for perl6. IE, by the time I know what I'm doing the 3 months will be up. Do any of you have any thoughts? 16:28
Aankhen`` avar: Go away, I'm working on subs. :-P
16:28 franck___ left
ac I mean, worried about Haskell being responsible for perl6 community attrition 16:28
avar Aankhen``: :)
Aankhen`` Er.
avar dlo: You can work on KindaPerl6, it's pretty entry level:) 16:29
Aankhen`` tries to figure out why .emit_lisp fails on a ::Sub object.
avar what are you getting?
fail how?
16:29 amnesiac left
dlo avar: ... googleing kindaperl6 16:29
Aankhen`` Can't call method "emit_lisp" without a package or object reference at lib-kp6-mp6-p5/KindaPerl6/Visitor/EmitLisp.pm line 198, <> line 1.
ruoso keeps getting back to the idea of using libparrot...
in C
Aankhen`` avar: One moment, I'll pastebin.,
s/,//
avar dlo: what languages do you know? 16:30
dlo: search.cpan.org/~avar/KindaPerl6-0.001/
lambdabot Title: &#198;var Arnfj&#246;r&#240; Bjarmason / KindaPerl6-0.001 - search.cpan.org
dlo avar: I'm proficent in perl5, but I also know php (big help), and I've touched a score of other languages, but I'm not proficent in any of them.
pasteling "Aankhen" at 122.162.198.101 pasted "HALP" (15 lines, 571B) at sial.org/pbot/27876 16:31
[particle] dlo: i believe you can do good things with that donated time, even if you're just "ramping up". the question is, will the returns be enough to satisfy you? 16:32
avar I don't think $.parameters is a instance of something that can be emitted with .emit_lisp
[particle] dlo: want a pugs commit bit?
avar what does --ast say abou tit?
Aankhen`` It's a ::Var, which can certainly be emitted. 16:33
16:33 weinig left
Aankhen`` If I remove the special case for the ::Sub, it falls through to the not-yet-implemented part without any problems. 16:33
dlo [particle]: Well, I don't want to spend 2 months ramping up and then spend 1 month coding. I want to put in enough effort that something gets done, but I can't afford to get caught up in a long term project ( > 3 months).
avar What's on lib-kp6-mp6-p5/KindaPerl6/Visitor/EmitLisp.pm:198 ?
Also it would be nice to pass the kp6-mp6-* stuff through perltidy
dlo: What are you interested in? 16:34
dlo [particle]: as for pugs or kp6, I'm in inquring.
avar dlo: There'l plenty of stuff you can do in that timeframe
dlo avar: What I want is perl6 in a format that the average perl5 programmer can use and be productive in an production envirorment.
16:35 ferreira left
avar try cpan -f -i KindaPerl6; echo 'say "foo"' | kp6 | PERL5LIB=$(kp6 -lib) perl 16:35
dlo avar: I'm really sick of reading "perl" right next to "shell scripting" languages on job applications :)
[particle] dlo: have you cross-platform experience, or have you ever put anything on cpan? kp6 has just hit cpan, but isn't portable yet
avar dlo: I've been meaning to make that more friendly, see also the top of pugs.blogs.com
so if you want to make kp6 more UI friendly that would be neat!:) 16:36
pasteling "Aankhen" at 122.162.198.101 pasted "EmitLisp.pm:198" (72 lines, 2K) at sial.org/pbot/27877
dlo [particle]: I'm tenatious, but as for cross-platform experience, my C skills are a bit weak.
avar [particle]: Not portable, do you mean the dependency on gnu make?
[particle] dlo: i'm talking more about portable perl and make experience
avar: yep
avar uh, which of these is line 198? 16:37
Aankhen`` avar: That's line 198 passed through perltidy.
pmurias dlo: re C kp6 is writen mostly in a subset of perl 6, there is only one option C part
dlo [particle]: if you mean by portable, by "use English" then yes. I have expierence on FreeBSD and Win32.
[particle] avar: and things like C<PERL5LIB=foo bar/baz/quux> aren't portable either
avar Aankhen``: Munge the makefile so it passes stuff through | perltidy, that'll make things understandable
[particle]: You can use -I
dlo As for cpan distributions, my work contract really would be that happy about that, but that's been rectified.
avar [particle]: And I don't think the GNU make-dependant stuff runs during install
but sure, could be fixed up:) 16:38
Aankhen`` avar: Er, actually, that isn't line 198.
dlo I produced Proc-Backoff :)
16:38 fax joined, pmurias left
[particle] avar: setting an env var on the same line as running a program won't work on win32. in general, using ExtUtils:: modules would benefit kp6 portability 16:39
avar [particle]: I'm using ExtUtils, just not for everything:)
but yeah, it needs fixing
pasteling "Aankhen" at 122.162.198.101 pasted "EmitLisp.pm:198 (really)" (1 line, 956B) at sial.org/pbot/27878
Aankhen`` avar: There you go.
And I'm not touching the Makefile.
avar [particle]: I mean you can use -I instead of PERL5LIB
dlo [particle]: I can run make, however I wouldn't be "immediately" useful in cross platform coding.
[particle] dlo: i think you should have a pugs repository commit bit. send me your email address, and i'll make it so
avar Aankhen``: TOUCH THE MAKEFILE 16:40
:)
just add | perltidy
Aankhen`` No!
I'm mortally afraid of breaking things.
16:40 rindolf left
[particle] avar: yes, but i wasn't sure if that was the only non-portable bit. i just saw it in the blog posting. if you need somebody to test portability on win32 with ms tools, i can help 16:40
ruoso fglock, I was trying to particle at #parrot... maybe the libparrot idea is not that absurd 16:41
try to use the rules-to-C thing
fglock ruoso: cool
how about threads, GC, ... ? 16:42
ruoso ahm... forgot to ask about it
dlo [particle]: I sent you my email via a private IRC window...
[particle] threads subsystem design is happening soon, gc api is pretty stable 16:43
ruoso have a headache talking about the same thing with the same people in different channels ;) 16:44
Aankhen`` avar: Okay, using $.arguments.block.emit_lisp instead worksā€¦ so apparently Subs can't be emittedā€¦ hmm. 16:45
[particle] dlo: i didn't get the message. are you registered with nickserv? if not, you can't do private messages
Aankhen`` avar: Wait, further narrowing it down. 16:46
dlo ah, I'm not sure I'm registered with nickserv, I thought all #irc channels were connected (only 4 people on irc.perl.org#perl6 :)
[particle]: you can reach me at [email@hidden.address]
allbery_b no, they're only connected if the servers are linked. irc.perl.org isn't part of this irc network 16:47
(this is freenode, irc.perl.org is on magnet)
allbery_b is on magfnet, freenode, and cognet
dlo eyes cross...
ok
allbery_b try the /links command to see what IRC servers are part of the network you're on
dlo allbery_b: I'm using pidgin, I get a greyed out "Unknown command." 16:48
ruoso fglock, do you know what is the status of the rules-to-C thing from pmurias?
allbery_b oh, yeh, pidgin lacks most of the irc protocol
dlo sigh.
[particle] dlo: invite sent!
fglock ruoso: i haven't tried it, but it's a good start
[particle] dlo: i use pidgin too
/msg nickserv identify <password> # to register... 16:49
avar Aankhen``: USE PERLTIDY
Aankhen`` avar: Okay, the problem was with trying to call emit_lisp on a non-existent $.invocant
For some reason it didn't read my mind. :-S
avar perl also supports #line and such so you can have the error message say where it was in the file it was emitted from 16:50
that would probably be worse in our case though
Aankhen`` puts all that on avar's TODO list.
dlo I think dloca is already in use by someone else. 16:51
16:51 dlo is now known as dlocaus
[particle] hrmm, sending this commit bit is taking longer than i expected... 16:53
[particle] kicks commitbit
dlocaus [particle]: I have to get back to work, I'll contact you later? 16:56
16:57 jeffreykegler joined 16:59 melissa joined 17:00 pmurias joined 17:01 pmurias_ joined, pmurias left
pmurias_ ruoso: you inquired about the rule 2 C translator? 17:02
17:02 pmurias_ is now known as pmurias, dlocaus left
pmurias it can parse simple stuff, but better Match object creation is needed 17:02
17:03 bjoern_ joined
pmurias it emitts perl5 code which uses Inline::C 17:03
ruoso pmurias, hmm... I see.. 17:04
what I'm trying to sort out is how could we use the parrot runtime with libparrot 17:05
but use directly in C
pugs_svnbot r18287 | Aankhen++ | [kp6-lisp]
r18287 | Aankhen++ | * added newline in generated output after code block variable declarations.
r18287 | Aankhen++ | * added "binding" for subs.
r18287 | Aankhen++ | * implemented a simple &emit_lisp for ::Sub and ::Signature objects.
r18287 | Aankhen++ | * added KP6-SIGNATURE to export list.
r18287 | Aankhen++ | * removed PAD slot from KP6-CODE, added SIGNATURE slot.
r18287 | Aankhen++ | * added MAKE-KP6-SUB wrapper function.
r18287 | Aankhen++ | * added KP6-SIGNATURE class in Signature.lisp
diff: dev.pugscode.org/changeset/18287
lambdabot Title: Changeset 18287 - Pugs - Trac
bjoern_ Hi, I have this code: [[[
token CR { <[\x0D]> }
token LF { <[\x0A]> }
token CRLF { <CR> <LF> }
rule foo { bar | <CRLF> }
my $content = "\x0d\x0a";
my $match = $content ~~ m:keepall/<foo>/;
say $match.perl;
]]]
my version of Pugs says this does not match. However, if I change foo so it reads: '''rule foo { <CRLF> }''' the string matches. why is that?
(I tested this with 6.2.13 on Win32)
avar ruoso: See Parrot::Embed, but you need to know the pir stuff to use their rule compiler 17:06
ruoso avar, the idea is use a rules to c from kp6
like kp6-parrot 17:07
but...
pugs_svnbot r18288 | avar++ | EmitLisp, not Lisp
diff: dev.pugscode.org/changeset/18288
lambdabot Title: Changeset 18288 - Pugs - Trac
ruoso kp6-c-parrot
kp6 emitting C with parrot as runtime
pmurias got to leave& 17:08
fglock bjoern_: my pugs doesn't like the keepall option 17:09
[particle] bjoern_: if you're using the cpan release of 6.2.13, that's very old. i recommend you svn co from the repo and build with ghc on windows 17:10
fglock token foo { 'bar' | <CRLF> } works 17:12
it's probably a PCR bug :( 17:13
bjoern_ The keepall can be removed, same effect. I'm using a win32 build from 2007-05-31; I also tried a month old svn version, but it's generating plenty of error messages...
If this is a bug, is there any way to rewrite the grammar to the desired effect? 17:15
fglock can you replace 'rule' with 'token' ?
pugs_svnbot r18289 | Aankhen++ | [kp6-lisp] 17:16
r18289 | Aankhen++ | * reworked MAKE-KP6-SUB into a macro.
r18289 | Aankhen++ | * added KP6-CHECK-ARGUMENTS generic function (always returns T for now).
diff: dev.pugscode.org/changeset/18289
lambdabot Title: Changeset 18289 - Pugs - Trac
avar hello world for subs works!:) 17:17
17:17 weinig joined
avar prods fax:) 17:17
Aankhen`` avar: Arguments are ignored though.
Working on it.
avar neat
fax What is hello world like now? 17:18
17:18 masak left
avar sub hello { "hi" } say hello() 17:18
fax :o
cool
bjoern_ I'm not sure, this was really a part of a hello world grammar to explore rules... 17:19
What I tried to do was make a grammar for the IETF ABNF format and convert the parse tree into some other format.
avar ABNF?
bjoern_ www.ietf.org/rfc/rfc4234.txt 17:20
avar Files=19, Tests=40, 20 wallclock secs ( 0.25 cusr + 0.29 csys = 0.54 CPU)
the lex-sub* tests pass now
Aankhen`` avar: I'm impressed.
17:22 arrowax joined
avar a loot of stuff would work if we had methods, like 17-closure.t 17:22
Aankhen`` Yeah, I'm sure. :-P
avar :)
fax :[ 17:23
I forget how to run it
something like: echo "sub hello { "hi" } say hello()" | pugs run_kp6_mp6_perl5.pl ?
avar perl script/kp6 -lisp | sbcl
fglock bjoern_: it's not a pugs bug - replace 'rule' with 'regex'
rule has a special meaning 17:24
bjoern_ Could you say how the special rules for 'rule' affect the case above? 17:25
fax works :D
17:25 baest joined
avar bjoern_: <ws> is inferred on whitespace and it backtracks 17:25
fglock it means that spaces really mean ' '
avar mm 17:26
Aankhen`` Hmm.
fglock HHmmm
Aankhen`` In order to implement proper arguments, I'm going to end up writing a pattern matcher.
[particle] hrr
bjoern_ So where are the spaces in `rule foo { bar | <CRLF> }`?
(there are four, but which of them are significant for this?) 17:27
moritz_ bjoern_: all of them are interpreted as <.ws>
fglock bjoern_: rule foo {bar|<CRLF>} works
moritz_ bjoern_: because rule implies :sigspace 17:28
bjoern_ I see. I had assumed this is concerned with inter-word spaces, like in rule { a b c }. thanks.
[particle] rule foo { bar | <CRLF> } == regex foo {<.ws>bar<.ws>|<.ws><CRLF><.ws>}
Aankhen`` [particle]: Stop smirking at me. 17:29
[particle] :D better? 17:30
Aankhen`` mutters something rude.
Coke o_O
fglock rule also implies :ratchet, so it's actually a toke 17:31
token
moritz_ fglock: but token doesn't imply <.ws>, right? 17:32
bjoern_ I guess then the solution would be to set <.ws> to "nothing" since/so-long-as I take care of all the white space handling in the grammar?
fglock you should just use 'token', i think 17:33
or 'regex', when you need backtracking
token ws { '' } does work :P 17:34
17:35 Psyche^ joined 17:41 lumi left
fax avar: Should I try to get something from t/kp6/ to work or is there anything I could do to help? 17:46
avar fax: getting t/kp6/ stuff to work is the best you can do atm I think
17:46 bjoern_ left
avar fax: if you do perl compile*lisp* you get sbcl errors where the tests fail to compile, that's a good place to start 17:47
17:47 Patterner left, Psyche^ is now known as Patterner
avar fax: If you look at the -perl output you can see roughly how it works there 17:48
-perl | perltidy
cpan -i Perl::Tidy
buubot avar: Sorry, I couldn't find a module that matched: -i
17:48 lumi joined
avar fax: and prove -r t-lisp-exe runs the tests 17:49
buubot: stfu
buubot avar: stfu is a6.vox.com/6a00b8ea0717f31bc000c22527717e8fdb-pi
fax cool, I'll try
avar fax: and Runtime/Perl5/ for how some of the stuff works there 17:51
fglock Junction needs some VAR() kind of escape, in order to allow (1|2).perl
avar -ast | perltidy is also useful for debugging
fax: might want to try t/kp6/45-for.t :) 17:53
17:58 jisom joined
[particle] obra: commitbit is acting up. i can't successfully send an invite, the browser times out with "The connection to the server was reset while the page was loading." 18:00
obra any feather admins around? 18:07
[particle] juerd: ping 18:09
diakopter: ping
iunno any others
Juerd pong 18:10
Ugh 18:12
Someone didn't complete a debian installation 18:13
[particle]: It works but is slow 18:14
[particle] juerd: i'll try again 18:15
Juerd obra: Apparently commitbit uses a lot of memory, possibly allocating and deallocating a lot.
[particle] now it worked, and quickly.
Juerd++
Juerd obra: It's very slow whenever there are memory performance issues. Like when that DIMM was broken, and now, when the wrong libc was installed.
obra: And much slower than all the other things on feather 18:16
obra I think that has a lot to do with how it uses the sqlite db 18:17
is there a postgres on feather?
fglock obra: hi! 18:19
Juerd obra: Not yet :)
obra hey fglock!
Juerd: that _may_ be an answer
Juerd obra: If I install postgres, can you do the rest? 18:20
If it doesn't work, it can be removed just as quickly
obra let me try one other thing first?
the other thing requires a lot less work on my part ;)
Juerd Sure :) 18:21
I can't actually test if the solution works, I just realised :) 18:22
obra heh
does it feel better now? 18:24
Juerd I have no idea :( 18:25
obra fglock: how's things?
[particle]: how does it feel now?
I cleaned sessions from the db and vacuumed it. saving um. 80% of the db size
Juerd It feels fast
obra ok :) 18:26
Juerd Takes 5 seconds to invite someone. That's faster than a few minutes ago.
Still not very fast, but it's not something you do a lot anyway :)
obra *nod*
18:27 Jedai joined
fglock obra: fine; doing some planning again 18:28
obra cool. what's coming up?
fglock Perl 6 threads would be next 18:29
Juerd obra: 80%, hmm :)
fglock not sure how to implement yet,
Lisp might be it,
obra Juerd: lots of old sessions
fglock AST transforms would help fake it
it would be nice if Parrot already had concurrency 18:30
this is not a showstopper for STD in kp6, 18:31
but it is needed by TimToady's own implementation
obra *nod* 18:34
18:38 ruoso left, prefiks left 18:40 IllvilJa left
fglock i'm going to Lisbon.pm social meeting in a while 18:41
brb &
18:44 Torment left
pmurias fglock: i think coro and threads are seperate things 18:51
18:52 IllvilJa joined, blindfish left
pmurias i find it all Spec/Concurrency strange it mentions both Co-Routines and Corutines 18:55
fax eek! 18:56
pmurias they use diffrent syntaxes
the first sub and produce the other coro and yield
fax v6/v6-KindaPerl6/lib-kp6-mp6-p5/KindaPerl6/Visitor/EmitLisp.pm is generated by running mp6.pl ? 18:57
pmurias the first don't use threading i'm not sure about the other
fax: yes
unfortunatly
fax ah It's because it's generated from Perl6 code 18:58
(I suppose)
pmurias yes
kp6 can't compile itself yet so MiniPerl6 is used
which is a kp6 ancestor 18:59
it's compiled from a same named file in lib 19:01
fax perl mp6.pl # so this should do it?
19:01 fglock left
fax If I am in pugs/v6/v6-KindaPerl6 19:01
(It's running for ages, seems like it's not doing anything?) 19:02
19:04 jhorwitz left
pmurias fax: no 19:05
fax: use make kp6_mp6
perl mp6.pl is waiting for input on stdin
19:08 avar left, avar joined 19:10 takanori_ left, takanori joined 19:11 jwest- joined 19:15 justatheory left 19:27 thoughtpolice left 19:28 hiarc left 19:31 chris2 joined 19:32 lifo joined 19:33 lifo left 19:36 pmurias left 19:37 jwest- left, kjwcode joined
fax :D 19:49
I made 45-for.t work
avar: Do you know Array.lisp 19:50
Is it :fill-pointer t needed?
Because I tried removing it so I could do (adjust-array (kp6-value self) (1+ index)) when you try to assign something into an array which hasn't enough space yet 19:51
if :fill-pointer t is needed then I could replace it with vector-push-extend or something like that
19:53 thoughtpolice joined 19:55 thoughtpolice left
pasteling "fax" at 137.195.250.2 pasted "modification to Array.lisp" (14 lines, 501B) at sial.org/pbot/27881 19:55
19:56 drupek12 left
fax should I commit ? 20:00
(I added a for loop macro and made the EmitLisp thing emit that as well) 20:01
avar: you around :S
[particle] fax: it's easier to ask forgiveness than permission 20:02
fax haha
20:02 dduncan joined
fax ok you're right it's easily modified 20:02
pugs_svnbot r18290 | fax++ | Added kp6-for-->-single macro to Interpreter.lisp, made EmitLisp generate that. 45-for.t now succeds. 20:05
diff: dev.pugscode.org/changeset/18290
lambdabot Title: Changeset 18290 - Pugs - Trac
20:06 thoughtpolice joined
[particle] fax++ 20:07
fax *g*
20:17 rindolf joined 20:23 mj41 left 20:30 barney left 20:32 sunnavy_ joined 20:35 sunnavy left, sunnavy_ is now known as sunnavy 20:42 sunnavy left, sunnavy joined 21:05 Limbic_Region joined 21:18 TJCRI left 21:19 dduncan_ joined 21:26 dduncan_ left, dduncan_ joined 21:27 dduncan_ left 21:32 cmarcelo left, dduncan left, lichtkind left 21:34 dduncan joined 21:45 rindolf left 21:50 japhb joined
pugs_svnbot r18291 | Aankhen++ | [kp6-lisp] 21:53
r18291 | Aankhen++ | * got sub signatures (sort of) working. limitations: no checking of types, no slurpy parameters, must add a useless variable at the beginning.
r18291 | Aankhen++ | * replaced ARRAY and HASH slots in KP6-SIGNATURE with a single VALUE slot which consists of conses whose CAR is 'POSITIONAL or 'NAMED and CDR is a variable name.
r18291 | Aankhen++ | * added WITH-KP6-ARGUMENTS macro for validating and binding arguments.
r18291 | Aankhen++ | * added KP6-SIG-ITEM wrapper function.
r18291 | Aankhen++ | * changed $.block.emit_lisp in ::Sub to $.block.emit_body since we take care of declarations and signatures already.
diff: dev.pugscode.org/changeset/18291
lambdabot Title: Changeset 18291 - Pugs - Trac
22:00 justatheory joined
pugs_svnbot r18292 | Aankhen++ | [kp6-lisp] 22:05
r18292 | Aankhen++ | * obviated the need to prepend a throwaway variable to sub signatures.
diff: dev.pugscode.org/changeset/18292
lambdabot Title: Changeset 18292 - Pugs - Trac
22:08 Coke left
pugs_svnbot r18293 | Aankhen++ | [kp6-lisp] 22:08
r18293 | Aankhen++ | * fixed PERL->DISPLAY for KP6-ARRAY.
diff: dev.pugscode.org/changeset/18293
lambdabot Title: Changeset 18293 - Pugs - Trac
22:11 REPLeffect_ joined 22:14 iblechbot left 22:15 kanru joined 22:16 lisppaste3 left 22:17 mj41 joined
Aankhen`` burns out halfway through trying to convert GLOBAL.lisp to use MAKE-KP6-SUB. 22:21
I need sleeeeeeeeeeeep.
22:21 Aankhen`` left 22:23 lisppaste3 joined 22:25 polettix joined, drupek12 joined 22:30 elmex left
pugs_svnbot r18294 | lwall++ | [STD] now autolexes (sort of) everything under expect_term 22:47
diff: dev.pugscode.org/changeset/18294
lambdabot Title: Changeset 18294 - Pugs - Trac
22:49 arrowax left, Jamtech joined 23:00 pravus joined 23:01 wilx left 23:04 rafl left 23:05 Helios- left 23:10 Gothmog_ left 23:12 chris2 left
avar fax: :) 23:15
23:28 [particle1 joined 23:31 Gothmog_ joined 23:37 Caelum left 23:44 [particle] left 23:48 rafl joined, Helios- joined