pugs.blogs.com | pugscode.org | pugs.kwiki.org | paste: sial.org/pbot/perl6 | <stevan> Moose... it's the new Camel ":P | .pmc == PPI source filters!
Set by Alias_ on 16 March 2006.
nothingmuch hola KingDiamond 00:13
isn't it very late over there? or is it already early? 00:14
heh
00:17 rmoriz joined
rmoriz hi 00:18
FurnaceBoy ola 00:24
TimToady ruoso: to do good translation you need everything from surface syntax to deep semantics. If you try to simplify it to one level you're throwing away information that can potentially allow someone to write a better translator that has a better idea when to preserve surface forms and when to abandon them. 00:28
11:09 < nothingmuch> my $x = $x; # the second $x is outer 00:30
nothingmuch: re: my $x = $x; # the second $x is outer
the comment is incorrect.
you have to say $OUTER::x to get the outer one. That's something we changed from Perl 5. 00:31
declarators always introduce names immediately rather than at the statement boundary.
off to work on jitting my speech... 00:33
00:38 jroc joined
nothingmuch TimToady: i was talking about Perl 5 =) 00:48
00:50 Khisanth joined
audreyt ingy: libapi.com 00:51
00:53 TreyHarris joined
TreyHarris should i expect "pugs -e 'say "hi"'" to take several seconds, and the "make test" to have taken 12 hours, or do i have something wrong with my install? 00:56
Khisanth you didn't specify the speed of the system executing things :) 00:58
00:59 clkao joined
TreyHarris Khisanth: it's a powerbook g4. so a relatively recent but not particularly fast machine. 00:59
in general, code seems to run about as fast as i can eyeball it and decide what it should be doing. so it runs about as fast as the virtual p6 interpreter in my brain :-) 01:10
obra TreyHarris: how old a copy of pugs are you using? 01:19
TreyHarris: Powerbook G4 what ;) that's a pretty wide range. 01:20
TimToady nothingmuch: okay, I see now that it wasn't clear. 01:21
TreyHarris obra: 1.5 GHz, 512MB 01:24
pugs 6.2.11 01:25
obra ok. both reasonably recent. 01:28
Audrey will likely be back within the hour and have more useful commentary than I do
01:33 mjl69 joined
meppl gute nacht 01:37
audreyt TreyHarris: try thunk - we've optimized it a lot. 01:40
TreyHarris: svn.openfoundry.org/pugs/ 01:41
s/thunk/trunk/
wow, they just published a "Learning Haskell" book in Japanese today 01:42
and the 1st chapter starts with my quote :D
ayrnieu what is your quote? 01:43
audreyt What's faster than C++, more concise than Perl, more regular than 01:44
Python, more flexible than Ruby, more typeful than C#, more robust than
Java, and has absolutely nothing in common with PHP? It's Haskell!
ayrnieu neat :-)
miyagawa OMG 01:46
obra wow. cool
miyagawa that's neat
clkao WHOOT 01:47
audreyt looks like a good book
zhobson 01:52
TreyHarris is the SVN server down, or is it at some URL other than svn://svn.openfoundry.org/pugs/? 01:53
audreyt http:// not svn://
TreyHarris oh. i tried that first, and gave up after several minutes went by with no activity 01:54
audreyt weird, it worksforme 01:55
TreyHarris oh, it's coming
just very.... very.... slowly
audreyt sorry about that...
TreyHarris about 9600 bps or so, sure it isn't on the other side of a modem? :-)
audreyt adds "source snapshot server" to the list of todos
:/ 01:56
02:01 amnesiac joined 02:20 stevan joined 02:36 clkao_ joined 02:47 clkao_ is now known as clkao 02:48 DesreveR joined
theorbtwo audreyt++ # "...and has absolutely nothing in common with PHP? It's Haskell!" 02:54
03:08 tennin joined 03:22 justatheory joined 03:29 bane joined 03:43 FurnaceBoy_ joined 03:45 qu1j0t3 joined 03:53 hcarty joined, hcarty left 04:26 tennin left 04:27 FurnaceBoy joined 04:33 clkao joined 04:47 mako132_ joined 04:58 ast[away] joined 05:07 clkao_ joined 05:10 Quell joined, FurnaceBoy joined 05:18 kanru joined
TimToady audreyt: ping 05:22
audreyt TimToady: pong -- I got the slides tech working
TimToady ++
audreyt perlcabal.org/~autrijus/tokyo/takahashi.js 05:23
oh wait, a sec, I need to port to your .js
TimToady cue Jeopardy music... 05:24
audreyt done
TimToady same?
obra "I'll take javascript for 1000, Larry"
audreyt TimToady: perlcabal.org/~autrijus/takahashi.js 05:25
drop-in replacement for your takahashi.js 05:26
05:26 Khisanth joined
audreyt TimToady: (refresh) 05:27
(18883 bytes)
TimToady has a bug 05:28
audreyt fixed that
(18884 bytes)
TimToady repeats first ~ line
audreyt yeah, fixed :)
refresh
TimToady seems to drop middle line of "syntax should no tdangle..." 05:32
audreyt worksforme...
you changed .xul? 05:33
TimToady current .xul on wall.org/~larry/stumpjp.xul 05:36
audreyt 403
a+r?
TimToady try agin 05:38
audreyt got it
perlcabal.org/~autrijus/tokyo/takah...tumpjp.xul 05:41
perlcabal.org/~autrijus/tokyo/takah...kahashi.js
wolverian (doesn't work here) 05:43
hm, wait, yes it does. nevermind
audreyt sure, it's lacking css
wolverian yeah. it looked weird. :)
audreyt (uploaded that too, for none($Larry)) 05:44
wolverian hrm, the title is strange 05:45
{{img src=... }}
audreyt yeah.
Sporx can fix all that
we (as in, ingy) are working on it
05:51 clkao joined
TimToady it's kinda badly busted now. titles carry over from one slide to another, and title space never goes 0. 05:53
audreyt yup
TimToady plus I have titles with 4 or more lines... 05:54
audreyt fixing... 05:55
05:55 GeJ joined
audreyt TimToady: fixed... 05:58
(same two urls)
hm
I only handle 4 lines at the moment
maybe it's better to split the long lists
subtitles with 6+ lines doesn't really read well :/ 05:59
gaal morning!
audreyt hi gaal!
gaal is it on purpose that the title page is blank, and need to click to get first text?
audreyt I don't know, ask larry... 06:00
TimToady it's the first picture that never works. If you back up to it you'll see it's just a trash picture. 06:01
audreyt ah right.
(battery at 4%, can go out at any moment)
gaal any word on NA hackathon? 06:03
06:05 FurnaceBoy_ joined, Quell joined 06:25 fordinal joined 06:28 GeJ joined
Quell bbl 06:29
06:32 bernhard joined 06:42 nirgle joined
fordinal hi all 06:46
nirgle hi
fordinal I have one question. I plane one bigger project, but is perl6 implementation in pugs stable for generic coding? 06:50
nirgle couldn't tell you, i've only been half following perl development over the years 06:51
audreyt fordinal: if it is for production use, then no. 06:52
probably wait until we get Perl6->perl5 compiling working reasonably well
currently it's more for exploring how Perl 6 works like, not for deploying in production
fordinal hmm... I have now one application framework programmed in perl5, but new version is comming soon. and perl5 is not suitable fot it 06:53
obra what sort of apps?
PerlJam why is perl5 not suitable? 06:54
fordinal i need bytecode (for future), full support for object programming, etc...
and no problematic threads 06:55
06:55 iblechbot joined
audreyt hm. 06:55
audreyt suggests OCaml
(the only language I know of that supports all three)
obra why do you need the bytecode? 06:56
fordinal one feature is for me excelent from parrot: function programmed in another language can be used in parrot
bytecode for distribution modules without source code
PerlJam PAR?
audreyt hm there's also Erlang :)
obra It's usually easy to turn bytecode back into source 06:57
PerlJam continues to advocate perl :)
audreyt PerlJam++
PerlJam fordinal: perl5 code *can* be compiled to bytecode you know.
fordinal but turn bytecode back into PIR source, not? :)
PerlJam: yes? give me link 06:58
audreyt PerlJam: not platform-independent tho :)
fordinal: pp -f Bytecode yourprogram.pl 06:59
then
./a.out
fordinal turn into another language is not possible now. here is 10 peoples who can write good in perl for 5-6 years, and learning another language...
audreyt link is search.cpan.org/dist/PAR/ # my previous pet project
PerlJam fordinal: it's not possible now with parrot either.
fordinal PelJam: what is not possible now with parrot?
07:00 KingDiamond joined
PerlJam fordinal: because hardly anyone is doing that part of the work! 07:00
audreyt fordinal: parrot's threading is currently not designed and not implemented :/ 07:01
PerlJam fordinal: All VMs have the potential to be mass-translated back and forth between languages that target the VM ... if only there are people working on the translation
fordinal audreyt: but pugs has threading implemented 07:02
audreyt fordinal: right, because we use GHC threading
and we have bytecode (in YAML format)
fordinal audreyt: yep :)
audreyt: in sources of pugs is some cool examples
audreyt but we don't yet have full object programming (missing roles support), and perl5 embedding leaks memory, and the runtime speed is rather slow. 07:03
we are fixing them, but it's probably not going to work great today for production use )
:)
fordinal audreyt: yes, but I can programming in perl6 and in future migrating all code easy to parrot, or no?
PerlJam audreyt: I'd almost advocate using pugs in production anyway for certain problem domains. 07:04
audreyt fordinal: yes, but you won't get sane threading until parrot implements them
07:04 Khisanth joined
audreyt fordinal: but once parrot does, sure, you'd get that 07:04
fordinal Is there any timeplan or roadmap for parrot? i can't find any deadlines with date. 07:07
audreyt looks at obra 07:08
07:08 pjmiguens joined
audreyt obra is listening to ingy's talk :) 07:09
obra fordinal: It's an opensource project.
PerlJam heh
obra The variables on software are: 07:10
1) Time
2) Quality
3) Money
4) Hackers
We mostly have fixed 2 and 3.
To help drop 1), if you could add 4), people would be thrilled 07:11
fordinal obra: yes, I know that is it a opensource project. next version of our application framework will be opensource :)
obra: can I help with perl5 coding? hmm? 07:12
obra can you help parrot by writing code in perl5?
One thing parrot could really use is a more comprehensive test suite. 07:13
fordinal: what's interesting about your app framework?
fordinal with anything with my perl5 skills
audreyt fordinal: I'd say a lot of www.parrotcode.org/todo are perl5 tasks
pjmiguens I'm search for a modules that i can use to manage GPO's on win xp systems, anyone knows? 07:15
obra pjmiguens: you might want to try #perl
pjmiguens: or searching for "Win32" on search.cpan.org
pjmiguens i have already search for 07:16
07:16 ghenry joined
pjmiguens but ok 07:16
fordinal At first it was an flexible CMS. Now its an XML application framework for just about anything we need. 07:17
audreyt fordinal: in that case... I'd say still write the next version in Perl 5, but experiment the design with Pugs's Perl 5 runtime technologies 07:19
(like Moose.pm and Pugs::Compiler::Rule etc)
and as Pugs's p6->p5 compilation ability gradually matures 07:20
you can replace parts of your program and write it in perl 6 via "use v6-pugs;"
but you can still access to the wonderful XML modules available from CPAN 07:21
PerlJam fordinal: or wait a couple of years to have a production perl6 compiler you can use. :-)
audreyt because $your_favourite_xml_module is great
(leaving it as a variable to avoid holy wars)
PerlJam pmichaud tells me that there's a 70% probability that he'll have a parrot based perl6 compiler by June. Applying the random-fudge factor, I'd guess that there's more like a 40% probablity for June and a 90% probability for September/October 07:23
07:24 marmic joined
audreyt . o O (Not to be contrary, but he said the same thing, last March.) 07:24
PerlJam audreyt: he moves much better on the perl6 front if you pry him away from the paying gigs and his pmwiki work. (Or make perl6 a paying gig I guess) 07:26
audreyt *nod*
PerlJam Now that there's a perl6 parser in parrot/languages/perl6, maybe we won't be bottlenecked by a small number of people. 07:29
audreyt indeed
PerlJam (and particle++ for the apparent heroics trying to get a test framework for it)
07:29 drrho joined 07:35 pjmiguens left 07:37 xinming joined, clkao_ joined
obra pmwiki? 07:39
PerlJam obra: www.pmwiki.org 07:43
07:43 nothingmuch joined
nothingmuch gaal, audreyt: seen the new smoke server crap i've been working on? 07:44
07:47 ko1_ joined, mj41_ joined 07:53 khisanth_ joined 07:58 clkao joined
audreyt nothingmuch: no... url? 08:03
nothingmuch one sec 08:04
nothingmuch has to run it on woobling 08:06
08:13 KingDiamond joined
nothingmuch morning KingDiamond 08:13
08:27 xinming joined 08:29 kane_ joined 08:31 khisanth_ joined 08:45 bsb joined 08:50 khisanth__ joined
nothingmuch sooo.. 08:52
nirgle is reading about light sabers on wikipedia
nothingmuch audreyt: ping 09:03
webmind bored? 09:04
nothingmuch gaal: ping
pasta.woobling.org:3000/
09:10 khisanth__ joined 09:20 ruz_ joined 09:25 KingDiamond joined
wolverian whoa, feather has vim7 beta :) 09:32
nothingmuch beta? isn't it alpha/ 09:36
nothingmuch really likes MacVim 7
unicode + popups for completion
wolverian it says beta 09:38
2;0 wolverian@feather:~$ vim --version
VIM - Vi IMproved 7.0b BETA (2006 Mar 24, compiled Mar 27 2006 13:41:56)
nothingmuch coolness 09:39
gaal nothingmuch++ # smoke comparison! w00t
09:40 khisanth__ joined
nothingmuch it's also has "real" storage 09:40
i.e. you upload serialized $model->structure
not an html
gaal who renders the html?
nothingmuch so there's no XSS
the server
with caching
otherwise you couldn't compare
gaal that used to be slowass
nothingmuch the slowass part was the YAML loading 09:41
gaal but maybe w/yaml-syck it isn't?
yay
nothingmuch the whole pugs test suite graph on my box takes ~14 seconds to draw
which is bad
and consolidated will take more
gaal used to take more, no?
nothingmuch but if it's cached then it's OK i guess
it takes several mins for the yaml to load the model
integral nothingmuch: perhaps cache the YAML as a Storable blob when it is uploaded? 09:42
gaal anyhow doh, I'll ci something that uses yaml-syck in pugs if available right away
nothingmuch integral: uploading is through Data::Serializer
and it's stored in a storable blob on the server
integral ah 09:43
nothingmuch dev.catalyst.perl.org/repos/Catalys...mokeServer
there's even a demo DB with it
the one you see on the site =)
anyway, look at the upload script
and at Test::TAP::Model::Smoke
which should be renamed to SmokeReport
anyway, i have multi page hexdumps to read and compare 09:44
so ta ta
tell audreyt though
i want all your feedbacks
FYI categorization will be tag oriented
and smokes will be tagged e.g. 'project.pugs platform.linux core.jsperl revision.xxxx'
then you can categorize by "project", "platform" etc etc 09:45
or filter by any of these
oh, btw
compare is broken in one case (tth bug)
pasta.woobling.org:3000/compare?smo...p;smoke=14
smoke 14 contains one less test
so the data is non existent 09:46
and thus the template explodes
and lastly, with a fix that will happen to DBIC soon (actually SQL::Abstract) the whole smoke list will be one SQL query 09:47
so the performance should vastly improve
(currently $smoke->tags is a separate query, but the prefetching gets confused due to several joins over the same table)
09:48 ko1_ is now known as ko1_away
rgs Test::TAP::Model::Smoke ? I wonder whether this could be used for blead 09:48
09:57 nnunley joined, khisanth_ joined
nothingmuch rgs: per'aps 09:58
it's really just Test::TAP::Model->structure with a bit of meta data
tags, date, title, etc
the title in the smoke report defaults to "smoke" in my example data
but should probably be $manual_user_input || $sensible_default 09:59
tags encompass all the attribute meta data type things
that is, not date, not duration, not test info (numbers, etc)
rather things that tell the user what this smoke report checked
(e.g. perl with mad skillz and threads) 10:00
the test links won't work btw ;-) 10:01
i have to find a sane semantic for them
and should deactivate them till then
perhaps Model::TestFiles ought to be customizabel 10:02
so that it can rely on e.g. Model::SVN
(which actually exists =)
10:06 linyhire joined, linyhire left 10:12 elmex joined
gaal argh yaml.pm's not being oopy is giving me teh pain 10:13
$yaml_provider->Load doesn't
10:35 LordArthas joined 10:36 LordArthas left 10:44 KingDiamond joined
nothingmuch wolverian: DIR => $self->{tmpdir} ? $self->{tmpdir} : undef seems redundant 10:54
(defined $self->{tmpdir} ? ( DIR => $self->{tmpdir} ) : ())
wolverian hum.. what?
nothingmuch err, not wolverian 10:55
wolverian :)
nothingmuch oi vey
wrong channel
wolverian no worries
nothingmuch wishes audreyt would comment already ;-) 10:58
11:13 kane__ joined 11:15 kanru joined 11:21 chris2 joined
svnbot6 r9792 | fglock++ | PCR - subrule backtracking - tests pass; fixed warning in perl5() method 11:28
11:29 ko1_away is now known as ko1_ 11:36 bsb_ joined 11:44 kallianka joined 11:46 ko1_ is now known as ko1_away, ko1_away is now known as ko1_ 11:58 Aragone is now known as Ara4n 12:08 mugwump_ joined 12:17 KingDiamond joined 12:18 nirgle left 12:23 Khisanth joined 12:35 ko1_ is now known as ko1_away 12:46 KingDiamond joined 12:53 xinming joined 12:54 tomyan joined 12:57 mako132_ joined 12:59 kolibrie joined 13:09 hugues23 joined 13:12 Limbic_Region joined 13:35 xinming_ joined 13:44 clkao joined 14:03 mkirank joined 14:06 mkirank left 14:11 vel joined 14:37 KingDiamond joined 14:52 pmichaud joined 14:56 b_jonas joined 15:10 xinming_ joined 15:13 pmurias joined
pmurias hi all 15:13
svnbot6 r9793 | clkao++ | Initial Data::Bind, doing simple scalar bindings. 15:15
15:18 FurnaceBoy joined
kallianka Hello, how do I open a file for writing in perl6/pugs? 15:20
svnbot6 r9794 | clkao++ | Call positional bind in one go.
audreyt kallianka:
my $fh = open($filename, :w);
kallianka audreyt: thx 15:21
where can I read about how to use the builtins?
It's a bit difficult to guess ;) 15:22
15:27 hexmode joined
kallianka anyway, i've just written my first perl6 program (that's not a test of perl6 but solves a problem), it was fun! 15:27
15:38 Aankhen`` joined 16:03 elmex joined
nothingmuch audreyt: ping 16:07
audreyt: pasta.woobling.org:3000/list?tag=&a...orm+result
gaal: pasta.woobling.org:3000/list?tag=&a...orm+result
(tag grouping)
audreyt kallianka: woot! 16:08
kallianka: docs/other/porting_howto 16:09
brace for very bad idea 16:17
(coming to pX space)
svnbot6 r9795 | audreyt++ | * Evil plans to use Perl 6 Rules as Templates (i.e. structure-to-string
r9795 | audreyt++ | rendering engine), so that Anything You Can Render, You Can Parse Back!
16:37 saorge joined
pmurias audreyt: cool, IMHO very perlish 16:41
audreyt thanks! also cf Template::Extract and Template::Generate on CPAN from yours truly 16:42
16:44 KingDiamond joined
stevan audreyt: ok, thats both very cool, and very scary at the same time :) 16:47
which is to say "I like it very much" 16:48
gaal audreyt: Template::Patch 16:51
nothingmuch: whee
audreyt: actually I encountered a weird problem working on that 16:52
prolly a bug in YAML::Syck
audreyt oh?
I've been fixing Y::S bugs left and right :)
gaal no simple test case, sec. 16:53
audreyt complex test cases are fine
gaal audreyt: I'm using 0.38; anything to upgrade to? 16:54
audreyt 38 is latest 16:55
gaal audreyt: wget feather.perl6.nl/~gaal/tmp/Template...ch.tmp.tgz 16:57
please build, and run prove -Ilib t/01-basic.t 16:58
that dumps a structure with Data::Dumper, YAML, and YAML::Syck
Y:S elided some details :) 16:59
the weird thing about the structure is that parts of it were created with Class::Accessor::Ref
(Dan Kogai)++ # Just::Another::Perl::Hacker 17:01
17:08 nothingmuch joined
nothingmuch audreyt: ping ping ping ping ping 17:10
gaal nothingmuch: moose moose moose moose moose
nothingmuch how goeth it?
gaal I think it starts with "a one, a two, a one two three four" 17:11
nothingmuch gaal: force audreyt to go to the smokeserver dem 17:12
o
she keeps saying "url?" when I'm not around ^H 17:13
and then not backlogging
audreyt nothingmuch: I did look at it thx
nothingmuch any feedback?
gaal nothingmuch: I got degumptionated re porting everything to best<YAML::Syck YAML> because they don't have class methods. :( 17:14
nothingmuch =(
audreyt pmichaud++ # fruitful discussion on #parrot re auto-translate-parsec-to-p6rules 17:15
gaal wow!
nothingmuch audreyt: no feedback?
gaal תשומי
audreyt (summary: experiment with the style of midstream {return...} with a subset of p6)
17:15 fglock joined
audreyt so instead of relying on TGE for any sort of post processing 17:15
nothingmuch gaal: כן 17:16
audreyt just write in parsec style and impl the subset of p6 used inside {return...} (bindings, $<var>, $0, $<var>[0], conditionals)
nothingmuch gaal: אני לא צריך את זה לעצמי אז אני רוצה ביקורת
pmichaud probably also want $()
audreyt so P::C::R can impl that in p5 and PGE in pir
but the point is that we can (again) have a shared p6 grammar
(across hs/p5/pir) 17:17
I really worry about forking into three grammars
as was what seemed to about to happen
which would be very bad :)
pmichaud it may still happen; but I agree it would be very bad
audreyt I try to not let very bad things happen if I can help it. :) 17:18
pmichaud best avoided
pmurias hi fglock
pmichaud I'm not sure what you mean exactly by "write in parsec style" 17:19
gaal off to dinner with a moose & 17:21
PerlJam pmichaud: He chose though primitives because they are isomorphic to haskell code structure (and semantics). So he can continue writing code that looks and acts like haskell, but is really p6ish. (At least that's what I think)
pmichaud perljam: but emulating haskell may not be the best thing to do for the perl 6 grammar in general 17:22
audreyt er, not emulating haskell at all
emulating parsec, yes. 17:23
namely, rules should return the thing they match
pmichaud okay, emulating parsec might not be the best thing to do for the perl 6 grammar in general then :-)
audreyt in addition to the hash of arrays of hashes of how they match
(or, if we are faily sure we don't care how they match, discard that also)
17:23 lisppaste3 joined
audreyt s/faily/fairly/ 17:24
if I asked for a <number>, I should be able to numify it 17:25
using (some subset of) p6 alone
not forced to write verbose code in each runtime's native assembly.
(as is the case for p5 in PCR and pir in PGE)
(and Hs in Parsec and whatnot in js.) 17:26
if it's not a good idea for perl6 grammar... I want to know why :)
pmichaud I agree that it will be good to know why. I'm just not wanting anyone to get the impression that I've committed to any particular approach 17:27
I do think that implementing the perl 6 primitives is an excellent idea
PerlJam audreyt: forget other runtimes! Parrot is all that matters for perl6 ;-)
audreyt pmurias: <audreyt> ... experiment with the style ... :) 17:28
er sorry
pmichaud: <audreyt> ... experiment with the style ... :)
pmichaud good enough for me :-)
I'm going to now return to studying tge 17:29
audreyt enjoy :)
(and I'm going back to sleep)
gaal: I'll debug the YAML::Syck thing later... can you mail the files that I'm supposed to wget to my mailbox?)
pmichaud yes indeed. I really was working on ways to fix up the tge syntax, so this was quite timely -- thanks, audrey
audreyt np :) I'm very, very happy you are on IRC. 17:30
fglock back
pmichaud yes, I've gotten a lot of my other (paying) items off of my plate, so I have more IRC-time available again
audreyt fglock: see backlog for not duplicating code between the three rule runtimes 17:31
pmichaud <audreyt>(summary: experiment with the style of midstream {return...} with a subset of p6)
fglock backlogging... 17:32
audreyt s/duplicating code/forking perl6 grammars/
gaal audreyt: sent
audreyt gaal++ # thx
fglock: also, for the runtime, I've got clkao doing perl5/Data-Bind/ 17:33
pmichaud audreyt: on thinking about it more, I'm comfortable with single-{} for p6prim
audreyt pmichaud: woot
fglock: so p6 calling conventions (that := thing) can reasonably be emulated on p5 space 17:34
more details later after I get some sleep at last :)
pmichaud audreyt: sleep well
audreyt (also Matz will be joining us in realspace tomorrow, and Sasada in cyberspace too about porting p6 callconv to YARV) 17:35
PerlJam audreyt: how much do you sleep btw? You were awake before I went to sleep last night and I assume you haven't slept in the mean time.
audreyt PerlJam: 3hr and 4hr respectively
was also <6 for the week before that
so I'm deeply sleep deprived
fglock the idea is to use the underlying implementing language in return blocks?
audreyt fglock: the idea is to _not_ use the underlying implementing language
pmichaud fglock: the idea is to have an easily-implemented subset of perl 6
gaal seeply dleep predrived 17:36
fglock so it's ok to source filter (for example)?
pmichaud whatever the underlying implementation wants to do to handle that subset is up to the implementor
audreyt fglock: sure, preferably Filter::Simple::Cached
fglock: but the idea is to scan for {...} and srcfilter (kluge, whatever) it into the :P5 style you are using 17:37
and find a small subset of p6 that this transform is 1)easy 2)fast
pmichaud: exactly.
pmichaud and limit ourselves to that subset in the shared grammar
fglock ok - that's easy for p6->p5 (lrep already works this way) 17:38
audreyt fglock: yes, guess where I drew my inspiration from :)
pmichaud yes, and I think it'll be relatively easy to do in Parrot
audreyt but the point is so I can port parsec in Pugs.Lexer and Pugs.Parser
and Pugs.Rule.Token
automagically into this kind of rules
so we get a parse-perl6-the-way-pugs-does-it-now
fglock there are some things in PGE that are not implemented yet in PCR - like <after...> and char classes - should I implement those first?
audreyt fglock: lookaheads are probably going to be useful... but not critical. char classes, definitely. 17:39
pmichaud for the statement_end rule I just used <after> because it was handy -- there may be a better way to do it
(after is a lookbehind)
audreyt oh right. they take reverse meanings
parsec doesn't do lookbehinds, so my transformation won't generate them 17:40
fglock a shared grammar is definetly handy - and it works as a living specification
audreyt (it also commits by default for quantifiers, so the transformation will have a lot of : symbols)
fglock how about the p6 AST?
pmichaud committing is a good thing
audreyt fglock: I need to study your document about it tomorrow 17:41
fglock audreyt: you mean ruoso's?
audreyt er, right
I was deep in ruby's
(they have this NODE structure that serves as AST) 17:42
(and YARV compiles it to instructions (insn) kinda like PBC)
and the design seems to be very sane.
the NODE is fairly removed from the parse tree 17:43
so much so that there's no trivial B::Deparse from it
(although certainly would be still pa^Hossible)
so I agree with pmichaud's idea that although a rule can return an AST, its Match is still valuable
eg for other languages that want to parse a chunk of code "as perl 6sees it" but want to generate other ASTs from the match structure. 17:44
17:44 manbou joined, KingDiamond joined
audreyt (currently the Parsec implementation doesn't have this information, which makes debugging kinda frustrating at times) 17:44
pmichaud we definitely cannot throw away the match structur
e
:keepall requires it
and there are other reasons for having it. Not everything is an AST :-) 17:45
audreyt sure, by using rules we are already committing to not throw away the match:)
pmurias audreyt: any docs on the NODE structure?
audreyt (that's why this result object thing gets introduced during Mt.Arbel hackathon as a parallel-track info channel)
pmichaud btw, I really like the result object
good job to whoever thought it up 17:46
took all of about 20 minutes to add to pge :-)
audreyt curtsies
though the sugar around it is primarily TimToady's (and also lambdamoose's) :) 17:47
pmichaud I still need to fold in the re-wording revisions to S05 for that
audreyt (sugar, as in automagic coercion and <(...)>)
nod
pmichaud I still need to implement <( and )>, too 17:48
lunchtime here
back after lunch
audreyt pmurias: www.namikilab.tuat.ac.jp/~sasada/pr...nodes.html
fglock goes back to class in 1h :( 17:49
audreyt wow, 3am. 17:50
I probably shouldn't go on awake like this :)
audreyt waves and tries to get some sleep
fglock exactly 12h from here - you are in the other side of the world
or I am
audreyt we both are
:)
17:51 siosiosios joined 18:16 arcady joined 18:21 mewo2 joined
mewo2 Anybody about? 18:22
integral hello!
mewo2 Ah, hello. 18:23
integral a maths student?
mewo2 Ex-student, but yes.
You don't happen to know how long Pugs' make smoke is supposed to take, do you?
integral 100-300 minutes I'd guess 18:24
what's your proc speed and memory?
(m19s28.vlinux.de/cgi-bin/pugs-smokeserv.pl is the current smoke report page)
mewo2 1.2 GHz G4 PowerPC, 512 MB RAM.
It's been running for about 10 hours now. 18:25
integral ah, yes, it does take ages on that hardware. 18:26
I've a 1GHz PPC 1.25GB box, and I've not had it finish in any reasonable time yet :-)
there's a smoke report at 1605 minutes 8-)
mewo2 Ah, fair enough. 18:27
integral ppc is unfortunately quite poor for haskell performance
mewo2 Will let it do its thing then.
Yeah, I've noticed that in the past.
Not sure my compiler-fu is strong enough to do anything about it. 18:30
18:32 fglock left
mewo2 It doesn't seem to be using any CPU though. 18:35
18:38 fridim joined
fridim hi 18:39
integral mewo2: hmm, there may be an infinite loop. it happens occasionally, which test? 18:41
18:41 larsen_ joined
mewo2 ext/libwww-perl/t/getprint.t 18:41
It's been sitting on that for at least an hour. 18:42
integral hmm, yes, I get a hang without any apparant output, weird 18:43
mewo2: hmm, it finally ran for me. Choked on my proxy settings though 18:50
mewo2 On PPC, or on something else? 18:51
integral yeah ppc/darwin 18:52
mewo2 I don't have any proxy settings, so it shouldn't be that.
integral but it only took a few minutes, not hours :-/ 18:53
personally I'd kill that particular test now
mewo2 Yeah, I think it's had its chance.
19:03 rindolf joined
rindolf Hi all! 19:03
audreyt: here?
PerlJam rindolf: asleep
rindolf PerlJam: I see. 19:04
19:10 zellyn joined 19:11 zellyn left 19:13 kral joined, heng joined 19:16 marmic_ joined 19:17 nnunley joined 19:28 zgh joined 19:34 KingDiamond joined 19:39 avar joined
pmurias rindolf: hi 19:57
audreyt: thanks, the reference proved helpfull 20:02
20:24 larsen joined 20:28 ruz_ is now known as ruz 20:33 marmic joined
pmurias heard the audreyt podcast at perlcast.com 21:04
theorbtwo Oooh, I should remember to do that on the next commercial break. 21:09
21:12 iblechbot joined
theorbtwo Wow, audrey's audio is /really/ bad. 21:14
21:14 SamB joined 21:19 tomyan left 21:31 bernhard joined
Juerd 4444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444444 21:35
LeTo thu shallst nevva sleep on thye keybrd 21:37
21:44 nothingmuch joined
nothingmuch møøse 21:44
21:44 ruoso joined
ruoso se pergunta se deve seguir o BOFH e incentivar isso 22:03
oops
22:32 pmichaud left 22:46 mugwump_ is now known as mugwump 22:52 drrho joined
mugwump builds ghc6 23:36
ghc-pkg: cannot find package plugins no such key: 'rpath_blib' 23:38
Are those errors known ? That's on ghc 6.4
23:39 dormon joined 23:42 siosiosios joined 23:54 mugwump joined