Perl 6 - the future is here, just unevenly distributed

IRC log for #perl6, 2007-09-10

Perl 6 | Reference Documentation | Rakudo

| Channels | #perl6 index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:29 Lunchy joined #perl6
00:41 nothingmuch gaal_: moose
01:23 agentzh gaal, nothingmuch: elk :)
01:26 nothingmuch hola agentzh
01:26 nothingmuch long time no chat
01:51 bsb moosen
01:59 agentzh nothingmuch: *nod*
02:01 nothingmuch hola bsb, likewise =)
02:08 jisom joined #perl6
02:12 weinig joined #perl6
02:17 f0rth joined #perl6
02:22 Squern joined #perl6
02:40 REPLeffect joined #perl6
03:05 Partizx joined #perl6
03:17 lyokato joined #perl6
03:23 amnesiac joined #perl6
03:42 buetow joined #perl6
04:14 gaal yo yo yo
04:15 gaal ell oh ell, it takes ghc 6.6.1 apparently to compile ghc g.7 nowadays
04:17 allbery_b yep.  annoying
04:18 gaal as long as it doesn't take 6.8, we're good
04:19 allbery_b actually, last I tried it, it wouldn't build itself
04:19 gaal also i noticed I typed "g" instead of "6". it may be the four hours of driving, but I think my typing may be deteriorating
04:19 gaal that's what happens when you neglect your haskell
04:46 IllvilJa joined #perl6
05:16 xinming_ joined #perl6
05:22 devogon joined #perl6
05:44 Ashizawa joined #perl6
05:46 baest joined #perl6
05:48 masak joined #perl6
06:04 GeJ joined #perl6
06:32 JarJarBi1ks joined #perl6
07:22 elmex joined #perl6
07:29 iblechbot joined #perl6
07:33 Aankhen`` joined #perl6
07:35 drrho joined #perl6
07:39 flokuehn joined #perl6
07:44 Daveman joined #perl6
07:45 rfordinal_ joined #perl6
08:09 moritz_ good morning everybody ;)
08:11 masak good morning moritz_
08:26 DarkWolf84 goodmorning moritz_
08:28 nipra joined #perl6
08:28 ruz joined #perl6
08:34 cognominal_ joined #perl6
08:34 franck___ joined #perl6
09:02 fglock joined #perl6
09:22 Juerd Heh, I made a mistake in crontab
09:22 Sizur joined #perl6
09:22 Juerd Which made feather0 start the backup program every hour...
09:22 Juerd It doesn't finish in an hour
09:22 Juerd So 3 backup processes were running.
09:22 Juerd Has anyone noticed that feather became slower?
09:22 moritz_ ;)
09:22 Juerd :)
09:23 moritz_ I didn't use it ;)
09:23 Juerd I did, and haven't noticed
09:23 moritz_ cool ;)
09:23 masak joined #perl6
09:25 Juerd I discovered it only because I saw the almost-continuous 60 Mb/s data flow in the charts
09:29 chris2 joined #perl6
09:49 pugs_svnbot r17732 | fglock++ | [kp6] fixed namespace in APPLY
09:49 pugs_svnbot diff: http://dev.pugscode.org/changeset/17732
09:49 lambdabot Title: Changeset 17732 - Pugs - Trac
09:56 Juerd http://juerd.nl/berg.png  # feather0 backup gone bezerk
09:58 moritz_ that looks a lot like RRD-Tools ;)
09:59 moritz_ or collectd (which uses rrdtools): http://collectd.org/
09:59 lambdabot Title: collectd - The system statistics collection daemon
10:00 Juerd moritz_: cacti, which uses rrdtools :)
10:00 fglock i'm wondering - if kp6 were ported to run in pugs, how would it detect side-effects of BEGIN blocks
10:04 moritz_ fglock: why would it be different from what it is using now?
10:05 fglock kp6-perl5 uses custom-built containers which do logging, but pugs already has it's containers
10:06 moritz_ er.. what exactly do you mean by "run in pugs"?
10:07 moritz_ just that pugs executes the kp6 source files, or something else?
10:07 fglock run perl6-in-perl6 over pugs
10:07 Sizur left #perl6
10:09 fglock when 6-in-6 finds a BEGIN block, it needs to eval() it and emit code to replicate the side-effects at INIT time
10:13 fglock running in pugs has the same problem as running in bootstrapped kp6
10:14 cognominal_ joined #perl6
10:18 fglock re what it is using now - kp6 is implemented over mp6, just like pugs is implemented over haskell - this makes it very easy to access low-level data structures
10:19 fglock but 6-in-6 should work with plain perl6 data
10:20 moritz_ so the problem is that kp6 relies on the underlying semantics of mp6?
10:20 Ashizawa_ joined #perl6
10:21 fglock kp6 relies on the underlying semantics that was built using mp6/perl5/whatever, which is not plain Perl 6 semantics - it was extended to allow BEGIN blocks to work
10:22 fglock the problem is, can we do that without extending Perl 6?
10:29 fglock the kp6 extension is the Pad object:
10:30 fglock you can do Pad.eval( $code )
10:30 fglock and then Pad.side_effects returns a list of variables that were modified
10:32 moritz_ so Pad has to be implemented in pure perl 6, right?
10:32 fglock yes, in order to run in pugs
10:32 fglock otherwise it would have to be implemented in haskell
10:34 fglock which is bad, because you would also need parrot, javascript and jvm versions
10:38 moritz_ which means the desired solution is really a perl 6 one, right?
10:38 fglock yes
10:38 moritz_ so that the AST is transformed to a BEGIN-free AST
10:38 fglock the current solution is really a perl6-over-miniperl6 solution
10:39 fglock which would use pugs as a VM
11:16 pugs_svnbot r17733 | fglock++ | [kp6] .side_effects is a Pad method
11:16 pugs_svnbot diff: http://dev.pugscode.org/changeset/17733
11:16 lambdabot Title: Changeset 17733 - Pugs - Trac
11:45 theorbtwo joined #perl6
11:47 kanru joined #perl6
11:52 IllvilJa joined #perl6
11:54 |Jedai| joined #perl6
11:56 drupek12 joined #perl6
12:00 cognominal_ joined #perl6
12:06 theorbtwo joined #perl6
12:13 franck___ joined #perl6
12:14 franck___ joined #perl6
12:15 ruoso joined #perl6
12:18 polettix joined #perl6
12:31 franck___ joined #perl6
12:38 * ruoso notices a big changeset after 4 days off
12:38 * ruoso also notices run_tests_kp6_kp6 fails some tests now
12:38 pugs_svnbot r17734 | fglock++ | [kp6] Pad.eval_ast() - refactor some vm-specific code out of COMPILER
12:38 pugs_svnbot diff: http://dev.pugscode.org/changeset/17734
12:38 lambdabot Title: Changeset 17734 - Pugs - Trac
12:39 moritz_ ruoso: it never worked completeley
12:39 ruoso moritz_, yeah it didn't
12:39 ruoso s/didn't/did/
12:39 moritz_ which revision?
12:39 ruoso it was working last week
12:39 fglock ruoso: i think Pad can be used to abstract the VM
12:39 ruoso fglock, yep... that's the core idea of that diagram I wrote.
12:40 fglock ah :)
12:40 ruoso fglock, but you think we should do that on kp6 already?
12:40 fglock yes
12:40 ruoso shouldn't we leave that for kp6+1
12:40 ruoso ?
12:40 fglock we need a working prototype
12:41 drupek12 joined #perl6
12:41 jhorwitz joined #perl6
12:41 ruoso my idea was to do that using kp6 as runtime
12:42 fglock ruoso: you can still use kp6 as a runtime, this is just internals
12:43 ruoso fglock, I mean, changing that in kp6 would be a large refactoring
12:56 fglock the compiler structure will not change, i'm just moving the vm-specific code to a single place
13:06 cmarcelo joined #perl6
13:09 awwaiid joined #perl6
13:18 cognominal_ joined #perl6
13:48 vel joined #perl6
13:56 ferreira joined #perl6
14:03 amnesiac joined #perl6
14:08 masak joined #perl6
14:21 pugs_svnbot r17735 | fglock++ | [kp6] finish separating mp6/kp6 runtimes
14:21 pugs_svnbot diff: http://dev.pugscode.org/changeset/17735
14:21 lambdabot Title: Changeset 17735 - Pugs - Trac
14:30 zamolxes joined #perl6
14:31 fglock obra: ping
14:34 TJCRI joined #perl6
15:02 weinig joined #perl6
15:05 hcchien joined #perl6
15:13 iblechbot joined #perl6
15:13 pmurias joined #perl6
15:13 pmurias hello
15:13 fglock hi
15:14 masak howdy
15:16 araujo joined #perl6
15:23 pmurias fglock: will need to specify whether GLOBAL:: gets prepended to modules names when they are encoded into perl5 modules
15:24 pmurias if 'module Foo' ends up GLOBAL::Foo or Foo
15:24 fglock it's just Foo
15:25 pmurias so it's a bug in Global that vars get mangled?
15:26 pmurias nice :) it's fixed
15:26 masak that's what I call service
15:33 pugs_svnbot r17736 | pmurias++ | [kp6] 51-Test.pm GLOBAL:: is no longer appended unnecessarly
15:33 pugs_svnbot diff: http://dev.pugscode.org/changeset/17736
15:33 lambdabot Title: Changeset 17736 - Pugs - Trac
15:35 rafl joined #perl6
15:36 pugs_svnbot r17737 | pmurias++ | [kp6] GLOBAL::Test renamed to Test
15:36 pugs_svnbot diff: http://dev.pugscode.org/changeset/17737
15:36 lambdabot Title: Changeset 17737 - Pugs - Trac
15:40 REPLeffect joined #perl6
15:45 DarkWolf84 joined #perl6
15:48 broquaint joined #perl6
15:52 f0rth joined #perl6
15:53 fglock pmurias: there is a problem with using kp6 as perl6-in-perl6, that it requires the containers to set the modified-flag
15:54 fglock so we can't just run kp6 over pugs, for example
15:57 pmurias fglock: one would need wrapers over pugs containers
15:57 fglock but then, you would be using pugs as just another VM
15:58 pmurias would kp6 under pugs generate perl5 code?
15:59 moritz_ if you don't modify it any further: yes
15:59 fglock BEGIN blocks would generate perl6 code, but the compiler would generate whatever
15:59 pmurias i see
16:00 pmurias what's the benefit in running kp6 on pugs?
16:01 fglock it's the same thing as running kp6 on kp6 - it's 6-in-6
16:01 fglock running on pugs keeps it honest
16:01 fglock no kp6-specific hacks
16:03 fglock i think there is something wrong with requiring a layer between kp6 and perl6
16:05 pmurias fglock: mental layer?
16:05 fglock i mean, we should be able to use unmodified Perl 6 containers
16:07 pmurias one could keep a md5 checksum for every variable and check if it's modified
16:08 pmurias not sure how fast/slow would it be
16:08 moritz_ if I'd compile a program like this: 'my $time = BEGIN { time }' in Perl 6, should $time be startup time of the program, or the compilation time?
16:08 fglock compilation time
16:08 moritz_ ok
16:08 PerlJam (compilation time as of the execution of that statement :)
16:09 pmurias fglock: or have all access to outer lexical be made via COMPILING which would return wrapers
16:09 fglock also:  my $time; BEGIN { $time = time }
16:09 moritz_ right
16:10 moritz_ Juerd, diakopter: the SVN web front end seems broken
16:10 moritz_ Warning: License check failed! in /data/home/audreyt/pugs/docs/feather/repos/errors/404/index.php on line 0
16:10 moritz_ at least for 404 errors
16:10 barney joined #perl6
16:11 Juerd whuh
16:11 Juerd license... check...?
16:11 fglock pmurias: i think the current algorithm is efficient enough, the problem is that it requires more access to the VM than Perl 6 supports
16:12 Juerd ...failed...?
16:12 Juerd moritz_: url?
16:12 moritz_ Juerd: http://svn.pugscode.org/pugs/src/perl6/Perl-6.0.0-STD.pm
16:13 Juerd I really have no idea which thing requires this license, who got it, and why it's broken
16:13 moritz_ it's that fancy svn frontend
16:14 moritz_ I think diakopter++ installed it, but I'm not entirely sure
16:14 moritz_ http://moritz.faui2k3.org/irclog/out.pl?channel=perl6;date=2007-07-11#i_60074
16:14 lambdabot Title: IRC log for #perl6, 2007-07-11, http://tinyurl.com/252nfb
16:15 Juerd diakopter: I think you can fix this :)
16:15 pmurias fglock: the methods i suggested are probably any faster but they would require less support from the vm
16:15 Juerd (I hope you can...)
16:16 pmurias fglock: at least the checksum method
16:16 fglock pmurias: we could the way you implemented Bind in the perl5 emitter
16:16 fglock calling MODIFIED explicitly
16:17 diakopter Juerd: oh; heh.  I guess they don't want to renew our license. :(
16:17 diakopter <sigh>
16:17 pmurias fglock: MODIFIED is called explicitly only for Bind
16:17 fglock s/could/could do it/
16:17 moritz_ diakopter: but why is it just for 404 errors?
16:17 pmurias i thought other ways of modification are more difficult to handle
16:17 fglock yes, but in 6-in-6 we'd have to call it explicitly every time
16:17 [particle] moritz_: that file was renamed, is that part of the problem?
16:17 pmurias k
16:18 pmurias i don't think kp6-in-pugs is a priority
16:18 pmurias ?
16:18 pmurias s/?//
16:18 moritz_ [particle]: no, arbitrary non-existing files produce that error
16:18 pmurias it would restrict kp6 code base to kp6/pugs feature set intersection
16:19 moritz_ pmurias: don't do that, pugs has some weird OO bugs
16:19 Juerd diakopter: The license file suggests it should expire not before October...
16:20 fglock i mean kp6-in-perl6, where perl6 = any( kp6, pugs, p6parrot )
16:20 diakopter Juerd: well, I disabled it for now anyway
16:21 diakopter back to boring old non-xslt html
16:21 diakopter ;)
16:21 PerlJam Why is it that we need a "browser interface" to subversion?
16:22 moritz_ PerlJam: no need, it's just nice to have eye candy ;)
16:23 fglock pmurias: it should work on anything that can eval perl6 code (so we can build Pad objects)
16:26 pmurias fglock: i think kp6-in-kp6 is the most important right now
16:26 diakopter PerlJam: I'm afraid you weren't around to voice any objections to the proposal to implement that, back when I asked.
16:27 pmurias the ran on any( kp6, pugs, p6parrot) would IMHO require a lot of workarounds
16:27 pmurias as they each implement a different feature set
16:27 fglock i don't think you can go very far without portability
16:28 PerlJam diakopter: Had I been around, I would have asked the same question  :)
16:30 pmurias fglock: does portablility across diffrent perl6 compiler actually give us anything
16:30 pmurias ?
16:31 pmurias * compilers
16:31 moritz_ pmurias: robustness
16:32 pmurias how?
16:32 fglock it gives you... perl6-in-perl6
16:33 pmurias pugs and p6parrot aren't perl6 for that definition of perl6
16:34 pmurias as they don't implement the whole synopsis
16:34 moritz_ but kp6's aim is to implement p6 in a subset of p6, afaict
16:34 moritz_ and p6parrot would give you speed, presumably ;)
16:35 fglock yes, and we've achieved that goal
16:35 fglock now it's time to start 6-in-6
16:35 fglock which is not exactly kp6-in-kp6
16:37 pmurias completing the bootstrap and changing the name is a good starting point IMHO
16:37 fglock bootstrapped kp6 is still not 6-in-6
16:37 pmurias why?
16:38 fglock because 6-in-6 should not depend on vm-specific code
16:38 pmurias fglock: you have to, to some extend
16:39 pmurias emitters have to be vm-specific
16:40 pmurias targeting a second backend would seperated it more cleanly
16:40 fglock pmurias: i guess i'll start an experiment
16:41 pmurias fglock: what sort of?
16:41 fglock targetting pugs is good because you can't easily sneak haskell code in
16:41 pmurias emitting pugs code?
16:42 fglock i'll try writing the kp6 low level libs over pugs
16:43 pmurias ok
16:44 weinig joined #perl6
16:46 obra fglock: pong
16:46 obra sorry for the delay
16:46 fglock np
16:48 fglock obra: http://svn.pugscode.org/pugs/v6/v6-KindaPerl6/docs/IDEAS-HACKATHON
16:48 lambdabot http://tinyurl.com/ys259y
16:48 obra looking
16:49 obra so. you'd talked about there being 3 possible ways forward in your mail
16:49 fglock yes,
16:50 fglock follow the roadmap as-is - fix some bugs, merge with the STD grammar, then optimize
16:50 fglock or, bootstrap 6-in-6 before proceeding with the plan
16:50 fglock or work on the parrot backend
16:51 obra ok. why is your confidence in the existing roadmap shaken?
16:51 fglock there is a new idea, which is to bootstrap over pugs
16:52 fglock i noticed kp6 needs a new container model, because what we have today is too tied to the underlying vm
16:52 fglock this is a portability problem
16:53 fglock and it makes it difficult to have 'real' perl6-in-perl6
16:53 fglock the current roadmap doesn't address that
16:53 obra *nod*
16:54 justatheory joined #perl6
16:54 obra Personally, I'd love to see an implementation that's actually fully using the standard grammar.
16:54 obra but I want to understand what really makes sense for kp6
16:54 fglock we could just fix this small problem, and then proceed with the plan
16:56 fglock ruoso and pmurias really want a bootstrapped kp6
16:56 fglock and i don't want to stop them :)
16:56 obra ok.
16:56 obra Good :)
16:56 mattwg joined #perl6
16:56 obra fixing the small problem seems clever.
16:57 obra proceeding with the plan rather than getting distracted by something else shiny feels clever, too
16:57 obra unless you feel like the current plan will end in sadness
16:58 fglock it could make sense to fork the project for a while
16:58 * [particle] wishes pugs repo supported branches
16:58 fglock and work both in bootstrapping and STD
16:59 fglock it's kind of forked already
16:59 [particle] but not so easy to manage merges
17:01 obra how much forking is really required for both projects?
17:02 ruoso [particle], I'm having success with the two branches merging so far
17:02 ruoso because the codes aren't that differente
17:02 nothingmuch /w 24
17:03 fglock we've been merging back and forth in the last week or so
17:03 ruoso obra, the fork only happens in the runtime part
17:03 ruoso as kp6-mp6 uses explicitly mp6 runtime
17:04 ruoso while kp6-kp6 uses kp6 runtime
17:04 ruoso which means calling conventions and object meta protocol
17:04 ruoso s/object meta/meta object/
17:04 obra Is there a good visiolization of this?
17:05 ruoso obra, not really...
17:05 ruoso everytime we draw some diagram
17:05 ruoso it looks like kp6+1
17:05 ruoso instead of document kp6 itself
17:06 fglock obra: it's still compatible with http://svn.pugscode.org/pugs/v6/docs/kp6.jpg
17:07 ruoso fglock,
17:07 ruoso hmm
17:07 ruoso after merge I get a Undefined subroutine &Main::say called at lib-kp6-mp6-p5//KindaPerl6/Grammar.pm line 64,
17:07 fglock ruoso: i think i found a way to make the runtime compatible with plain perl6, so it would possibly run in pugs
17:08 ruoso fglock, hm?
17:08 fglock ruoso: looking
17:08 ruoso kp6-mp6 fails one test right now
17:08 fglock ruoso: we can call 'MODIFIED' explicitly, instead of hiding it inside the Container code
17:09 ruoso kp6-kp6 fails in the use of lib-modules
17:09 ruoso and it's because we need to compile lib-modules to two different targets
17:09 ruoso we need a lib-modules-kp6-p6
17:09 ruoso p5
17:09 ruoso we need a lib-modules-kp6-p5
17:09 ruoso and a lib-modules-mp6-p5
17:10 fglock ruoso: i'd rather have completely separate directories, this is confusing me
17:10 ruoso fglock, but lib-modules will be the same
17:10 ruoso it's just the compiled versions that are different from each other
17:10 ruoso I mean
17:10 ruoso they shouldn't
17:10 ruoso hmmm
17:11 ruoso why is test 51 failing with kp6-kp6?
17:13 ruoso fglock, I was wrong... there's no need for two lib-modules
17:13 fglock obra: i think we will proceed with the plan *and* get distracted
17:14 obra sounds good :)
17:14 fglock we probably need a roadmap for kp6-kp6
17:15 ruoso fglock, that's just "create tests that reproduces the features of kp6-kp6"
17:15 pugs_svnbot r17738 | ruoso++ | [kp6] merge kp6-mp6 to kp6-kp6 -r17697:HEAD
17:15 pugs_svnbot diff: http://dev.pugscode.org/changeset/17738
17:15 lambdabot Title: Changeset 17738 - Pugs - Trac
17:19 ruoso fglock, it's kinda hard to write down a ROADMAP (besides what is in docs/KP6-Bootstrap/Notes
17:19 ruoso fglock, it's kinda hard to write down a ROADMAP (besides what is in docs/KP6-Bootstrap-Notes.txt
17:19 ruoso because we don't know exactly what's missing
17:19 ruoso I really think it's just a matter of converting the runtime files
17:20 ruoso to the new calling convention
17:24 pugs_svnbot r17739 | fglock++ | [kp6] fixed kp6-mp6 runtime
17:24 pugs_svnbot diff: http://dev.pugscode.org/changeset/17739
17:24 lambdabot Title: Changeset 17739 - Pugs - Trac
17:28 fglock ruoso: re MODIFIED, how would kp6 run in pugs, for example?
17:29 ruoso pugs would have a namespace management that would take care of it
17:30 ruoso and parrot already does it
17:30 ruoso the Emitter (if any) should take care of getting that values for the specific runtime
17:31 ruoso in kp6+1 we would implement that completely in p6
17:31 ruoso and each runtime would optimize it for its own runtime
17:32 ahmadz joined #perl6
17:32 ahmadz hi there
17:32 moritz_ hi ahmadz
17:32 ahmadz hi moritz ;-)
17:33 Psyche^ joined #perl6
17:33 ahmadz quick question: i tried to compile pugs today and it told me i need ghc 6.6.1 [i had 6.6 before and it used to work]
17:34 moritz_ ahmadz: things change ;)
17:34 DarkWolf84 ~from some time
17:34 fglock ruoso: so we would use MODIFIED for backends that don't support it natively
17:34 ahmadz yeah they sure do ;-)
17:34 DarkWolf84 I had such problem
17:34 ahmadz ghc take a lot of time to compile...
17:35 DarkWolf84 get the binary then
17:35 moritz_ ahmadz: DarkWolf84 is right ;)
17:35 ruoso fglock, but that would be used just in that backend's implementation
17:37 fglock which is true for kp6-perl5 and for a possible kp6-pugs and kp6-p6parrot
17:37 ruoso not for kp6-parrot
17:37 ruoso parrot takes care of it by itself
17:37 fglock i mean kp6-p6parrot
17:37 [particle] :)
17:37 ruoso a
17:37 ruoso I see
17:38 fglock perl6-in-parrot is just plain perl6
17:38 fglock it doesn't keep track of side-effects
17:39 ruoso parrot does
17:39 ruoso yep
17:39 ruoso the same for kp6-parrot
17:40 fglock so 6-in-6 will always need an emulation layer
17:40 ruoso yes...
17:40 ruoso kp6 would be a runtime provider for p6-p6
17:42 ferreira obra: ping
17:42 ruoso kp6-parrot would provide a runtime for pk6+1 to run on top of parrot
17:42 obra ferreira: pong
17:43 ferreira I thought about a proposal for a Perl 6 microgrant, but I am not sure about its validity and details.
17:44 obra rough description?
17:44 ferreira Could you take a look before I officially submit it to perl6-microgrants@perl?
17:44 obra sure, though that just goes to me and acme :)
17:44 obra submitting a bad one won't result in us turning you down ;)
17:45 obra or rather
17:45 obra it won't stop you from being able to submit a new one after our response
17:45 ferreira well, then I will post it. But I would look to keep it open for finishing details.
17:46 PerlJam ferreira: I'd like to read it :)
17:46 obra ok
17:47 PerlJam obra: what's the turn-around time on responses?
17:47 fglock obra: in short, i will keep the roadmap as-is, and the big changes will go into an alternate implementation and will be merged later
17:50 ahmadz is it normal to get a lot of warnings when compiling pugs on feather?
17:50 moritz_ I get some on my maschine
17:50 moritz_ about 20..30 lines per compiled file
17:51 obra fglock: seems sane
17:51 obra PerlJam: varies from 5 minutes to never
17:51 ferreira obra: I posted it
17:52 ruoso fglock, i found why the 51-Test was failing
17:53 ruoso the -Ilib-modules-p5 was missing in the run_tests calll
17:54 ruoso but there's still a test failing in both kp6-mp6 and kp6-kp6
17:54 pugs_svnbot r17740 | ruoso++ | [kp6] fixed run_tests_kp6_kp6 to -Ilib-modules-p5
17:54 pugs_svnbot diff: http://dev.pugscode.org/changeset/17740
17:54 lambdabot Title: Changeset 17740 - Pugs - Trac
17:55 obra ferreira: will discuss with acme
17:56 ferreira thanks
17:57 vel joined #perl6
17:57 ruoso fglock, BTW... kp6 can't compile its own grammar anymore
17:58 ruoso fglock, the problem is in lit_code
17:58 ruoso ops
17:58 ruoso lit_object
17:59 ruoso does it means the ::Object() constructor ?
17:59 fglock looking
18:00 fglock ruoso: this seems to be unchanged
18:07 ruoso fglock, take a look at the new test
18:08 ruoso grammar/08
18:08 ruoso it fails with a weird symptom
18:08 pugs_svnbot r17741 | ruoso++ | [kp6] a new test failing with a weird symptom
18:08 pugs_svnbot diff: http://dev.pugscode.org/changeset/17741
18:08 lambdabot Title: Changeset 17741 - Pugs - Trac
18:12 fglock ruoso: the named params implementation was not finished
18:12 ruoso fglock, is ::Lit::Object a new object?
18:12 fglock yes, it calls the constructor
18:13 fglock currently parses using mp6-ish syntax
18:13 ahmadz_ joined #perl6
18:13 ferreira pugs: my @a = <a b c>; my @x = <x y>; @a Z @x
18:13 exp_evalbot OUTPUT[(("a", "x"), ("b", "y"))␤]
18:14 DarkWolf84 now it works with :
18:14 DarkWolf84 ?
18:14 ferreira pugs: my @a = <a b c>; my @x = <x y>; @a Z (@x, *)
18:14 exp_evalbot OUTPUT[(("a", "x"), ("b", "y"), ("c", Inf))␤]
18:15 ruoso fglock, it seems that ::Lit::Object doesn't have a method "true" that returns true
18:15 DarkWolf84 strange
18:15 DarkWolf84 why is inf there
18:16 ferreira It must have been ( ("a", "x"), ("b", "y"), ("c", "y") ) by the rule of extending arrays. Or not?
18:17 DarkWolf84 hm
18:17 fglock ruoso: why do you need .true?
18:18 ruoso fglock, wait... maybe I had a wrong premise
18:18 ruoso I thought it was matching
18:18 ruoso but maybe the match is indeed failing
18:19 ferreira Also, in run.pugscode.org, evaling <a b c> Z <x y> gives a wrong answer ( (<a x>), (<b y>), ('c', undef) ). That means pugs there is outdated while the Perl 6 bot may be using newer (fixed) sources.
18:20 DarkWolf84 ?eval my @a = <a b c>; my @x = <x y>; @a Z (@x, *);
18:21 DarkWolf84 why evalbot is not responding?
18:22 ferreira which implementation evalbot uses?
18:22 DarkWolf84 maybe the old evalbot is gone for some reason
18:23 ferreira if the extended list (@x, *) is not treated lazily, it recurses forever. pugs (in run.pugscode.org) also has problems with 1..*
18:23 moritz_ pugs: my @a = <a b c>; my @x = <x y>; @a Z (@x, *);
18:23 exp_evalbot OUTPUT[(("a", "x"), ("b", "y"), ("c", Inf))␤]
18:24 ferreira pugs: my @ints = (1..*); @ints.first
18:24 exp_evalbot OUTPUT[pugs: out of memory (requested 1048576 bytes)␤]
18:24 ferreira well, here is the same. It probably is trying to flatten the infinite list.
18:24 moritz_ pugs: my @ints := (1..*); print @ints[0]
18:24 exp_evalbot OUTPUT[1]
18:25 moritz_ pugs: my @ints = (1..*); print @ints[0]
18:25 exp_evalbot OUTPUT[pugs: out of memory (requested 1048576 bytes)␤]
18:25 ruoso fglock, it seems that the problem is in the val_int toke
18:25 ruoso token
18:25 ferreira pugs: my @ints := (1..*); @ints.first
18:25 exp_evalbot OUTPUT[pugs: out of memory (requested 1048576 bytes)␤]
18:26 moritz_ ferreira: it seems that the lazyness of lists stands on shaky grounds
18:28 danga joined #perl6
18:28 ferreira I thought this laziness was for free when writing pugs in Haskell. I believed we should only take care to not do something that flattened the infinite list, like producing one as the final result which would be dumped (crashing the interpreter).
18:29 moritz_ ferreira: I think it's not so easiy to use builtin types for p6 land...
18:29 REPLeffect joined #perl6
18:29 moritz_ ferreira: it all has to be mapped to pads etc
18:30 riffraff joined #perl6
18:30 ferreira fglock: how about the construction of these infinite at v6, mp6, kp6?
18:30 ferreira What are those pads? I only knew the term in connection with lexical pads or something like that.
18:31 fglock ferreira: only PIL-Run supports that
18:31 moritz_ ferreira: lexical pads, yes
18:31 fglock PIL-Run = re infinite lists
18:32 moritz_ basically environment and their variables
18:32 fglock yes, Pad is the high level representation of a lexical pad, it's used by the kp6 compiler to track things defined in BEGIN blocks
18:33 fglock home &
18:34 ferreira moritz_: and... Sorry, I haven't understood why storing the objects in pads did not allow to take advantage of the Haskel features
18:34 pugs_svnbot r17742 | ruoso++ | [kp6] syntax error message more usefull
18:34 pugs_svnbot diff: http://dev.pugscode.org/changeset/17742
18:34 lambdabot Title: Changeset 17742 - Pugs - Trac
18:35 fglock s/kp6/perl6/
18:35 moritz_ ferreira: I don't either, but if it were that simple it would be implemented
18:35 moritz_ ferreira: but maybe the problem is something different...
18:36 moritz_ ferreira: btw .first is not a vaild array method
18:36 jisom joined #perl6
18:37 ferreira pugs: my @ints := (1..*); @ints.start
18:37 exp_evalbot OUTPUT[pugs: out of memory (requested 1048576 bytes)␤]
18:37 moritz_ maybe .start tests if length >= 0 ;)
18:40 rindolf joined #perl6
18:42 cmarcelo joined #perl6
19:01 Schwern joined #perl6
19:25 ferreira left #perl6
19:43 ferreira joined #perl6
19:47 cmarcelo joined #perl6
19:56 polettix joined #perl6
20:04 weinig_ joined #perl6
20:06 xinming joined #perl6
20:23 thoughtpolice joined #perl6
20:31 ferreira left #perl6
21:19 wolverian S02 should be named Types, at least in parentheses..
21:27 weinig joined #perl6
21:28 stevan_ joined #perl6
21:35 weinig_ joined #perl6
21:37 DaGo joined #perl6
21:39 Caelum what does "Setup: cannot satisfy dependency pugs-hsregex-any" mean? do I need another ghc lib?
21:43 Limbic_Region joined #perl6
21:51 TimToady @tell ferreira pugs does not yet implement Range objects, nor does it really treat * as anything other than Inf, so 0..* is pretty much doomed at the moment
21:51 lambdabot Consider it noted.
21:57 Limbic_Region TimToady - how was Vienna?
21:58 Limbic_Region oh, and you're photo gallery appears to be b0rk
22:02 Juerd your
22:08 amnesiac mine?
22:17 TimToady yeah, has been b0rk since I the last disk upgrade...lack of round tuition
22:17 TimToady Vienna was lovely
22:18 TimToady strange, but lovely
22:24 REPLeffect strange in a good, or bad way (or, just strange?)
22:26 justatheory joined #perl6
22:32 Caelum Anyone know what library I need for this to go away: "Setup: cannot satisfy dependency pugs-hsregex-any"
22:33 TimToady strange in a way that only a lot of money over a long period of time spent by many architects of violently clashing tastes can accomplish
22:34 justatheory joined #perl6
22:36 riffraff joined #perl6
22:37 justatheory joined #perl6
22:51 justatheory joined #perl6
23:30 justatheory joined #perl6

| Channels | #perl6 index | Today | | Search | Google Search | Plain-Text | summary

Perl 6 | Reference Documentation | Rakudo