Perl 6 - the future is here, just unevenly distributed

IRC log for #perl6, 2008-04-29

Perl 6 | Reference Documentation | Rakudo

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

All times shown according to UTC.

Time Nick Message
00:13 aindilis joined #perl6
00:32 stevan_ joined #perl6
00:50 AgeofRuin joined #perl6
00:51 AgeofRuin can anyone point me in the direction of a good tutorial for perl
00:53 Eevee perl 5?
00:53 Eevee or 6
00:54 AgeofRuin 6
00:55 Eevee I haven't seen anything very in-depth or tutorial-like online, but there's this: http://www.oreilly.com/catalog/059600737X/
00:55 lambdabot Title: O'Reilly Media | Perl 6 and Parrot Essentials
00:55 AgeofRuin thanks
00:56 Eevee np
01:00 alanhaggai_ joined #perl6
01:01 obra Eevee: that book is..old. and has been freed. there's a copy of it in parrot svn
01:01 Eevee oh seriously
01:02 Eevee I've just been watching parrot/perl6 from the outside so I can never tell what's up to date
01:11 Torment joined #perl6
01:14 Grrrr joined #perl6
01:24 yewenbin joined #perl6
01:25 Schwern joined #perl6
01:34 justatheory joined #perl6
01:59 [particle] joined #perl6
02:11 [particle] joined #perl6
02:42 quarkcs joined #perl6
02:49 alanhaggai__ joined #perl6
02:53 [particle] joined #perl6
03:02 [particle1 joined #perl6
03:08 [particle2 joined #perl6
03:09 [particle3 joined #perl6
03:17 d4l3k_ joined #perl6
03:18 baest joined #perl6
03:39 Auzon joined #perl6
03:47 [particle] joined #perl6
03:49 alanhaggai_ joined #perl6
04:11 [particle2 joined #perl6
04:16 [particle2 joined #perl6
04:22 [particle] joined #perl6
04:26 [particle] joined #perl6
04:28 Psyche^ joined #perl6
04:34 [particle1 joined #perl6
04:41 Zaba joined #perl6
05:03 alanhaggai joined #perl6
05:26 alanhaggai_ joined #perl6
05:34 Caelum joined #perl6
06:19 IllvilJa joined #perl6
06:44 Eidolos joined #perl6
07:03 iblechbot joined #perl6
07:05 elmex joined #perl6
07:11 jan_______ joined #perl6
07:22 jisom joined #perl6
07:23 alc joined #perl6
07:28 quarkcs joined #perl6
07:41 Zaba_ joined #perl6
07:48 Zaba__ joined #perl6
08:10 meppl joined #perl6
08:23 ruoso joined #perl6
08:51 masak joined #perl6
09:03 IllvilJa joined #perl6
10:34 wknight8111 joined #perl6
10:45 penk left #perl6
10:46 rdice joined #perl6
11:06 riffraff joined #perl6
11:07 riffraff hi
11:10 syle joined #perl6
11:19 drfaustus joined #perl6
11:20 alc joined #perl6
12:25 alanhaggai joined #perl6
12:31 masak joined #perl6
12:47 renormalist joined #perl6
12:52 alc joined #perl6
13:07 Limbic_Region joined #perl6
13:08 Limbic_Region ?eval sub foo { state $bar = "one"; my $temp = $bar; $bar = "two"; return $temp; } my $msg = foo() ~ foo(); say $msg;
13:14 Limbic_Region it has been too long - what is the proper fu to get the evalbot to run p6 code?
13:15 rdice joined #perl6
13:17 masak Limbic_Region: see the topic
13:18 masak either `pugs:`, or `kp6:`, or `rakudo:`, or...
13:19 Limbic_Region danke
13:19 Limbic_Region pugs: sub foo { state $bar = "one"; my $temp = $bar; $bar = "two"; return $temp; } my $msg = foo() ~ foo(); say  $msg;
13:19 exp_evalbot OUTPUT[oneone␤]
13:19 Limbic_Region when did pugs implementation of state variables get broken?
13:24 jhorwitz joined #perl6
13:41 TJCRI joined #perl6
13:55 iblechbot joined #perl6
14:11 alester joined #perl6
14:33 chris2 joined #perl6
14:45 yewenbin joined #perl6
14:49 mncharity joined #perl6
14:55 mncharity TimToady: re S06#Macros "syntax tree form is generally preferred because it allows the parser and debugger to give better error messages.  Textual substitution on the other hand tends to yield error messages that are opaque to the user. Syntax trees are also better in general because they are reversible, so things like syntax highlighters can get back to the original language and know which parts of the derived program come from which
14:58 mncharity textual substitution can yield good error messages - you just use a string class which remembers where its various characters came from.  Similarly, that history permits showing the user the original language and where it came from.  To get similar results from syntax trees, also have to teach them to preserve history, else you have a tree, and you know where parts were originally from, but you have no idea how they were massaged into
14:59 mncharity now I love tree macros, but find string macros are much maligned.
14:59 mncharity ;)
15:01 mncharity wrote such a string class years ago for p5.  performance wasn't even horrible.
15:04 mncharity hmm, STD.pm regex rules are unchanged from last week.  guess I'll move downstream.  rx ir and emitter.
15:14 pmurias joined #perl6
15:17 pmurias mncharity: it might make sense to turn rules into NFAs instead of turning them into code directly
15:19 mncharity hi pmurias,
15:19 pmurias hi
15:19 mncharity you mean in the current elf parser (STD_red), or in general?
15:20 pmurias i'm having doubts if it would really be a good idea
15:20 pmurias but in an ast transformation stage
15:23 mncharity I was puzzling a few minutes ago how to go forward with parsing.  some options include () create a p6 version of STD_red (handwritten or metaprogrammed subs-or-methods); ()
15:23 Schwern joined #perl6
15:24 mncharity () continue to use STD_red while STD matures; ()
15:24 mncharity ()
15:25 mncharity () try to massage STD.pm (STD_deconstructed or STD_scrape)? into a form which can be run on a regex engine.
15:26 mncharity actually, the "set up for metaprogramming" and "use engine rather than hack" are somewhat orthogonal.
15:26 pmurias metaprogrammed = code-generated
15:26 pmurias ?
15:26 pmurias ;)
15:27 mncharity yeah.  STD_deconstructed or STD_scrape -> manageable p6 code.
15:27 mncharity the equivalent of  gimme5 et al in src/perl6
15:27 mncharity and gimme
15:30 pmurias the first option dosn't seem very good
15:34 pmurias i think the choice depends on how long will Larry's effort to bootstrap STD.pm take
15:34 mncharity re first option, has advantages of low development risk (known to work, known to be fast enough), completes bootstrap (ie, so elf could become a cpan module).  disadds of lots of manual grunge work.
15:35 pmurias do you understand how to implement the longest token stuff?
15:36 mncharity re depends, indeed.  and whether the resulting performance permits it to be used for everything, as the day-to-day parser, or just as a bootstrap tool.
15:36 mncharity re "longest token stuff", no.  I've been waiting for it to work&mature before looking closely at it.
15:36 kanru joined #perl6
15:38 mncharity it's still an open question whether the longest token stuff will be a prerequisite for a "good" parse, for some value of good, or whether one can continue fudging it.  my impression is one can start making inroads into the pugs t/ while fudging it, or I might prioritize it differently.
15:39 pmurias i read the TDFA paper, but don't full understand what parts of perl6 rules get turned into a DFA and when
15:39 * TimToady is back, finally...
15:39 TimToady STD.pm has the changes in, but not checked in because I have my match object contstruction in little bits at the moment
15:40 TimToady but I could probably check in STD.pm separately
15:40 TimToady (was out yesterday at the Dr, but that's overwith now)
15:41 mncharity re depends, another issue is how large a role the ruby-ness of STD_red is playing in people not using elf yet.  If it was p6, and thus emitted p5, it wouldn't be an issue.  but if "elf isn't usefully running against t/ yet, so it's hard to tell what's working" is the bottleneck, then replacing STD_red wouldn't actually help.
15:42 mncharity re little bits, :)  np.  not worth extra effort to buy a day or two.  there is other stuff which needs banging on.
15:43 pmurias i think it's the problem of elf not being complete enough
15:44 TimToady I also changed the fate/binding parameterage, so I need to check in changes together, unless people want to see the newest STD.pm just for reference
15:44 TimToady but I'm probably still the only one actually trying to run STD5_run...
15:45 TimToady which probably means the right thing is to just check in a snapshot of where I am regardless of match breakage
15:45 [particle1 TimToady++
15:48 pugs_svnbot r20316 | lwall++ | [STD] various tweaks suggested by mncharity++
15:48 pugs_svnbot r20316 | lwall++ | [gimme et al] new parameter passing regime
15:48 pugs_svnbot r20316 | lwall++ | [cheat] now emits intermediate multi-faking rules as proto tokens
15:48 pugs_svnbot r20316 | lwall++ | [Cursor5] half-refactored match object construction
15:48 pugs_svnbot diff: http://dev.pugscode.org/changeset/20316
15:48 lambdabot Title: Changeset 20316 - Pugs - Trac
15:49 TimToady mncharity: I'm sure there are some questions you had about STD that I didn't entirely address yet, but I've forgotten the ones that didn't result in a tweak to STD
15:50 TimToady a busy weekend followed by a day under anesthesia will do that to you...
15:50 kolibrie TimToady: I tried running STD5_run once, but have not gotten back to it yet - will need to at some point, but I have no problem with it being broken for a few days
15:50 pugs_svnbot r20317 | putter++ | [elf_e] Non-scalars, @a and %h, in my() and has(), are no longer left uninitialized.  But only in elf_e - elf_e_nomoose doesn't do defaults yet.
15:50 pugs_svnbot diff: http://dev.pugscode.org/changeset/20317
15:50 lambdabot Title: Changeset 20317 - Pugs - Trac
15:51 mncharity re anesthesia, eep
15:52 TimToady ah, yes, it was regex_assertion:sym<.>, which is already handled by regex_assertion:method
15:53 meppl good night
15:56 mncharity good night meppl :)
15:57 meppl ;)
15:57 pmurias TimToady: is longest token stuff fully speced out or still being experimented upon?
15:57 justatheory joined #perl6
15:58 TimToady it's mostly specced out, though I'm still discovering some interesting holes as I implement
15:58 pugs_svnbot r20318 | putter++ | [elf] EmitRegexYare.pm sketch: expand_backtrack_macros() translated into p6.
15:58 pugs_svnbot diff: http://dev.pugscode.org/changeset/20318
15:58 lambdabot Title: Changeset 20318 - Pugs - Trac
15:58 TimToady speaking of which, I also didn't answer the []: question...
15:59 TimToady or the "veryify" question--but I think pge is probably wrong there
15:59 TimToady *verify, grr
16:00 TimToady one LTM area that's not specced yet is when a subrule can be a part of a longer token, if the subrule is partly pure and partly impure
16:00 pugs_svnbot r20319 | pmurias++ | [elf] removed .node_name
16:00 pugs_svnbot diff: http://dev.pugscode.org/changeset/20319
16:00 lambdabot Title: Changeset 20319 - Pugs - Trac
16:05 TimToady I think the meaning of []: is clear--you can't backtrack into the [], since it's the previous atom controlled by a bare ':' backtrack controller
16:06 pmurias mncharity: does /misc/pX/Common/yet_another_regex_engine work for you?
16:06 TimToady and the q about ' : ' in isolation is just the same backtrack controller applies to previous atom
16:06 pmurias mncharity: i mena pass any tests
16:06 TimToady and that's why if: is wrong, because there's no backtracking possible into 'f'
16:10 mncharity pmurias: why r20319?
16:11 mncharity re "yet_another_regex_engine work for you?", yes, on both 5.8.8 and 5.10.  not for you?
16:12 TimToady mncharity: did I answer all your q's?
16:13 TimToady well, all your STD q's, anyway...
16:13 pmurias mncharity: no
16:13 mncharity re q's, sorry, got distracted.  let's see...
16:13 pmurias mncharity: it dosn't work
16:14 mncharity re yare doesn't work, ok, that will be third on my stack to explore.
16:14 mncharity why r20319?
16:14 TimToady shave *my* yak! shave *my* ya!   :)
16:14 TimToady *yak
16:14 mncharity lol :)
16:14 pbuetow joined #perl6
16:14 TimToady take your time really, going off to read perlmonks...
16:15 pmurias if i type "prove t/re_tests.t" i get mostly errors
16:29 pasteling "mncharity" at 71.232.7.10 pasted "/usr/bin/prove t/re_tests.t 96.25% okay." (15 lines, 1.3K) at http://sial.org/pbot/30912
16:30 mncharity pmurias: what does your  perl -w Regexp_ModuleA.pm --test  output look like?
16:34 mncharity re node_name, that namespace ("CompUnit", etc) is used to describe nodes independently of what package they happen to appear in.  Eg, by IRx1_InfoForP5.pm.  Though not much code uses it yet.
16:37 lisppaste3 joined #perl6
16:43 mncharity pmurias: or you can use  perl -w Regexp_ModuleA.pm --repl  to run single cases.
16:44 mncharity http://svn.pugscode.org/pugs/misc/pX/Common/yet_another_regex_engine/README
16:44 lambdabot http://tinyurl.com/6cevls
16:52 pugs_svnbot r20320 | putter++ | [elf_e] expand_backtrack_macros() now works.  Tweaked p5 prelude regexp helpers.
16:52 pugs_svnbot r20320 | putter++ | Reverted r20319 - though not yet much used, the node_name names are needed.
16:52 pugs_svnbot diff: http://dev.pugscode.org/changeset/20320
16:52 lambdabot Title: Changeset 20320 - Pugs - Trac
16:55 pugs_svnbot r20321 | putter++ | [elf] Reverted r20319 - though not yet much used, the node_name names are needed.  r20320 intended to revert it, but didn't.
16:55 pugs_svnbot diff: http://dev.pugscode.org/changeset/20321
16:55 lambdabot Title: Changeset 20321 - Pugs - Trac
16:59 pmurias pugs: class Foo::Bar {};my $foo = Foo::Bar.new();say $foo.WHAT
16:59 exp_evalbot OUTPUT[Foo::Bar␤]
16:59 mncharity pmurias: re node_name, if you think it should be renamed, or really think it should be gone, just let me know.
16:59 pmurias pugs: class Foo::Bar {};my $foo = Foo::Bar.new();say $foo.WHO
16:59 exp_evalbot OUTPUT[*** No such method in class Foo::Bar: "&WHO"␤    at /tmp/1Mh5fVGNA7 line 1, column 48 - line 2, column 1␤]
17:00 pmurias what is the short name for an object?
17:00 pmurias * prototype object
17:01 pmurias mncharity: compile "abc" failed: Not a HASH reference at Regexp_ModuleA.pm line 2189
17:02 * ruoso wonders if people is following the polymorphism thread in p6-lang
17:02 mncharity pmurias: perl -v ?
17:02 pmurias 5.8.8
17:02 mncharity very odd
17:03 mncharity hi ruoso.  when you would like to start compiling p6 to smop, just let me know.
17:03 ruoso mncharity, I still need to write p6opaque...
17:03 ruoso but the compilation could start earlier actually...
17:04 ruoso you wouldn't be able to test it, tho
17:04 ruoso but it would already allow writing the p6 types in p6
17:04 ruoso which is a requirement for SMOP
17:04 alanhaggai_ joined #perl6
17:05 ruoso the classes available in the global namespace
17:05 ruoso like String
17:05 ruoso will be the ones implemented in Perl 6
17:05 ruoso SMOP might use a low-level implementation when you simply do "foo"
17:05 zostay_ joined #perl6
17:05 ruoso but the Integer type needs to be there...
17:06 pmurias mncharity: wouldn't it be possible to retrieve node_name from .WHO or .WHAT?
17:12 mncharity ruoso: ok.  I started on http://svn.pugscode.org/pugs/misc/elfish_smop/ , but got distracted.  Just let me know when it would be good to have it working.
17:12 lambdabot Title: Revision 20321: /misc/elfish_smop
17:13 barney joined #perl6
17:13 ruoso mncharity, oh... that one can be working right away...
17:13 ruoso it doesn't even depend on p6opaque...
17:14 ruoso and it could be a way to benchmark how bad SLIME would perform ;)
17:15 mncharity pmurias: re 'wouldn't it be possible to retrieve node_name from', yes, but then, eg, for CompUnit, you would need to trim off the IRx1:: or Whatever::It::Happens::ToBe::  prefix yourself.
17:16 gbacon joined #perl6
17:17 mncharity ruoso: could you set it up so that given the target code, it actually compiles and runs?  that would let the generated code be tested, even when it's "not quite identical, but believed equivalent to" the target.
17:18 ruoso mncharity, actually... I won't be able to do it right now...
17:18 ruoso real life struggles
17:18 ruoso but if you like... it won't be that hard making it work...
17:18 mncharity ah.  :/  ok.  whenever you get a chance.
17:21 mncharity since it's main purpose is to help you pursue smop, it's usefulness/interestingness rather depends on your planning to pursue smop. :)
17:21 mncharity I'm not sure anyone else is set up to pursue it at the moment.
17:22 mncharity bummer about not getting funding.
17:22 mncharity audreyt had the same problem. :/
17:23 mncharity ah, TPF.  sigh.
17:29 rdice sorry, what?
17:29 ruoso mncharity, I've re-posted SMOP to TPF GC
17:30 ruoso rdice made a good job in dealing a new process in the GC
17:30 rdice we're trying to make things run better.
17:30 ruoso the proposals will be open for community comments and peer review
17:30 rdice and ruoso, I understand you made a new submission.
17:30 ruoso rdice, yes... I did...
17:30 rdice We were talking about that yesterday.
17:31 rdice And we wanted to make sure that it received appropriate public comment.
17:31 rdice I have to admit, I try to stay out of the GC's business as much as I can.  It is semi-autonomous and has its own rules and regulations.  But based on recent experiences they have asked for my help in making some improvements.
17:32 rdice So these things should work their way through the system over the next few months.
17:33 * ruoso always read garbage collection for GC for a few seconds....
17:36 ruoso rdice, it's in much better shape from what I could see... good job...
17:37 rdice Please give 99% of the credit to ambs. :-)  I just help him where I can.  Wish I could do more, but I have no many different things in TPF that need attention that none of them get as much as what would make things perfect.
17:37 eternaleye joined #perl6
17:38 sunnavy_ joined #perl6
17:38 pugs_svnbot r20322 | putter++ | [STD_red] Unbreak STD_red_run read-eval-print-loop.
17:38 pugs_svnbot diff: http://dev.pugscode.org/changeset/20322
17:38 lambdabot Title: Changeset 20322 - Pugs - Trac
17:42 [particle] ambs++ rdice-- :P
17:43 sunnavy_ joined #perl6
17:45 Eevee so if I were potentially interested in contributing..  which of the myriad of perl 6 interpreters in progress should I be looking at
17:49 obra That's a good question. Right now, Rakudo has the most momentum
17:49 obra Though if working on the test suite floats your boat, that's a nice, shared resource
17:50 Eevee rakudo's in the parrot repo and the test suite is largely in pugs, right?
17:50 * ruoso later &
17:51 obra correct
17:52 [particle] the test suite is *entirely* in pugs
17:52 [particle] t/spec/ is the official test suite, and many tests in t/ need to be migrated there
17:54 rdice [particle] Thanks.
17:55 [particle] anytime, rdice :)
17:55 Eevee what's in parrot/languages/perl6/t/spec then?  just an older copy of pugs/t/spec?
17:56 [particle] an svn co
17:56 [particle] check languages/perl 6/Makefile
17:56 [particle] *perl6
17:57 Eevee oh, ha
17:58 Eevee how complete is the test suite?
17:59 [particle] it's hard to say, but probably about 25%
18:00 [particle] we have a GSoC student, Auzon, working this summer on adding ~1000 tests
18:05 mncharity re 25%, I'd guess more like 5%.  or less.  it's currently more of a "feature exists" checklist, than a "complex feature is handling all the edge cases right" validation suite.
18:06 Eevee ouch
18:06 Eevee do any implementations do much not covered by the test suite?
18:07 alinbsp joined #perl6
18:07 mncharity p6 is a very big and complex language.
18:07 mncharity re do much not covered, no,
18:07 Eevee I know, terrible question, but it's hard to get a grasp of where things are
18:08 mncharity while not quite test driven development, there has been a focus on avoiding regressions, so features tend to get test cases.
18:09 mncharity re grasp, one might look at...
18:11 mncharity /me struggles to find the smoke test url...
18:14 mncharity wee... ok, http://m19s28.dyndns.org/cgi-bin/pugs-smokeserv.pl will
18:14 lambdabot Title: Pugs Smoke Reports
18:14 Eevee intuitive
18:14 mncharity give you some idea what is definitely not working.
18:15 Eevee are there smoke tests for rakudo, etc?
18:16 mncharity other than that, looking at the tests, and at the p6 modules in ext/ .
18:16 mncharity re "smoke tests for rakudo", I don't know.  anyone?
18:16 elmex joined #perl6
18:16 Eevee google's not turning much up except "we need rakudo smoke tests"
18:17 mncharity so that's for pugs.  for kp6, which is much more limited, the kp6 t/, and itself (it (is intended to)? self-compile).
18:18 [particle] eevee: auzon will be concentrating on making sure the test suite covers the spec, which is the best way imo
18:18 [particle] i'm talking to microsoft about using their infrastructure for parrot and rakudo smokes
18:18 mncharity for elf, nothing yet but it's own code (it self compiles) and a list of passing tests.
18:18 mncharity microsoft, eep
18:19 ispy_ joined #perl6
18:21 alinbsp joined #perl6
18:22 alinbsp left #perl6
18:22 mncharity TimToady: ok, so the residue looks like  (1) regex_mod_internal missing nth x etc.  (2) regex_metachar:sym<$> needs an alternative to the <before>, so it can match at eos (eor?:). (3) regex_assertion <alpha-[x]> doesn't seem to parse without an additional regex_assertion:ident clause. (4) re regex_assertion:sym<.>, <.ident> is parsed as a method?!?
18:22 Eevee wish all this stuff were more consolidated
18:22 [particle] Eevee: what stuff? the info you're asking about?
18:23 mncharity re ext/, that's http://svn.pugscode.org/pugs/ext/
18:23 lambdabot Title: Revision 20322: /ext
18:23 [particle] consult the perl 6 wiki, and if it's not there... it's a wiki!
18:23 Eevee everything related to perl 6, really
18:23 mncharity once upon a time it was mostly running.  current mileage may vary.
18:23 pmichaud "everything related to perl 6" is pretty big.  But I think the general consensus is that we'd like that to be organized on the wiki
18:24 obra Eevee: helping us get things back to a single coherent whole would be higely useful
18:24 pmichaud extremely useful, yes
18:25 pmichaud I tend to find the perl6 wiki start page a bit overwhelming, also
18:25 Eevee there is also the opposite problem of taking down everything that's moved or outdated, which is a bit harder
18:25 [particle] well, it can be marked for review
18:25 obra starting to make a list would be a reasonable beginning
18:26 mncharity http://www.perlfoundation.org/perl6/index.cgi
18:26 lambdabot Title: Perl 6 / Perl 6
18:28 mncharity and feedback would be useful too.  there's some concern (well, *I'm* concerned;) that socialtext is currently too weak a wiki engine for the task (eg, no discussion/Talk pages associated with each "article" page).
18:28 ispy_ mncharity: I wouldn't have chosen SocielText, honestly.  For that exact purpose.
18:28 pmichaud is there an rss feed for the Perl 6 wiki ?
18:28 mncharity s/is/may be/
18:28 Eevee haha, did anything ever come of that old offer for someone to write a wiki in perl 6?
18:29 ispy_ I have been looking for a project for perl6, but I might be over booking myself.
18:29 obra the community needs to outgrow the wiki that was set up for free before it's really worth the effort of replacing it ;)
18:29 mncharity re 'wiki in perl 6', :)  not that I know of.  elf might actually be strong and fast enough to do it now
18:29 ispy_ I am planning on re-writing my news portal in Perl6
18:29 elmex_ joined #perl6
18:30 mncharity s/to do it/to make it feasible/
18:30 ispy_ obra: I think we should go ahead and prepare a more scalable solution now, before it becomes a more difficult task later.
18:30 pmichaud I see a mythical "we"
18:30 ispy_ haha...
18:32 Eevee what is elf exactly
18:34 mncharity obra: re 'the community needs to outgrow the wiki', the problem is growing a wiki community is a very non-trivial and fragile endeavor.  so having engine problems... "we'll live on mustard gas for a while, until we out grow it, and then consider moving to oxygen".
18:35 [particle] if rakudo supported sockets, i think it'd be possible to write a wiki engine. actually, with mod_perl6 already working, you won't need sockets
18:36 Juerd Is there a feature matrix for rakudo somewhere?
18:37 [particle] there's a roadmap!
18:37 mncharity even on mediawiki, with a focus and great variety of things intended to help make community easy and attractive, most wikis fail to achieve community critical mass and eventually evaporate.
18:37 [particle] http://svn.perl.org/parrot/trunk/languages/perl 6/\\
18:37 [particle] grr
18:38 [particle] http://svn.perl.org/parrot/trunk/languages/perl6/ROADMAP
18:38 Juerd [particle]: I'm looking for an itenerary, not a roadmap :)
18:39 TimToady mncharity: re (1), I think nth and x are just dups of ** internally, so not necessary, and dubious because they imply a .*? in any case
18:39 Eevee [particle]: why sockets?  can always cgi it
18:39 Juerd BTW, re wikis, I have a creole wiki grammar
18:39 Juerd But I never tried to compile it because I don't know where to start
18:40 Juerd http://feather.perl6.nl/~juerd/creole
18:40 [particle] Eevee: yes, i was thinking about the whole engine. don't need sockets.
18:40 TimToady re (2), I didn't think $ was necessary because you can't have a regex at end-of-file, but then I thought about indirect <$rule>, which can be a string, so yes, it's missing.
18:40 TimToady re (3) the intent is that if you're going to do set manipulation you must start with <+ or <-
18:41 Eevee maybe I should give that a try
18:42 TimToady re (4), the current regex_assertion parsing .ident as a P6 expression, though a postfix would probably be more appropriate there.
18:43 mncharity obra: wrapping up, my one (few?) attempts to edit the wiki bogged down on "what I really want to do is start an out-of-band discussion, not deal with throwing away or substantially changing someone else's nicely edited text".  left me with a "don't bother trying that again" feeling, despite a long and strong inclination towards wikis.
18:44 Auzon joined #perl6
18:44 mncharity or some such
18:45 [particle] Eevee: i suggest you try mod_perl6
18:45 Eevee oh, wait a minute
18:45 Eevee are there any database bindings for perl 6?
18:47 pugs_svnbot r20323 | lwall++ | [STD] couple more tweaks from mncharity++
18:47 pugs_svnbot diff: http://dev.pugscode.org/changeset/20323
18:47 lambdabot Title: Changeset 20323 - Pugs - Trac
18:48 pmichaud mncharity: can the out-of-band discussion go on a mailing list somewhere?
18:48 [particle] Eevee: there's no DBI yet
18:48 pmichaud mncharity: yes, I agree it's not as nice as having the oob discussion in the same place as the wiki
18:48 Eevee not even wrappers for an api?
18:48 Eevee that would make a perl 6 wiki a little more problematic
18:48 mncharity re (1), looks like http://svn.perl.org/viewvc/parrot/trunk/t/compilers/pge/perl6regex/rx_modifiers needs to be tweaked then.
18:48 lambdabot Title: 1 [ 6 parrot 1 ] 1 35 Log of /trunk/t/compilers/pge/perl6regex/rx_modifiers 30, http://tinyurl.com/6rqbtd
18:49 * [particle] tries to remember if 'system' is implemented in rakudo
18:49 [particle] one could *at least* use svn as a wiki backing store with system
18:51 pmichaud fwiw, PmWiki uses text files as its database.  Works well, for small scales.
18:51 pmichaud although there are some PmWiki installations with 200K+ pages
18:51 Eevee ouch
18:52 mncharity Eevee: re db, elf compiles to p5. :)  I've intentionally kept it slightly inconvenient to inline p5, just to avoid sloppiness in doing the bootstrap, but would be happy to change that. :)
18:53 mncharity might start writing p6 the way you would want to, and then complain to me and rakudo folks when things don't work. :)
18:54 pmichaud yes, exactly
18:55 pmichaud I'm still a bit slow about responding to things (dayjob), but bug reports and feature requests definitely get higher priority than other things
18:57 Eevee well!  maybe I will give it a shot
18:57 Eevee might just have to go with flatfiles to keep it simple  (fsvo "simple")
18:57 Juerd BTW, which channel is most appropriate for Rakudo discussion? #perl6 or #parrot?
18:57 [particle] i think flatfiles will work well
18:58 TimToady [particle]: there is no system() in p6
18:58 [particle] juerd: i prefer #perl6
18:58 [particle] TimToady: what's the equivalent?
18:58 TimToady it is renamed to run() to avoid return value confusion
18:58 [particle] ah, right. thx
18:59 mncharity pmichaud: re "go on a mailing list somewhere?" and "not as nice as", that seems a really big barrier.  even just requiring logging in and account creation is can be deadly, albeit unavoidable until the community is large enough for 24x7 spam watch.  hmm, except for "a few writers are collaborating here on a project" use, vs "the community is maintaining this".
18:59 Eevee TimToady: return value will dtrt when cast to boolean?
18:59 TimToady documented in S29:1815
18:59 TimToady Eevee: yes
18:59 Eevee hooray
18:59 TimToady success should always be true
19:01 elmex_ joined #perl6
19:02 pmichaud mncharity: ultimately I agree with you that the wiki as currently implemented isn't exactly conducive to what we're trying to accomplish
19:03 Eevee what does the wiki really need
19:03 Eevee besides talk pages
19:05 pmichaud at this point I think it needs reorganization.  It feels like there's too much competing to be on the top level page
19:05 mncharity pmichaud: I don't know that I've ever seen real data, but my fuzzy impression is the issue is people do escalating commitment.  With first edits rarely being major.  A spelling correction noted in passing, leads to a sentence edit, leads to... etc.  Or minor tasks - url mentioned on irc gets added to a page, etc.  So one wants that slippery slope as smooth as possible. :)
19:06 mncharity Eevee: there was talk of perhaps tweaking socialtext to include Talk pages.  it might even not be hard.
19:09 mncharity re what else does the wiki really need, hmm.  I kind of got stuck on the talk issue.  drat, don't remember, sorry.
19:10 Eevee I doubt hacking talk pages into any wiki engine would be hard
19:12 mncharity re (3), ah, ok, so http://svn.perl.org/viewvc/parrot/trunk/t/compilers/pge/perl6regex/rx_subrules?revision=22145&amp;view=markup needs to be tweaked.  <alpha-[Jj]>+
19:12 lambdabot Title: 1 [ 6 parrot 1 ] 1 35 View of /trunk/t/compilers/pge/perl6regex/rx_subrules 30, http://tinyurl.com/5ssqaf
19:16 mncharity TimToady: re (4), "the current regex_assertion parsing .ident as a P6 expression, though a postfix would probably be more appropriate there.", what about <.before \w+?>  ?
19:17 mncharity it seems . should be ? and ! like, no?
19:17 ruoso joined #perl6
19:17 mncharity anyway, I have a workaround.  so resolution is not pressing.  just flagging it as a potential issue.
19:21 IllvilJa joined #perl6
19:26 pugs_svnbot r20324 | putter++ | [STD_red] Tweaked regex parsing to reflect STD.pm changes.
19:26 pugs_svnbot diff: http://dev.pugscode.org/changeset/20324
19:26 lambdabot Title: Changeset 20324 - Pugs - Trac
19:27 aindilis joined #perl6
19:32 mncharity pmurias: re perl -v, what OS?
19:34 pmurias gentoo linux
19:35 pugs_svnbot r20325 | nothingmuch++ | remove overloading from Data::Capture
19:35 pugs_svnbot diff: http://dev.pugscode.org/changeset/20325
19:36 lambdabot Title: Changeset 20325 - Pugs - Trac
19:36 mncharity and -w didn't have any complaints?  (5.10 should be making (perhaps bogus) "Variable "$s" is not available" (and "$x") complaints).
19:37 pmurias i'll eat some food, as it's getting cold &
19:38 mncharity ok.  yeah, I seem to have missed brkfast+lunch.  bbiab. :)
19:40 meppl joined #perl6
19:40 mncharity with 5.8.8,  perl -w Regexp_ModuleA.pm --repl  should be -w silent and well behaved.
19:44 pmurias i get a prompt what should i type to check if it works
19:45 pmurias ?
19:46 mncharity if abc is failing,  perhaps  abc  and then a string of  xabc  ?
19:46 TimToady mncharity: that must be written <.before(/\w+/)>.  the . means "call this in method syntax" which happens to have the side effect of not capturing because . is not alphabetic
19:47 TimToady not that <.before ...> makes any sense anyway...
19:47 TimToady but the idea is that if it looks like a method call, it is one.
19:47 pasteling "mncharity" at 71.232.7.10 pasted "perl -w Regexp_ModuleA.pm --repl abc" (9 lines, 288B) at http://sial.org/pbot/30913
19:50 TimToady but yes, that contradicts what S05 says currently.  :)
19:50 TimToady so I guess we can still decide based on whether there's a ( after the ident
19:51 pmurias mncharity: have you ci'ed everything? please check sv(n|k) status?
19:53 mncharity re "side effect of not capturing", S05, :)  ok, np - I left the "( after the ident" hack in std.rb for now.
19:53 pmurias mncharity: checking if it works on feather
19:53 mncharity pmurias: rechecking...
19:56 mncharity re status, rm -rf'ed, checked out fresh, moved to /tmp, and run as not-me user.
19:56 pugs_svnbot r20326 | lwall++ | [STD] allow both <.foo regex> and <.foo(@args)>
19:56 pugs_svnbot diff: http://dev.pugscode.org/changeset/20326
19:56 lambdabot Title: Changeset 20326 - Pugs - Trac
19:56 armagad joined #perl6
19:59 pugs_svnbot r20327 | lwall++ | [STD] but also allow <.$indirect()> forms of method call...
19:59 pugs_svnbot diff: http://dev.pugscode.org/changeset/20327
19:59 lambdabot Title: Changeset 20327 - Pugs - Trac
20:01 TimToady now I can picture someone expecting <.foo: 1,2,3> to work the way they expect it to, which says maybe that our current use of : is wrong in assertions...
20:02 mncharity pmurias: that said, Regexp_ModuleA has been observed to cease working if "should have no effect" code reordering is done.  a check for that specific case should warn you the perlbug workaround has failed.  but I could well believe some other bug is being tickled.  p5 support for anon subs seems not entirely robust.  and I'm 64 bit to your 32.
20:04 TimToady which would mean we'd need to write <panic: 'message'> with quotes rather than relying on > for final quote
20:05 TimToady that seems a bit saner than the current use of :
20:06 TimToady pmichaud: see above speculation about <foo: args>
20:06 pmurias mncharity: fail on feather too
20:06 mncharity re yare, so there's a development risk for the current effort to add the yet_another_regex_engine core to elf.  any flakiness may continue to exist.  Long ago I looked at the two 5.10 warnings, but the simpler of the two seemed to be a 5.10 bug.
20:07 mncharity re feather, I start to say :(, but that should be a :) .
20:07 mncharity Juerd: after all these years... can I get a feather account? :)
20:08 pmurias mncharity: why a :) ?
20:08 pmurias * fails
20:08 mncharity a replicable problem is a happy problem
20:08 pmurias true
20:09 pmurias maybe emitting an NFA (with extra bits) wouldn't be a bad idea after all
20:11 pmichaud TimToady: I don't have a problem with <foo: 'args'>
20:11 * [particle] finds peace in <foo: 'args'>
20:13 pmichaud that does seem to fit the indirect method syntax somehow, also
20:13 mncharity re alternate core, oh, certainly.  I'm just concerned with bootstrap.  have working code, in a domain famous for "it's way harder to get this stuff working than you think", so using it.  would be nice to use 5.10 reentrancy if its bugs are avoidable, or have an engine coded in p6 which can be compiled down to C, or... etc.
20:13 pmichaud might not be a desirable pun
20:13 pmurias s/bad/that bad/
20:15 mncharity s/an engine/a vm engine/
20:15 Zaba_ joined #perl6
20:21 pmurias mncharity: are you familiar with NFAs and DFAs?
20:30 mncharity re familiar, not really.  the hard part is all the optimization games.  I've read the odd paper, but no more.  actually, my hope/intent is that p6 moves towards providing a general parser.  ie, there's intense grammar analysis, a potentially heterogeneous mix of parsing strategies is emitted, but the user doesn't have to be care about parsing theory when writing a grammar.
20:32 mncharity no "oh, I have to rewrite this all as (non) left-recursive because the engine is doing x".  you should just be able to write a grammar, run it, and if it's not fast enough, ask for hints on where the time is going and how to improve.
20:39 elmex joined #perl6
20:39 pmurias mncharity: a subcall for each char would be too slow?
20:42 mncharity at runtime, definitely.  p5 subcall is ~few hundred us.  a C subcall would be faster.  but the p5 engine does significantly better.
20:44 mncharity 200 microseconds means you only get 5000 of them per second.
20:45 pmurias i thought about changing the transitions array of NFA state to a closure
20:45 pugs_svnbot r20328 | lwall++ | [STD] syntax of <panic: "message"> now requires quotes
20:45 pugs_svnbot diff: http://dev.pugscode.org/changeset/20328
20:45 lambdabot Title: Changeset 20328 - Pugs - Trac
20:46 mncharity using closures on backends (eg, a hypothetical Common Lisp backend) which can aggressively compile
20:46 mncharity things down could be a win.
20:47 mncharity on p5... you would have to profile it.
20:47 mncharity you're trying to avoid all the object creation of the (current?) STD5 ?
20:50 mncharity my hope is once the first rx engine in elf is working, it will be easy to experiment with adding additional ones.  much of the IR analysis will be shared, and you may be able support the same runtime or compile time apis, and thus can start with partial implementations (either only supporting some features, or only working when some problematic rx constrict doesn't appear).
20:50 pmurias no just thinking about the simplest way to make the NFA none regular,
20:51 pmurias and cheat around unicode
20:52 mncharity ah, ok.
20:52 Juerd If Perl 5's regex engine source was less messy I'd say: look at Perl 5's regex engine for inspiration re cheating around unicode
20:53 mncharity hi Juerd.  so what's the current recipe for requesting a feather account?
20:54 mncharity re p5 engine, I've not see it since the 5.10 changes & cleanup.  worth looking at just to appreciate the scale of the task.  though my fuzzy recollection is it's actually a bit dumb in some respects.
20:56 mncharity Juerd: nm, found it.  tnx
20:56 Juerd mncharity: Email me your real name and preferred username
20:56 mncharity thanks
21:00 mncharity bbl
21:02 pmurias mncharity: once you get yare working with elf it should be possible to benchmark other idea's against it
21:03 pmurias anyway i should go to sleep now as i don't think clearly enough now ;) &
21:17 mncharity that's the hope.  we'll see.  g'night. :)
21:17 pugs_svnbot r20329 | nothingmuch++ | split Capture into ::Overload and normal class, and use it in prepare_binding in Data::Bind
21:17 pugs_svnbot diff: http://dev.pugscode.org/changeset/20329
21:17 lambdabot Title: Changeset 20329 - Pugs - Trac
21:17 mncharity http://eprints.utas.edu.au/644/
21:17 lambdabot Title: UTas ePrints - A New General Parser for Extensible Languages
21:17 mncharity http://citeseer.ist.psu.edu/tratt04compiletime.html
21:17 lambdabot Title: Compile-time meta-programming in Converge (ResearchIndex)
21:17 mncharity random papers
21:19 mncharity Ian Lewis ("A New General...") is shifting to packrat.
21:19 pugs_svnbot r20330 | nothingmuch++ | split Capture into ::Overload and normal class, and use it in prepare_binding in Data::Bind
21:19 pugs_svnbot diff: http://dev.pugscode.org/changeset/20330
21:19 lambdabot Title: Changeset 20330 - Pugs - Trac
21:34 pugs_svnbot r20331 | nothingmuch++ | all_variable_names fixes
21:34 pugs_svnbot diff: http://dev.pugscode.org/changeset/20331
21:34 lambdabot Title: Changeset 20331 - Pugs - Trac
21:51 eternaleye_ joined #perl6
21:57 [particle1 joined #perl6
22:15 luqui joined #perl6
22:28 Jamtech joined #perl6
22:32 smtms joined #perl6
22:36 bsb joined #perl6
23:10 luqui joined #perl6
23:29 Limbic_Region joined #perl6

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

Perl 6 | Reference Documentation | Rakudo