pugscode.org/ | nopaste: sial.org/pbot/perl6 | pugs: [~] <m oo se> (or rakudo:, kp6:, smop: etc.) || We do Haskell, too | > reverse . show $ foldl1 (*) [1..4] | irclog: irc.pugscode.org/
Set by TimToady on 25 January 2008.
00:16 aindilis left, aindilis joined 00:21 aindilis left, aindilis joined 00:37 aindilis left, aindilis joined
TimToady cognominal_: both the table and the section agree that list prefix is looser than comma. I don't see a problem. 00:41
00:42 thoughtpolice joined 00:43 thoughtpolice left 00:44 perljunkie left, thoughtpolice joined 00:49 cmarcelo joined 00:51 aindilis left, aindilis joined 00:55 kanru left 00:57 peeps[work] left 01:00 nipotaway is now known as nipotan 01:01 cnhackTNT joined 01:15 lyokato joined 01:16 aindilis left, aindilis joined 01:18 LimbicRegion joined 01:20 aindilis left, aindilis joined
diakopter pugs: sleep 1 01:22
exp_evalbot OUTPUT[*** Unsafe function 'sleep' called under safe mode␤ at /tmp/xleOeBb6xu line 1, column 1 - line 2, column 1␤]
diakopter oh yeah
01:22 tobeya left 01:25 Limbic_Region left, LimbicRegion is now known as Limbic_Region
katie_d Sorry if this is a stupid question, but my knowledge of C is still very shallow. I'm trying to read the #define statements in config_h.in and there are a bunch of statements like "typedef @nv@" -- what do the @ symbols mean? I've never seen that in C code before. 01:42
[particle] you probably want #perl, that is if you're looking at perl 5's source 01:43
however, it's likely that what you're looking at is a template, and @foo@ is a template variable that will be replaced with some value
i'm not familiar with perl 5 source, so i can't be certain 01:44
katie_d [particle]: I was looking at Parrot sources. But thanks anyway -- it probably is a template variable. Do you happen to know what it's used by?
[particle] heh, i'm quite familiar with parrot source :)
which file?
parroters hang out on irc.perl.org#parrot btw 01:45
katie_d config/gen/config_h/config_h.in
[particle] ok, that's a template file for the configure system 01:46
katie_d I was trying to figure out what PARROT_API was for, and grep pointed me there
[particle] do you use vim?
01:46 justatheory left
katie_d particle: no I use emacs 01:46
[particle] ok. that'll work to, i think. 01:47
make tags
i only know how to navigate a tag stack in vi, but i'm sure you can do it with emacs
you can put your cursor on PARROT_API, press a magic combination of buttons, and arrive at it's declaration/definition 01:48
in vim, it's Ctrl-]
katie_d Wow :) Thanks. I didn't know I could do that in emacs. I've been using Eclipse for that sort of thing, but for some reason the search feature is broken, so I've just gone back to using grep
btw, what did you mean the parroters don't hang out in here? Is this mostly for pugs or something? 01:49
[particle] anyway, that's a template file. after Configure.pl finishes running, a header file will be generated that contains #define PARROT_API 01:50
it's so we can specify which functions are exported
so embedders and extenders can use them 01:51
see include/parrot/config.h about line 60
01:52 obra left, obra joined
katie_d Thanks particle. 01:54
[particle] glad to help! 01:55
01:55 Quaver left
diakopter [particle]: I noticed you put your name under the GSoC items. Do you think they should be near the other Parrot/Perl 6 items (at the bottom of that page)? 02:01
02:02 mj41__ joined 02:04 aindilis left, FurnaceBoy left, qmole left, IllvilJa left, Botje left, Qiang left, [particle] left, alester left, kolibrie left, pugsbot_r17185 left 02:05 cognominal_ left, kst left, ycheng left, cmarcelo left, thoughtpolice left, Patterner left, mncharity left, Zygo left, jrockway left, marmic left, jjore left, ead left, drbean left, penk left, rakudo_svn left, demq left, xinming left, GeJ left, pmichaud left, agentzh left, buubot left, shachaf left, nipotan left, silug left, Limbic_Region left, gbacon left, sordith left, bbkr__ left, zev left, rhr left, mugwump left, arguile left, Caelum left, allbery_b left, buu left, pasteling left, kcwu left, lambdabot left, Helios left, IRSeekBot left, peepsalot left, Jedai left, integral left, charsbar left, pugs_svnbot left, SamB left, Lunchy left, rafl left, cnhackTNT left, lisppaste3 left, wknight8111 left, avar left, revdiablo left, ting left, yahooooo left, Daveman left, meteorjay left, stevan_ left, nothingmuch left, moritz_ left, nnunley left, spinclad left, TimToady left, ruz_ left, TreyHarris left, wolverian left, SubStack left, Alias_ left, mj41 left, arcady left, katie_d left, Khisanth left, RayMagini left, whiteg left, cj_ left, cxreg left, jnthn left, mtve left, Southen_ left, Tene left, zostay left, RichiH left, mj41__ is now known as mj41 02:06 ting joined, jnthn joined, wolverian joined 02:08 Alias_ joined 02:09 meteorjay joined 02:11 cnhackTN1 joined, Alias_ left, Alias__ joined, ting left, TimToady joined, cnhackTNT joined, lambdabot joined, arcady joined, lisppaste3 joined, Helios joined, wknight8111 joined, drbean joined, avar joined, katie_d joined, Khisanth joined, RayMagini joined, penk joined, pmichaud joined, buubot joined, rakudo_svn joined, shachaf joined, demq joined, nipotan joined, xinming joined, GeJ joined, silug joined, agentzh joined, IRSeekBot joined, revdiablo joined, ting joined, peepsalot joined, yahooooo joined, Daveman joined, whiteg joined, cj_ joined, Jedai joined, integral joined, stevan_ joined, nothingmuch joined, charsbar joined, ruz_ joined, SamB joined, zostay joined, Lunchy joined, TreyHarris joined, spinclad joined, Alias_ joined, SubStack joined, nnunley joined, irc.freenode.net sets mode: +o TimToady, Southen_ joined, moritz_ joined, cxreg joined, pugs_svnbot joined, Tene joined, mtve joined, rafl joined, irc.freenode.net sets mode: +o Tene 02:12 avar left, TreyHarr1s joined, Daveman left, revdiablo left, Tene left, Alias_ left, nnunley left, nnunley joined, TreyHarr2s joined, cnhackTNT left, cmarcelo joined, thoughtpolice joined, Patterner joined, mncharity joined, Zygo joined, jrockway joined, marmic joined, jjore joined, ead joined, eternaleye joined 02:13 avar joined, aindilis joined, FurnaceBoy joined, qmole joined, IllvilJa joined, Botje joined, Qiang joined, [particle] joined, alester joined, kolibrie joined, pugsbot_r17185 joined, cognominal_ joined, kst joined, ycheng joined, Limbic_Region joined, gbacon joined, sordith joined, bbkr__ joined, zev joined, pasteling joined, buu joined, kcwu joined, arguile joined, allbery_b joined, Caelum joined, rhr joined, mugwump joined, ting_ joined 02:17 TreyHarris left, TreyHarr1s is now known as TreyHarris 02:19 TreyHarr2s left 02:20 eternaleye left 02:21 eternaleye joined 02:22 [particle1 joined 02:23 ting left 02:30 Southen joined 02:31 mugwump left, rhr left, zev left, bbkr__ left, sordith left, Limbic_Region left, arguile left, Caelum left, allbery_b left, buu left, pasteling left, kcwu left, gbacon left, Limbic_Region joined 02:34 gbacon joined, sordith joined, bbkr__ joined, zev joined, pasteling joined, buu joined, kcwu joined, arguile joined, allbery_b joined, Caelum joined, rhr joined, mugwump joined, bbkr__ left, bbkr__ joined 02:35 zev left, zev joined, mugwump_ joined, mugwump left 02:36 [particle] left 02:41 thoughtpolice left, thoughtpolice joined 02:42 dalek left 02:48 dalek joined 02:50 pugs_svnbot left, Lunchy left, Jedai left, rafl left, SamB left, peepsalot left, Helios left, lambdabot left, charsbar left, integral left, IRSeekBot left 02:51 charsbar joined 02:54 Helios joined 02:55 lambdabot joined, IRSeekBot joined, peepsalot joined, Jedai joined, SamB joined, Lunchy joined, pugs_svnbot joined, rafl joined, integral joined 02:56 IRSeekBot left, Lunchy left 02:59 Lunchy joined, IRSeekBot joined 03:00 Southen_ left 03:21 stevan__ joined 03:25 rimad left 03:26 rimad joined 03:29 alc joined 03:31 stevan_ left 03:46 aindilis left 03:51 rindolf joined 03:57 Jamtech joined 03:58 jjore is now known as zz_jjore, zz_jjore is now known as jjore, Daveman joined 04:07 alester_ joined, Jamtech left 04:09 lyokato left 04:12 wknight8111 left 04:15 mncharity left 04:19 RayMagini left 04:22 RayMagini joined 04:31 dalek left, stevan__ left, integral left, pugs_svnbot left, peepsalot left, Jedai left, rafl left, SamB left, lambdabot left, RayMagini left, bbkr__ left 04:32 rhr left, sordith left, arguile left, Caelum left, allbery_b left, buu left, pasteling left, kcwu left, gbacon left, alester_ left, IRSeekBot left, charsbar left, mugwump_ left, Southen left, pugsbot_r17185 left, kolibrie left, Qiang left, qmole left, alester left, FurnaceBoy left, Botje left, cognominal_ left, kst left, IllvilJa left, ycheng left, avar left, rindolf left, eternaleye left, jjore left, marmic left, Patterner left, Zygo left, ead left, jrockway left, cmarcelo left, Lunchy left, Helios left, thoughtpolice left, agentzh left, rakudo_svn left, pmichaud left, penk left, GeJ left, buubot left, shachaf left, demq left, xinming left, nipotan left, drbean left, silug left, Daveman left, TreyHarris left, TimToady left, moritz_ left, spinclad left, ruz_ left, nothingmuch left, yahooooo left, SubStack left, lisppaste3 left, ting_ left, nnunley left, mtve left, cxreg left, whiteg left, katie_d left, arcady left, Khisanth left, zostay left, cj_ left, alc left, Limbic_Region left, Alias__ left, cnhackTN1 left, lumi_ left, Auzon left, bockmabe left, orafu left, araujo left, ruoso left, BinGOs left, gugod left, r0bby left, cosimo left, lidden left, PerlJam left, baest left, Juerd left, Maddingue left, rimad left, mattz left, cookys left, kane__ left, Eidolos left, tewk left, ilbot2 left, jql left, diakopter left, sunnavy left, thepler left, obra left, clkao left, felipe left 04:33 TimToady joined, d4l3k_ joined, RayMagini joined, alester_ joined, Daveman joined, rindolf joined, alc joined, rimad joined, stevan__ joined, IRSeekBot joined, Lunchy joined, integral joined, rafl joined, pugs_svnbot joined, SamB joined, Jedai joined, peepsalot joined, lambdabot joined, Helios joined, charsbar joined, thoughtpolice joined, mugwump_ joined, bbkr__ joined, rhr joined, Caelum joined, allbery_b joined, arguile joined, kcwu joined, buu joined, pasteling joined, sordith joined, gbacon joined, Limbic_Region joined, Southen joined, eternaleye joined, ting_ joined, ycheng joined, kst joined, cognominal_ joined, pugsbot_r17185 joined, kolibrie joined, alester joined, Qiang joined, Botje joined, IllvilJa joined, qmole joined, irc.freenode.net sets mode: +o TimToady, FurnaceBoy joined, avar joined, ead joined, jjore joined, marmic joined, jrockway joined, Zygo joined, Patterner joined, cmarcelo joined, nnunley joined, TreyHarris joined, mtve joined, cxreg joined, moritz_ joined, SubStack joined, spinclad joined, zostay joined, ruz_ joined, nothingmuch joined, cj_ joined, whiteg joined, yahooooo joined, agentzh joined, silug joined, GeJ joined, xinming joined, nipotan joined, demq joined, shachaf joined, rakudo_svn joined, buubot joined, pmichaud joined, penk joined, Khisanth joined, katie_d joined, drbean joined, lisppaste3 joined, arcady joined, Alias__ joined, cnhackTN1 joined, obra joined, lumi_ joined, Auzon joined, bockmabe joined, orafu joined, araujo joined, ruoso joined, cosimo joined, mattz joined, BinGOs joined, gugod joined, jql joined, r0bby joined, lidden joined, cookys joined, felipe joined, clkao joined, thepler joined, sunnavy joined, diakopter joined, ilbot2 joined, tewk joined, Eidolos joined, kane__ joined, baest joined, Maddingue joined, PerlJam joined, Juerd joined 04:34 d4l3k_ is now known as dalek, thoughtp1lice joined 04:35 demq left, demq joined 04:38 rindolf left 04:46 rakudo_svn left, GeJ left, agentzh left, pmichaud left, Helios left, drbean left, xinming left, Lunchy left, shachaf left, thoughtpolice left, penk left, buubot left, nipotan left, silug left 04:47 araujo left 04:48 Limbic_Region left, Helios joined, Lunchy joined 04:49 shachaf joined 04:50 drbean joined, thoughtpolice joined, penk joined, agentzh joined, silug joined, GeJ joined, xinming joined, nipotan joined, rakudo_svn joined, buubot joined, pmichaud joined 04:51 thoughtpolice left 04:57 lyokato joined 04:59 cnhackTN1 left 05:00 cnhackTNT joined 05:07 stevan__ left 05:08 stevan_ joined 05:10 cmarcelo left 05:40 iblechbot joined 05:41 stevan_ left 06:01 Auzon left 06:02 Auzon joined 06:39 cnhackTNT left 06:43 alester_ left 06:45 BinGOs_ joined 06:51 BinGOs left 07:17 FurnaceBoy left, BinGOs_ is now known as BinGOs 07:20 rakudo_svn left, xinming left, GeJ left, pmichaud left, agentzh left, buubot left, penk left, nipotan left, silug left 07:23 GeJ joined 07:24 penk joined, agentzh joined, silug joined, xinming joined, nipotan joined, rakudo_svn joined, buubot joined, pmichaud joined, Alias__ left, Alias__ joined 07:25 baest left 07:26 baest_ joined, baest_ is now known as baest 07:30 iblechbot left 07:34 Alias_ joined 07:35 Alias__ left, Lunchy left, Helios left, pugs_svnbot left, Jedai left, peepsalot left, rafl left, SamB left, lambdabot left, integral left 07:37 Helios joined 07:38 Lunchy joined, integral joined, lambdabot joined, peepsalot joined, Jedai joined, SamB joined, pugs_svnbot joined, rafl joined, integral left, integral joined 07:39 Patterner left, marmic left, jjore left, eternaleye left, Zygo left, ead left, jrockway left 07:40 eternaleye joined 07:41 Psyche^ joined, Psyche^ is now known as Patterner 07:42 Zygo joined, jrockway joined, marmic joined, jjore joined, ead joined, Zygo left 07:47 ikeda joined 07:55 ikeda left 08:14 hareldvd joined, hareldvd left 08:16 thoughtp1lice left 08:17 Aankhen`` joined 08:24 rby joined 08:27 ikeda joined 08:37 ikeda left
cognominal_ Tim Toady, I beg to differ, the = must bind looser than the comma to bind the whole unparenthesed list. I cant make sense of the sentence "the right side is looser than comma" because the right side is made of comma operator. 08:40
rakudo uses the S03 precedence table and gets it wrong too. One must parenthesize like in Perl5 08:42
08:45 masak joined 08:49 IllvilJa left 08:51 DarkWolf84 joined, araujo joined 08:55 ikeda joined 08:56 masak left 09:07 IllvilJa joined 09:10 laye joined 09:17 agentzh left 09:18 agentzh joined 09:27 laye left 09:28 ikeda left 09:38 ebassi joined 09:56 Schwern joined 10:02 ruoso left 10:10 Schwern left, Schwern joined 10:11 Schwern left, Schwern joined 10:22 hanekomu joined 10:24 hanekomu left 10:25 hanekomu joined
nothingmuch @tell mncharity no, it's that you can't efficiently pass inherited attributes around if your parent<->child relationships aren't statically known, you need to dispatch them lazily. 10:46
lambdabot Consider it noted.
nothingmuch @tell mncharity for instance if you have a global index inherited/synthesized attribute, and some Node { has @.children } the value of that attr varies for each @.children[$n] based on the value of $n - 1, and it must be resolved 100% dynamically 10:47
lambdabot Consider it noted.
nothingmuch @tell mncharity furthermore there is no data in $node.children[$n] linking it to $node or vice versa. in MO each invocant on the stack is asked whether or not it is the parent of $some_child via a cached method call 10:48
lambdabot Consider it noted.
spinclad nothingmuch: 1) not statically known because 'has @.children' is added at runtime? 2) value varies based on index - 1 (why '- 1'?) because array contains independent containers, or because (being added at runtime) nothing is known about it statically? 3) no data in $node linking to .children[$n], no static data? since @.children itself links to each child (just not statically)... 4) or is my confusion more profound? 11:24
11:26 ruoso joined
spinclad II) why, when you see 'Node { has @.children }', can you not infer some things statically? because the API for the attribute definition (going through the MOP) is dynamic, not closed, and so not knowable statically? 11:31
(/me -> bed, i'll backlog) 11:36
11:37 alc left 11:46 cmarcelo joined 12:19 masak joined 12:28 nipotan is now known as nipotaway 12:33 zblaxell joined 12:44 lyokato left 12:48 ebassi is now known as ebassi|lunch 12:53 pjcj left, pjcj joined 13:09 xinming_ joined 13:12 xinming left 13:28 ebassi|lunch is now known as ebassi 13:38 ebassi left 13:44 mncharity joined 13:51 simcop2387-vnc joined 13:53 simcop2387 left 13:54 alester_ joined 13:56 ebassi joined, alester_ left 13:57 simcop2387-vnc is now known as simcop2387 14:00 meppl joined 14:13 qmole left, qmole joined
mncharity two interesting papers: www.cs.ucla.edu/~awarth/papers/dls07.pdf from www.cs.ucla.edu/~awarth/ometa/ , and www.chrisseaton.com/katahdin/ 14:39
lambdabot mncharity: You have 3 new messages. '/msg lambdabot @messages' to read them.
14:41 eternaleye left
mncharity nothingmuch: re AG engines can't do p5/p6 because 'you can't efficiently pass inherited attributes around if your parent<->child relationships aren't statically known, you need to dispatch them lazily.', err, AG engines can't memoize? That doesn't sound right. 14:45
14:47 Psyche^ joined
mncharity re 'there is no data in $node.children[$n] linking it to $node', ah, ok. So one needs a system which permits node annotations. Eg, oo on prolog (eg, Logtalk or FLORA) rather than just prolog. I'm surprised none of the AG engines permit term annotation. I don't understand the "or vice versa." part. 14:49
So I'm still fuzzy on the constraints, but thanks. Interesting.
ruoso . o O ( AG? ) 14:50
mncharity ruoso, how goes the smop "before and after" example?
oh, hi ruoso :)
ruoso mncharity, hi...
mncharity Attribute Grammar (in the more recent instantiation of that phrase)
ruoso ahh....
ok... 14:51
14:51 sh joined
ruoso the example is not there yet.. 14:51
mncharity en.wikipedia.org/wiki/Attribute_grammar
ruoso mostly because I still need the lowlevel types (method, attribute, code...)
s/types/implementations/
mncharity www.perlfoundation.org/perl6/index....p_language is where it is landing? 14:52
lambdabot Title: SMOP s1p Language / Perl 6
ruoso yes...
but anyway... the code -> frame translation can be made in sm0p terms
I mean
the body of a method is already a solved issue
mncharity re low-level types, hmm.
oh, could you do a fib() translation? 14:53
ruoso maybe...
would you like to put the example code in the s1p page?
sh Does anyone know how to fix this pugs internal error: Invalid grammatical category: "Bool" 14:54
From SVN revision 20063 14:55
I get it during make of pugs 14:56
mncharity sub fib($n) { if $n ==0 { 0 } elsif $n == 1 { 1 } else { fib($n-1) + fib($n-2) } }
or whatever variation makes it easiest for your
sh I have a pugs binary, but when I run it, I get the above error
mncharity sh: it's the current way the pugs build is failing :(
it's being worked on
sh Thanks mcharity
mncharity hmm, anyone know a version number to suggest for a working pugs?
sh Good question, I'd be interested too! 14:57
mncharity I fuzzily recall there is a way to ask the evalbot...
pugs: 3
exp_evalbot RESULT[3]
mncharity pugs: $VERSION
exp_evalbot OUTPUT[*** ␤ Unexpected end of input␤ expecting "::"␤ Variable "$VERSION" requires predeclaration or explicit package name␤ at /tmp/vLoV8UcmPB line 1, column 9␤]
mncharity pugs: $PUGS_VERSION
exp_evalbot RESULT[\"Perl6 User\'s Golfing System, version 6.2.13, October 17, 2006 (r18093)"]
mncharity oooh, that's old.
r18093
sh Yeah, probably does not compile with GHC 6.8.2 14:58
ruoso mncharity, Ok... the interesting thing here... is that you need an object that implements "if" and "=="
mncharity right. 6.6.1 is your best bet.
ruoso mncharity, after that... you just need to pile frame nodes in the sm0p sense...
mncharity ruoso: :/ The compiler can't compile it into something simple before handing it to smop? 14:59
sh Ok, I was already out to buy champagne after the pugs binary compiled with GHC 6.8.2, but then noticed it does not work at all
mncharity s/The/My/
ruoso mncharity, you can compile to a set of sm0p blocks
"if" and "==" must still be implemented anyway (but that's not related to your example... it's a general task) 15:00
mncharity sh: yeah. :( not quite yet. got noticably closer yesterday thanks to gbacon.
ruoso mncharity, so you can just name a public symbol (like SMOP__OP__if) with an API you can define, and use it inside a sm0p block 15:01
mncharity ruoso: say I'm an optimizing compiler ("mncharity, you are an optimizing compiler"), couldn't if and == be inlined, and the $n type be native int?
15:02 rdice joined, hanekomu left
sh mncharity: Thanks for your help! 15:02
ruoso the $n type must probably be native int.. (it's the only one implemented)... 15:03
mncharity sh: np. any help debugging would be most welcome. :)
ruoso mncharity, the only thing that is defined is that the code you generate must comply to the www.perlfoundation.org/perl6/index....tation_api
lambdabot Title: SMOP Interpreter Implementation API / Perl 6, tinyurl.com/yv6sur 15:04
ruoso you can use SLIME... which is the www.perlfoundation.org/perl6/index....ementation
lambdabot Title: Default SMOP Interpreter Implementation / Perl 6, tinyurl.com/3328jg
sh mncharity: :-) I'm unfortunately not in a position to make a contribution in terms of actual code; I'm happy to beta test though
mncharity err, "SLIME" is already used as the name of a very nice lisp environment on emacs. :) 15:05
15:05 Patterner left
SamB yes it is 15:05
mncharity sh: re beta test, ok, that would be useful. stay tuned... :) 15:06
ruoso mncharity, re slime... it's not really a problem... as it's an internal thing that probably no one will notice...
mncharity ok :)
sh mncharity: Thanks! I'm looking forward to it
mncharity ruoso: basically, not having a clear direction in which to push on parsing this morning, I thought I'd poke at the ast phases. given two paragraphs, one p6, one smop, I could explore turning one into the other. as a "somewhat different that emitting to p5/ruby" case, to help shake down architecture choices. 15:08
ruoso mncharity, please... keep trying on that... 15:09
mncharity err, 'keep trying on ' which? 15:10
15:11 tooo joined
gbacon was hoping someone had a brilliant flash and fixed pugs 15:12
mncharity lol 15:13
:/ 15:14
15:14 chris2 joined 15:15 Auzon left
ruoso mncharity, on smop ;) 15:16
mncharity Juerd: so it looks like the pugs evalbot is running r18093. Does that mean something broke it in r18094, or that feather's ghc got upgraded? Hmm, I'm basically just asking if feather's ghc version changed, and if so, when. Thanks. :) 15:17
15:17 pmurias joined
mncharity addressed to Juerd, but anyone with a feather account could check... 15:17
hi pmurias 15:18
pmurias mncharity: hi
mncharity ruoso: ah, ok :)
15:19 simcop2387 is now known as boobot, Psyche^ left, boobot is now known as simcop2387
gbacon mncharity: it's a weird thing.. when a Haskell program typechecks, it almost always runs as expected too 15:20
15:20 stevan_ joined 15:23 masak left
mncharity gbacon: let me guess, and the corollary is, when it doesn't, you're then out on a very non-common-case branch, with no tools? 15:23
I note that since the error still occurs with empty string input, it might perhaps be an initialization phase which is choking. Which suggests one might approach it by shutting down parts of initialization (as they're not going to be needed) and see when it stops failing? 15:26
sordith simcop2387: i see you
simcop2387 hello sordith
sordith hi! 15:27
gbacon yeah, I'm thinking it's a prelude problem
parsing it, I mean
sordith simcop2387: sorry for my private message earlier. did not want to ask on this channel for permission to do so. was curious about your opinion regarding a problem on another channel 15:28
mncharity err, really? The point which fails in the build is the prelude precompilation to yaml. Which just runs ./pugs. If you go... hmm... one sec...
simcop2387 sordith: lets take this back to pm
sordith roger that
mncharity gbacon: If I echo -n > src/perl6/Prelude.pm, and make, all the files named /Prelude/ seem to be updated, seemingly with the null input. But the Bool error still occurs. 15:32
gbacon nods 15:33
mncharity s/seemingly with the null input/ well, ./src/Pugs/Prelude.hs is has the "", the others, binary I just checked timestamps on and briefly eyeballed.
15:33 iblechbot joined 15:34 [particle1 is now known as [particle], rby left
mncharity which suggests there is no trace of Prelude left in pugs itself. 15:34
and then ./pugs -c -e "" fails, suggesting it's not input either. 15:35
which leaves internal initialization, no?
diakopter mncharity: I think when the rebuild fails on feather, it keeps the Last Known Good build. 15:37
15:39 FurnaceBoy joined
mncharity diakopter: right. I was wondering if feather is still using ghc 6.6.1 , in which case it's worth exploring around r18094 to see what changed in Pugs. Or whether feather's ghc got upgraded, in which case being at r18093 is just an artifact. 15:42
hmm... having looked at svn log, 15:45
r18094 was a big change to _runpugs_. So it looks like being stuck at r18093 is indeed an artifact, but of runpugs breaking. 15:46
So the evalbot isn't the right guide to when pugs was last working.
And we have another todo item, fixing evalbot.
or instead, I could simply be babbling confusedly. runpugs isn't the evalbot. sigh. so rewinding... 15:48
Looking at the svn log, it's not clear why the evalbot might have stopped updating at r18093. So perhaps someone with a feather acount could check if the ghc version was changed, and if so, when. 15:50
Just to perhaps provide a bit on additional insight into what is going on with pugs.
*of
15:52 cj_ is now known as cj
mncharity and if feather has any kind of build log, knowing what the failures were would be great too. 15:53
rhr fwiw, I use 6.6.1 and my pugs binary is r19220 - I haven't tried to build it since then
it wirks fine
*works
mncharity Good to know. Thanks! :) 15:54
gbacon rhr: it'd be interesting to know whether you can build the current revision 15:55
well, whether the resulting image can at least hobble along 15:56
rhr gbacon: let me make a copy of my repo and give it a try 15:57
mncharity My vague impression is it was the start of upgrading to 6.8.2 which broke it. 15:59
Reading through log, "hey! I forgot all about perl5/Pugs-Compiler-Perl6/ !".
so many p6 implementations and fragments...
pasteling "rhr" at 65.94.10.253 pasted "pugs build failure" (22 lines, 1.1K) at sial.org/pbot/30422 16:01
rhr that's just the end
16:02 Aankhen`` left
nothingmuch mncharity: memoization is not as efficient 16:02
"real" ag engines linearize the both-ways recursion
16:02 Aankhen`` joined
nothingmuch inherited attributes are passed only to the children that will need them for the synthesized attrs 16:03
and this is done in advanc e
whereas for dynamic languages you go down to the synth attr at the bottom, see what inherited attrs it needs, and then climb back up to the parent
pugs_svnbot r20064 | pmurias++ | [kp6] --profile option 16:08
diff: dev.pugscode.org/changeset/20064
lambdabot Title: Changeset 20064 - Pugs - Trac
mncharity ok, so I hypothesize pugs on 6.6.1 broke around r19956 - r19962, in the attempt to upgrade it to 6.8.1. Which suggests r19955 is the most recent working version. 16:10
pmurias: one thing you might add to kp6 is the ability to load one of it's own yaml dumps. then you could parse, dump, save that, and repeatedly rerun your emitter, without having to parse again. You would know better than I if that would actually be useful. 16:12
rhr I can test that. what's the most efficient way to get that rev from svn?
mncharity svn up -r19955 ?
rhr mncharity: OK, thanks
mncharity double checks... yep 16:13
rhr: thanks! :)
rhr it seems to be building OK [ 35 of 109] 16:14
mncharity gbacon: so it might be worth eyeballing svn diff -r19955:19963 ? to see if something got broken in the upgrade attempt? 16:20
I note svn diff -r19958:19959 modified Types, but I don't see anything obvious. 16:21
16:21 Psyche^ joined, Psyche^ is now known as Patterner
pmurias mncharity: thought about it too, the time to recompile the file on changes would still be annoing 16:21
sh I tried to also build r19955 using ghc-6.6.1 on Fedora 8, but it fails after building dependencies: "Setup: cannot satisfy dependency pugs-HsSyck-any"
mncharity oh, right, the emitter is kp6. :/ well, could always use pugsx1/misc/winter_jig/backend_bare_perl5/ ;) 16:22
gbacon mncharity: looks like a bunch of renamings 16:24
pmurias mncharity: no, i'm talking about parsing the changed file
rhr r19955 works, r19956 fails for me
pasteling "rhr" at 65.94.10.253 pasted "r19956 failure (end)" (11 lines, 890B) at sial.org/pbot/30423 16:25
mncharity sh: re 'fails after building', :( re 19955-19956 transition, good to know.
gbacon rhr: and your attempt to build HEAD failed also, right? 16:27
mncharity re pugs-HsSyck-any, i wonder if you svn up to the version you had working, whether you can actually build it from scratch. 16:28
pmurias mncharity: with --profile you can see what kp6 spends time on 16:29
pugs_svnbot r20065 | pmurias++ | [kp6] kp6 doesn't print what's doing with profiling on
diff: dev.pugscode.org/changeset/20065
lambdabot Title: Changeset 20065 - Pugs - Trac
mncharity The reason for the question is on my fedora 8, I noticed I now only have Cabal 1.2.3... and I'm not sure that's what ghc 6.6.1 came with. Perhaps the fedora upgrade to 6.8.2 breaks pugs via Cabal. In which case I would expect an attempt to do a make clean rebuild of -r19220 to fail similarly. 16:30
pmurias: :) neat! instrumentation++
pmurias ;) 16:32
gbacon mncharity: when we get it working, we really need to look at overhauling the pugs build
via cabal, that is 16:33
mncharity hmm, r19810 edited AST.hs
gbacon as it sits, it's a pretty big mess
mncharity that could be neat. 16:34
gbacon mncharity: well, cabal ought to be doing more of the heavy lifting 16:35
16:35 sordith_ joined
rhr gbacon: yes, see sial.org/pbot/30422 16:35
lambdabot Title: Paste #30422 from "rhr" at 65.94.10.253
gbacon and the cabal folks seem willing to support us where we need them
mncharity should also perhaps get a braindump from audreyt on whether, long term, the right thing is to incrementally improve pugs, or to rewrite. ie, I don't know if "pugs isn't architected the right way" played any role in it's getting stuck. 16:36
gbacon rhr: what version of Cabal do you have? ghc-pkg list Cabal
mncharity s/a hypothetical "pugs isn't.../
rhr Cabal-1.1.6.2
mncharity 1.2.3
curious 16:37
sh fwiw, I have Cabal-1.1.6.2
rhr I was doing a make realclean between each build
mncharity sigh. too many dimensions in this space.
sh mncharity: I yum remove'd ghc and ghc-6.8.2 and symlinked /usr/bin/ghc-6.6.1 to /usr/bin/ghc
sh is trying realclean now 16:38
mncharity starts a build...
sh gets still the same problem 16:39
I had this problem when I first tried it with ghc-6.8.1 a few weeks back, and found suggestions to edit some Cabal files and actually add dependencies 16:40
gbacon doesn't Pugs.cabal require at least Cabal-1.2? 16:41
mncharity [particle]: I had a dream last night that I woke up and there was a rakudo p6 -> yaml parser, and everything was happy. :) 16:42
SamB they split a lot of packages off from base recently
sh build_pugs.pl considers Cabal`1\.[1-9] to be $has_new_cabal 16:43
SamB: Yes, but I think this was after ghc-6.6.1, wasn't it? I"m using 6.6.1 right now 16:44
SamB yeah
gbacon grep cabal-version Pugs.cabal*
SamB I think so
16:44 barney joined 16:45 sordith left
SamB apparantly recent cabal has some kind of alternatives thing if you want to make the .cabal file work for both GHC versions... 16:45
16:45 sordith_ is now known as sordith
gbacon SamB: it's kind of freaky: cabal files are Prolog programs :-) 16:45
it fiddles with the values of flags until it finds a happy combination 16:46
sh So many programming languages mixed just to get a Perl 6 compiler...
gbacon rhr: the build error that you pasted is a cabal issue
SamB but doesn't prolog have a "cut" thing? 16:47
that makes things nasty?
gbacon SamB: yes, so not full-on, hairy, nasty, scary, Cirque du Soleil clown Prolog
cognominal_ sh, you want to be in too. :) are you bourne or c?
ruoso mncharity, www.perlfoundation.org/perl6/index....p_language 16:48
lambdabot Title: SMOP s1p Language / Perl 6
gbacon just the nice bits
ruoso in the end of the page, I added a full version of SMOP fib...
SamB and the kinds of problems posed to cabal-get are a lot simpler than those posed to prolog interpreters...
sh cognominal_: Hehe, I'm D
SamB they have a lot less search breadth, I believe
sh cognominal_: Yet another incomplete language 16:49
cognominal_ I think there is many D languages. /me checks the wp 16:50
indeed... 16:51
sh Hehe
cognominal_ "there is" or "there are"? btw 16:52
in french we use singular "il a plusieurs langages..."
s/il a/il y a/
sh there are 16:53
mncharity re lots of languages, prolog, I yesterday rediscovered svn.pugscode.org/pugs/misc/Grammars..._engine.pl :)
lambdabot tinyurl.com/28x3jn
gbacon thought ilya was Russian 16:54
PerlJam gbacon: he is. :) 16:55
gbacon PerlJam: I was kidding about cognominal_'s s/il a/il y a/
PerlJam I know. 16:56
sh Lol
PerlJam So much is lost in communication due to the limited bandwidth of IRC
luckily I hang around with a crowd where I can almost assume everything is sarcastic or condescending or whatever :) 16:57
cognominal_ silly me, I thought that was a non sequitur
gbacon cognominal_: well, it would have been better for me to have /me thought ilya was Russian, not French 16:59
it's all in the delivery *sigh*
cognominal_ you lost me 17:00
gbacon thought Ilya was Russian, not French 17:01
would that have been more understandable?
cognominal_ I don't think Ilya is French 17:05
PerlJam heh
cognominal_ ok you made me say it... :)
"il y a" is French but not Ilya 17:06
Juerd mncharity: Yes, ghc got upgraded.
mncharity ok, on an fedora 8, x86_64, I get a working pugs with ghc661-6.6.1-3.fc7 (for ghc), ghc682-6.8.2-8.fc8 (for Cabal 1.2.3), and pugs r19955.
17:07 tooo left
mncharity after hand repointing the /usr/bin/ghc(-pkg) symlinks at the 6.6.1 binaries. 17:07
17:07 eternaleye joined
gbacon cognominal_: yes, I realize that.. it was a (bad) joke 17:08
17:08 baest_ joined 17:09 barney left, baest left, rakudo_svn left, pmichaud left, buubot left, penk left, nipotaway left, silug left
cognominal_ ...better than the ones parrot pulls on me :) :( 17:09
17:10 barney joined, penk joined, silug joined, nipotaway joined, rakudo_svn joined, buubot joined, pmichaud joined, schmalbe joined
mncharity Juerd: around 2007-09-22 ? 17:10
Juerd mncharity: No, a week or so ago
mncharity (r18094)
rhr Juerd: could you update from docs/feather/syn_index.html on feather? It's supposed to happen automatically, but apparently hasn't happened for months 17:11
mncharity hmm. so we still don't know why evalbot is stuck.
gbacon mncharity: at least pugs is compilant again :-)
Juerd rhr: I don't know what the current way of making it happen automagically is
rhr: I have insufficient tuits to track everyone's changes to feather :(
rhr me either... 17:12
Juerd rhr: In general it works like this: someone does something, after a while loses interest or tuits, and it breaks and does not get fixed.
[particle] considers opening a tuit factory
Juerd rhr: I think we'll have to find ways of getting used to that.
Alternatively, stop setting up new stuff when you do have tuits. Things that don't exist, don't break 17:13
But that sounds even less attractive.
mncharity s/so we still don't know why evalbot is stuck./so we still don't know why evalbot got stuck./ we know why it can work at the moment. :)
*can't sigh
Juerd Anyone with tuits and clue can request sudo access.
17:14 ebassi left, ruoso left, jjore left, marmic left, ead left, jrockway left
mncharity lol 17:14
gbacon Juerd: short supply of such people :-)
Juerd gbacon: Yes. There are many people with clue in our community
Just not enough that also have tuits.
PerlJam Juerd: why would anyone want sudo access?
Juerd PerlJam: Because I maintain the base Debian system, and provide the machine itself
17:14 pmurias left
Juerd PerlJam: And I expect the community, the users, to handle everything else. 17:14
rhr well, it's not getting changed very often, so we probably don't need to resurrect the auto-update machinery. If someone with access could copy the current version that would be nice. 17:15
mncharity re "why would anyone want sudo access?", to have feathered round tuits?
Juerd I cannot be a regular sysadmin for a 100 login machine, sorry.
[particle] ...but...but...we'll repay you with kindness, you beautiful sysadmin, you! 17:16
Juerd And if I were a normal sysadmin for feather, I'd forbid almost all of what has happened to it :)
So nah, that's not the right formula for feather.
17:16 jrockway joined
mncharity I wonder if the tech is at the point we could do a feather with virtualization, and each service gets its own "box". 17:17
Juerd mncharity: We are at that point.
mncharity: There is a separate VM *already* especially for all the more important services
mncharity: But nobody has the time and knowledge to move those services there.
mncharity ah, ok. shiny. :)
Juerd There's also a separate VM especially for evalbots and the like, just so they can't crash or breach the other VMs 17:18
It was offline for a few weeks and nobody appears to have noticed.
PerlJam is especially glad he never asked for sudo access :)
Juerd mncharity: If you need a VM for anything special, just ask for it. 17:19
The hardware problem was solved
17:19 ebassi joined, ruoso joined, marmic joined, jjore joined, ead joined
mncharity thanks. np. just wondering if that would make your/our collective lives easier. sounds like it already is. :) 17:19
Juerd There's no longer a shortage of hardware resources
Now to fix the people problem... :)
Any suggestions?
mncharity sigh. 17:20
Juerd mncharity: No, it makes life harder.
mncharity: Each VM is yet another linux box to maintain.
17:20 jrockway left, baest_ left, sordith left, Aankhen`` left, FurnaceBoy left, simcop2387 left, IllvilJa left, integral left, pugs_svnbot left, peepsalot left, Jedai left, rafl left, SamB left, lambdabot left
Juerd But I'll gladly do it if other people take care of the actual services provided. 17:20
mncharity that's keeps being the issue doesn't it. the will to go out of business, and otherwise mangle things, is so strong.
ruoso mncharity, I had review the s1p code example... and now it really looks like how the code generated by s1p should look like...
Juerd mncharity: The fundamental problem is just people not being able to work on Perl 6. 17:21
mncharity ruoso: oops, sorry, popped it in my browser and got distracted. looking...
17:22 baest joined
Juerd mncharity: I have medical and personal problems myself, others may be kept occupied by their jobs, etcetera 17:22
17:22 jrockway joined
Juerd feather was originally a box you could compile your own pugs on. Someone thought it'd be great if feather just provided a *compiled* pugs. 17:22
17:22 sordith joined
Juerd That sort of thing is really nice while it lasts 17:22
But eventually it will break, and the person who set it up is no longer active. 17:23
I simply cannot maintain all of that. I tried, but I no longer will. Takes too much energy.
I tried for a few months only. Feather has survived for years, so it's not all the big deal it may appear to be right now.
17:24 baest_ joined, Aankhen`` joined, FurnaceBoy joined, simcop2387 joined, IllvilJa joined, integral joined, lambdabot joined, peepsalot joined, Jedai joined, SamB joined, pugs_svnbot joined, rafl joined, baest_ left, integral left 17:27 barney left, Psyche^ joined, Alias__ joined 17:28 Alias_ left, baest left, baest_ joined
mncharity Juerd: right. It has been, and is, a very nifty part of the p6 ecosystem. Defintely a source of warm fuzzies. :) 17:28
ruoso: eep. /me thinks I should have asked for something simpler. :) But I shall grovel over it. 17:29
17:29 alester is now known as AndyGibb 17:30 AndyGibb is now known as alester, integral joined
mncharity re "The fundamental problem is just people not being able to work on Perl 6.", hmm, well, that, and a enormous effort gets dissipated in our "lost in development path space" style. The confronted by so many options on what to eat, we starve. 17:36
17:36 chris2 left
PerlJam mncharity: who's lost? 17:37
mncharity Well, perhaps not the parrot folks. One may question the approach, but they have been slogging away at it. Everyone else though? 17:38
cognominal_ not lost, just scattered... 17:39
mncharity hmm, lost vs scattered... 17:40
lost has a nice implication of lack of situational awareness (eg, failing to utilize existing tools), lack of clear view of how the development path will play out, and... there was something else... 17:42
blank mind. so scattered, definitely. but I think also lost. 17:43
17:43 devogon joined, Patterner left, Psyche^ is now known as Patterner
mncharity lost: not quite knowing where you are, where you have been, how you got here, where you are going, how you are going to get there, or how to get back to where you have been. ;) 17:44
oh, and why you are going there. (I've had some odd conversation fragments with at least vaguely p6-ish people, in which key parts of "the whole point" have been missed). 17:51
ruoso mncharity, that example is good because it exercises the use of inner blocks... 17:52
mncharity ok :)
ruoso and the capturize thing 17:53
which is what replaces registers in SLIME
17:53 sh left 17:54 Psyche^ joined
ruoso and in case you didn't notice... the method was implemented using continuation passing style... 17:54
which is just one of the options
mncharity lol
17:56 Patterner left, Psyche^ is now known as Patterner 17:59 Jamtech joined 18:00 Jamtech left 18:08 Schwern left 18:10 Jamtech joined 18:17 Psyche^ joined
cognominal_ is postcircumfix syntax allowed for functions call/methods call? if so I suppose the infix < necessitates a space. 18:24
TimToady that is correct
cognominal_ which statament is correct? 18:25
TimToady when S03 says "the right side of =", it is not talking about the comma, but the precedence context in which the comma is parsed
that infix:['<'] necessitates a space
foo()<bar> is a has subscript on whatever foo() returns 18:26
*hash
and that context is looser than comma
(for list assignment, but not for item assignment) 18:27
cognominal_ ok
must reread S03. but my last question was is foo< bar > equivalent to foo( 'bar' ) 18:28
?
TimToady no
cognominal_ ok
TimToady and in fact foo< bar> is likely to be illegal syntax, since a function without () is treated as a listop, and listops require a space also 18:29
cognominal_ so foo< bar > does not parse?
ok
18:30 qmole left
TimToady on the other hand, .method<foo> is fine, since .method assumes no args in the absence of () 18:30
but it's still a hash deref
not an arg to .method
cognominal_ ok
18:30 qmole joined
TimToady mncharity: sorry, have had negative tuits recently for responding... 18:31
cognominal_ so without the preceding @ sigil or a [], every allowed circumfix defaults to be a hash deref 18:32
TimToady the problem with a backtracking STD is that it's very difficult to find an ordering that emulates longest token without introducing lots of negative lookaheads to rule out a longer token
and negative lookaheads imply you know exactly what language your rule is being used as a part of 18:33
18:33 Patterner left, Psyche^ is now known as Patterner
TimToady the same problems for casting into LR etc 18:33
the same problems for using AG
the whole point of STD is to get past having to know what language the current rule is being used in 18:34
mncharity hi. so completely ignoring when code blocks get run, I gather the negative lookaheads are needed to avoid successful but incorrect parses?
TimToady but yes, it will also run a lot faster with LTM (I hope)
mncharity :)
ruoso home &
mncharity by ruoso, tnx
18:35 ruoso left
TimToady cognominal_: I didn't understand your question 18:35
mncharity: I know there are places where the current STD falsely recognize a shorter token if you assume alternatives backtrack 18:36
mncharity and that mis-recognition doesn't cause a later parsefail?
(just to make sure I understand)
TimToady might not 18:37
18:37 ebassi left
mncharity ok 18:37
TimToady in theory
mncharity hmm.
TimToady for example, if you only recognize part of an infix
then the rest might be recognizes as prefix to the next term
mncharity token level longest isn't a problem. it's the non-local longest-ness. 18:38
cognominal_ TimToady, thx. I think I got it.
mncharity s/non-local in grammar/
TimToady yes, and even within a known grammar, the transitive calculation of token pieces hoisted from subrules can be non-trivial to understand the implications of 18:39
mncharity ok
hmm
err, wait. I'm not so much worried about what the right language design is, as "if I take this as a shortcut, will I get burned when trying to parse peoples' not-too-hairy p6 code". 18:40
TimToady you might, without a scattering of negative lookaheads, or reordering/splitting rules 18:41
I already had to split the pair recognition into two subrules and put at opposite ends of noun for that reason
I'm sure we'll run into many similar situations 18:42
so <fatarrow> and <colonpair> have to be recognized early/late with respect to similar looking nouns
mncharity ok. sigh. thanks. 18:43
TimToady if | were LTM rather than backtracking, that wouldn't matter
mncharity right. but have backtrackers. don't know if we have an LTM yet? 18:44
TimToady Perl 5 thoroughly explored the space of evil hand-coded lexers, and Perl 6 is trying to make the opposite mistake
mncharity lol one for the quote book
TimToady mostly have LTM
still doesn't handle <foo>* <bar> quite right 18:45
the other major infelicity is down in the matchify code, which doesn't properly track rule levels, so confuses priors that should be subrules from those that should result in a list within the top match object
but that's not related to LTM 18:46
when you run STD_run it uses gimme5 -fo which fails over to a backtracking match, so it tends to hide the current problems in LTM 18:48
I run with -fo when I'm trying to debug LTM, because it makes it fail faster 18:49
*without
mncharity :) 18:51
18:51 peeps[work] joined
mncharity Turns out the effort to get a yaml parse tree from parrot/rakudo has stalled out for lack of tuits. 18:52
:(
So I'm back puzzling over what approach to take, trying to get a parser.
TimToady well, there are always a surplus of antituits
mncharity can I combine them with scattered tuits and get energy? 18:53
TimToady depends on if you know how to photosynthesize from hard gammas
mncharity lol :) 18:54
ah, what a zoo.
TimToady the only thing keeps us all going is stupid persistence 18:55
and the main danger signal I saw in your README a week or so ago was that you sounded like you'd give up after a day or two if you didn't get immediate gratification :)
mncharity I'm not sure what the smiley is for wry weary grin.
TimToady yeah, there should be a smiley for gritted teeth... 18:56
mncharity If one defines gratification as other people being involved, then I'm afraid, yes. I'm done with solitary speculative p6 coding exercises. Need customer/colleagues. 18:57
TimToady on the other hand, everyone comes and goes on various rhythms, so we work with that too...
cognominal_ TimToady, in S03, list prefix precedence "The right side is looser than comma." ~~ s/right/left/ ? 18:58
TimToady and it's just everyone's responsibility when they're "on" to recognize when two other people are not in sync when they could be...
cognominal_ or is stupid persitence from me?
TimToady the right side of a list prefix is looser than comma 18:59
print 1,2,3 the comma is tighter than the right side of the print
so the print governs the comma
again, "the right side of" is talking about the context, not the contents 19:00
cognominal_ ok, I am dense 19:01
anyay I got the idea
TimToady basically, 1,2,3 is in the identical state after either "print " or "@foo ="
mncharity was just thinking "invisible prefix tokens" would be much easier to visualize if they were, well, visible.
19:03 bocksp joined
cognominal_ :) 19:03
TimToady which prefix tokens are you thinking are invisible? 19:04
mncharity list prefix
TimToady if you think of a list prefix as kinda like a (, it's really the other end that is an invisible ) 19:05
cognominal_ the context act slike a invisible token
TimToady 1 + 2 * 3
that's what precedence is all about
it's where all the invisible (...) go
at least with list prefix the implied ( is next to the operator 19:06
unlike the * above
anyway, the listops vs commas work exactly the same in Perl 5 19:07
the old examples of print 3,4, sort 2,1 19:08
cognominal_ but the = has a precedence wich varies with the context ?
s/wich/which/
TimToady $x= is a different prefix than @x=
cognominal_ indeed
TimToady that's different from P5, yes
mncharity re rhythms, part of the problem is that while I believe I *should* be spending significant time on p6 at the moment, and basically am, I am failing to get myself in consultant mode, so it's unclear if that time is being spent at all well. I knew I had a dependency on parrot yaml, but wasn't tracking progress. A resource dependency on the parrot folks, but didn't 19:09
make a point to know what I needed, and why it was important.
cognominal_ in lay man words $a = 1, $b = 1 and @a = 1, 2 both dwim
TimToady yes. we recently re-adjusted the dwimmery to simplify it 19:10
mncharity s/to know/to let them know/
we seem so close to making rapid progress, but how to close that gap... I've notes from 2005 that could have been written last week. 19:12
TimToady mncharity: of course, letting them know why you think it's important still doesn't necessarily make it move up their priority list, since parrotfolk are naturally VM-centric rather than language-centric
mncharity sure. necessary, not necessarily sufficient.
TimToady mncharity: well, I hate to sound cruel, but maybe we'd be better off if you had stuck around a little more in the meanwhile to keep us on track... 19:13
mncharity well, perhaps not necessarily necessary, but... not doing the explanation indicates not really trying.
TimToady but, again, I do understand the come-and-go thing, and don't mind working within that constraint 19:14
mncharity yeah. sigh,
TimToady it's the nature of volunteer work that it involves volunteers...
and nobody can keep going at their best pace constantly
esp if they have to earn a living elsewise 19:15
and/or have other interests they must pursue to stay sane
or their health fails
all of which slow me down
it's a wonder I get anything done at all :) 19:16
mncharity re wonder, :)
re 'come-and-go thing', basically, it repeatedly seems I'm no longer on the critical path, so I go elsewhere, but later it turns out things got stuck. If I really truly believed parrot would have a usable p6 by year end, my time would be better spent in OLPC land, or generating income. 19:18
Not fully believing that, I'm trying to find some why to get non-parrot p6 to a self sustaining and moving state, quickly, before I really do have to go generate income. 19:19
*way 19:20
19:21 Aankhen`` left
TimToady well, I think trying to find a consensus view of AST is very valuable to all projects 19:25
just as the STD project started when I realized the various parsers were diverging rather than converging
mncharity hey, redsix was no different than pugs. :) 19:26
TimToady AST is not where my brain is right now, and I'm glad to see anyone else's brain there
mncharity well, modulo the oop.
re ast, yeah, but the real question is social. we have, for instance, kp6. and pugs, and redsix, and v6 mumble, and... . But don't have people writing p6 code for them. I know why *I'm* not, and hypothesize the same obstacles have discouraged others, but... that doesn't mean removing some or all of them, and which, will be sufficient to start having people care about my code. :) 19:29
TimToady indeed, you have project management skills that are lacking in most of the rest of us 19:31
I certainly learned long ago that I had to delegate all that
which is the only project management skill I have...
mncharity :)
cognominal_ redsix? 19:34
mncharity misc/pX/Common/redsix/redsix 19:35
ruby implementation of p6 from... 2006? pugs-based grammar in mutant p5+subrules engine, ruby+p6 runtime. 19:36
cognominal_ mncharity, I am knee-deep in the rakduo compilation chain, I may able to write your yaml dumper within a week. No promise.
mncharity I think it's high-water mark in test passing was something like 20%? so weaker than pil2js, or pil-run. 19:37
but mostly due to lack of pursuing "the slog to get tests to pass". 19:38
cognominal_ probably easier than the subclassing stuff where I hit wall after wall. And no one is interested to check-in the patch that works
19:38 eternaleye left
mncharity re knee deep in compilation chain, great image. 19:38
cognominal_ jonathan is not around . I could convince him
De Gaulle said "Qd vous ete dans la merde jusuq'au cou, je n'en ai que jusqu'au genou" 19:39
when you have shit up to the neck, I am only knee deep in it
19:40 eternaleye joined
cognominal_ mais je suis seulement un homme grand et pas un grand homme. 19:40
depending on the place of the adjectif "grand" it means you are really tall or metaphorically tall 19:41
mncharity re week, I have to ponder. I had planned on a "through the end of March" time frame. I'm not comfortable getting exceptional effort from folks when I'm not fully clear on how it will pan out. 19:42
re subclassing stuff? 19:43
cognominal_ well, subclassing pmc to make them real p6 classes.
mncharity recently encountered the concept of being hardly able to refuse to follow someone into a filled basement to address a plugged sewer drain, when the person your following is half your height. ;) 19:44
cognominal_ also but the hash pmc have the same name as the Perl hash class. "extended method" are called in the pmc instead of the p6 class :( 19:45
s/also//
mncharity ah, ok.
cognominal_ I asked in #parrot and filled bugs but I think they know the problems but are after architectual stuff that may or may not fix the problems 19:46
mncharity what's the next part, the thing which getting this done depends on? 19:47
cognominal_ you mean your yaml dumper?
I think nothing but my buddin skill
mncharity err, no, if the subclassing stuff started working this afternoon, 19:48
what would that enable you to do this evening?
cognominal_ well, I may fix it by inserting some code in the right place so I get to understand the compiling chain which makes possible to write your dumper 19:49
mncharity I guess the question is, is it "I have this fix, and it's languishing", or "I have this fix, and I can't make progress on X unill...
cognominal_ so, it is probably easier to me to work on your stuff
I don't know if my fix will work for the subclassing. 19:50
mncharity ok, /me confused. :)
cognominal_ pardon my french :)
mncharity having a yaml dumper depends on this subclassing patch? or the hash name decolliding? 19:51
cognominal_ no
19:51 kane__ left
cognominal_ I don't think so. 19:51
mncharity ok. so I understand the have patch part, and basically what it's intended to address. I'm missing the connection to yaml dumping.
cognominal_ none 19:52
mncharity lol
cognominal_ except it makes me understand the compilation chain
that's the connexion
mncharity yaml dumping is something entirely different you could work on, instead of pursuing patch, and whatever next task depends on patch.
re understanding, ah... 19:53
cognominal_ yes
19:53 schmalbe left
mncharity just curious, is there are particular next task which depends on patch? or is the patch simply general moving forward. 19:53
cognominal_ chasing the subclassing also involves getting in lot of C and pir code so you give me the occasion to focus only on the compilation chain 19:54
well, i want to make rakudo move forward but the parrot team is now involved in implementing role 19:55
19:55 araujo left
cognominal_ and you, you want to get away without doing parsing? 19:56
mncharity well, I'd be very happy to have a yaml dump from rakudo. let's see, what are the risks... 19:57
re "and you, you want to get away without doing parsing?",
my objective is for people to be able to quickly compile, and have run fast, p6 written as if it were basically p5-semantics wrapped in p6 syntax. 19:58
19:58 araujo joined
mncharity to make it possible to start coding bigger things in p6. specifically, parts of p6, like the prelude, and regex engine. 19:59
the advantage of writing it in p6 being, even if one want's to through it away later, in favor of p6 which uses more language features, that it can be run on other p6-ish things, rather than being tightly tied to a particular parser-compiler-runtime-baselanguage. 20:00
but that's a project risk there. if you build a ball field where people can code simple (but richer tha kp6) p6, and make it easy to use (err, rather, quick to use - error reporting is likely to be suboptimal) and fast, will they come to use it. 20:02
other possible parsers are STD_red, and redsix. Both in ruby. Both written by me. Both as yet unused by anyone else. Rakudo in contrast has the nice property of being a multiperson active effort that people care about. 20:04
20:04 buchetc joined
mncharity another parser is kp6, and one option is to do the same fast compile and run, but for kp6 dialect. If that attracts interest, that would be sufficient to show this is a right path. But if not, one doesn't learn much. 20:05
cognominal_ I just want to make rakudo as a perl5 with a better syntax. But the parrot team is aiming for mmd and roles. Static typing is not even at the horizon. Not that should be very difficult compared to role and mmd 20:06
mncharity hmm, that might be an interesting low-hanging fruit. have to play with the new --profile.
cognominal_ higher hanging than I tought 20:07
as always
mncharity I actually like the mmd and roles approach. I don't mention them only because they are fairly straightforward on ruby, and p5, backends. 20:08
Also like macros, but can work around their absence with metaprogramming the p6 files.
20:09 pmurias joined
mncharity pmurias: re --profile, can you give a rough rule-of-thumb on where kp6 spends its time? 20:09
s/kp6, and one option is to do the same fast compile/kp6, and one option is to do the same fast compile (well, slowish parse?) 20:10
20:10 rdice left 20:11 REPLeffect joined
mncharity background, we have lots of starts at Prelude code. Which got stuck on pugs not being able to deal with them. It would be nice to have something which could start running them, and allow work on them to proceed. 20:12
also assorted fragments from the various backends.
I don't believe kp6 is a sufficiently large dialect for that. Eg, assignment would be nice. 20:13
hmm, that's not fair. binding is basically working for assignment in kp6. hmm, why *exactly*, in what way, is kp6 dialect insuficient for prelude work...? 20:14
pmurias: so, any particular reasons why you're not writing big chunks of "p6" code to be run by kp6? 20:15
:) 20:16
pmurias was backlogging ;) 20:17
mncharity: kp6 has current high startup cost 1.3 sec (haven't looked yet why, it didn't the last time i added profiling) 20:18
mncharity so anyway. some risks of spending time on the yaml dumper are... () it's not clear anyone but me wants to use it, which makes me a pointfailure () the result might be such that one doesn't want to use it? hmm, why... narrow coverage or bugs? could manage that risk by running --target-parse over the t/spec and all t/ and see how it does. hmm, that's an action item for me. 20:19
pmurias the parsing takes most time on bigger tests
mncharity: two reasons kp6 has a slow parsers and i have to be preparing for my end of secondary education math exam 20:21
mncharity re 1 sec startup, pugs is similar, but people have still tried to use it...
LOL
and giving kp6 a faster emitter,runtime wouldn't help with the slow parse.
ok, thanks. :) 20:22
trying to figure out what a sort of kp6-like p6 implementation would need to do differently to get used more. can't do anything about math exams, but can aim to be a faster parse. :) 20:23
that's another risk which could be checked - time rakudo --target=parse. make sure it's sufficiently fast. 20:24
sigh, I'm in "oh, no, you really shouldn't" mode.
cognominal_: it would be great if you could get the yaml dumping working. 20:25
If I had it right now, I expect I would be using it right now.
If you have any questions about what I'd like to do with it, or anything else, please just ask. 20:26
pmurias mncharity: the yaml-ast could be fed to kp6 quite easily 20:27
i think it's mostly an issue of fixing Pads 20:28
mncharity what's unfixed about them?
20:28 lichtkind joined
pmurias evaluator 20:29
mncharity and before that, a first question of, given your experience and timing measurements, would loading from yaml actually be helpful?
pmurias mncharity: i didn't make much measurements
mncharity right
hmm...
ok. so need fuzzy/hypothetical. "evaluator"? 20:30
mncharity looks at a dump...
20:31 justatheory joined
pmurias it's easy to recreate a closure for evaling code in so they get their scopes right 20:31
mncharity eep. Dumping code and helping it will work when it gets loaded. 20:33
*hoping
s/code/runtime subs/
pmurias mncharity: i can try to get the loading of ast if it would be of help
mncharity I think that's one for the interesting idea list. The next time someone has to wrestle with kp6, we have this as an option. 20:34
err, the "interesting idea but let's not worry about it now" list.
the having to undump working subs converted it, for me, from a "oh, this should be easy", to "oh, lot's of pain, and it may never work". 20:35
pmurias mncharity: the can be recreated 20:36
mncharity but with only a fuzzy/hypothetical need, the effort can't be written off as "it's easy, let's just do it so we have it around later, in case it's needed". spending real time which could instead go elsewhere, needs a real need, a real payoff, and we don't really have one for this, no? 20:38
one buys software with chunks of peoples' lives. which is a commodity one really wishes to spend carefully. 20:39
even if doing so is not traditional :)
friend with 6 hours of meetings a day comes to mind. dilbert land. 20:40
pmurias mncharity: you have to get the ast from somewhere in the first place 20:41
gbacon meetings--
mncharity right. so the only role for a kp6 undumper would be to run the kp6 parser once, and the emitter multiple times (without having to rerun the parser).
I don't think it feasible to generate dumped kp6 ast from another, non-kp6 source. 20:42
which isn't enough payoff to be worth the effort, no?
pmurias usefull for example when working on the emitter and recompiling the whole of kp6
mncharity indeed
pmurias mncharity: why? 20:44
i mean why it's not feasible
pmichaud (from scrollback: "parrot team is focusing on mmd and roles") I'm not sure that's an entirely accurate assessment. At the moment jonathan is the one with the most available tuits for doing work, and mmd and roles seems to be his particular focus at the moment. But it would be a mistake to claim that this is a purposeful direction on the part of the team.
in my case, I'm low on tuits, but my focus will be on the tools and getting basic parsing to work and cleaning up the class/object implementation a bit 20:45
cognominal_ pmichaud, when you get time , please read my bugs report and commit my patch :) 20:46
I will be happy to contribute to you focus
s/you/your/
pmichaud okay. but see "low on tuits" above. :-)
mncharity pmichaud: re focus, thanks :) hmm, maybe that paragraph should be cut and pasted to a wiki somewhere... 20:47
cognominal_ so the "when you get time"
pmichaud actually, things should get drastically better now (cross fingers, knock on wood, etc.) Paula _finally_ made it to her 2nd round of chemotherapy yesterday (four weeks late) and she seems to be having almost no debilitating side effects. 20:48
cognominal_ ho, I forgot about that. Good luck
pmichaud thanks
pmurias mncharity: writing a solid perl5 rule engine would solve most of kp6 problems 20:50
but is a high time investment
cognominal_ it is missing an operator grammar parser. 20:51
I mean the perl5 engine
mncharity re 'no debilitating side effects', that's really great. having seen alternatives.
pmichaud (paragraph cut and paste to wiki) -- yes, it might be good to do that. I think that many times people interpret progress in a given area (e.g., mmd, roles) as being a deliberate choice of focus among the leaders. But in many cases the things that move forward are simply those that scratch a particular implementor's itch. 20:52
and so the reason that rakudo has junctions, objects, mmd, roles, etc. is because we have one productive developer who really wants to focus on that. It's not that we're deliberately targeting that -- that's just the way the tuits fall. (I suspect this is true for many volunteer-based projects.) 20:53
cognominal_ suffering of severe atopic dermitis that is itchin as hell, I hate that expression. :)
pmichaud and yes, I'm _very glad_ that jonathan is doing the work he is doing -- it's all helpful.
cognominal_ if I scratch my itch, I bleed.
pmichaud cognominal_: sorry for the expression, I'll replace it with another if you have one handy for my future use :-) 20:54
mncharity lol
cognominal_ this is not the bleeding edge I want to be in. 20:55
TimToady I have been extremely itchy lately too, and was ruining some of my skin till I started putting steroids on it instead of scratching...
the weirdest thing is to have an itch inside your lung--there's no way to scratch it even if you decided to... 20:56
pmichaud oh, I've had that
cognominal_ beware of regular use of topical steroids, it ruined my skin.
TimToady I'm very careful with steroids
pmichaud itch in lung is awful -- keeps feeling like I needed to cough something up
cognominal_ occasionnal use is great so if you can keep it so.
TimToady I basically ruined a cornea with steroids once (by necessity), so I'm quite familiar with the local and systemic effects of steroids, and just how much to use and how to taper off. 20:57
back when I got shingles in my right eye, steroids were all there were for that. 20:58
(acyclovir was only in clinical trials at that time)
cognominal_ hopefully for me there is now new molecules for my AD like tacrolimus, and old systemic medecine newly applied to AD like methotrextate.
methotrexate! 20:59
TimToady my brother was on methotrexate for many years for psoriasis, but recently had to quit 21:00
21:01 knewt_ joined
TimToady anyway, eventually got the eye fixed with a cornea transplant, so that worked out, eventually... 21:01
21:01 knewt_ left
cognominal_ hopefully we don;t have to integrate gattaca to program perl 21:01
TimToady were eventually == after 14 years...
I don't think my genome is terribly close to perfect... 21:02
mncharity pmurias: re 'writing a solid perl5 rule engine would solve most of kp6 problems', well, there's misc/pX/Common/yet_another_regex_engine/Regexp_ModuleA.pm, but I wouldn't call it solid. Passes almost all the p5 re_tests, and their p6 variants, but leaves lots of gaps, and known problems. And while it has commits, i'm not sure it's set up to do ratcheting. 21:08
s/but leaves/but that leaves/ 21:09
I'm not suggesting it's less solid than kp6... but likely differently non-solid, which could obviously be a difficulty in swapping it in for kp6's. 21:11
pmurias mncharity: kp6 is really minimalistic
mncharity it's support for grammar declarations is a definite kludge, very fragile, just enough to be able to write them when you know what you are expecting to parse them. just a couple of line hack. which could be addressed, but fyi. 21:12
re minimalistic, it might indeed be feasible. I suspect so. I'd be glad to help someone else do it. It's just never reached the top of my own todo list. But I'd be glad to help. 21:13
pmurias mncharity: do you think it will be fast enough 21:15
is it ratchety?
mncharity it's not at all ratchety, but its also quite fast. 21:19
cd misc/pX/Common/yet_another_regex_engine 21:22
prove t
you used to be able to swap it in as the pugs regex engine, but I suspect that's bitrotted. 21:23
21:24 kyrbe joined
mncharity oh, but it looks like the t/regex tests can be run directly. as long as they haven't changed much in form (it's a "file scraper" kludge). 21:24
README has instructions, and a project history. 21:26
biggest pain was p5 5... 5.6? 5.8? was proving fragile, making one not want to refactor, because things might then undebuggably break. 21:27
21:28 kyrbe left
pmurias it look a bit opaque... 21:29
* looks
21:30 kane_ joined
mncharity the p5 code, yes, rather. 21:33
21:34 justatheory left 21:37 luqui joined
mncharity big picture is there's a backtracking engine factory, Match, a regex AST, which emitters which invoke the factory, a p5 syntax bootstrap, a p6 syntax bootstrap, and grammar api (implementation messy, but grammar interitance (p5) works, and rules are both executable, and objects (and so can be used for analysis and optimization)). 21:38
so it does a lot, but yes, the "can't refactor without introduction subtle p5 bug tickling which the test suite has insufficient coverage to detect" left things ugly. 21:39
Thus my hope to rewrite it in p6, or at least ruby. :) Though now with 5.10... who knows, perhaps would could simply do a cleanup path without danger. 21:41
s/would/one/
s/path/pass/ sigh
(btw, apropos STD and redsix, I misremembered. redsix was the year before. it was the yet_another_regex_engine which had me flailing about for a p6 grammar). 21:43
and correct OPP.
21:43 justatheory joined 21:52 silug_ joined
pugs_svnbot r20066 | buchetc++ | [INSTALL] update ghc 6.6.1 workaround 21:57
diff: dev.pugscode.org/changeset/20066
lambdabot Title: Changeset 20066 - Pugs - Trac
21:58 alester left 22:09 silug left 22:12 silug_ left
pugs_svnbot r20067 | putter++ | misc/winter_jig/sixen/sixcmd: Deal more gracefully with the user not having created a rakudo exe. And with an invalid implementation being requested. 22:27
diff: dev.pugscode.org/changeset/20067
lambdabot Title: Changeset 20067 - Pugs - Trac
22:28 eternaleye left 22:30 pmurias left 22:34 Chillance joined
pugs_svnbot r20068 | putter++ | misc/winter_jig/ast_rakudo/parse_t_spec.t: Created. Runs rakudo --target=parse against pugs t/spec/ .t files. 22:34
r20068 | putter++ | Currently 35 of 128 files are successfully parsed. One non-terminates (S29-conversions/ord_and_chr.t).
diff: dev.pugscode.org/changeset/20068
lambdabot Title: Changeset 20068 - Pugs - Trac
22:39 Quaver joined 22:44 iblechbot left 22:47 Jamtech left 22:51 japhb joined 22:52 cmarcelo left 22:56 buchetc left
pugs_svnbot r20069 | buchetc++ | [t/spec] coro examples 23:04
diff: dev.pugscode.org/changeset/20069
lambdabot Title: Changeset 20069 - Pugs - Trac
23:15 Limbic_Region joined 23:18 nipotaway is now known as nipotan 23:21 alester joined
bockmabe Hey, who's the current Pumking? 23:23
TimToady of which pump? 23:31
bockmabe perl6 23:32
TimToady perl6 is too large for one pumpking to think about, so we have many of them 23:33
Limbic_Region TimToady - perl 6 still binds rules on streams correct?/
bockmabe Ah, I see.
I was trying to see if there is a correlation between the IRC social network in #perl6 and the Pumpkingyness of those in #perl6. 23:34
TimToady Limbic_Region: that is currently specced to be mediate by cat objects that emulate lazy strings
*mediated
Limbic_Region effectively yes though 23:35
TimToady there needs to be some way to do it
Limbic_Region have you read the pm threads on the topic?
TimToady bockmabe: um, in #perl6 a pumpking is anyone who will pick of one of the balls and run with it for a while 23:36
bockmabe Gotcha.
TimToady Limbic_Region: not that I recollect
Limbic_Region: recent threads? 23:37
meppl good night
TimToady like Why does my Perl regex substitution for linebreak fail? 23:38
Limbic_Region no, not recent threads
sorry, am bouncing betwen channels
TimToady - if you are interested, I could /msg you the ones in particular I was thinking of
23:39 penk left
TimToady well, I do have plenty of other things to think about with what little brain I have at the moment, so I don't really have cycles for handholding PMers at the moment... 23:40
23:40 meppl left
TimToady I barely have enough wind to walk at a steady pace currently, and it's slowing me down rather... 23:41
Limbic_Region TimToady - handholding PMers isn't why I was going to suggest it. Perhaps fresh ideas about implementation is all
23:41 japhb left
TimToady as long as there's some way eventually for a regex to ask for more string when it runs out, it doesn't much matter how 23:42
but lazy strings are not really fundamental to p6, I don't think
that's why we have Cat objects
Limbic_Region ok 23:43
TimToady which will make an iterator look like a flat string, hopefully, to the extent necessary
cattable semantics may end up leaking back to normal strings if there's no overhead, but I'm not mandating it for now 23:44
originally we were just going to allow @foo ~~ /pattern/ to autocat, but decided it was a bad idea 23:45
23:53 silug_ joined 23:55 MarcWeber joined, MarcWeber left

Powered by Cro and the Raku® Programming Language.

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

Thank you!