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.
ingy hola 02:13
anyone smart around? 02:14
;)
ingy shachaf: you're smart. are you around? 02:14
shachaf ingy: I don't know if I'd agree with that, but I'm around, at least. 02:15
ingy: Leaving in a few minutes, though.
ingy :) 02:16
shachaf ingy: Why?
ingy shachaf: I want to have a subset of python that I can compile to Perl 02:17
ingy do you know how I might do that? 02:17
I either need to override python compilation to bytecode 02:18
or find some parser generator to use
obviously I'd like a solution with the least work and also least language dependency
if you come up with any ideas, let me know 02:19
ingy fyi, everybody... shachaf is *extremely* smart :) 02:20
[particle] ingy: there's a subset of python running on parrot... languages/pynie 02:23
you could take the generated AST and transform it to perl 5
svn.perl.org/parrot/trunk/languages...Grammar.pg for the grammar 02:24
lambdabot tinyurl.com/2rfba6
[particle] i'm kinda close to getting the parrot AST to dump in YAML 02:25
you could take that and run
shachaf: sorry, i know how smart you are... hope i'm not stepping on your toes :) 02:26
ingy thanks [particle]! 02:43
[particle]: we should have a hackathon on some Sundays 02:44
ruoso wonders how hard would it be to write a ANSI C grammar... 12:52
ruoso considers writing a tool that analises the public headers of a library in order to verify ABI incompatibilities between versions...
pmurias ruoso: you could borrow it from somewhere 13:04
pmurias or do you want a perl6 rules instead of yacc & lex? 13:05
ruoso actually... I was thinking it probably better to do it from gcc directly... 13:06
integral parrot already has a c99 impl in it's languages/ dir 13:07
ruoso my idea is to parse a set of header files and generate an ABI description
then get that file from two different versions of a library 13:08
to check for incompatible changes...
integral was thinking of using LLVM's clang for something close to that
ruoso but oops... I've just realised I must have a C++ header parser... 13:09
the thing is that I'm packaging a library that uses non-public headers of another library... and it would be very important to notice when there are incompatible changes for headers the author of the other library doesn't care about... 13:10
pugs_svnbot r20025 | pmurias++ | [kp6-mp6like] methods and subs 13:15
diff: dev.pugscode.org/changeset/20025
lambdabot Title: Changeset 20025 - Pugs - Trac
ruoso hmmm... I was recommended to avoid trying that from inside gcc... it seems I would need a C/C++ grammar after all... 13:44
integral, do you think it would be possible to extend parrot's C99 grammar to support C++? 13:45
ruoso integral, btw... I didn't know llvm... it seems a very interesting project... 13:47
in fact... it seems that they already have what I want... 13:51
llvm.org/docs/TableGenFundamentals.html
lambdabot Title: TableGen Fundamentals
[particle] C99 is C++ 14:06
ruoso oh.. 14:12
ruoso always get confused by this...
[particle], "C99 introduced a number of new features that C++ does not support" -- en.wikipedia.org/wiki/C%2B%2B#Incom...ity_with_C 14:20
[particle] hrmm, i wasn't aware of that 14:22
ruoso++
avar C++ has never been a superset of C 14:43
Patterner int class; /* eat that C++ */
avar int *class = malloc(1); /* woo! */ 14:45
pmurias avar: would you be interested with continuing the clisp backend if kp6 bootstraped itself with a limited perl5 backend (equivalent to mp6 semanticly)? 14:50
avar Maybe:)
but bootstrap would be ++
Gothmog_ int i = 4 //* boom */ 2; 14:51
avar oh that's nasty 14:52
although the latest C standard supports // commends
*ents
Gothmog_ One could write funny programs that way which compile in both C and C++ and do completely different things. :)
Yeah.
Gothmog_ wam.tgoedderz.de/~tobias/cvscpp.c 15:00
ruoso I've just found another runtime that looks just like smop.... www.iolanguage.com/ 17:10
lambdabot Title: io
[particle] ah, i remember looking at that some time ago 17:11
ruoso the only difference seems to be the fact that they don't support different object representations...
pugs_svnbot r20026 | gbacon++ | FIXME: ! -e third-party/hsregex/LICENSE -> install fail (Cabal-1.3.3) 17:34
diff: dev.pugscode.org/changeset/20026
lambdabot Title: Changeset 20026 - Pugs - Trac
pugs_svnbot r20027 | lwall++ | [STD5] all trace info now goes to stderr 20:17
r20027 | lwall++ | [STD5_run] now simply dumps the return value
r20027 | lwall++ | [Cursor5] now refrains from flattening structure of $/
diff: dev.pugscode.org/changeset/20027
lambdabot Title: Changeset 20027 - Pugs - Trac
pugs_svnbot r20028 | buchetc++ | [t] S17 yield test 22:09
diff: dev.pugscode.org/changeset/20028
lambdabot Title: Changeset 20028 - Pugs - Trac
mncharity ingy: re python grammar, you've looked at pypy, yes? 23:17
mncharity pmurias: re Emit::AstYaml, yes, I added that. currently used by winter_jig. 23:20
mncharity ruoso: re 'how hard would it be to write a ANSI C grammar', for preprocessed C, not too hard. with backtracking you can even look just at the file itself, without obtaining types from headers. but for api work, you probably need preprocessor commands too. those... well, a few heuristics take care of a large majority of cpp usage. 23:23
but you are better off using an exiting tool.
mncharity oh, re 'C++ header parser', that's an entirely different kettle of fish. more like a tractor trailer of fungus. 23:24
to a first approximation it's impossible to do right, so everyone does hacks. so find an existing hack. 23:25
mncharity pmurias: re 'Emit/MiniPerl6Like.pm', have you looked at misc/winter_jig/backend_bare_perl5/ ? if bootstrap isn't part of your vision, six_kp6_bare_p5 and emit_from_kp6 may be useful. kp6 -Cast-yaml does the parse, and the emit is done in p5. so it's faster and easier to write. and doesn't use visitor injection, so you can inherit from the emitter to create derivatives. 23:29
question/comments would be most welcome. see emit_from_kp6_pl_generate, a work in progress emitter to simple but fast p5. 23:30
avar: ^^ same observations re cl. 23:32
[particle]: re "i'm kinda close to getting the parrot AST to dump in YAML", yay. looking forward to it. 23:34
avar mncharity: you mean the yaml stuff? 23:36
mncharity cj: re KindaPerl6::Visitor::Emit::Generic, the backends are reaching out to an mp6 runtime "dodge the bootstrap" file, but other than that, my fuzzy impression is there is little sharing/sharable code, in part because there is not much compilation or analysis happening. but it might be useful to add helper methods to the ast nodes. all the emitters are injected into them. :/ 23:38
avar: well, yaml decoupling kp6 parse from emit. so the emit can be done in p5. for which there is now assorted support in misc/winter_jig. 23:39
cj: but once parrot can dump an ast as yaml, I'm not sure how much role kp6 keeps playing going forward. 23:40
avar It might have a role as a p6-in-p6 compiler, but if you just want a yaml ast from it you're probably better of using parrot 23:41
mncharity (aside from being a valuable source of code to scavange of course) 23:41
mncharity avar: once parrot exposes yaml, then we're finally able to do full backends. those are not hard. we've just been long wedged by the absence of a parser willing to report on oo. with backends, we can actually start writing non-trivial perl6 implementation components in p6. which makes an incremental bootstrap less interesting. 23:44
avar Yes, parrot is turning out to be somewhat interesting after all, recently 23:45
Is somebody working on dumping yaml? 23:46
mncharity kp6 is parser, compiler infrastructure, emitters, runtimes. runtimes can be reused, the emitters are more easily written in p5 or a fuller subset of p6, the infrastructure can benefit from a reimplementation pass, and the parser becomes less interesting if parrot is usable. 23:47
the raw code, and certainly the concepts, underlying all those parts can be reused of course.
re someone, [particle]. 23:48
"i'm kinda close to getting the parrot AST to dump in YAML"
not sure what that means for when.
but I hope to roll something like misc/winter_jig/kp6_ast for it 23:49
so it's easy to metaprogram p5 backends. 23:50
though since the parrot ast is likely to be in greater flux than the kp6 ast has been, the approach taken may be somewhat different, to reduce maintenance cost. 23:51
mncharity gets warm fuzzies from seeing STD5_run edited. it's been rather a while since my p6-universe code was visibly banged on by anyone else. ;) 23:54
cj: once the parrot ast is out, it would be nice to start building p5 classes around it for doing semantic analysis. 23:59
ingy hi mncharity
mncharity hi ingy
ingy pypy eh?
cj mchhmm?
heya ingy!
ingy cj!

Powered by Cro and the Raku® Programming Language.

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

Thank you!