pugscode.org/ | nopaste: sial.org/pbot/perl6 | pugs: [~] <m oo se> (or rakudo:, kp6:, elf: etc.) (or perl6: for all) | irclog: irc.pugscode.org/
Set by Tene on 29 July 2008.
TimToady ruoso: "These variables are compile-time aliases to $+! and $+/" 00:31
no, they can't be
ruoso what's the difference?
TimToady $+ is short fot $CALLER:: and excludes the one in the current scope 00:32
$+! can only refer to one of your caller's $! variables
and since $! is defined in every scope, it's always the first caller
ruoso but when you set $! it propagates to the caller, 00:33
TimToady no
ruoso missed something...
TimToady more likely, fail() sets CALLER::CALLER::<$!>
if you want to set your caller's $! directly, you must set $+! 00:34
ruoso hm... I think I got it...
TimToady the whole point of giving each scope it's own copy of these variables is to keep them straight; there can be no aliasing
many of P5's problems stem from too many routines trying to deal with global $_ and $! 00:35
anyway, context vars never use the + in their own scope, but just the bare variable 00:37
ruoso and functions that read by line lookup $+/ to see the contextual setting for that variable
TimToady which is especially important since the rw-ness differs from $foo and $+foo
araujo greets TimToady
TimToady ruoso: I don't follow that last idea 00:40
TimToady greets araujo right back 00:41
araujo :-]
ruoso confusing $/ meaning... 00:42
but anyway...
a regex call sets $+/ which is then available in the caller...
the same way that happens with $! 00:44
TimToady yes
bbl & # dinner 00:49
ruoso TimToady, do every block define $! and $/? or are just Routines that define them? 00:59
pugs: { my $a = "foo"; { $a ~~ /o/ }; say $/ }; 01:01
p6eval pugs: OUTPUT[Error eval perl5: "if (!$INC{'Pugs/Runtime/Match/HsBridge.pm'}) {␤ unshift @INC, '/home/evalenv/pugs/perl5/Pugs-Compiler-Rule/lib';␤ unshift @INC, '/home/evalenv/pugs/third-party/Parse-Yapp/lib';␤ eval q[require 'Pugs/Runtime/Match/HsBridge.pm'] or die $@;␤}␤'Pugs::Runti...
ruoso rakudo: { my $a = "foo"; { $a ~~ /o/ }; say $/ }; 01:03
p6eval rakudo 29834: OUTPUT[Null PMC access in get_string()␤current instr.: 'print' pc 12259 (src/gen_builtins.pir:7802)␤]
ruoso rakudo: { my $a = "foo"; { $a ~~ /o/; say $/ }; };
audreyt pugs: { my $a = "foo"; { $a ~~ m:P5/o/ }; say $/ }
p6eval rakudo 29834: OUTPUT[o␤] 01:04
pugs: OUTPUT[o␤]
ruoso so it seems that $/ is defined by routines and not by every code
so CALLER is not simply each code... but each routine
audreyt sub f { if 1 { g() } } 01:06
for &g, the caller is f, not the if block, afaik.
ruoso that actually makes things much more complicated 01:07
audreyt what you have in mind is context()
see S06:2113, S06:1958 01:08
caller() is a filtered context, so to speak.
it discard the non-routine parts
does that make things easier? 01:09
ruoso yeah... it does...
but maybe we should store the enclosing routine for each context 01:10
audreyt *nod*
bbiab
whereiskurt ?eval say "What does TimToady eat for dinner? :-)" 03:46
gaal ruoso: you were looking at Signature objects? Perl6::Signature on cpan 05:18
lambdabot gaal: You have 1 new message. '/msg lambdabot @messages' to read it.
gaal ported from pugs
more or less
and nothingmuch++ and clkao++ have insane glue to make those actually available as sub declarations in perl5 05:19
hmmm though apparently that's not on cpan yet? 05:20
audreyt it's code2.0beta.co.uk/moose/svn/Perl6-S...ure/trunk/ 06:42
lambdabot Title: Revision 5096: /Perl6-Signature/trunk
nothingmuch gaal: in moose + jifty repos I think 07:14
pugs_svn r21789 | lwall++ | [STD vs t] more Mutually Assured Construction 08:10
audreyt *lol* 08:11
clkao gaal: perl6::declare 08:36
i was expecting more craziness for yapceu but nothingmuch is bailing out 08:37
masak hm, is there an 'is inline' trait on subs? 08:45
audreyt Yes, S06:1681 08:57
masak audreyt: thanks 09:00
pugs_svn r21790 | audreyt++ | * Pugs.Lexer: "\d123" must now be written as "\c123" according to S02 v133. 09:10
dduncan on that note, what word does the 'c' refer to? 09:13
dduncan eg, how is a c for base-10 or decimal self-documenting? 09:14
or is it deCimal?
ludan hi
Tene hi ludan 09:15
Patterner c)haracter? 09:20
masak feels like a category error next to \b, \x and \o, in that case 09:21
gaal Moose! 09:35
speckbot r14571 | audreyt++ | * S02: A few more C<...> an C<<...>> blocks, Contributed by John M. Dlugosz++. 09:39
speckbot r14572 | audreyt++ | * Typo spotted by John M. Dlugosz++: 09:44
r14572 | audreyt++ |
r14572 | audreyt++ | method close is export () { ... } # Wrong
r14572 | audreyt++ | method close () is export { ... } # Right
audreyt it means we've moved from decimal to centimal ;) 09:53
masak 100t! 09:55
ruoso work & 10:07
gaal For any base, you're always at base 10. Except unary. 10:44
audreyt_ gaal++ # That's true even for quater-imaginary bases 10:57
avar audreyt++ # Yay, she's back! 11:54
lambdabot avar: You have 1 new message. '/msg lambdabot @messages' to read it.
avar lambdabot: @messages
lambdabot audreyt said 22h 17m 45s ago: community.haskell.org/~audreyt/re-e....05.tar.gz # should build everywhere
avar Yeah I saw that, the OBJECT= stuff works but I'm a bit afraid of modifying the gettext stuff
still, I guess it's needed, win32 doesn't have libint.h or whatever that gnu stuff was
gaal So... still no luck with the smoke. Once a single test hits the timeout, subsequent ones all time out. 12:03
I've added a ghetto "killall pugs" when the test doesn't terminate on time (getting the pid of the real bad pugs is tricky, as it's created deep in the harness) 12:04
anyone know a portable way to get all my child pids + their $*PROGRAM_NAMEs? 12:05
ruoso gaal, I'm pretty sure there isn't one 12:07
except for libraries that abstract different implementations 12:08
gaal looks at the plumbing 12:09
beh, we'd need a patch to Straps.pm:651 to record the pid 12:14
straps uses global filehandles, joy :/ 12:15
Maybe it's time for T::T::Model to use TAP::Parser... let's see if that's any better. 12:16
gaal growls. Straps calls itself obsolete and directs us to TAP::Parser,, but TAP::Parser doesn't come with something that actually runs a test. 12:18
avar gaal: how do you make the kids? fork? 12:20
gaal open |
avar the only portable way is making a record when you make the kids
ew
gaal which returns a pid, iirc
open | is actually pretty portable.
gbacon in the Pugs build, are the profiling libraries strictly necessary? 13:35
INSTALL talks about a prof make target, but is that not required? 13:36
gaal I doubt they're required. 13:37
gaal But I recommend trying :) 13:38
audreyt "make prof" should now do a ./Setup configure -p --user 13:39
and then ./Setup build
feel free to hack it in. :)
gbacon: do you have a commit bit?
masak how do I get the name of the Role that results from a <=> or a cmp? 13:43
perl6: say (3 <=> 5).WHAT 13:44
p6eval rakudo 29834: OUTPUT[Order␤]
..elf 21790, pugs: OUTPUT[Int␤]
masak rakudo gives Order, but I'd like to get "Increasing" somehow
ah, .perl 13:47
perl6: say (3 <=> 5).perl
gbacon audreyt: yes!
p6eval rakudo 29834: OUTPUT[Order::Increase␤]
..elf 21790, pugs: OUTPUT[-1␤]
gbacon audreyt: right, but running perl Makefile.PL runs cabal install -p 13:48
audreyt yes, "just in case" 13:49
a better way would be when you "make prof"
go back and detect and re-build the libraries taht needs to be profiled
try hacking Makefile.PL line 167ish and add a "pugs-prof" target?
and you can take off the default -p in Makefile.PL
makes sense?
gbacon nods 13:50
gbacon audreyt: I dug through the svn logs, but I'm still very curious as to how you fixed the "invalid grammar category" error that we were getting when trying to build the prelude 13:52
pasteling "rhr" at 65.94.11.172 pasted "pugs' Makefile.PL fails" (44 lines, 2.3K) at sial.org/pbot/31806 13:53
rhr is anyone else getting this error?
gbacon rhr: you might try installing cabal-install by hand 13:59
rhr I tried that, it doesn't help :( 14:00
audreyt rhr: oy. 14:07
if you can, rm -rf ~/.cabal and ~/.ghc and perl Makefile.PL again
most likely though, "ghc-pkg unregister --user zlib" would be sufficient.
(or "ghc-pkg unregister zlib") 14:08
sorry for being mostly away... preparing for a job interview coming up in 1.5 hours 14:09
rhr audreyt: I have a working pugs from a few days ago, so no hurry :) 14:12
audreyt okies :)
rhr audreyt: the rm method worked, thanks. have a good interview! 14:19
audreyt thanks!
gnuvince @seen audreyt 15:40
lambdabot audreyt is in #xmonad, #ghc, #haskell and #perl6. I last heard audreyt speak 1h 21m 8s ago.
audreyt yes? 15:47
gbacon audreyt: good interview?
[particle]1 i'm expecting her to reply, "actually, i'm in it now :)" 15:48
audreyt actually I'm in it now. 15:49
[particle] :P
gnuvince audreyt: nothing, I was just very happy to see you were back when I read that blog post of yours last week. 15:50
pmurias ruoso: hi 17:43
lambdabot pmurias: You have 3 new messages. '/msg lambdabot @messages' to read them.
ruoso hi pmurias
pmurias ruoso: how long does the SMOP talk take you? 17:44
ruoso it took me a little more than half an hour
gbacon audreyt: would you prefer to keep profiling enabled in the default pugs build? 17:47
audreyt gbacon: no, I don't think so
esp. on some platform prof and thread are mut. excl. 17:48
and I'd much rather thread than prof
pmurias ruoso: did anyone ask about polymorphic eval? ;) 17:49
ruoso pmurias, not really.. .but the audience was not as qualified as I would expect in yapc::eu 17:50
pmurias i'll read a bit about Glib or choose a different example 17:53
audreyt yay I got the job. :) see you tomorrow *wave* 17:54
pmurias *wave*
[particle]1 congrats audreyt 18:06
pugs_svn r21791 | gbacon++ | Add prof make target for profiling build. 18:09
pmurias ruoso: got a documentation pointer to how glib object look like? 18:29
ruoso in terms of data strucutre? 18:30
no, I don't...
pmurias i'll edit it out then
pmurias the SMOP talk will be my first talk, so any advice is appreciated ;) 18:32
ruoso ;) 18:37
pmurias getting cat food&
pugs_svn r21792 | lwall++ | [STD] let EXPR parse term after infix with a rule specified by the infix op 19:14
r21792 | lwall++ | (considered other kludges but this preserves 1-pass parsing)
r21792 | lwall++ | switch , parsing to using this mechanism plus <nulltermish>
r21792 | lwall++ | fix .= parsing by forcing <dottyop> as next term
r21792 | lwall++ | [pre_post.t] can't leave the space out between "is Foo" and the class block
r21792 | lwall++ | or it is interpreted as an arg to the Foo pseudo-adverb
pmurias mncharity: hi 19:21
mncharity hi pmurias 19:26
pmurias mncharity: i haven't seperated EmitSM0P yet, as it will be obsoleted soon 19:32
mncharity puzzled ("seperated EmitSM0P"?), but *nod* 19:33
I'm exploring getting STD's tryfile to accept the elf sources. 19:34
pmurias separated, sorry 19:35
mncharity: that should force removing =~ 19:38
mncharity pmurias: wasn't the spelling... I just don't know, or have forgotten, what EmitSMOP is being separated from, and with what objective.
pmurias from elfX into elfish/smop 19:39
mncharity re =~ ... do you know if STD will tolerate them if it sees an explicit =~ definition?
pmurias so it could be integrated into elf
mncharity: =~ is and explicit error
s/and/an 19:40
mncharity :( ah well. maybe =~~ ... 19:41
pmurias you now my attitude towards =~ 19:42
* know
ruoso pmurias, when you finish the slides you can add a second link to them in the wiki
pmurias ruoso: ok 19:45
ruoso: i'm currently thinking what to change 19:46
mncharity pmurias: re know, yes :) will go away as part of merging in regex/p engine. 19:47
ruoso pmurias, you can always throw that slides away if you like to... ;)
pmurias i 19:48
'm actually hoping to finish m0ld before YAPC::EU
Mold too 19:49
ruoso that would be very cool 19:50
pmurias, btw... have you seen the wiki page on lexical scopes?
TimToady mncharity: STD does not currently do anything with an infix:<foo> definition because pmurias got distracted :) 19:52
mind you, m0ld is a fine thing to be distracted by... 19:53
mncharity lol 19:56
TimToady but STD is up to parsing 98.44% of t/, and several of the remaining errors have to do with not mutating the grammar on operator defs
TimToady nap & # stayed up too late hacking last night... 20:00
mncharity re 98.44%, !!! There's a Perl 6 project milestone of note. 20:03
TimToady
.oO(still one percentile short of Ivory Soap) # rethunk
pmurias it might be that we need more cunning tests 20:05
mncharity more cunning tests++ 20:07
mncharity hmm, might call it "accepting" instead of "parsing"? I always find it very hard to tell if parses are correct until they are used, running, and passing tests. 20:09
re "Ivory Soap", their line is n % "pure". Where "pure" is without meaning. So you could feel free to declare STD's purity to be 99.99999%. 20:11
pmurias my limited experience with pixie is that the resulting AST is bug-riden, but it's hardly resonable to expect untested things to be correct
mncharity "it's hardly resonable to expect untested things to be correct", ooooo, I like that quote. Nice alternative to "if it hasn't been tested, it doesn't work". 20:14
ah, pixie! that's what I was looking for... 20:16
pmurias mncharity: pixie is currently broken as STD remove a {*} per rule 20:37
mncharity pmurias: np. i'm going to try a STD_red like approach - getting the Match tree and groveling over it, rather than using make(). 20:50
gaal mncharity: moose! 21:00
pugs_svn r21793 | pmurias++ | [m0ld] calls with positionals only are translated to bytecode 21:02
pmurias mncharity: using make is better as you might want to start modifing the grammar at some point from the actions
otoh you should be able to convert from one approach to the other easily 21:07
mncharity gaal: moose :) 21:09
mncharity pmurias: *nod*. at least for now, with a semi-static grammar, the advantage of greater debuggability outweighs future flexibility. 21:16
pugs_svn r21794 | pmurias++ | [m0ld] named arguments 21:44
gbacon transparency++ 21:51
mncharity pmurias: before I start exploring, any thoughts on what "Unable to parse class definition at ../../misc/elf/elf_g_src/EmitSimpleP5.pm line 946: ------> expecting nibbler (alt 02)" means? 946 is end of file. 22:03
oh, and the =~ have all gone away. ;) 22:04
(haven't measured the performance hit yet :/ )
TimToady probably a runaway string 22:09
mncharity yep. unescaped { in "". 22:10
dduncan question: regarding base-10/decimal numeric literals, what word does the 'c' prefix stand for? is it deCimal or something else?
TimToady admittedly could use a better message :)
your question was answered last night 22:11
but you vamoosed too soon :)
dduncan I'll check the logs ...
TimToady it's 'c' as in 'characgter'
we can't use \d because that already means something else in regex
and the idea is that you should be concentrating on producing the character, not on its radix 22:12
dduncan okay
TimToady on top of that, nobody should be using decimal for that anyway except in a few cases
but almost all the decimal characters people know are already better encoded some other way 22:13
\c9 == \t for instance
and unicode has pretty much settled on hex
dduncan now looking at the log, I saw the answer, and also the comment: feels like a category error next to \b, \x and \o, in that case 22:14
TimToady anyway, was just looking for a way to sneak in the decimal notation, and it turned out that current \c usage had holes in the right place
sure, but those are exceptional in not being base 10, and we have ten fingers
it just points out that you can't actually create the Perfect Language 22:15
dduncan indeed
pmurias gbacon: what did the transparency++ comment refer too?
dduncan but one can try 22:16
TimToady (as in the Search for the Perfect Language, Umberto Eco)
even trying has its dangers, if you don't understand the tradeoffs
gbacon pmurias: mncharity's "at least for now, with a semi-static grammar, the advantage of greater debuggability outweighs future flexibility."
TimToady and one of them in this case is the current cultural meaning of \d, which I don't wish to break
dduncan generally you just want to make the language good for the task its intended for 22:17
I'm not arguing for anything different than what you did
TimToady and in that larger scheme of things, I doubt people will be using \c that often
and when they do, it'll more often be \c[LINE FEED] 22:18
or \cG
and every now and then \c[13,10]
dduncan as for now, I'm trying to use Moose for the first time, and am learning how it is, and isn't like Perl 6 or Perl 5
pugs_svn r21795 | pmurias++ | [smop] hand compiled the m0ld example from the wiki and added as test 22:19
TimToady gotta commute, bbiab &
pmurias gotta sleep, bbiab& ;)
pugs_svn r21796 | putter++ | [elf_g] STD's src/perl6/tryfile r21792 now accepts the elf sources (no parse errors are reported). 23:19
r21796 | putter++ | One acceptable regression - t/xx-uncategorized/deref_doesnt.t now fails differently.
mncharity Next steps are () Get a Match tree dump from STD (in the same format as STD_red_run --format=p5a). STD_blue? () New IRx1_FromAST2. <> Elf bootstraps on STD_blue. Then... elf accepting, running, STD. Yielding bootstrapped front-end. 23:28
g'night all &

Powered by Cro and the Raku® Programming Language.

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

Thank you!