»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg camelia perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend!
Set by moritz on 3 May 2013.
sorear nqp: my int $x; say($x) 00:01
camelia nqp: OUTPUT«0␤»
sorear nqp: my int $x; say($x)
camelia nqp: OUTPUT«0␤»
00:09 raiph left 00:12 raiph joined
sorear heh. I'm reading documents on delimited continuations in Haskell... Pugs is listed as a major user :D 00:17
ssutch hmm. now im stumped. i need to add comment support to this grammar, comments can appear any where whitespace can. 00:20
i guess i have to switch back to token, add back all the <.ws> but make my own .ws that includes '// comments' 00:21
seems like there should be a better way 00:22
sorear ssutch: just make your own ws
00:23 ecocode left
sorear if you override ws, rule will call your override instead of the builtin 00:23
ssutch oh, cool
can i do something like token ws { <.super> | <comment> }
sorear I'm not sure if it's actually possible to do that 00:24
but there's not much point since the default is "token ws { <!ww> \s* } 00:25
"
ssutch ok 00:26
00:27 konundra joined
ssutch what is the difference between <.ws> and <ws>? 00:27
colomon . means it doesn't capture 00:30
o/ 00:31
colomon hopes he is right. :)
masak you are :)
00:32 konundra left
masak "don't store in a keyed slot in the match object" 00:32
ssutch r: gist.github.com/samuraisam/5786535 00:33
camelia rakudo b2072f: OUTPUT«｢␤ // import some shit␤ // more shit // and more shit␤ package fart ;␤｣␤ proto => ｢｣␤␤»
ssutch my ws just seems to gobble everything up
dalek p: 70b445b | sorear++ | t/jvm/01-continuations.t:
Test dynamic binding interaction
 00:34
p: 6cfee5a | sorear++ | docs/continuations.pod:
Updates for documentation
sorear dinner&, then implement GatherIter
(and probably find more bugs) 00:35
colomon sorear++ 00:36
ssutch how can i make this rule stop when it reaches a newline: token comment { '//' [.]*? <.nl> $ } 00:49
it just keeps ticking right along after hitting a <.nl> (which is defined as { \xA | "\r" \xA | "\r" | "\f" })
colomon try \v (vertical whitespace) instead? 00:50
though the $ there worries me
00:50 risou_awy is now known as risou
ssutch how so? 00:51
colomon what are you trying to test there?
newline followed by end-of-string?
ssutch no 00:52
yeah that doesn't make a whole lot of sense, having that ther
colomon if you take out the <.nl> and $, you can use $$ (end-of-line) there
00:53 kivutar left
ssutch hey! :) 00:54
that works great
colomon probably should have used it a few times in the ABC grammar. :)
ssutch so now i have:
token comment { '//' [.]*? $$ }
token ws { <!ww> [\s | <.comment>]* }
which appears to play well with all the rules 00:55
(c/c++ style // comments)
using the $$ seems a lot more efficient to match
(watching the Grammar::Tracer)
colomon cool 00:56
it certainly is a nice concept to have in your toolkit 00:57
01:00 FROGGS_ joined 01:04 FROGGS left
timotimo_ ssutch: what is [.]*? supposed to do? shouldn't you be able to do .*? instead? 01:12
maybe it was bigger once
ssutch ill try it out
it appears to work the same 01:13
timotimo_ i don't think it makes an actual difference 01:17
maybe the resulting AST is a bit smaller 01:18
hm, the dumping of QAST::Regex nodes seems a bit boring
masak 'night, #perl6 01:19
timotimo_ ssutch: fwiw, if you change '(' ... ')' into '(' ~ ')' [...], you'll get nicer error messages when the closing part is missing 01:34
r: say "(foo" ~~ /'(' \w+ ')'/;
camelia rakudo b2072f: OUTPUT«Nil␤»
timotimo_ r: say "(foo" ~~ /'(' ~ ')' [\w+] /;
camelia rakudo b2072f: OUTPUT«Unable to parse expression in ; couldn't find final ')'␤ in any FAILGOAL at src/stage2/QRegex.nqp:1037␤ in regex at /tmp/jsCohX6DZ_:1␤ in method ACCEPTS at src/gen/CORE.setting:10509␤ in method ACCEPTS at src/gen/CORE.setting:683␤ in block at /tmp/jsCohX6DZ_…
ssutch oh cool 01:35
timotimo_ r: given "(foo" { when /'(' ~ ')' [\w+] / { say "test" } default { say "yup" } }
camelia rakudo b2072f: OUTPUT«===SORRY!===␤Confused␤at /tmp/qaRBXGFeb1:1␤------> when /'(' ~ ')' [\w+] / { say "test" } ⏏default { say "yup" } }␤ expecting any of:␤ postfix␤ statement end␤ statement modifier␤ statement modifie…
timotimo_ r: given "(foo" { when /'(' ~ ')' [\w+] / { say "test" }; default { say "yup" } }
camelia rakudo b2072f: OUTPUT«Unable to parse expression in ; couldn't find final ')'␤ in any FAILGOAL at src/stage2/QRegex.nqp:1037␤ in regex at /tmp/fkjBYDR09G:1␤ in method ACCEPTS at src/gen/CORE.setting:10509␤ in method ACCEPTS at src/gen/CORE.setting:683␤ in block at /tmp/fkjBYDR09G…
timotimo_ and now you need to wrap it in "try" :|
i'm too tired to decide if this is something bad or good.
01:40 konundra joined
ssutch is there a way to get the full path of the current file $?FILE is just the base name 01:55
01:55 raiph left
timotimo_ do you want to see my very ugly solution to figuring out what file & line my function was called from? :D 01:57
ssutch hah, yes 01:58
timotimo_ github.com/timo/perl6-vortrag/blob...te.pm6#L49
but as you can see it's also relative paths
which would give you only the basename + extension for your "main script" anyway
ssutch looks like i can go say IO::Path.new($?FILE).absolute 01:59
timotimo_ r: say $?FILE.path.absolute;
camelia rakudo b2072f: OUTPUT«IO::Path</tmp/wRG91sC32C>␤»
sorear ssutch: the canonical way to write that is token comment { '//' \V* }
ssutch r: say $?FILE.path.absolute.directory
sorear \N* rather
camelia rakudo b2072f: OUTPUT«/tmp␤»
02:00 raiph joined
ssutch token comment { '//' .*? $$ } is what i have 02:00
so it should just be token comment { '//' \N* } 02:01
sorear yeah, that's much slower
ssutch what is the difference if you don't mind?
sorear \N* means "match a sequence of non-newline characters"; it's compiled to a single op 02:02
.*? $$ means "match 0 characters. check for newline. if you don't find one, repeat with +1 character"
it requires going through the backtrack engine 02:03
i suppose the long way could be optimized to the short fast way, but that's not done yet
sorear attempts to figure out how to build a rakudo-jvm against HEAD NQP 02:04
ssutch wow, thanks, sorear 02:12
sorear wow? 02:14
raiph r: sub foo () { \(:bar, :baz) }; say foo<baz>
camelia rakudo b2072f: OUTPUT«True␤»
raiph r: sub foo () { %(:bar, :baz) }; say foo<baz> # also works 02:15
camelia rakudo b2072f: OUTPUT«True␤»
raiph r: sub foo () { (:bar, :baz) }; say foo<baz> # could this work, and would it compile time check?
camelia rakudo b2072f: OUTPUT«postcircumfix:<{ }> not defined for type Parcel␤ in method gist at src/gen/CORE.setting:10161␤ in method gist at src/gen/CORE.setting:893␤ in sub say at src/gen/CORE.setting:11047␤ in block at /tmp/JahlZPe2Ha:1␤␤»
timotimo_ r: sub foo () { (:bar, :baz).{} }; # is this in rakudo yet?
camelia rakudo b2072f: ( no output )
timotimo_ r: sub foo () { (:bar, :baz).{} }; say foo<baz> # is this in rakudo yet?
camelia rakudo b2072f: OUTPUT«postcircumfix:<{ }> not defined for type Parcel␤ in method gist at src/gen/CORE.setting:10161␤ in method gist at src/gen/CORE.setting:893␤ in sub say at src/gen/CORE.setting:11047␤ in block at /tmp/a03Z1vxXCj:1␤␤»
ssutch i dunno, just interesting 02:16
raiph timotimo_: snap ;)
timotimo_ i thought .{} was intended to be a nice contextualizer, like %(...) would be 02:17
ssutch how do i pass a smart matcher to dir, it takes a named parameter called test, so i am trying it like this dir 'dirname', :test</proto$/> to find all things that end with proto 02:18
timotimo_ :test</proto$/> will turn that into a string
colomon to dir method? 02:19
timotimo_ r: :test(/proto$/).perl.say;
camelia rakudo b2072f: OUTPUT«"test" => regex(Mu : Mu *%_) { ... }␤»
timotimo_ r: :test</proto$/>.perl.say;
camelia rakudo b2072f: OUTPUT«"test" => "/proto\$/"␤»
ssutch ahh
sweet, thanks
timotimo_ <...> is, in most places, the new quotewords syntax:
r: <foo bar baz>.perl.say 02:20
camelia rakudo b2072f: OUTPUT«("foo", "bar", "baz")␤»
sorear has an up-to-date sorear/rakudo fork
ssutch right 02:21
02:23 JimmyZ joined
raiph "a Parcel ... lets you access the named [elements] by their name" (from S08) 02:23
timotimo_ it does? oh! 02:24
er, hold on, how does it interact with multiple elements of the same name?
raiph Reading specs to see what the deal is re parcels; followed link to a 2009 dialog; apparently "spec" isn't abbr for "specification" but rather "speculation"... :) 02:31
timotimo_ :3
ssutch shouldn't slurp take an IO::Path, in addition to a string. seems silly to write (slurp $path.Str)
colomon maybe? but there's definitely a slurp method on IO::Path 02:32
r: say dir 02:33
camelia rakudo b2072f: OUTPUT«IO::Path<star> IO::Path<src> IO::Path<.subversion> IO::Path<.bashrc> IO::Path<nom-inst1> IO::Path<toqast> IO::Path<test3.pl> IO::Path<.profile> IO::Path<t> IO::Path<nom-inst2> IO::Path<nom-inst> IO::Path<toqast-inst> IO::Path<toqast-inst2> IO::Path<examples> IO::Pa…
colomon r: say "test3.pl".path.slurp
camelia rakudo b2072f: OUTPUT«No such method 'slurp' for invocant of type 'IO::Path'␤ in block at /tmp/FgIstDplmx:1␤␤»
colomon :\
timotimo_ r: say "test3.pl".IO.slurp;
camelia rakudo b2072f: OUTPUT«use Test;␤␤{␤ "fooxbar" ~~ /x/;␤ is $/.prematch, "foo", ".prematch works";␤ is $/.postmatch, "bar", ".postmatch works";␤␤ is ("foo" ~~ /(f)/).kv.join("|"), '0|f', '.kv sees positional';␤ is ("foo" ~~ /$<x>=[f]/).kv.join("|"), 'x|f', '.kv sees names';…
timotimo_ hum.
labster .IO.open.slurp? 02:34
raiph r: say slurp("test3.pl")
timotimo_ er, wha?
camelia rakudo b2072f: OUTPUT«use Test;␤␤{␤ "fooxbar" ~~ /x/;␤ is $/.prematch, "foo", ".prematch works";␤ is $/.postmatch, "bar", ".postmatch works";␤␤ is ("foo" ~~ /(f)/).kv.join("|"), '0|f', '.kv sees positional';␤ is ("foo" ~~ /$<x>=[f]/).kv.join("|"), 'x|f', '.kv sees names';…
timotimo_ r: "test3.pl".IO.open.read().chars
camelia rakudo b2072f: OUTPUT«Not enough positional parameters passed; got 1 but expected 2␤ in method read at src/gen/CORE.setting:11184␤ in block at /tmp/0h98j2DU9H:1␤␤»
labster yeah, that might need fixed 02:35
timotimo_ anyway, i should be heading towards bed.
colomon labster: IO::Path.slurp should be there, no? I don't see it in the spec, either, but....
labster Perhaps, but I just have decide which is appropriate for IO::Handle, and which for IO::Path 02:36
I don't think the whole "one giant IO object" thing worked out so well. 02:37
colomon that one seems like it should be on both, IMO 02:38
labster slurp, spurt, and open, should be in both 02:39
anything else? 02:40
colomon lines? 02:41
labster lines currently requires an open file handle, but that could be changed too. 02:42
oh, never mind, .get opens it for .lines 02:43
colomon the thing about slurp, spurt, and lines is they all make sense on an entire file
labster I've been thinking about defining .succ and .pred for IO::Path, too.
timotimo_ based on dir()? 02:44
labster Either that, or just inherit from Str.succ. I haven't decided.
colomon .... what's the use case?
labster r: say "file01.txt".succ 02:45
camelia rakudo b2072f: OUTPUT«file02.txt␤»
colomon that does seem pretty handy
labster Yeah. based on dir() would be the most magical, but it would also be slow, as it would have to read the directory and sort on each successive call. 02:46
colomon I think the succ thing is better if it *isn't* based on dir. 02:48
timotimo_ agreed
timotimo_ disappears
labster maybe in a module then :)
good night timotimo_
colomon o/ 02:49
isn't there a readdir method for iterating through a directory? 02:50
labster contents
the same as dir(), it's a lazy list. 02:51
But no, we don't have directory handles like Perl 5.
colomon contents is the sub/method name? 02:52
labster yes. Unless you can think of something better. 02:53
colomon how's it different from dir?
labster dir() === IO::Path.contents
Is there a use case for P5 style readdir? 02:54
colomon labster: not that I can think of, if dir is lazy.
I just thought it was there. Might have just done too much p5 programming lately. :) 02:55
labster opendir ==> readdir ==> closedir is one of my less favorite parts of p5
Well, I have a whole bunch of path methods to add now: chmod, rename, link, symlink, slurp, spurt, lines, succ, pred 02:56
colomon labster: it's brill compared to the C equivalent, isn't it?
labster I think I'm glad I'm so ignorant of C. 02:57
02:57 RedditAnalytics left
colomon C is opendir, readdir, closedir too, actually. 02:58
let's celebrate our dir
labster reads rosetta code 02:59
by Jove, you're right!
03:07 tgt left
ssutch how can i tell which line/character grammar.parse fails on 03:08
colomon grammar debugger? 03:19
oooo, rakudo-jvm builds again, and does 2 ** 3 correctly! 03:20
03:22 woosley1 joined
JimmyZ ssutch: github.com/jnthn/rakudo-debugger 03:23
sorry, it's github.com/jnthn/grammar-debugger
ssutch JimmyZ: i am using that, it's just a little impractical
JimmyZ oh 03:24
ssutch my test files are ~500kb so it takes like 5 mins
to get to the place where it ies
03:25 SamuraiJack joined 03:27 benabik left 03:33 preflex left
sorear o/ labster 03:34
colomon \o sorear
03:35 preflex joined, ChanServ sets mode: +v preflex
Teratogen I would think that threads would have been built into the core of Perl 6 from the beginning. 03:37
Why try to bolt it on afterwords?
JimmyZ Teratogen: Patch welcome 03:39
Teratogen it's too late to patch
it should have been in the specification!
lue perlcabal.org/syn/S17.html # since 13 June 2005, according to the top of the file 03:40
Teratogen: ^^^
(yes, it needs plenty of work, but it is there) 03:41
labster \o sorear 03:44
sorear Teratogen: are you here to be productive or just to complain? 03:45
labster hyper ops, async, hyper, and race have been in the specs for a long time.
Teratogen i am here to be productive! 03:46
JimmyZ Ok, then it's not late
Teratogen I just want threads in Perl 6 so I can write Android apps in Perl 6!
labster tl;dr: none of us were smart enough to figure out parrot threads, so we're trying to build a thread implementation on a platform we do understand. The implementations should inform the specification, but the basic ideas are already included.
sorear Teratogen: niecza has had threads for years 03:47
real, working threads
Teratogen cool
sorear and you just ignore them and pretend they don't exist
it makes me sad/mad
labster correction: sorear++ and au++ are smart enough to implement threads.
lue
.oO(It just so happens that Parrot's threads were apparently problematic, thus rakudo hasn't implemented S17 fully)
 03:52
geekosaur I thought parrot didn't have weortking threads until recently 03:53
*working
and the current state of them is deemed unsatisfactory
(i.e. they may work for some things but are not yet viable for rakudo)
sorear weird threads 03:54
lue I've never been too involved with rakudo or parrot, so I'm not the most reliable source on the subject :) . 03:55
geekosaur my impression was that parrot had an impossible thread spec that wanted to be all things to all people, and within the past 6 months or so it was thrown out and some kind of actual thread implementation was added. but as sorear said, "weird threads" 03:58
...so, if you want Rakudo to have a working thread implementation, either help Parrot get real threads or help get rakudo-jvm going :) 03:59
(or one of the other backends e.g. MoarVM) 04:00
..or use Niecza which has threads now 04:01
sorear I see there's still plenty to do with rakudo-jvm 04:04
(It crashes when trying to print sorry messages)
geekosaur tbh I didn't think it worked at all yet, they were still working on nqp-jvm? 04:05
hackathon may have gotten more done than I realized
sorear I spent the hackathon working on nqp-jvm 04:06
got it passing all tests by the end
geekosaur ah
sorear hmm, we might need to split class files eventually 04:09
the constant table of CORE.setting.class is more than half full! D:
36,416 entries 04:10
Teratogen you mean constants like 3.14159?
sorear Teratogen: docs.oracle.com/javase/specs/jvms/s...l#jvms-4.4 04:12
Teratogen neat 04:20
04:29 Patterner left 04:31 Psyche^ joined, Psyche^ is now known as Patterner
sorear blegh 04:42
.tell jnthn so uh.. does take have to be a control exception? because supporting continuation operations within an exception handler = </3. also I'm not immediately seeing how to catch the take exception 04:48
yoleaux sorear: I'll pass your message to jnthn.
sorear i guess making the nqp exception engine not a continuation barrier would be horrible, but not *unspeakably* horrible 04:58
05:05 JimmyZ left
diakopter sorear: my intuition tells me that would add a lot of checks.. am I far off? 05:17
05:19 Teratogen left
sorear not so much "add lots of checks" as "rewrite nested loops as goto soup" 05:21
.seen jnthn 05:22
yoleaux I saw jnthn 13 Jun 2013 12:18Z in #perl6: <jnthn> airport, stuff...be back sometime &
diakopter merge the frames? I thought nqp/rakudo's compilers put all scopes in their own frames
sorear I'm not sure what frames have to do with this
what problema re you trying to solve? 05:23
diakopter I didn't see how you could goto across frames
sorear diakopter: the exception engine is java, not nqp 05:26
05:29 Teratogen joined
diakopter is that different from the "nqp exception engine" you mentioned above? 05:30
sorear the nqp exception engien is written in java
github.com/perl6/nqp/blob/master/s...g.java#L50 05:31
diakopter well so far you've answered none of my questions directly, and I can't seem to infer what you want me to infer from your indirect answers; afk&
sorear this needs to be modified to not be a continuation barrier
i'm answering to the best of my ability, please ask better questions 05:32
ssutch r: '23.0' ~~ /<[1..9]>\d+!\B/ 05:33
camelia rakudo b2072f: ( no output )
ssutch r: '23.0' ~~ /<[1..9]>\d+!\B/
camelia rakudo b2072f: ( no output )
ssutch r: '23.0' ~~ /<[1..9]>\d+\B/ 05:34
camelia rakudo b2072f: ( no output )
ssutch hm, trying to match 1..9 that doesn't have a period at the end 05:36
sorear .ask jnthn Why does dieInternal return a value? 05:37
yoleaux sorear: I'll pass your message to jnthn.
ssutch to differentiate between ints and floats
05:42 kaare_ joined 05:43 dylanwh left, dylanwh joined
sorear ssutch: <!before '.'> *but* you don't have to do this 05:43
ssutch: it's better to use a proto regex and let longest-token semantics do the distinguishing 05:44
ssutch okay
sorear if you have [[ proto token foo {*}; token foo:A { abc }; token foo:B { def } ]] then it acts a lot like [[ token foo { abc | def } ]] 05:45
ssutch sorear: gist.github.com/samuraisam/5790917
sorear ssutch: as a matter of style, I prefer to use the bare :foo form rather than :sym<foo> *unless* <sym> is used in the body 05:47
ssutch ahh okay, that makes sense
sorear ssutch: so somewhere higher up you want [[ proto token value {*}; token value:int { <int-lit> }; token value:float { <float-lit> }; token value:bool { <bool-lit> } ]] 05:48
ssutch so what i did before was token constant { [<int-lit> | <float-lit> | <bool-lit> | <str-lit> | <ident>] } 05:49
sorear or you could flatten it out. [[ proto token value {*}; ...; token value:oct { '-'? 0 <[0..7]>* }; ...; token value:sym<true> { <sym> }; ... ]
05:51 JimmyZ joined, Chillance left
ssutch playing with this 05:57
this is what i have now: gist.github.com/samuraisam/5790917 05:58
05:59 bluescreen10 joined, bluescreen10 left, bluescreen10 joined
sorear why are nuls special-cased in the grammar? 06:00
06:00 bluescreen10 left 06:01 bluescreen10 joined 06:02 bluescreen100 left
ssutch in the strings? 06:02
sorear yes 06:04
ssutch it was in the ebnf for this language
sorear I see you've hit on a very important and non-obvious grammar optimization (using a repeated negated character class to quickly slurp up runs of ordinary characters in a string) 06:05
ssutch that's good 06:07
06:08 zby_home_ joined
ssutch ok so this is what i have now: gist.github.com/samuraisam/5790917 it didn't fix the issue with matching a part of a float as an int 06:08
however, if i change token int-lit:sym<dec> to { '-'? <[1..9]>\d* <!before '.'> } it works just fine 06:10
sorear strange. it most definitely SHOULD work
ssutch shrugs
sorear r: grammar G { token TOP {^ <num> .* }; proto token num {*}; token num:int { \d+ }; token num:flt { \d+ '.' \d+ } }; say G.parse('12.34abc') 06:11
camelia rakudo b2072f: OUTPUT«#<failed match>␤»
sorear wat 06:12
r: grammar G { token TOP {^ {say 1} <num> .* }; proto token num {*}; token num:int { \d+ {say 2} }; token num:flt { \d+ '.' \d+ } }; say G.parse('12.34abc')
camelia rakudo b2072f: OUTPUT«1␤#<failed match>␤»
ssutch le bug? 06:13
sorear r: grammar G { token TOP {^ <num> {say 2} .* {say 3} }; proto token num {*}; token num:int { \d+ }; token num:flt { \d+ '.' \d+ } }; say G.parse('12.34abc') 06:14
camelia rakudo b2072f: OUTPUT«#<failed match>␤»
sorear r: grammar G { token TOP {^ <nuxm> .* }; proto token nuxm {*}; token nuxm:sym<int> { \d+ }; token nuxm:sym<flt> { \d+ '.' \d+ } }; say G.parse('12.34abc') 06:15
camelia rakudo b2072f: OUTPUT«｢12.34abc｣␤ nuxm => ｢12.34｣␤␤»
sorear r: grammar G { token TOP {^ <nuxm> .* }; proto token nuxm {*}; token nuxm:int { \d+ }; token nuxm:flt { \d+ '.' \d+ } }; say G.parse('12.34abc')
camelia rakudo b2072f: OUTPUT«#<failed match>␤»
sorear n: grammar G { token TOP {^ <nuxm> .* }; proto token nuxm {*}; token nuxm:int { \d+ }; token nuxm:flt { \d+ '.' \d+ } }; say G.parse('12.34abc')
camelia niecza v24-76-g3e65d84: OUTPUT«｢12.34abc｣␤ nuxm => ｢12.34｣␤␤»
06:15 xinming left
sorear apparently rakudo doesn't like the :foo short form very much 06:15
so much for my style advice - sorry 06:16
ssutch yeah, i had to change it back
(no problem, i appreciate all the help)
sorear anyway my example demonstrates rakudo preferring :sym<flt> when appropriate because it's longer
23:15 <+camelia> rakudo b2072f: OUTPUT«｢12.34abc｣␤ nuxm => ｢12.34｣␤␤» 06:17
so.. the FUNCTIONALITY works
[\d+]? is kind of silly, btw 06:18
ssutch don't need the brackets
sorear I mean you can just write \d*
\d+? is something very different, don't write that 06:19
(I beleive it's a non-greedy quantifier)
ssutch hah, okay, thanks
sorear + (1 or more) * (0 or more) ? (0 or 1)
(0 or 1) times (1 or more) = (0 or more) :)
06:23 twigel left 06:24 cognominal joined
eternaleye ssutch: sorear: The short :foo form interacts poorly with Actions in my testing, and I've also found that declaration order changes the behavior of proto/multi tokens in a grammar (!!!) 06:27
In ways that look like they aren't consistently LTM, no less. 06:28
cognominal Backto Paris. Nancy french workshop was nice. lizmat and wendy were great as usual. Too bad jnthn and masak could not make it. We did the Chartreuse Party place Stanislas after an impressive son et lumère.
www.lorraineaucoeur.com/evt-10359/s...atuits-ete
ssutch r: '23.23' ~~ /'-'? <[1..9]>\d* <!before '.'>/
camelia rakudo b2072f: ( no output )
sorear eternaleye: per spec, declaration order is used as a tiebreaker. I think rakudo is being a bit eager to call ties.
eternaleye Mm 06:29
ssutch r: say '23.23' ~~ /'-'? <[1..9]>\d* <!before '.'>/
camelia rakudo b2072f: OUTPUT«｢2｣␤␤»
sorear ssutch: 3 isn't a period, so 2 is not-before-period
eternaleye ssutch: Because of ^^^, try putting the 'float' rule/token above the 'int' one and see if that lets you leave off the <!before>
sorear cognominal: have you seen jnthn anywhere recently?
oh
I should read more before commenting 06:30
cognominal sorear, I was not in YAPC::NA
sorear cognominal: jnthn has been irritatingly quiet on #perl6 today, I briefly wondered (because I didn't read your comment well) if that was because he was busy at FPW 06:31
ssutch how can i get that pattern to continue matching 06:32
eternaleye r: say '23.23' ~~ m:ratchet/'-'? <[1..9]>\d* <!before '.'>/ 06:33
camelia rakudo b2072f: OUTPUT«｢23｣␤␤»
eternaleye Hmm
ssutch well, i want it to fail because that's a float
eternaleye Yeah
sorear eternaleye: suggest not making the before and after the same
eternaleye sorear: Ah 06:34
sorear ssutch: you're building increasingly ugly workarounds for LTM problems
eternaleye r: say '23.25' ~~ m:ratchet/'-'? <[1..9]>\d* <!before '.'>/
camelia rakudo b2072f: OUTPUT«｢25｣␤␤»
ssutch sorry, LTM?
sorear ssutch: longest token matching
ssutch: the rule that says if you can match 2 with one rule or 2.5 with another, take the longer match 06:35
sorear needs to do dynamic-wind-y stuff for tc.handlers 06:36
ssutch ah ok, well, i guess i need a workaround for now
eternaleye r: grammar Foo { proto token num { * }; multi token num:sym<float> { '-'? \d* [ '.' \d+ ]? }; multi token num:sym<int> { '-'? \d+ } }; say Foo.parse( '23.25', :rule<num> )
camelia rakudo b2072f: OUTPUT«｢23.25｣␤␤»
sorear ssutch: try flattening it (instead of having value go to int, and int go to {oct,dec,hex}, do it all at the value level) 06:38
and I second eternaleye's suggestion to play with the order
eternaleye r: grammar Foo { proto token num { * }; multi token num:sym<float> { '-'? \d* [ '.' \d+ ]? }; multi token num:sym<int> { '-'? \d+ } }; class FooActions { proto method num { * }; multi method num:sym<float> { make 'Float: ' ~ $/.Str }; multi method num:sym<int> { make 'Int: ' ~ $/.Str } }; say Foo.parse( '23.25', :rule<num>, :actions( FooActions.new ) ).ast
camelia rakudo b2072f: OUTPUT«Cannot call 'num:sym<float>'; none of these signatures match:␤:(FooActions : Mu *%_)␤ in method num:sym<float> at src/gen/CORE.setting:440␤ in any !reduce at src/stage2/QRegex.nqp:667␤ in any !cursor_pass at src/stage2/QRegex.nqp:631␤ in regex num:sym<float> at…
eternaleye r: grammar Foo { proto token num { * }; multi token num:sym<float> { '-'? \d* [ '.' \d+ ]? }; multi token num:sym<int> { '-'? \d+ } }; class FooActions { proto method num { * }; multi method num:sym<float>( $/ ) { make 'Float: ' ~ $/.Str }; multi method num:sym<int>( $/ ) { make 'Int: ' ~ $/.Str } }; say Foo.parse( '23.25', :rule<num>, :actions( FooActions.new ) ).ast 06:39
camelia rakudo b2072f: OUTPUT«Float: 23.25␤»
eternaleye I always forget the signatures on Actions.
ssutch moving the order around made it work 06:40
eternaleye ssutch: FYI, note the :rule I passed to .parse. This will make testing your grammar VASTLY easier.
ssutch (with <!before '.'>)
eternaleye: thanks for the tip
eternaleye You can use that to test individual rules, while still doing actions as normal, rather than trying to test everything via either TOP or smartmatching. 06:41
sorear ssutch: does it also make it work without the <!before> ? 06:42
06:42 benabik joined
ssutch sorear: yes, it works fine without the <!before> 06:43
well, for this section, running a larger test now (it goes really slowly with Grammar::Tracer on) 06:45
06:53 nitestryker_ joined
ssutch well, shoot, now it doesn't work with hex eg 0xFFF 06:53
an issue with LTM? putting <!before <[xX]>> at the end of the float rule makes it work 06:55
eternaleye ssutch: Well, my float rule also matches int. 06:56
Remove the [ ]? from around the second part.
ssutch: The issue is likely that float (which is first) matches the leading 0.
ssutch yes
eternaleye ssutch: As a rule of thumb, put the *most restrictive* rule first.
Since 0xFFF won't match things int or float match, but int or float will eat it up, put hex before them. 06:57
My mnemonic is 'Picky eaters get first pick' 06:58
ssutch ok,
eternaleye Once LTM is fully working, it'll be unneccessary, but for now ordering things like I describe mostly does the job. 06:59
ssutch ok, that can work if i flatten it, is it possible to use proto token implementations as a part of other rules eg proto token x { * }; token x:y { … }; token t <x:y>
because i need to use int-lit elsewhere besides just in constant 07:00
07:01 cognominal left 07:02 cognominal__ joined
eternaleye r: ( grammar { token x { * }; multi token x:sym<y> { <z> }; proto token z { * }; multi token z:sym<foo> { 'blah' }; } ).parse( 'blah', :rule<x>, :actions( (class { proto method x( $/ ) { * }; multi method x:sym<y>( $/ ) { make $<z>.ast }; proto method z( $/ ) { * }; multi method z:sym<foo>( $/ ) { make "Success, bwahaha!" }; } ).new ) ).ast 07:03
camelia rakudo b2072f: OUTPUT«===SORRY!===␤Method 'rxtype' not found for invocant of class 'Integer'␤»
07:04 cognominal__ left 07:05 cognominal joined
dalek p: 8db2525 | sorear++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/ (2 files):
Add a simple way to set exception verbiage at runtime

Set NQP_VERBOSE_EXCEPTIONS in your environment. This is not the ideal interface, but it will do for now.		 07:06
p: 4deb1c7 | sorear++ | src/vm/jvm/ (2 files):
Refactor compilation of continuation-aware ops, allow any op to be easily designated so
eternaleye r: grammar A { token x { * }; multi token x:sym<y> { <z> }; proto token z { * }; multi token z:sym<foo> { 'blah' }; }; class AActions { proto method x( $/ ) { * }; multi method x:sym<y>( $/ ) { make $<z>.ast }; proto method z( $/ ) { * }; multi method z:sym<foo>( $/ ) { make "Success, bwahaha!" }; }; A.parse( 'blah', :rule<x>, :actions( AActions.new ) ).ast
camelia rakudo b2072f: OUTPUT«===SORRY!===␤Method 'rxtype' not found for invocant of class 'Integer'␤»
eternaleye r: grammar A { proto token x { * }; multi token x:sym<y> { <z> }; proto token z { * }; multi token z:sym<foo> { 'blah' }; }; class AActions { proto method x( $/ ) { * }; multi method x:sym<y>( $/ ) { make $<z>.ast }; proto method z( $/ ) { * }; multi method z:sym<foo>( $/ ) { make "Success, bwahaha!" }; }; A.parse( 'blah', :rule<x>, :actions( AActions.new ) ).ast 07:08
camelia rakudo b2072f: ( no output )
eternaleye r: grammar A { proto token x { * }; multi token x:sym<y> { <z> }; proto token z { * }; multi token z:sym<foo> { 'blah' }; }; class AActions { proto method x( $/ ) { * }; multi method x:sym<y>( $/ ) { make $<z>.ast }; proto method z( $/ ) { * }; multi method z:sym<foo>( $/ ) { make "Success, bwahaha!" }; }; say A.parse( 'blah', :rule<x>, :actions( AActions.new ) ).ast
camelia rakudo b2072f: OUTPUT«Success, bwahaha!␤»
eternaleye ssutch: ^^^ calling a proto from one branch of a proto
*another
ssutch cool 07:09
07:10 cognominal left 07:12 bbkr_ joined 07:15 ivan``_ joined 07:16 ivan`` left, bbkr left, diederich left, cosimo left, huf left, p5eval left, Hor|zon left, silug left, silug_ joined, huf joined, Hor|zon joined, cosimo joined, p5eval joined, skids left 07:17 skids joined
ssutch hot damn, it parses the spec file 07:17
labster \o/ 07:19
sorear "the" spec file?
\o/
07:20 cognominal joined
diakopter sorear: lololol 07:20
labster We need a grammar to parse the synopses and automatically generate a P6 implementation. 07:22
eternaleye labster: Well, I think that's called 'irc://freenode/net/#perl6' 07:23
:P
*freenode.net
FROGGS_ .oO( use English? this could generate grammars for all programmings languages )
sorear o/ FROGGS_
FROGGS_ hi sorear
labster morning FROGGS_
FROGGS_ hi labster :o)
hi all
sorear FROGGS_: p5 already has an English.pm module! we could import that with v5.pm! 07:24
FROGGS_ sorear: but it totally does something else
I already have a tiny bit of that
sorear FROGGS_: I found myself wondering earlier today if v5.pm would allow you to use p5 code together with my delimited continuations stuff :) 07:25
FROGGS_ well, what exactly is your "delimited continuations stuff" ? 07:26
but basically everything that nqp offers is somehow available in v5
(in case the Perl 5 syntax isnt to restrictive) 07:27
too*
sorear nqp-jvm: my $cont := nqp::continuationreset(0, {3 * nqp::continuationcontrol(0, -> $c { $c })}); say(nqp::continuationinvoke($cont, {25})); 07:28
camelia nqp-jvm: OUTPUT«No registered operation handler for 'continuationreset'␤ in compile_op␤ in as_jast␤ in as_jast␤ in as_jast_clear_bindval␤ in <anon>␤ in <anon>␤ in compile_var␤ in as_jast␤ in as_jast␤ in <anon>␤ in <anon>␤ in compile_op␤ in as_jast␤ in as_jast␤ in <anon>␤ …
sorear pff, does ot seem to have rebuilt
FROGGS_ would it say 75? 07:29
I'm not sure that I understand what it does
sorear sorear$ ./nqp -e 'my $cont := nqp::continuationreset(0, {3 * nqp::continuationcontrol(0, nqp::null(), -> $c { $c })}); say(nqp::continuationinvoke($cont, {25}));'
75
07:29 cognominal left
sorear github.com/perl6/nqp/blob/master/d...ations.pod 07:30
JimmyZ yeah!
sorear o/ JimmyZ
FROGGS_ reads
JimmyZ sorear: aloha 07:31
sorear wonders if yeah! was meant for sorear
JimmyZ yes 07:32
congratulations on your continuation work :P 07:33
FROGGS_ hmmm, I dont understand it ó.ò 07:34
it feels a bit macroish
07:35 woosley1 left 07:37 cognominal joined
dalek p: 986b39d | sorear++ | docs/continuations.pod:
Add a conjecture for improving the nqp::continuationclone API
 07:39
07:40 snearch joined 07:48 cognominal left
sorear nqp-jvm: my int $i; try { ""~[]; CATCH { nqp::resume($!) unless $i++ } } 07:54
camelia nqp-jvm: OUTPUT«No registered operation handler for 'resume'␤ in compile_op␤ in as_jast␤ in as_jast␤ in <anon>␤ in <anon>␤ in compile_op␤ in as_jast␤ in as_jast␤ in <anon>␤ in compile_all_the_stmts␤ in as_jast␤ in as_jast␤ in <anon>␤ in compile_all_the_stmts␤ in as_jast␤ …
sorear This is nqp version 2013.02.1-50-g0968572 built on JVM 07:58
that's what I get from the evalbot :/
07:58 aghbas_ left
sorear ponders fixability 07:59
08:01 tomyan joined 08:07 ecocode joined 08:12 JimmyZ left 08:13 tomyan left 08:17 JimmyZ joined 08:25 tomyan joined 08:36 FreedomKnight joined 08:39 woolfy joined 08:42 ssutch left
woolfy At the French Perl Workshop, somebody told me he likes Parrot. Mostly the champagne-like wine that is made in the Jura. www.jura-vins.com/details-domaine-p...UR,189.htm 08:50
OK, I am bringing Perl-wine to Perl-meetings. Now I have to drag around champagne-like bottles with Parrot as well? 08:51
Liz just tells me no. The wine-domain is 300km south of here, just a bit too much of a detour on our way home tomorrow.
sorear just a bit :) 08:53
TimToady at the moment we're a couple hours north of Toulouse, and I finally have ssh access again... 09:00
nwc10 on a train? 09:03
TimToady in a timeshare 09:08
woolfy TimToady in Toulouse? So you could maybe have visited the French Perl Workshop... boohooo.... (or is it Toulouse, North Carolina or something like that).
TimToady the schedule wouldn't have worked, since Aron had to be Nice yesterday
after coming from Cambridge
woolfy It was only 936 km, just a 10 hour drive. :-) 09:09
nwc10 that sounds far more relaxing 09:10
woolfy I know, Toulouse is farther from Nancy than from Nancy to Berlin.
TimToady well, we drove yesterday for about 10 hours already...
woolfy Ow wow, you must be completely sore...
TimToady well, my clutch leg ought to be sore from an hour in a traffic jam, but it doesn't seem to be 09:11
woolfy But you know, to know that you are in the same country that I am in and it would just to be too much of a hassle to come over and give you a hug... Raaahh!
TimToady c'est la vie
woolfy Biensur!
nwc10 TimToady: do you dive a manual at home, or is it just a holiday treat? 09:12
sorear
.oO( TimToady rites manuals, he doesn't ride them... )
TimToady I haven't had a manual for several years now, but drove one for most of my life 09:14
TimToady learned to drive in a manual, even 09:15
nwc10 which, I assume, is as unusual in the US as it is common in Europe
TimToady I'm sure if I hadn't been walking all over Texas and England the last two weeks, my clutch leg would have been much sorer 09:16
woolfy You did walk quite a bit in Austin. Just the walks between the two buildings of the YAPC were a nice exercise. 09:17
09:19 bbkr joined, tomyan left 09:20 dmol joined
bbkr hi. is there a way to access raw bytes in IEEE754 format from Num type? maybe through some NQP tricks? 09:20
09:28 JimmyZ left 09:29 lizmat_ joined, lizmat left 09:30 tomyan joined
lizmat_ good morning #perl6! 09:31
github.com/perl6/specs/issues/54 Auth/Ver information is UNIT based, but specced as class etc. Propose -unit-
lizmat_ hopes TimToady will have some time to look at issues 09:32
sergot hi ! o/
09:33 atroxaper joined 09:39 tomyan left, Timbus left 09:40 Timbus joined 09:41 tomyan joined
sorear hi lizmat_! hi sergot! 09:43
FROGGS_ lizmat_: I commented 09:45
lizmat_ FROGGS++
09:45 lizmat_ is now known as lizmat
sorear .tell jnthn observation: Perl6/Grammar.class has nearly 100x as many bytes of serialized objects as Perl6/Actions.class. Possibly worth looking into? 09:45
yoleaux sorear: I'll pass your message to jnthn.
09:50 spider-mario joined 09:53 JimmyZ joined 09:59 kaare_ left, dmol left 10:07 raiph left 10:09 tomyan left 10:10 PacoAir joined 10:20 ecocode` joined
masak lizmat: also worth thinking about in the memespace around github.com/perl6/specs/issues/54 is how Perl 6 sees the distinction between "modules" (essentially the 'unit' keyword you're proposing) and "distributions" (a CPAN concept which can contain zero or more modules). 10:22
10:24 ecocode left 10:26 sivoais left 10:27 sivoais joined 10:29 SamuraiJack left
sorear masak! o/ 10:34
sorear has made changes in the exception system and is getting failures that seem almost unrelated
10:36 JimmyZ left 10:42 tomyan joined, JimmyZ joined 10:43 pmurias joined 10:48 tomyan left 10:55 imarcusthis- joined
moritz back 11:02
11:05 dmol joined
sorear Oops :D 11:19
11:19 rindolf joined
sorear QAST::VarWithFallback was... not *quite* following the stack rules 11:20
didn't notice until I made die continuation-transparent
11:24 FreedomKnight left
lizmat masak: a unit may contain multiple modules. A distribution may contain multiple units. It's up to the author whether to separate modules over multiple units (allowing them different auth/ver information), or keep them in one unit (forcing the modules to share auth/ver infomation). 11:31
JimmyZ sorear: I had a question ...
github.com/perl6/nqp/blob/9514dbe8....java#L307 Is it still used? 11:32
lizmat the advantage of keeping them in one unit, is that these modules can easily share variables and internal subroutines 11:33
sorear JimmyZ: yes
JimmyZ: the stage0 compiler generates that
it won't be used after the next bootstrap update
JimmyZ sorear: thanks, I'm just curious
sorear good :) 11:34
JimmyZ :P
lizmat sightseeing&
11:36 woolfy left
sorear ah-double-ha. there's an overeager optimization here 11:36
sorear fixes
timotimo_ optimizers are hard 11:38
dalek p: 3b780bd | sorear++ | src/vm/jvm/ (3 files):
Change interface of exception system to support eventual continuation transparency
 11:45
timotimo_ so, nqp-jvm has a MapIter and a GatherIter now, so for loops and gather for loops should work now. what else can i do so far? last time i tried i could say "hello world", but say 1 would cause a huge exception 11:47
and instantiating classes with the default constructor wouldn't work either, because of nqp::getoutercapture or something similar
sorear nqp has neither of those 11:49
those are rakudo things
11:49 kaare_ joined
sorear and GatherIter is still not quite there 11:49
timotimo_ er, right. why do i get those two confused? i should know better
12:01 JimmyZ left 12:02 fgomez left
dalek p: 8975957 | sorear++ | / (2 files):
Make exception handlers continuation-transparent
 12:16
12:35 JimmyZ joined 12:48 bruges_ left, bruges joined 12:49 benabik left 12:56 ecocode` left 12:59 ggoebel_ joined 13:01 ggoebel left, ggoebel_ is now known as ggoebel
sorear \o/ just ran a gather block successfully 13:08
TimToady \o/
FROGGS_ sorear++ # \o/ 13:09
13:11 rindolf left
colomon \o/ 13:11
13:12 SamuraiJack joined
timotimo_ oh, look at that, this folder has a file called "gpn13-perl6.mp4" in it. i wonder what it is ... :P ftp://media.ccc.de/events/gpn/gpn13/ 13:13
(unfortunately in german and with pretty hard to read text)
bike_tour& 13:14
sorear sorear$ ./perl6 --ll-exception -e 'my @l := gather { take 1; say "hi"; take 2 }; say @l.shift while @l' 13:15
1
hi
2
JimmyZ push push :P
sorear .tell jnthn Mission accomplished! github.com/sorear/rakudo/commit/ec48e7 # I'd like to know if I can do without p6handletake 13:18
yoleaux sorear: I'll pass your message to jnthn.
sorear JimmyZ: there's my commit 13:19
JimmyZ wheee!! 13:20
sorear now um 13:21
there are a few other things wrong with rakudo-jvm
sorear digs in
sorear would like a jnthn or a pmichaud to discuss exception handling factoring with D: 13:28
13:32 daniel-s__ is now known as daniel-s
pmurias sorear: what is continuationclone needed for? 13:34
sorear pmurias: there's a test in t/jvm/01-continuations.t that fails without it 13:35
a callframe is not a completely immutable object, and it needs to be duplicated if you want to invoke the same continuation twice at the same time 13:36
otherwise the callstack gets screwed up because the caller pointers form a loop 13:37
pmurias what I'm asking for is what Perl 6 feature requires invoking the same contiunation twice
* continuation 13:39
sorear none
but NQP is not just for Perl 6 13:40
pmurias it could be usefull for Ruby etc.
sorear ...I shouldn't be so defensive.
pmurias is thinking how to implement the opcodes on node.js 13:41
sorear don't worry if you can't; I don't expect all VMs to provide precisely the same set of fancy control ops 13:42
everybody that supports p6 needs to provide at least coroutines 13:43
it just happens that the JVM implementation could be extended to unrestricted multiprompt delimited continuations at little marginal cost, so I did so
we might have nqp::coroutine ops at some point. for now, the expectation is #?if jvm 13:44
anyone know offhand which CLA I need for Rakudo? 13:47
pmurias one shot contiunations (one without clone) are basically coroutines, I just have to figure out what delimited coroutines amount to ;)
sorear pmurias: Lexotic gather/take 13:48
13:50 birdwindupbird joined 13:51 tgt joined
sorear labster++ # found the link in your blog 13:52
[ www.perlfoundation.org/attachment/legal/cla3.pdf ] 13:54
13:54 woosley1 joined 13:57 woosley1 left
pmurias sorear: re docs/continuations.pod in line 39 "manually add the prompt" it's unclear what prompt refers too 14:13
14:17 tomyan joined
sorear pmurias: s/prompt/reset/ 14:20
they're synonyms in the literature.
dalek p: ffee47a | sorear++ | docs/continuations.pod:
Fix inconsistent terminology. pmurias++
 14:21
14:23 snearch left
pmurias sorear: could you supply an example how to simulate shift using control/reset? 14:30
sorear pmurias: when you want to use the continuation, instead of doing continvoke(cont, arg), do reset({ continvoke(cont, arg) }) instead 14:34
14:35 cognominal joined 14:37 spider-mario left
sorear if we were representing continuations as functions, you could do sub shift(&f) { control(sub (&rawc) { f(sub ($a) { reset({ rawc(a) }) }) }) } 14:38
14:43 guru joined 14:44 guru is now known as Guest36735, Guest36735 is now known as ajr_ 14:45 amoe_ joined 14:47 amoe left
pmurias sorear: nqp-on-jvm keeps it's own frames rather than using java ones? 14:50
sorear pmurias: we use both java frames and CallFrame objects 14:51
there's no cps conversion here, calls and returns are native java calls and returns
14:53 ggoebel left
pmurias jvm allows you to manipulate the stack? 14:53
sorear only by serious abuse of exceptions 14:54
14:54 ggoebel joined
pmurias and how do you recreate the stack, fake things with CallsFrame's? 14:55
diakopter re-entry points to routines, aiui
sorear control throws a StackSaveException 14:56
all generated routines have StackSaveException handlers that reify the Java frame as an object (which is not the same as the CallFrame)
reset catches the exception, and bundles the reified frames into a continuation object
diakopter wait, sorear's not usually up this... early.... I mean late. :)
sorear invoke calls the first method in the reified chain frame, passing it the reified frame chain 14:57
JimmyZ \o diakopter
sorear the generated code then recreates the current frame, and recursively reestablishes the stack
pmurias what's the overhead of this on normal code? 14:59
sorear one ifnonnull opcode in the function prologue 15:00
(well, and an aload for the argument)
15:09 domidumont joined 15:10 ggoebel left 15:17 pmurias left, zby_home_ left
[Coke] Happy father's day to moritz, TimToady, pmichaud, colomon and everyone else. 15:18
15:19 zby_home joined 15:22 zby_home left
colomon [Coke]: thanks! 15:30
colomon got a father-and-son set of melodicas for Father's Day. en.wikipedia.org/wiki/Melodica 15:31
15:31 rindolf joined
TimToady got to take a long nap, which is about the height of my ambitions today :) 15:36
15:40 domidumont left, rindolf left, domidumont joined 15:46 risou is now known as risou_awy 15:48 risou_awy is now known as risou, fhelmberger joined 15:50 risou is now known as risou_awy 15:51 risou_awy is now known as risou 15:57 JimmyZ left 16:00 fhelmberger left, census joined, fhelmberger joined 16:01 zby_home_ joined 16:07 fhelmberger left 16:14 raiph joined 16:23 atroxaper left 16:24 Gothmog_ left, pmurias joined 16:25 risou is now known as risou_awy
[Coke] again bemoans debugging utilities for nqp. 16:41
16:43 guru joined, ajr_ left, guru is now known as Guest59128 16:47 cognominal left
[Coke] I have 2 tcl programs that differ only by whitespace, and am trying to figure out why one explodes about "Can only use get_how on a SixModelObject". 16:47
(without recompiling nqp to insert debugging output)
16:52 raiph left, Guest59128 is now known as ajr_ 16:55 scottp left 16:56 SamuraiJack left, scottp joined 16:59 raiph joined
[Coke] is there a way in NQP to get the name of the sub you're executing? 17:05
nqp: say('word:sym<{*}>'); 17:06
camelia nqp: OUTPUT«word:sym<{*}>␤»
[Coke] nqp: say('word:sym<" ">:') 17:07
camelia nqp: OUTPUT«word:sym<" ">:␤»
sorear nqp: sub bob() { say(nqp::getcodename(nqp::curcode())) }; bob() 17:11
camelia nqp: OUTPUT«bob␤»
17:15 birdwindupbird left
dalek p: 05231f5 | sorear++ | src/vm/ (7 files):
Add file names and line numbers to NQP-jvm backtraces when possible
 17:22
[Coke] trying to add a debug attribute to my actions class so I can conditionally say "you're in this action method and were passed this value" - just adding the debug attribute is causing a compilation error. wtf.
nqp: class barf {␤ has $.debug := False; ␤ method eek() {say(3)} } 17:23
camelia nqp: OUTPUT«Unable to parse expression in blockoid; couldn't find final '}' at line 3, near "has $.debu"␤current instr.: 'panic' pc 14721 (src/stage2/gen/NQPHLL.pir:5232) (src/stage2/gen/NQPHLL.nqp:279)␤»
[Coke] r: class barf {␤ has $.debug := False; ␤ method eek() {say(3)} } 17:25
camelia rakudo b2072f: OUTPUT«===SORRY!===␤Cannot use := to initialize an attribute␤at /tmp/snuKMfRC6C:2␤------> has $.debug := False⏏; ␤ expecting any of:␤ postfix␤»
[Coke] r: class barf {␤ has $.debug = False; ␤ method eek() {say(3)} }
camelia rakudo b2072f: ( no output )
17:34 ssutch joined
timotimo_ nqp: class barf { has $.debug; method eek() { say(3) } } 17:34
camelia nqp: OUTPUT«Unable to parse expression in blockoid; couldn't find final '}' at line 2, near "has $.debu"␤current instr.: 'panic' pc 14721 (src/stage2/gen/NQPHLL.pir:5232) (src/stage2/gen/NQPHLL.nqp:279)␤»
timotimo_ strange
[Coke] wanders off to do something fun instead of fight with nqp. 17:38
sorear that's because nqp only supports $!. 17:42
[Coke] nqp: class barf {␤ has $!debug := False; ␤ method eek() {say(3)} } 17:47
camelia nqp: OUTPUT«Error while compiling block : Error while compiling block (source text: "{\n has $!debug := False; \n method eek() {say(3)} }"): Error while compiling op bind (source text: ":="): First child of a 'bind' op must be a QAST::Var␤current instr.: '' pc 48612 (src/stage2/QAST.p…
sorear and you can't bind in a class body. 17:50
have to do the binding in method new
[Coke] nqp: class barf {␤ has $!debug; ␤ method eek() {say(3)} } 17:51
camelia nqp: ( no output )
[Coke] nqp: class barf {␤ has $!debug; ␤ method new() {$!debug := True} method eek() { $!debug && say(3)} }; barf.new().eek() 18:06
camelia nqp: OUTPUT«Unable to parse expression in blockoid; couldn't find final '}' at line 4, near "method new"␤current instr.: 'panic' pc 14721 (src/stage2/gen/NQPHLL.pir:5232) (src/stage2/gen/NQPHLL.nqp:279)␤»
18:09 sftp left
sorear missing newline between new and eek 18:09
18:09 sftp joined
sorear also, nqp news look like this: 18:10
ok, I don't know as well as I thought 18:12
18:12 rindolf joined
[Coke] gist.github.com/coke/cdc0d82fd63e0282038c - there's no concat that isn't commented out in that file. weird. 18:14
ah. might be an old action. now to figure out how QAST::Op works. :| 18:15
sorear++ for help so far. 18:16
lee_ i did some comparisons with scala, and summing 1..1000000 takes about 5s vs 10s in rakudo-jvm 18:23
not bad!
after the first run scala keeps a "CompilerServer" running in the background and it takes about 1s 18:24
er CompileServer 18:25
pmurias pastie.org/8049400 18:28
I get an error when running perl Configure.pl --gen-parrot 18:29
used to work before
which is caused by --prefix="/home/pawel/..." instead of --prefix="/home/pawel/..." being passed to Configure.pl in parrot 18:34
18:37 ajr_ left
pmurias and the whole bug was solved in parrot 8 days ago 18:38
moritz doesn't see the difference 18:40
FROGGS_ there is no difference 18:49
18:50 census left
timotimo_ ~ doesn't matter if you're black or white ~ 18:51
19:00 tomyan left 19:02 raiph left
ssutch what are any resources for how grammar/actions and their interaction works? 19:05
running the debugger, i can step into actions, but kinda lost at what to do once there 19:06
btw, a grammar that parses all of protocol buffers IDL: github.com/samuraisam/p6-pb/blob/m...Grammar.pm (!)
timotimo_ ssutch: when the token/rule/whatever is finished matching, the same-named method in the action class will be called and it will be passed the $/ variable 19:07
then, you'd usually use the "make" function to set what the resulting object's .ast value is
19:07 drin_m left
ssutch when i run 'say $/' in the debugger, it just says (Any) 19:07
pmurias FROGGS_: re there is no difference - you mean to the build process? 19:08
timotimo_ did you put the $/ into the method signature?
ssutch method proto ($/) { make $<proto>.ast; }
timotimo_ okay. in that case, perhaps your version of rakudo is too old; there recently was a fix that made $/ and $! visible to the repl and debugger 19:09
ssutch ah ok, perhaps, i am running the latest star on this machine
FROGGS_ pmurias: the --prefix-es you posted... they are the same 19:10
timotimo_ star is too old, yes
ssutch timotimo_ yes, you're right; i added my $f = $/ and got something from it
timotimo_ the fix went in a few days or maybe a week after star was released
19:12 census joined, telex left 19:14 telex joined
pmurias FROGGS_: sorry 19:16
FROGGS_: a --prefix with quotes is not accepted by the parrot nqp fetches
because of a bug 19:17
19:17 sunnavy left
FROGGS_ pmurias: ahhh, I see 19:18
19:18 bonsaikitten left 19:19 bonsaikitten joined 19:20 Ben_Goldberg joined, woolfy joined 19:24 domidumont left 19:28 [particle] left 19:30 Gothmog_ joined 19:34 raiph joined 19:36 census left
ssutch so i think it makes sense to turn the Grammar into a PB::Package class, which in turn has PB::Messages, enums, options, etc 19:39
how do i carry that $package throughout the actions then? do i just set $package to an instance variable on the actions class? 19:40
19:42 SamuraiJack joined 19:45 [particle] joined
timotimo_ you mean use the grammar to generate an instance of the PB::Package class? 19:46
ssutch yeah
timotimo_ ah
that makes sense, yeah
ssutch and eventually, take PB::Package and turn it into native-feeling perl module 19:47
19:48 sunnavy joined 19:51 cognominal joined 20:02 kaare_ left
ssutch so does this seem right: method package ($/) { make PB::Package.new(name => $<dotted-ident>.Str); } 20:03
20:03 raiph left, spider-mario joined 20:05 abnorman left
ssutch r: gist.github.com/samuraisam/5793230 20:06
camelia rakudo b2072f: OUTPUT«(Any)␤»
ssutch not what i wanted, but on my system gives "use of uninitialized value of type Any in string context in block at lib/PB/Actions.pm:23"
timotimo_ i think your $/.ast is faulty, but i'm not sure. 20:07
ssutch yeah, kinda still bumping into things whilst wandering around in the dark 20:08
timotimo_ i think you can turn the [...] in your proto token into a $<foo>=(...) 20:09
it should be easier still to use an actual proto token for your proto token 20:10
FROGGS_ that $/.ast is faulty
should be like $<proto>=[<message>..., and then in actions: make QAST::Op.new( :op<call>, :name('&infix:<,>'), $<proto>.ast ) 20:11
20:11 Rix left
timotimo_ wait what? when did we start building qast? 20:11
ssutch 0_0 20:12
FROGGS_ ohh, nvm
timotimo_ :D
ssutch so $<statements>=[<message> | <package> | <import> | <option> | <enum> | <extend> | <service> | ';']* 20:13
20:13 pmurias left
ssutch then i can go $<statements>.ast.flat; 20:13
FROGGS_ yeah, and in the method proto: $<statements>.ast.flat
right
timotimo_ yup. not 100% sure how it'll handle the ";", it would probably give you an (Any) in your list of ast's 20:14
FROGGS_ you have to change method TOP afterwards
timotimo_ watch out, though, the .flat could flatten the asts messages, packages etc
ssutch r: gist.github.com/samuraisam/5793230 20:16
camelia rakudo b2072f: OUTPUT«===SORRY!===␤Redeclaration of symbol PB::Package␤at /tmp/aktXiRtFoe:110␤------> class PB::Package ⏏{ ␤ expecting any of:␤ scoped declarator␤ constraint␤ postfix␤ statement end␤ statement mod…
[Coke] ("s not accepted) - are you getting the latest parrot?
ssutch r: gist.github.com/samuraisam/5793230
camelia rakudo b2072f: OUTPUT«(Any)␤»
[Coke] ISTR I just chased this down so that you could do something like --prefix="/path with a space/" and have it work.
ssutch ok i made those changes, $statements= and $<statements>.ast.flat 20:17
timotimo_ ah, hold on 20:18
20:18 cognominal left
timotimo_ i think you need $<statements>>>.ast instead 20:18
$<statements> will be a list of match objects, rather than an actual match object, no?
r: <foo bar baz>.>>.uc.perl.say 20:19
camelia rakudo b2072f: OUTPUT«===SORRY!===␤Confused␤at /tmp/Llggt7RJ_X:1␤------> <foo bar baz>.⏏>>.uc.perl.say␤ expecting any of:␤ dotty method or postfix␤»
timotimo_ r: <foo bar baz>>>.uc.perl.say
camelia rakudo b2072f: OUTPUT«("FOO", "BAR", "BAZ")␤»
timotimo_ std: <foo bar baz>.>>.uc.perl.say 20:20
camelia std d4cc5ab: OUTPUT«ok 00:00 41m␤»
timotimo_ that's a rakudobug, then? hm. wonder if i'd be able to figure it out.
ssutch alright, shoot
i have to go do family things, i'll be back 20:21
20:21 Rix joined
timotimo_ sure :) 20:21
ssutch "fathers day" 20:22
[Coke] anyone see anything obviously wrong with the action here: 20:23
github.com/partcl/partcl-nqp/blob/...ns.pm#L281
timotimo_ do you get a specific error from that? 20:24
[Coke] gist.github.com/coke/cdc0d82fd63e0282038c - not that I know that's where the error is coming from.
20:27 lizmat left
[Coke] yes, that's where it's coming from. 20:29
FROGGS_ [Coke]: you can try &concat instead
timotimo_ ah, indeed that sounds like a good plan 20:30
[Coke] looks like it might be a confusion of an old PAST style syntax.
20:31 woolfy left
[Coke] Could not find sub &concat. 20:32
trying to ditch the call and just use concat.
FROGGS_ hmmm
[Coke] docs/qast.markdown - "the ops themselves are documented separtely" - is this referring to docs/nqp-opcode.txt ? 20:33
FROGGS_ [Coke]: is this about string concat?
[Coke] I believe it is trying to QAST-wrap a call to nqp::concat, aye. 20:34
FROGGS_ [Coke]: if yes, you could call &infix:<~> on all parts at once
timotimo_ try nqp::say($past.dump); after that action method?
[Coke] where is, e.g. "nqp::concat" documented? 20:35
timotimo_ oh, you want to do nqp::concat 20:36
- QAST::Op(concat) nqp::concat(\"hoo\", \"blop\")
that's how you do a nqp::concat in qast
it's not op<call>, it's op<concat>
[Coke] there's a concat and a concat_s - do either of those save the value in one of the operands? 20:37
20:37 PacoAir left 20:38 pmurias joined
timotimo_ i can't even find the nqp::concat op :| 20:38
20:43 dmol left
pmurias [Coke]: I don't think the ops are documented anywhere, I mostly search the test suit and read the source 20:43
[Coke] pmurias: I was afraid that was the answer. 20:45
FROGGS_ please try to call the infix, that is what rakudo and v5 does 20:46
[Coke] pushed a fix to eliminate the call. no clue if it's working, but gets me past the current problem. 20:47
20:48 dmol joined
timotimo_ [Coke]: can you give me a brief summary of why i should be interested in tcl? 20:49
[Coke] very unlikely. 20:50
ssutch hah.
[Coke] its parser is very different from other languages. it was, for a while, *the* embedding language. 20:51
masak waves from back-in-Sweden o/ 20:52
FROGGS_ hi masak
20:53 cognominal joined
lue hello world o/ 20:53
[Coke] timotimo_: only reason I'm working on it was that I thought it would be an interesting language for multi-language support on parrot. that never really went anywhere, though I got it to the point where you could run quite a bit of the tcl test suite.
lue Would it be better to move S03/Adverbs to after terminator precedence, or to make it a =head2 ? (I prefer the first choice because Adverbs are not a precedence level) 20:54
20:54 woolfy joined 20:55 lizmat joined
masak lue: I'm not sure such a change is worth the dust it'd kick up in the blame-log. 20:59
that's just my very uninformed initial opinion, though.
lue Well, considering the Adverbs section describes them as "at a pseudo-precedence level slightly tighter than item assignment", I feel that at least the =head1 was a typo that should be =head2 21:00
(I just suggested the move because it's not even in the precedence table at the top of the spec) 21:01
21:05 MrMeek-afk joined 21:07 MrMeek left 21:08 woolfy left 21:09 fgomez joined 21:13 zby_home_ left, twigel joined 21:15 bruges left 21:17 bruges joined 21:25 woolfy joined 21:32 woolfy left, lizmat left, lizmat joined
lizmat r: 42.classify: { $_ }.say # why shouldn't this work? 21:32
camelia rakudo b2072f: OUTPUT«Block.new()␤Nominal type check failed for parameter '&t'; expected Callable but got Bool instead␤ in method classify at src/gen/CORE.setting:1344␤ in block at /tmp/pDg3yuJ0Ok:1␤␤»
lizmat r: 42.classify:({ $_ }).say # I meanyt why shouldn't this work? 21:33
camelia rakudo b2072f: OUTPUT«===SORRY!===␤Unable to parse expression in signature; couldn't find final ')'␤at /tmp/R14I_lSmnN:1␤------> 42.classify:(⏏{ $_ }).say # I meanyt why shouldn't th␤ expecting any of:␤ colon pair␤ signature␤»…
lizmat r: 42.classify({ $_ }).say # I meant why shouldn't this work?
camelia rakudo b2072f: OUTPUT«("42" => [42]).hash␤»
lizmat t/spec/S32-list/classify.t seems to think it shouldn't 21:34
pmurias [Coke]: I'll expand the test suit as part of my gsoc project which hopefully will make it a better reference ;)
[Coke] pmurias++ 21:35
lizmat I'm inclined to fix the test assuming it's ok that it works (albeit a bit nonsensical) 21:36
masak I also think it should work. 21:42
Perl generally doesn't have a bias against confusing an item with a collection.
21:46 cognominal left, cognominal joined
[Coke] what is the nqp equivalent of pir::find_dynamic_lex ? 21:52
timotimo_ git log -u -S to the rescue! 21:53
- my $found := pir::find_lex_relative__PPs(
+ my $found := nqp::getlexrel(
oh, wait, that's not it at all
masak :)
timotimo_ pir::find_dynamic_lex_relative__PPs -> nqp::getlexreldyn ; likely your op is called something similar
pir::find_dynamic_lex__Ps('$?REGEX')(self) -> nqp::getlexdyn('$?REGEX')(self) - sound good? 21:54
21:57 rindolf left
[Coke] timotimo_: you find that in a git history search? danke. 21:57
(someone should add that to docs/nqp-opcode.txt
dalek ast: a9f8006 | (Elizabeth Mattijsen)++ | S32-list/classify.t:
Simplification/Addition tests of classify() and .classify()
 22:05
22:08 raiph joined
lizmat too tired to work on categorize.t now, sleep& 22:11
[Coke] No registered operation handler for 'getlexreldyn'
timotimo_ nqp: nqp::say(nqp::getlexdyn('$?ROUTINE')); 22:17
camelia nqp: OUTPUT«Null PMC access in get_string()␤current instr.: '' pc 43 ((file unknown):148909852) (/tmp/vtHJHfoR4x:1)␤»
timotimo_ nqp: nqp::getlexdyn('$?ROUTINE'); 22:18
camelia nqp: ( no output )
timotimo_ nqp: nqp::getlexreldyn('$?ROUTINE');
camelia nqp: OUTPUT«Error while compiling block : Error while compiling op getlexreldyn (source text: "nqp::getlexreldyn('$?ROUTINE')"): No registered operation handler for 'getlexreldyn'␤current instr.: '' pc 48612 (src/stage2/QAST.pir:17766) (src/stage2/QAST.nqp:2981)␤»
timotimo_ mhm
[Coke] no doubt I'm missing a setup somewhere. 22:19
masak yes, hm, there's a file where you register these. 22:20
22:22 spider-mario left, lizmat left, nitestryker_ left 22:24 lizmat joined 22:26 dmol left 22:29 bbkr left 22:32 Rix left 22:35 pecastro_ joined 22:37 pecastro left
[Coke] I suspect I need a lesson on how to use the funky lexical stuff in nqp. 22:43
Ben_Goldberg .ping 22:50
yoleaux There is no ping command; nor can this be construed as a response.
timotimo_ :D
diakopter .yoleaux
speechless.
22:52 benabik joined
Ben_Goldberg I've got a silly question: how hard would it be to implement a 'comefrom' command in perl6? 22:52
22:54 frdmn left
dalek rl6-bench: 7db63d0 | (Geoffrey Broadwell)++ | TODO:
Add a TODO for IO tests
 22:55
22:55 frdmn joined
rl6-bench: a46b452 | (Geoffrey Broadwell)++ | timeall:
Add timeall infrastructure to support tagged tests; add automatically generated tags to tests at startup; add options (with corresponding docs) to list all known tags, list tests matching a tag query, and run tests matching a tag query
rl6-bench: 2b05b1e | (Geoffrey Broadwell)++ | bench:
Enable 'bench time' pass-through to timeall for --tests-tagged option
rl6-bench: 9a6ae7c | (Geoffrey Broadwell)++ | bench:
Tiny whitespace fix
masak Ben_Goldberg: pretty hard, I imagine. goes pretty deep into the kind of runtime semantics that isn't easily pluggable. 23:15
Ben_Goldberg: also, have you considered exactly what semantics you want this 'comefrom' keyword of yours to have?
23:15 Rix joined
masak Ben_Goldberg: for example, can there be more than one 'comefrom' statement referring to the same label? 23:16
...and what would that mean?
(and before you say "well, obviously...", consider whether it is so obvious) :)
timotimo_ may i suggest a comefrom_and_take($foo, $bar, $baz) syntax, so that you may take local variables over from the previous "position"?
masak wtf 23:17
...because 'comefrom' wasn't crazy enough already!? 23:18
diakopter haha.
masak timotimo_: that's not what lexical variables *mean*. 23:19
diakopter eye_gouge($just, $give, $up, $now, $compiler)
timotimo_ i'm glad i was able to improve on the crazy that is comefrom
masak all you did was smash Don Quixote real hard against another windmill.
timotimo_ use A_WHITEBOARD_FOR_ASSISTANCE_WHEN_READING_THIS_CODE; 23:20
good night :)
masak 'night, timo
sorear masak: multiple come-froms have a standard meaning. www.cse.unsw.edu.au/~malcolmr/inter...eaded.html 23:21
masak sorear: I'm aware of multi-threaded INTERCAL ;)
sorear: I would also have accepted "it randomly picks a destination" as a "correct" answer. :)
diakopter "Each variable has its own set of variables which are private to that thread" 23:22
surely there's a wrong word in there
masak even the variables have variables!
diakopter: it's well-founded because some variables can have the empty set :P 23:23
23:24 Ben_Goldberg left 23:25 BenGoldberg joined
lue just imagined some sort of C<use ONE_WORLD_GOV;> that makes every single variable global to the entire program, even if they weren't meant to be. 23:30
sorear bah. autosinking causes a huge explosion of QAST 23:47
(dump output) 23:48
diakopter sorear: your development cycle nowadays is curiously different from when you were doing niecza :)
at least, seems that way from a distance 23:49
back then, your computer was so slow you wrote bug-free code :P
sorear diakopter: my new computer can compile rakudo about as fast as the old one could compile niecza :P 23:58
diakopter haha.

Powered by Cro and the Raku® Programming Language.

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

Thank you!