Check your feather email | spec.pugscode.org | paste: sial.org/pbot/perl6 | pugs.blogs.com | www.treehugger.com/files/th_images/paradigm.jpg [set by audreyt on 2006-08-29 05:26:57 -0700]
Set by Akwa|user on 14 September 2006.
00:11 Zoffix joined
Zoffix Will scripts that run in Perl 5 run in Perl 6? 00:11
Patterner theoretically yes 00:12
Zoffix ok 00:13
00:13 Zoffix left 00:17 markstos joined 00:23 nekokak joined 00:30 hikozaemon joined 01:01 mr_goose joined 01:03 mr_goose left 01:27 azr|elious is now known as azr|el 01:37 nnunley joined 01:54 justatheory joined 01:59 xdg joined 02:21 revdiablo joined 02:22 f0rth joined 02:24 lambdabot joined 02:36 tcliou joined 02:47 kastol joined 02:59 KingDiamond joined 03:11 Aankhen`` joined 03:25 evilmnky joined 03:37 Daveman joined 04:05 ofer0 joined 04:19 kastol joined
svnbot6 r13587 | gaal++ | * Pugs.Val: add -funbox-strict-fields to let ValNative be what it says 04:44
r13587 | gaal++ | * castVal for ValUndef and ValNative
04:48 amv joined 04:53 buetow joined
ingy seen [particle] 04:57
jabbot ingy: [particle] was seen 2 days 10 hours 21 minutes 15 seconds ago
05:14 BooK joined
svnbot6 r13588 | fglock++ | [v6] 05:15
r13588 | fglock++ | - fixed prefix:<$> rule (added quotemeta);
r13588 | fglock++ | - Term table cleanup; disabled '$()';
r13588 | fglock++ | - fixed '${...}', '$[...]', ':$<...>', '$0'
r13588 | fglock++ | - fixed coercions in emitter
r13588 | fglock++ | [Pugs-Compiler-Rule]
r13588 | fglock++ | - added method Match.chars
r13588 | fglock++ | - RegexPerl5 accepts parameters i m s x ;
r13588 | fglock++ | [plain_regex]
r13588 | fglock++ | - code cleanup; added bug list
r13588 | fglock++ | - added ^^ $$ \n \N - TimToady++
r13588 | fglock++ | - /./ matches 2-char newline
r13588 | fglock++ | - <'literal'>
r13588 | fglock++ | - simple character classes, negated char classes
r13588 | fglock++ | [Pugs-Compiler-Python] 05:16
r13588 | fglock++ | - first commit (in misc/pX/Common)
r13589 | lwall++ | Renamed .words to .comb 05:21
r13589 | lwall++ | More Foo:: whackage and "method is export" insertage.
clkao Pugs-Compiler-Python! 05:22
05:29 mdiep_ joined 05:45 nekokak_ joined 05:53 marmic joined 05:55 hikozaem_ joined 06:00 Dr_Pi joined 06:32 Gothmog_ joined 06:40 iblechbot joined
svnbot6 r13590 | audreyt++ | * HsBridge: Improved dump of named arguments. 06:52
07:03 kastol joined 07:16 zakharyas joined 07:25 iblechbot joined 07:30 ludan joined 07:55 scw joined 08:01 nothingmuch joined 08:08 kane-xs joined 08:14 ruoso joined 08:47 drrho joined
svnbot6 r13591 | audreyt++ | * Remove t/Dialects now that our test suite is spec-based. 08:59
r13592 | audreyt++ | * "try {...}" now catches internalErrors as well as user errors. 09:05
r13592 | audreyt++ | * "slurp" now uses strict ByteString.readFile instead of the much
r13592 | audreyt++ | slower readFile.
r13592 | audreyt++ | * Infix "===" now correctly compares value (instead of pointer)
r13592 | audreyt++ | equivalence.
r13593 | audreyt++ | * "pugs -Ifoo -Ibar" now correctly lists foo before bar on @*INC.
r13593 | audreyt++ | (This was a regression from previous release.)
r13594 | audreyt++ | * Pugs.Compile: Misc cleanup on list compiling.
r13595 | audreyt++ | * Pugs.Eval: Allow "last" and "redo" in "repeat { ... }" loops. 09:08
r13595 | audreyt++ | * Also implement \($x, $y, |$z, $w) splicing.
r13596 | audreyt++ | * Allow 0_1234 to mean 0d1234; disallow "\_1234" to mean "\d1234". 09:09
r13597 | audreyt++ | * "my &infix:<foo>" without explicit associativity now 09:12
r13597 | audreyt++ | defaults to left-associative.
r13598 | audreyt++ | * On Unix systems, don't bork on source files with CRLF line endings.
r13599 | audreyt++ | * "my $x; { $x; my $x }" now fails with a sensible error message.
r13600 | audreyt++ | * Pugs.Parser: Rename parseHasParenParamList to ruleParenArgs 09:15
r13600 | audreyt++ | as we are now more careful about the distinction between
r13600 | audreyt++ | params (Signature) and arguments (Capture).
r13600 | audreyt++ | * "do { ... }" is now made a true block, by desugaring it into
r13600 | audreyt++ | "repeat { ... } while False". 09:16
r13600 | audreyt++ | * "f( :$x :$y :$z )" without commas between them is now legal syntax.
r13601 | audreyt++ | * Add missing svn:eol-style on test files. 09:19
r13602 | audreyt++ | * proto.t: Fix parsefail and don't (yet) treat proto as expressions. 09:22
r13603 | audreyt++ | * return_function.t: Use |(f) instead of *f() to agree with the new spec.
r13604 | audreyt++ | * splatty_in_caller_position.t: It's f(|@args), not f(*@args). 09:26
audreyt oy... rt.openfoundry.org crashed again, we're back to readonly 09:39
09:42 renormalist joined
nothingmuch 3/w 12 09:56
audreyt: if you're awake please op the opbots on #moose 09:59
clkao: you too
gaal audreyt: commas are mandatory between params in Signatures, right? 10:14
and another question about r13600, shouldn't a do{} block evaluate to its uh, last evaluated value? 10:15
10:17 elmex joined
svnbot6 r13605 | audreyt++ | * Pugs.Types: Allow $::x to mean exactly the same as $x. 10:59
r13606 | audreyt++ | * Add svn:eol-style to TASKS as well.
r13606 | audreyt++ | (r13604 accidentally included all tests in the previous triage run, 11:00
r13606 | audreyt++ | but I don't quite feel like reverting and committing them separately again... :-))
r13607 | audreyt++ | * temp.t: TODO a pseudo-regression -- this test was passing
r13607 | audreyt++ | because =:= wasn't implemented. ;)
11:01 chris2 joined
svnbot6 r13608 | audreyt++ | * symbolic_deref.t: $*IN should be compared with ===, not =:=. 11:03
audreyt gaal: right. and yes, a do block does that; so does "do repeat {...} while False". 11:05
svnbot6 r13609 | audreyt++ | * 1.WHICH should be 1, not undef.
r13610 | audreyt++ | * Perl6::FAQ::Capture: Use |$x not [,]=$x.
gaal audreyt: ah, I misparsed "true" in "true block". 11:07
svnbot6 r13611 | audreyt++ | * Pugs.Parser: Allow ::Int:: to mean the same thing as ::Int.
r13611 | audreyt++ | This is wrong -- It should be a Package object associated with
r13611 | audreyt++ | ::Int -- but that will need to wait until MI is rolled in.
r13612 | audreyt++ | * Minor: Give rules in Pugs.Lexer and Pugs.Parser.Literal names. 11:10
audreyt ah. "true" as in "last/redo/next works" 11:12
svnbot6 r13613 | audreyt++ | * Oops, typo in the previous commit.
r13614 | audreyt++ | * The main package is now ::Main, not ::main.
gaal *nod* 11:16
11:23 kanru joined 11:31 rafl joined
gaal lol. I'm reading the sources for syb stuff now. cast's implementation is amusing 11:55
cast x = r where r = if typeOf x == typeOf (fromJust r) ....
12:14 Limbic_Region joined
svnbot6 r13615 | audreyt++ | * PCR: Add the case of (""~~/<null>/) to t/03-match.t. 12:37
kolibrie slides from my talk (as delivered) are up: feather.perl6.nl/~kolibrie/extraction.xul 12:40
lambdabot tinyurl.com/rxtmp
svnbot6 r13616 | audreyt++ | * PCR: Regex wasn't allowed to match at end of string; this fixed it.
r13617 | audreyt++ | * 07-simple-multisubs.t: Reformat the sanity test somewhat.
12:40 rafl joined
kolibrie with soundtrack: feather.perl6.nl/~kolibrie/extraction.mp3 12:40
lambdabot tinyurl.com/femjh
12:40 rafl joined, iblechbot joined
kolibrie (now we just need to fix the slides that are wrong) 12:41
audreyt gaal: yeah, that's a neat trick to work around the lack of scoped tyvar 12:46
gaal audreyt: uh huh. 12:47
kolibrie audreyt: gaal: thank you very much for the last minute help with my slides - I think the talk went okay
audreyt with scoped tyvar (part of -fglasgow-exts) it'd be
cast :: forall a b. Typeable a, Typeable b => a -> b 12:48
er, -> Maybe b
cast x = r where r = if typeOf x == typeOf (undefined :: b) ...
12:48 juzek joined 12:49 juzek left
Limbic_Region hey - I mostly understand that aramaic er uh, haskell 12:57
audreyt nice! :) 12:58
gaal audreyt: ACK, ACK.
anyone know of a omnigraffle-like tool for linux or windows? 12:59
(free)
audreyt umbrello? 13:00
svnbot6 r13618 | audreyt++ | * More s/main::/Main::/ fixes.
gaal i'll try that, thanks. 13:01
Limbic_Region gaal - I think you would have more luck searching for visio rip-offs than omnigraffe
where rip-off == free alternative
audreyt aka dia, kivio
gaal i don't want something with very complex graphical caps. way I need is something that's smooth to work with and create an ordered graph 13:02
I mean I don't need very complex caps
Limbic_Region Graphviz perhaps? 13:03
gaal that's pleasant to enter data directly to from the gui
13:03 xinming joined 13:04 agentzh joined 13:09 fglock joined
svnbot6 r13619 | audreyt++ | * "sub ($x = 0 is copy)" is invalid; say "sub ($x is copy = 0)" instead. 13:09
Limbic_Region audreyt - for those following along at home 13:11
is that example assigning a value to $x if it already exists - or only giving it a default value if not defined?
lumi The latter, I think? 13:14
Limbic_Region lumi - I think so too but it was a leading question 13:15
does it only work if not defined?
IOW - can I do something like //= 0
gaal Limbic_Region: neither. it's predicated on whether $x was mentioned in the capture. 13:16
Limbic_Region gaal - oh, present or not
not - defined or not
Limbic_Region neverminds
gaal :($x = X) is equivalent to :($x? = X)
audreyt: does r13619 mean order of param parts is now specced? 13:17
are all parts order-strict? or do we need a classifier after all? 13:18
gaal might make do with vi x.dot and dotty x.dot, seeing as www.graphviz.org/Gallery/directed/unix.dot.txt is so elegant 13:19
lambdabot tinyurl.com/jbw8h
audreyt I think the =xxx part obviously follows the "is foo" part
gaal because the traits don't modify the default value, yes. 13:20
is there ever a case where :($x where {...} is ~Y~) and the trait modifies the closure? 13:21
audreyt not likely
so I think those two are order indep
gaal so, data ParamModifier = ModTrait Trait | ModConstraint Code | ... 13:22
and something to reassemble these parts.
fglock just checking: ':$<rule>' is 'rule => $($/<rule>)' right? 13:26
13:26 dakkar joined, crem joined
fglock and: 'after => :$<rule>' is 'after => { rule => $($/<rule>) }' ? 13:28
13:29 Odin-LAP joined 13:31 arhuman joined
arhuman hi 13:31
fglock or should that be ' :$$<rule> ' # a pair with string value 13:32
13:32 evalbot_r13586 is now known as evalbot_r13619
audreyt :$<rule> is 13:33
rule => $/<rule>
there's no $() part
13:33 vel joined
audreyt you need :$$<rule> for that yes 13:33
fglock does $() still exists?
audreyt it does 13:35
$() is short for $$/
@() is short for @$/
lambdabot Maybe you meant: . bf ft id pl v wn
audreyt %() is short for %$/
ok, they are not shorter
fglock :)
audreyt so maybe s/short/alternative spelling/ :)
13:36 lanny joined
obra $() sure is a lot nicer to read than $$/ 13:36
. o O { And some day will be abused to make code that compiles in perl6 and javascript }
audreyt yes, that's in fact my first reaction when that syntax emerged :) 13:37
obra Heh.
svnbot6 r13620 | lanny++ | [READTHEM] 13:39
r13620 | lanny++ | added wikibooks.org entry for online Haskell
13:40 baest joined 13:47 xerox joined
lanny So are things like rand() in S29 declared both method and sub to allow folks to multi on rand() without having to define a method on some class (to be rand()-ed)? Declaring method rand(Int:) and sub rand(Int) seems like the sub will never be invoked unless folks go out of their way. E.g., rand(7,) 13:49
13:51 arhuman left
audreyt I think it's jsut that TimToady havn't got around tuit on the method-is-export campaign 13:52
method rand () is export { } # will suffice
otoh we do need a multi rand () {} 13:53
obra it's really "is export" and not "is exported"?
audreyt which rdispatches to 1.rand
or "is exportable"? yeah, that's faq and I gave up long ago :) 13:54
obra yeah. ok. 13:55
lanny So why would we need a multi rand that re-dispatched? I would have thought the method rand() would have been found before you got to the sub for the redispatch back up to a method. 13:58
Since it's arity-1. Arity-1+ I can see.
audreyt the arity-1-sub-to-method-preference is specced, yes, and implemented 13:59
lanny I'm all for covering all the bases since it's good documentation. Just wondering if there are deeper reasons I'm missing.
audreyt but that was before we can export methods
and TimToady is still considering whether we can teach people to write $x.foo and don't expect foo($x) to work unless it's exported 14:00
lanny Aha.
audreyt in which case the preference (which causes quite some headache in f(this_is_a_call_that_does_different_things_in_itme_vs_list_context()).
...can go away. 14:01
(when interpreted as .f, the cxt is item, but when interp as f(), there may well be a slurpy &f in scope)
lanny Well I hope there's at last a 'use SlowLearner' because I would want foo($x) without having to import if there was a foo() in Perl5.
audreyt but you can't go back and reeval it as slurpy anymore
svnbot6 r13621 | fglock++ | v6 - implemented $(), @(), %(), :$$<...> 14:02
audreyt yes, which is why we export all p5-builtin-like methods such as rand.
the export is just a sugar multi that redispatch back into the method anyway
lanny Ok. Good to know I got it mostly right on motivation.
audreyt =) 14:03
lanny Must run girls to school. Thanks for the clue, audreyt.
14:04 iblechbot joined
svnbot6 r13622 | audreyt++ | * unTODO passing tests. 14:11
Limbic_Region audreyt - when is the ETA for new release?
number of failing tests seems to continue fluctuate and not only go down 14:12
OTOH (new tests)++
14:15 lisppaste3 joined 14:16 orafu joined
svnbot6 r13623 | fglock++ | v6 - s/Regex/Token/ in Term.pm 14:23
14:24 lollan joined
audreyt Limbic_Region: those are not new tests, merely parsefail tests made nonparsefail 14:24
triage eta another couple days 14:25
with chglog help perhaps this weekend
svnbot6 r13624 | fglock++ | Pugs-Compiler-Rule - refactored Rule grammar to use shortcuts, such as 14:29
r13624 | fglock++ | '{ return { after => :$$<rule>, } }' 14:30
fglock compiling the compiler with itself is funny, as you can use the newly implemented features in the source code 14:32
audreyt yeah :) 14:33
nice use, that
theorbtwo You're providing a pre-compiled version, I assume, to saneify the circularity of it all? 14:35
obra seen pdcawley
jabbot obra: pdcawley was seen 1 days 1 hours 22 minutes 42 seconds ago
theorbtwo gets On Trusting Trust flashbacks.
fglock theorbtwo: yes, there is a perl5 .pmc file, which is re-generated each time the p6 code is edited 14:37
theorbtwo Right-o.
fglock does <!aaa> matches 3 chars or one? 14:39
[particle] you mean <![aaa]> ? 14:40
audreyt isn't that negated 'aaa' ?
I mean <aaa>
fglock yes - 3 chars, I think
oh, right 14:41
audreyt but <aaa> here is named subrule
[particle] right, named subrule
that's where i was confused
fglock oh, I meant - <!<'aaa'>>
sorry
[particle] ! is always a zero width assertion
fglock [particle]: thanks! 14:42
audreyt if it boolifies to false, it can't have a from/to, I think
though I can argue the other way, but it makes it easier this way :) 14:43
(and if it boolifies to ture, then that [particle] said applies.)
TimToady ⅋ ZZZ
audreyt hi TimToady :) 14:44
TimToady mornin'
fglock so <!<'aaa'>> is the same as <!before <'aaa'>> ?
TimToady: hi 14:45
ajs_ I'm still having throuble with the captures threads. I think my big point of confusion is on what this means, or even if it works: |c = \$s
audreyt seems so... which makes S05:923 rather redundant...
ajs_: well, it havn't been decided yet whether |c can be assigned to 14:46
[particle] fglock, audreyt: indeed
audreyt i.e. if it' more like & or more like @
ajs_ audreyt: We're quoting Synopses by chapter and verse, now? I can't wait to see the guy that goes to basebal games with "S03:16" signs
audreyt :)
[particle] fglock: should your example be <!'aaa'> ? 14:47
ajs_ I see. well, still the real question is "what is \$s" is it a capture in the same sense that "\($s:)" is? is it just filling the scalar slot as some recent messages on the list have asserted? 14:48
14:48 ajs_ is now known as ajs_work
theorbtwo I'm not sure I know what the difference between those is. 14:48
audreyt ajs_work: yes.
theorbtwo Isn't \($s:) just filling the scalar slot? 14:49
audreyt currently as specced \$s \($s:) \($s) all bois down to the same thing
ajs_work theorbtwo: yes
fglock [particle]: I don't know if that works, maybe
PerlJam sticks a $ s ( ) and a : in a box and shakes up
ajs_work Wait... \($s) should be equiv to \(=[$s]) not \($s:)
audreyt if the unary-sub-makes-method preference goes away, then it may make sense to make \($s) mean \(: $s) but \$s still means \($s :) -- though I'm not sure about that.
s/unary-sub/unary-subcall/ 14:50
ajs_work So, \&f fills the scalar slot with the sub?
fglock [particle]: maybe the difference is that <!before ...> takes a regex argument, while <! ...> takes a single metasyntax term? 14:51
audreyt ajs_work: yes, and you take it out with $ not with &
[particle] fglock: perhaps... what would <!foo+bar-baz> do? is that equiv to <+!foo+bar-baz>? or more like (syntactically incorrect) <!(+foo+bar-baz)> 14:53
...basically, must it be a single metasyntax term
and if not, how does it interpolate
...a "metasyntax expression". 14:54
14:54 jferrero joined
ajs_work thanks audreyt. This makes more sense to me now, though \ seems pretty heavy-weight. 14:55
14:58 buetow joined
[particle] hrmm, reading, i believe ! is just like + or -, and takes a single term 14:59
fglock [particle]: fixed
svnbot6 r13625 | fglock++ | Pugs-Compiler-Rule - <! ...> takes a metasyntax expression (not a rule pattern)
[particle] what tests are you currently running? 15:00
t/regex/from_perl_rules?
fglock I'm running just the pcr tests, and checking if Test.pm compiles 15:02
audreyt ajs_work: if it's immmutable, we don't have to alloc the redundant storage at all...
fglock [particle]: I was fixing v6.pm hash/array, and got distracted :) 15:03
15:04 ofer0 joined
TimToady it's a bit odd that <!!foo> means something different than <?foo>. I wonder if we should have a different "suppress named capture" char. 15:11
Juerd Postfix ?, perhaps?
Prefix # 15:12
<#foo>
[particle] prefix:<.>
Juerd <.foo>? I'd expect that to call a method. 15:13
TimToady it's supposed to.
assuming that $_ is meaningfully set to something callable
Juerd Grin. Prefix / :)
</html>
[particle] i don't see 'A leading .' in the spec, perhaps it's in [AE]05 15:14
TimToady I get caught *every* time I write s/x/<foo>x</x>/
Juerd Isn't not-capturing like not-caring?
'cause perhaps you can sneak in Whatever here. 15:15
<*foo>
TimToady <=foo>
theorbtwo likes the Whatever idea.
A foo. Whatever foo happens to be around, I don't care.
15:15 c6rbon joined
TimToady "find me a rull ending in foo" 15:16
*rule
theorbtwo <.*zug>
TimToady <|foo> heh 15:17
15:17 vel joined
audreyt pipe is the new colon 15:17
zakharyas hah
obra . o O { Larry gets the pipe } 15:18
theorbtwo In the conservatory?
TimToady <|this> is not a pipe
[particle] is |$smoking
zakharyas foward --> slewis surely?
TimToady <^foo> 15:19
[particle] so, <!!foo> is legal?
audreyt TimToady: I've changed the tests such that the only remaining tests for [,] ensures it's noop-ish 15:20
TimToady yeah, you almost need |<<[,] to get anywhere...
audreyt and all magic interpolation replaced by |$foo and |<< @bar, %bax
TimToady that's fine.
audreyt ...which are, incidentally, implemented along with "my |x"
cool :) will commit+sleep in a bit 15:21
TimToady particle: why not? ! and ? are recursively defined.
theorbtwo |<< @bar ?
audreyt theorbtwo: |(@bar[0]), |(@bar[1]), ... 15:22
theorbtwo Oh, hyper-pipe prefix operator.
TimToady instead of making [,] do it magically
theorbtwo Yeah, seems more or less sane.
That is, it seems the saner of those two positions. 15:23
audreyt since [,] was losing any resemblance of reductional nature implied by [] :)
theorbtwo Oh, do Captures have a type slot -- can you ::|foo?
TimToady <??!!?!???foo> is theoretically legal.
or to raise putter's question, do they have a slot for the constructor you'd call to reconstruct the original (reversibility at stake here) 15:24
audreyt theorbtwo: I hope not... neither does it have a &|foo position.
svnbot6 r13626 | fglock++ | v6 - cleanup Term grammar
audreyt (brb) 15:25
PerlJam Could someone explain to me exactly what we gain by sigilizing Captures? 15:27
I'm just not seeing it.
15:28 lisppaste3 joined 15:32 chris2 joined
TimToady mostly we gain not having to write |$foo everywhere we can currently write |foo. But it might be a fair trade to require |$foo in a signature for the ability to write |func() in an rvalue. 15:33
Juerd TimToady: <*foo> doesn't say "anything ending in foo", as much as *@foo doesn't. Perl 6 doesn't have globs anyway :)
15:33 fglock left
audreyt otoh, if | is made special form just like the old splatty *, |<< will still work. 15:33
TimToady but * as in Whatever has strong connotations of maximizing whatever it is in a globby or Inf way. 15:34
Juerd TimToady: ($foo, *, $bar) = (5, 6, 7)
TimToady: I thought that that'd "capture" only 5 and 7.
TimToady |<< works as long as | parses as prefix:<|>
the operator in question doesn't have to pay attention to those "strong connotations" 15:35
:)
audreyt I like :(*|x) as alternative to :(\$x) but the latter is quite livable
and so indeed there doesn't seem to be a strong argument for sigilizing | 15:36
Juerd I read *|x as any(Whatever, x)
audreyt well, yeah, there's that too.
TimToady by \$x you mean a param that must be a capture?
as opposed to |$x which would capture all the rest of the params into $x?
audreyt the "take the unprocessed arglist and put them into this"
it would? 15:37
TimToady I'm saying if we demoted | to unary then |$x is the old \$x
in a declaration
audreyt ah right. 15:38
Juerd What's the use of that?
i.e. why not just use \ instead then
audreyt and it reads slightly better than $|x
and it reads slightly better than *|x
15:38 weinig|bbl is now known as weinig
TimToady don't want to put | into the twigil slot. 15:38
|$+capture
audreyt nodnod.
TimToady in declaration | is more like * slot
audreyt Juerd: the signature syntax tries to mimick the actual syntax in reduspatch
Juerd What is reduspatch? 15:39
audreyt sub (:$x) { redispatch(:$x) }
Juerd Oh, redispatch
audreyt: But why couldn't you use \ in both?
audreyt sub (|$x) { redispatch(|$x) }
uh because then it's not a redispatch anymore?
TimToady same reason you can't interchange * and & in C
Juerd I see. 15:40
audreyt prepares to revert the ruleSigil patch
TimToady not that C's declaration syntax is to be emulated here...
Juerd must be missing something sufficiently relevant :)
audreyt not having oto write |(f()) is a win indeed
TimToady of course, then someone will ask what "my |@foo =" means... 15:41
yeah, I've been worrying about |(f()) for a couple of days now... 15:43
audreyt the same question can be raised against *@foo anyway
TimToady I mean "my |@foo :="
audreyt it's @foo bound to Capture
which means you automagically gets interpolating access to its positionals... 15:44
my @foo is Capture
TimToady and %@foo gives you the named? :)
audreyt |%foo, yes
and |@@foo gives you access to raw feeds
TimToady and @@@foo gives you the, er...
svnbot6 r13627 | audreyt++ | * Temporarily, for the duration of this commit, | is made sigil. 15:46
r13627 | audreyt++ | It's going to be reverted in the next commit...
Juerd Hehe, nice commit message
15:46 buetow joined
ajs_work Question about List vs Lazy: is there a difference or should that be written "lazy List" as opposed to "eager List"? Reason I ask is that S06 says that List is always lazy, so I don't see why we're using "Lazy" in S29 15:46
TimToady the type hierarchy in .hs has List (Lazy...Eager) 15:47
audreyt S06 I think says List by default is lazy
TimToady but is there really an Eager type?
audreyt so maybe List (Eager)
15:47 justatheory joined
audreyt or is it a strictness annotation? 15:48
15:48 hexmode joined
audreyt Seq is eager, tough. 15:48
s/tough/though/ 15:49
TimToady it's a promise about side effects or something
how can you have a mutable eager thing?
so maybe Seq =:= Eager 15:50
audreyt indeed 15:51
fglock's plan (and the same in newVal) is have List present a immutable interface but maintain some Range/Seq/other-producer-things underneath 15:52
TimToady does that mean the prefix:<seq> is the new eager?
audreyt not unlike how Str maintains buf underneath
hm, that sees strangely familiar... 15:53
Seq(1..Inf) will do already no?
TimToady I guess I'm confusing laziness with immutability here... 15:54
something can be immutable but lazy
audreyt right
=<> for example.
TimToady ENOTENOUGHMUGSOFCOFFEE 15:55
ajs_work I think I might have gotten an answer, but I'm not sure ;)
It does seem, though, that List implies laziness, so at the very least, I don't think we need Lazy in S29 15:56
TimToady yes, but it's a lazy answer.
ajs_work heh
TimToady yeah, but is there any reason Eager is inside List in Types.hs?
15:57 kanru joined, prefiks joined
TimToady or maybe Eager is the subset of values that are evaluated Right Now. :) 15:57
ajs_work I'm not sure that I followed the mutable implies non-eager comment, so I don't think I can answer that.
TimToady actually, I was addressing the lady hunched over the mac 15:59
ajs_work I realized, I was just poking for more info on the nature of mutability, really. ;)
TimToady biab & 16:01
16:01 wilx left
audreyt well, not exactly hunching... 16:08
audreyt discovers OSX doesn't like 0bytes left in / and has to reboot
Juerd 16:10
Oops. Sorry for that :)
16:10 mauke_ joined
[particle] i thought you had zero bytes left in #perl6 :) 16:11
16:20 mauke_ is now known as mauke 16:23 fglock joined
svnbot6 r13628 | audreyt++ | * As promised, | is no longer a rule but a special form prefix. 16:24
r13629 | audreyt++ | * Implement splicing: \( |<< @array_of_captures )
audreyt sigh, s/no longer a rule/no onger a sigil./ 16:25
16:25 prefiks left
svnbot6 r13630 | audreyt++ | * Change the type "Pugs::Internals::VRule" back to "Regex". 16:26
16:27 Psyche^ joined
svnbot6 r13631 | audreyt++ | * Chase the s/Pugs::Internals::VRule/Regex/ change in tests. 16:27
16:29 penk joined
svnbot6 r13632 | audreyt++ | * Change [,] to the shiny new | and |<<. 16:29
16:31 penk joined
fglock how is :$$<foo> meant to be parsed? it is currently using a special rule in Term; are things like :$$$$<foo> valid? 16:34
audreyt well, :$$<foo> is no exactly specced 16:35
however, it parse the same as any consecutive sigils 16:36
$$$foo always binds tighter than anything else
i.e. $$$foo[0] is always ($$$foo)[0] and never $$($foo[0])
:eval my $foo = \10; :$$<foo> 16:37
?eval my $foo = \10; :$$<foo>
16:37 weinig joined, evalbot_r13619 is now known as evalbot_r13627
fglock :@%$<foo> 16:37
evalbot_r13627 Error: Unexpected ":$$<" expecting "::"
audreyt hm ,havn't caught up yet
TimToady "Gee, Dad, that is that swearing?" --Heidi
fglock so ':' just looks ahead for an identifier 16:38
TimToady (fyi, hacking [,] -> | in syns, btw)
audreyt &infix:<plus> means something extremely different from $infix:<plus> 16:39
the latter is a named pair literal
plus => $infix<plus>
I wonder if that form can find some better alternative...
theorbtwo That's somewhat scary.
audreyt and unimplemented by pugs as I can't seem to find a way to do it without massive lookahead 16:40
16:40 zakharyas left
theorbtwo Would there be a major problem with getting rid of the $infix:<plus> form, then? 16:40
16:40 Psyche^ is now known as Patterner
audreyt not sure. I mean, :$<foo> is a special case of that 16:41
:$/<foo> is not valid
$/:<foo> would be the currently specced equiv
another thought is to make :%h<foo> parse as foo=>%h<foo> 16:42
not (h=>%h)<foo> which is likely nonsensical
I think I've raised that with damian on Chicago and he agreed, bu tthen we got distracted into multi dispatch... 16:43
theorbtwo Gah, I need to pay more attention, it seems.
However, more importantly, I need to go eat, right now. 16:44
Later.
audreyt and I need to sleep :) 16:45
TimToady: best luck synhacking :)
fglock do you mean %h:<foo> ? 16:46
16:46 weinig is now known as weinig|bbl
audreyt fglock: I mean have :%h<foo> replace %h:<foo> 16:46
fglock so this is better left to implement after a few days :) 16:48
svnbot6 r13633 | fglock++ | Pugs-Compiler-Rule - Match.kv, Match.elems 17:05
r13634 | audreyt++ | * Pugs.Lexer: tryVerbatimRule now actually tries :) 17:09
r13635 | audreyt++ | * ruleSigiledVar no longer backtracks. 17:12
r13635 | audreyt++ | * Change existing uses of tryVerbatimRule back into verbatimRule.
ajs_work I've fired off a message to the list, condensing my question here into something much longer and rambling ;) 17:14
17:16 lanny joined
svnbot6 r13636 | fglock++ | Pugs-Compiler-Rule - $/{0} === $/[0] 17:16
ajs_work One question about mutability. I know we've had the "Str is immutable, does that mean I can't assign to .substr" conversation, but I don't recall what was decided.
gaal hello, hello. I'm running a smoke after a while of not running the whole thing - and I see the output's more verbose than it used to. is this intentional? 17:17
audreyt: any specific triaging help I can moose?
(I see we're not in mass-TODO phase yet) 17:18
lanny I got the feeling it was the changelog she was not looking forward to, gaal, but was only an impression from one comment when someone asked about ETA. 17:20
gaal yes, changelog's a daunting task
I'll help by submitting my stuff there, then, while the smoke completes :) 17:21
fglock ajs_work: lvalue substr changes the "container" contents (such as $str); whether or not it creates a new string depends on the actual backend implementation 17:22
17:29 mdiep joined 17:31 bernhard joined 17:40 ludan joined 17:48 Odin-LAP joined
svnbot6 r13637 | fglock++ | Pugs-Compiler-Rule - updated Match pod, TODO 17:48
17:50 pjcj joined 17:51 vel joined 17:53 pjcj joined
gaal @tell audreyt vardecl and particularly "my T ($x, $y)" doesn't use the Signature parser yet, does it? There are two items in the ChangeLog about this that may be misleading. 17:54
lambdabot Consider it noted.
svnbot6 r13638 | gaal++ | * ChangeLog: gaal's stuff
ajs_work fglock, so you are saying that $str.substr(0,1) = "A" is the same as $str = "A"~$str.substr(1) ? 17:56
Which would mean that lvalue methods can't be used on non-lvalues... which might be fine... just thinking it through. 17:57
I think regexp substitution is going to be a pain in the ass to write. 17:58
TimToady it was a pain to write in C...
ajs_work I just mean in terms of the replacement of multiple sub-sequences. It's going to have to re-allocated the string for each :g iteration 17:59
fglock ajs_work: yes - and the underlying implementation may decide whether to use mutating buffers or not
meeting & 18:00
18:00 fglock left
ajs_work Will there be mutating buffers with string semantics? 18:00
TimToady If you want something that varies, maybe you want a variable. :) 18:04
ajs_work eh?
It was my understanding, based on S06 that "my Str $str = "one fine day"; $str =~ s/<alpha>/{$1.uc}/g" would either fail, or re-assign $str a newly allocated Str several intermediate times. fglock's statement about the potential use of an intermediate mutating buffer instead was what lead to my question. 18:07
er :g 18:08
you know what I meant ;)
and ~~
and all that Perl 6 stuff... shoot me, I still have to write a lot of Perl 5 code ;) 18:09
TimToady I don't see anything in there that prevents it from doing it all with one copy operation.
"my Str $str" is merely a guarantee that $str will return a Str, not necessarily the same one you put there... 18:10
ajs_work Right... ok. 18:11
18:13 weinig|bbl is now known as weinig
TimToady I guess the interesting question is whether method substr($self: ...) can bind a new object to $self and have it propagate back into $str.substr 18:17
lanny Wouldn't that be $str .= substr(...)? 18:18
TimToady that would chop it down to a substr of itself, but not allow substitution of a substr 18:20
lanny ah. right. Forgot this was about lvalues.
TimToady anyway, commuting... & 18:21
18:29 MenTaLguY joined 18:41 fglock joined
svnbot6 r13639 | kolibrie++ | talk - Practical Extraction with Perl 6 18:42
r13640 | kolibrie++ | added talk to ChangeLog 18:48
ajs_work Was toying with \$foo in pugs. Seems that \ is just ignored. Interestingly the reverse is true: $$foo works, regardless of what's in $foo. 18:51
fglock is this right? 'Match does Str does Array does Hash' 18:54
'Match does Str|Array|Hash' ?
19:01 weinig is now known as weinig|away
kolibrie fglock: thanks for all your help with my talk. I'm going to try to correct any syntax errors I may have put in the slides 19:03
fglock kolibrie: it helped me debug the $/ shortcuts in v6.pm :) 19:04
kolibrie fglock: I think it was good for a number of people :)
I get to present it again this Friday at work 19:05
fglock kolibrie: I'm starting to reimplement map() so that it can run the match-to-HOH transform
kolibrie HOH?
fglock (hash-of-hashes) 19:06
kolibrie oh, awesome
fglock kolibrie: if you have some time, take a look at the new Pugs::Grammar::Rule2.pm
kolibrie fglock: I will 19:07
19:15 ofer0 joined
TimToady Match does Capture? 19:17
fglock yes, I think so
clkao fglock: what's with rule2 ?
fglock clkao: it uses some new shortcut syntax for getting/coercing match elements 19:18
ajs_work <strikethrough>Match does</strikethrough> Capture?
Or does Match have some extra state/methods not implied by Capture? 19:19
clkao where is rule.pmc compiled from these dys?
kolibrie ajs_work: Match is a pretty specific thing, Capture is very generic 19:20
19:20 vel joined
TimToady Match has .from and .to 19:21
fglock clkao: the source code is Rule2.pm - it has a different name so that it does not get recompiled by mistake - which could break everything :) 19:22
Rule2.pm is the actual p6 bootstrap - v6.pm just adds more syntax 19:26
clkao *nod* 19:27
svnbot6 r13641 | fglock++ | v6 - fixed compile-time objects; all methods should return objects 19:37
fglock Str('abc').eq( Str('xyz') ) returns BoolExpression( " 'abc' eq 'xyz' " ) - is there a node in Pugs for this? 19:39
I'd like to reuse the node names 19:40
wolverian throws a hissy java fit 19:41
fglock CapInternals.hs has 'ExpVal' - so maybe 'an expression that returns a bool value' would be 'ExpValBool' 19:50
bbiab
gaal fglock: the type is Val, ExpVal is an alias 19:56
Bool is PureBool
or PureBit maybe
CapInternals is no longer a very good reference, I'm afraid. a diverging approximation, if you will. 19:57
19:59 hexmode joined
[particle] @approximations.>>.diverge; 20:01
lambdabot Unknown command, try @list
wolverian heh. 20:02
fglock gaal: how would you name the node for an expression that returns a bool ? 20:11
gaal you mean boolean context, like ?EXPR ? 20:12
because I we don't have haskell-style type annotations...
20:13 ofer0 joined
gaal so if you do mean that, it's an Exp of the application of ? to another exp.... 20:15
fglock gaal: I have a Bool, which is the result of an expression (lazy evaluated)
gaal EControl (CCall "&prefix:<?>" some_expression) 20:16
oh
okay, well so?
it's just the expression 20:17
fglock yes, but I want to attach the type info to it - so I thought of naming the node something like ExpBool 20:18
gaal what does attaching the type mean here, an annotation that would allow a typechecker to perform some assertion? 20:19
because at the end, you probably have simply some Val
fglock yes, the main use for this is autoboxing (in perl5) 20:20
gaal EVal (VPure (PureBit True))
if the value is already reduced, then that's all the information you need 20:22
unless I'm not understandind where you are
I don't think Perl gives you a way to promise that an expression will yield some type. 20:25
fglock gaal: this is for low level types only - so the compiler can work as much as possible with unboxed values 20:26
so that it knows for example that 'abc'~'d' is still a string 20:27
gaal hm, we do have a ValNative that has PureBit too... but I'm still not sure I see how AST fits into this
er 20:28
NBit that is
but it looks to me you need to do some semantic analysis? maybe you're looking to build another tree, with more hints in it? 20:29
fglock yes - it is performing an ast transformation - at the code-generator level 20:30
gaal ah. you're more advanced than the pugs runcore then.
meaning, these AST types don't express what you want :/ 20:31
fglock oh, ok
20:34 Limbic_Region joined
fglock I have 3 kinds of bool: unboxed/native value (1 or 0), object (Bool), and unboxed/native expression result ( ?$a ) 20:38
[particle] is Bool implemented using bool? if so, then it's really two types 20:40
fglock [particle]: yes 20:41
hmm - it's really 4 types at compile-time: unboxed/boxed and lazy/eager 20:43
ugh - Pugs::Emitter::Perl6::Perl5::Bool::unboxed::lazy 20:44
in p6, if you create a named module inside of another module, does it inherits the base name? 20:46
like: Class X { Class Y { } } - is class Y actually X::Y ? 20:47
20:49 kastol joined 20:52 weinig|away is now known as weinig
svnbot6 r13642 | fglock++ | v6 - added Pugs::Emitter::Perl6::Perl5::Expression 20:56
[particle] gaal, did i mention to you that pmichaud's working on PAST now? 20:57
20:59 jferrero joined
gaal P-as-in-Parrot PAST? 20:59
[particle] yes.
it should express semantics of all parrot hlls
gaal I really don't know anything about that. Should probably educate myself! 21:00
[particle] fglock pointed me at p6ast_draft, and i forwarded a link to him
gaal the draft is outdated
[particle] well, that's what happens to drafts, isn't it :)
gaal even more than CapInternasls :)
especially re: values.
[particle] where's the latest info? 21:01
fglock [particle]: heh - it is in our heads
gaal src/Pugs/Val.hs, src/Pugs/Val/*.hs, src/Pugs/Exp.hs
simply ignore every block not starting with 'data' :-) 21:02
[particle] comments++ 21:05
svnbot6 r13643 | fglock++ | v6 - moving new emitter to pX/Common/Pugs-Compiler-Perl6 21:08
[particle] parrot/perl6 already uses p6 grammar syntax to define the grammar 21:09
is there interest in using same to generate grammars for haskell or p5 backends?
...allowing us to unify on a single grammar?
fglock [particle]: sure 21:10
this was my TPF proposal
[particle] oh? i had no idea.
gaal see also scw's misc/pX directory
(rules -> parsec) 21:11
[particle] gets easily lost in the pugs repo
21:12 weinig is now known as weinig|bbl
svnbot6 r13644 | fglock++ | v6 - moving new emitter to pX/Common/Pugs-Compiler-Perl6 (more) 21:14
r13645 | fglock++ | v6 - moving new emitter to pX/Common/Pugs-Compiler-Perl6 (done) 21:16
TimToady fglock: yes, class declarations default to "our". It's just the outer one that happens to be in the * package. 21:18
fglock TimToady: thanks 21:19
21:21 spoop joined 21:24 timbunce joined
fglock gaal: in short, v6.pm creates a syntax tree, very close to p6 syntax - the emitter transforms this tree into a compile-time tree, which knows how to stringify to perl5 21:29
gaal *nod*
fglock it is probably possible to do this in a single step 21:30
gaal yes, but then you lost portable macros in AST :) 21:33
okay, bedtime for me... good night &
fglock gaal: it is still an ast tree, I just need to keep node names compatible 21:34
it can be reblessed into another emitter class 21:36
21:36 MenTaLguY left
fglock actually, all the emitter needs to do is to bless the nodes in the syntax tree 21:37
which is actually a good idea, so the syntax tree don't get polluted with these Pugs::Emitter::Perl6::very::large::node::names::which::are::a::pain::to::write 21:41
21:48 larsen joined
svnbot6 r13646 | fglock++ | pX/Pugs-Compiler-Perl6 - added tests for ast->perl5 21:52
21:55 gunya joined
fglock when I do ' Bool::True == 42 ' - does it coerce 42 to Bool, or Bool::True to Int ? 22:08
?eval Bool::True == 42 22:09
wolverian the table in S03 indicates the latter 22:12
fglock wolverian: which section is that? 22:13
wolverian "Smart matching"
== seems to be the numeric comparator, in other words. 22:14
fglock found it - thanks!
wolverian you're welcome
22:22 weinig|bbl is now known as weinig 22:24 SubStack joined
svnbot6 r13647 | fglock++ | pX/v6 - more tests 22:24
fglock so far so good - it can make coercions, and it keeps the Type info
22:34 ludan joined
svnbot6 r13648 | fglock++ | pX/v6 - more compile-time evaluation & typing 22:39
22:39 fglock left 22:42 kastol joined 22:50 Aankhen`` joined 22:51 ludan joined 23:03 nperez joined 23:17 weinig is now known as weinig|away 23:23 mako132 joined 23:29 Shabble joined 23:40 aufrank joined 23:48 markstos_ joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!