pugs.blogs.com | pugscode.org | pugs.kwiki.org | paste: sial.org/pbot/perl6 | <stevan> Moose... it's the new Camel ":P | .pmc == PPI source filters!
Set by Alias_ on 16 March 2006.
00:12 Quell joined 00:14 neoesque joined
audreyt rehi 00:19
audreyt sees a _why in the house :)
stevan__ hallo
audreyt hi moosecamel :) 00:20
stevan hola lambacamel :)
audreyt what's up? :) 00:27
stevan not much, looking at Path::Class right now to see if we can use it to replace all our File::Spec stuff
boring I know
and watching Star Trek with my daughter :)
and you? how is $work?
audreyt seeing the end of the quantum tunnel 00:29
aufrank stevan: long time no see!
buu Yay startrek!
audreyt ghc ported to aix43; jifty ported to win32; the rest is just dressing
TimToady Ooh. Let's party!
aufrank parties 00:30
stevan grabs some ice cold Junctions from the fridge as passes them around
audreyt so the $job coding should be all done by tomorrow's Meeting
ooh the subtle taste of junctions
TimToady I think I'll steer clear of the rum barrel though... 00:31
audreyt right, you have you blue sugar cubes already...
TimToady We were just talking about red sugar cubes earlier today. 00:32
audreyt backlogeth
TimToady Though in this case the discussion was polio vaccines.
All my responses are in p6l...
aufrank there's a strange cat in my apartment!
audreyt yay S02+S12 bonus updates
TimToady well, how yay depends on what they say, presumably... 00:33
Actually, didn't need either rum nor sugar cubes this morning--was spray painting the family heirloom crib for the soon-to-be grandson, and got plenty high from the enamel fumes... 00:35
due in less than a week now, so I'll be disappearing to drive to Seattle any moment now... 00:36
plus or minus two weeks... 00:37
Hmm, we need fuzzy junctions...
stevan bushes the fuzz off his junction, tickles my nose when I drink it
buu Yay fuzzy junctions! 00:38
TimToady say $time < 1`week 00:39
Bool::Maybe
mugwump offers aroung the room the fresh meat of the giant aquatic brazilian centipede
TimToady Go not to the elves for advice, for they will say Bool::Maybe... 00:41
audreyt TimToady: congrats :) 00:48
mmm grandpa larry 00:49
buu What on earth is 1`week
aufrank week as a unit 00:51
buu Horrifying. 00:52
aufrank 60`seconds == 1`minute 00:53
maybe === , I don't know for sure
00:54 lumi joined
wolverian my $date_range = now - 1`week .. now + 2`month; # hm 00:56
01:00 penk joined 01:08 kanru joined 01:17 justatheory joined
audreyt goes to pre-meeting meeting & 01:19
01:31 drbean_ is now known as drbean 01:34 flounder99 joined 01:43 mako132_ joined 01:50 jserv-- joined 02:14 neoesque joined 02:27 bsb joined 02:55 arcady joined 03:05 flounder99 left 03:16 qu1j0t3 joined 03:46 mako132_ joined 03:50 multic joined 04:15 shachaf joined 04:51 jserv-- joined
aufrank audreyt: let me know if you're back after the meeting 04:58
04:58 kanru joined 05:02 kanru2 joined 05:13 KingDiamond joined 05:23 hikozaemon joined 05:37 kanru2 is now known as kanru
svnbot6 r10243 | scw++ | Pugs::Grammar::MiniPerl6 05:43
r10243 | scw++ | * list constructor accepted
r10243 | scw++ | * more careful escaped charactor in string
05:49 iblechbot joined 06:17 marmic joined 06:28 neoesque joined 06:33 ghenry joined 06:36 drbean_ joined 06:53 _seaker_ joined 06:55 renormalist joined 06:58 aufrank joined 07:06 neoesque joined 07:09 _seaker_ is now known as k00k, k00k is now known as c00k 07:20 drbean joined
gaal re: [email@hidden.address] !!!! (!) 07:32
# reduce in list context == scan
wolverian scanl or scanl1? (or r?)
gaal well, foldl or foldr has already been addressed 07:33
wolverian ah, I've missed that
gaal so presumably the same direction
wolverian how was it addressed
gaal wolverian: is [-] @l r or l? depending on &infix:<->
wolverian s/$/?/
mm, okay. thanks 07:34
how about the foldl/foldl1 distinction?
gaal I mean for reductions in general -- including scans, there's a default direction 07:35
now let me remember where it's controlled
L<S03/Reduction operators> 07:37
errr, I can't find discussion of fold direction there. hmm? 07:38
07:39 elmex joined
gaal "it must associate the same way as the operator used". but how do you control that? 07:39
oh, okay: it's in the prototype (S06) 07:41
(is assoc)
bsb gaal!
I thought: @b <== map { ... } <== @a 07:42
was like a scan
but haven't considered it in detail 07:43
gaal bsb: Hi! got your moosage. The postmark said May 3, so you do the math
um, isn't @b <== map {} <== @a... like a map? :) 07:45
bsb but lazy 07:46
aufrank that's how I parsed it when I read it
bsb or maybe not
gaal the thing with scans is the accumulator
bsb ah I remember now,
maybe map { state }, that's gross though
gaal reduce (+) [1..10] -- 10th triangular number 07:47
scan (+) [1...10] -- 10 first traingular numbers
bsb My original plan was @a <== { $^a + $^b } <== @b being scanish 07:48
gaal (I've simplified because it's fold with identity and direction blah blah, but that's the idea)
bsb see www.nntp.perl.org/group/perl.perl6....uage/20905
but that was canned. Sensibly too
gaal cute idea though! 07:50
ingy obra: ping 08:20
aufrank soliciting comments: aufrank.net/notebook/development/SoCapp.txt 08:35
gaal furthering .. Further - consider furthering .. Moreover 08:38
will be useful to the community even aspects - "even *if* aspects"
the list in the deliverables looks like requirements for something else, the thing built on top of your API 08:40
so maybe make more concrete what will be in the API itself 08:41
aufrank good point 08:43
thanks
08:46 OutThere joined
gaal sure! good luck with the application 08:46
aufrank thanks :)
are you in a position to apply? are you applying? 08:47
gaal not a student :)
aufrank that's why I didn't just ask what your project was :)
it's too easy to assume the rest of the world is in school when you are yourself
been bit by that one
gaal it's easy to wish more of the world was! 08:48
aufrank heh
spinclad audreyt: i've got updated JudyHS docs locally. next step: check in fresh 1.0.3 sources to thirdparty/, and apply my patch. 08:50
08:50 iblechbot joined
aufrank spinclad++ 08:53
Aragone aufrank: s/Grammars/Grammar/ in the first line of the summary there? 08:56
aufrank took out the 'a', but yeah. good catch, thanks. 08:57
Arathorn looks good to me, otherwise :) 08:59
(both in content and grammar, so to speak :)
aufrank =\
09:28 OutThere joined
audreyt rehi 09:29
spinclad: cool!
audreyt praises hot spring pools with wireless support 09:30
gaal thinks audreyt confused those with hot *spots* 09:31
audreyt no, I'm actually in a large container of sulphuric hydrogen oxide liquid... 09:33
scw: how did you test the MiniPerl6 grammar? 09:37
scw: can you check in a simple Makefile.PL or t/ to MiniPerl6 so that the process of .grammar->.pm is automated?
webmind hot springs with wireless? 09:39
awww :(
audreyt cf. redhanded.hobix.com/cult/rubyspameeting2006.html 09:40
webmind nice.... 09:41
wolverian "incremental tri-color write barrier GC" heh
audreyt looks forward to the Data.Rope library 09:42
based on boehm's www.cs.ubc.ca/local/reading/proceed...spe986.pdf
aufrank as in "enough to hang yourself with"?
;)
audreyt the unicode_string draft I did with Kogai already requires this kind of data structure
so it's nice to see something reusable :) 09:43
aufrank: also, it may be cyclic to tie knots, or something
aufrank =P
gaal I'm implementing TimToady++'s excellent scan idea 09:44
audreyt list-context fold? 09:45
gaal yes
audreyt cool
aufrank ok, updated: aufrank.net/notebook/development/SoCapp.txt
gaal++ 09:46
Arathorn++
audreyt aufrank: looks great! 09:47
gaal we don't have a good way to flip the assocativity of an operator, right? so controlling foldr and foldl is always by finding out the assoc of the present op and possibly running the list through reverse, right?
aufrank gaal had the magic suggestion 09:48
Arathorn gaal: is that scan being a reduce operator in a list context? or did it get refined further?
gaal well, I just asked about folds, even before the scan the question was pertinent 09:49
[] is our reduce metaop, with list context it encapsulates both scan vs. fold and r vs. l 09:50
(with "[] reverse") being the r<->l control :-)
Arathorn heh, nice
audreyt ponders &infix:<+> but assoc('right');
gaal yeah, that's what I was wondering about
Arathorn still seems slightly weird that say [+] (1,2,3) will presumably generate "136" rather than "6" 09:51
gaal but is our rec{ override }
audreyt can certainly be made to work
([+] 1,2,3).say 09:52
gaal haskell has visual rtl folds, our are logical
arrrr beh, I can't higher-order scanM and foldM because they have different sigs. Darn. 09:53
I don't think I can sweep this into Val being a list, either :( 09:54
aufrank submitted :)
gaal aufrank++
good luck again :)
aufrank thanks again 09:55
gaal actually, I *have* to sweep this list into a Val, come to think of it. good. 09:56
aufrank ponders submitting another one, GData client library in Perl 6 09:57
Arathorn what's gdata? the google xml format?
aufrank yeah, atom + namespace extensions to integrate events, contacts, and "messages" 09:58
where a message is any text sent over the wire that wasn't a person or an event
code.google.com/apis/gdata/index.html 09:59
09:59 chris2 joined
Arathorn nods 09:59
aufrank I actually think it's a pretty well thought-out extension to atom, which I think is a pretty well thought-out spec
Arathorn wonder if they'll submit it to ietf 10:00
aufrank I don't know if namespace extensions usually go that route 10:02
but the lockin isn't too bad, since the document will still parse as valid atom even if google takes there ball and goes home
s/there/their
Arathorn nods
aufrank wow, /me is fading. 10:03
spinclad yeah, /me too. no checkin tonight... 10:04
and aufrank: bon chance
audreyt spinclad: as soon as JudyHS iterating hits the repo, I think I'm going to switch Pugs internals to use that :) 10:06
aufrank: so, no PDL for lambdafolks? :) 10:08
spinclad question for the assorted svn-aware: judy has an svn repo on sourceforge. i think for third-party/ i should unpack their 1.0.3 tarball and check all that in, just as baseline for our project-local changes. sound sane? 10:09
audreyt well, not sure about "all that in"
maybe we just need src/. 10:10
spinclad src/ and doc/ 10:11
audreyt yes to doc/ext/{COPYRIGHT,LICENSE} 10:12
not sure about the others
your call really :) 10:13
aufrank audreyt: yeah, I just don't have those chops and didn't even really know how to write the proposal.
spinclad i need doc/ext/JudyHS_*, as that's my doc changes
10:13 nnunley joined
spinclad that's good enough to take patches against, which to send upstream 10:14
audreyt oh, sure then
so yes, check in pristine, then check in your mods
spinclad ok, thanks, will do. 10:15
to bad &
/bad/bed/ grr! 10:16
audreyt may be worthwhile to check the project svn vs the 1.0.3 tarball to see if anything noncompat has happened
g'nite :)
aufrank audreyt: did the update to PIL that was going to follow PILN ever happen? 10:17
spinclad last change mentioned in svn was about when 1.0.3 came out.. will fit a quick double check in before too long.
r48 came out later. will look. 10:20
& for reals 10:22
aufrank audreyt: was $meeting for $job, $cabal, or $parrot? 10:29
audreyt aufrank: $job 10:39
aufrank: the update to PIL is the same thing as Capturizing pugs 10:40
aufrank that is one connection I most certainly had not made
audreyt ok. pugs internals currently has various Syn nodes 10:41
to compensate for the fact that we didn't have macros that manipulate ASTs
aufrank ok
audreyt but now they can all be desugared at parser level 10:42
presumably
right?
aufrank that's what I'm working on!
svnbot6 r10244 | clkao++ | Make this like a real module.
audreyt exactly.
aufrank has the beginnings of Capture.pg sitting open in his editor
audreyt woot 10:43
so yes. instead of the "App" taking arbitrary Exps
nnunley pg for Perl Grammar?
audreyt nnunley: yeah.
aufrank yup 10:44
those postgres guys don't really need file extensions
so we called dibs
clkao ;)
audreyt instead of the App taking arbitrary Exps, the parser needs to emit PIL constructs
that is, explicit Call/Apply nodes taking Captures as arglists
svnbot6 r10245 | clkao++ | Correct & todo tests, retire parser.pl. 10:45
audreyt and this structure needs to be reflected back to p6 space as well
clkao scw: ;)
nnunley clkao++
aufrank ok
clkao it's now a real module
aufrank audreyt: that makes sense. and call/apply are 2 of the nodes in PILN?
audreyt right.
Call is anything that has a name 10:46
$foo.moose(3); blah(4)
Apply is one that has not
$foo(3); &blah(4)
in the current Exp/App node we confuse the two
resulting in various bad dispatch bugs
(for hopefully obvious reasons)
aufrank yeah, got that 10:47
audreyt so Capturizing involves 4 steps
first is to rework the Val type so that it agrees with PIL^N's object layout 10:48
(without this we can't easily reflect structures into Val space (aka object literals))
next is to rework Exp so it reflect the new calling conventions
next is to convert Pugs.Eval to rid of reduceSyn 10:49
moving them into macro space
and retain only the "dynamic" part in evaluator
all static parts are none of its business
finally, update the codegens to deal with serialized object space, instead of raw haskell yaml dumps 10:50
where "serialized object space" starts with ::Main
normally
and contains &Main::MAIN (main program)
et cetera.
most of this is already done in the PIL^N prototype
just need to cannibalize them into the core runtime 10:51
end of explanation. am I making sense? :)
aufrank yes
lots of it
though I initially read Capturizing as a process that would be done during every parse, and thought "all of this reworking! why not throw out the broken parts?" 10:52
then I realized you were describing a one-time throwing-out of the broken parts 10:53
audreyt yup :)
where "broken" means "fresh new calling convention as of December 2005"
but well, things change.
aufrank quicker around here than most places
audreyt indeed 10:54
aufrank got a minute to answer Signature.pg questions?
audreyt sure 10:55
aufrank can I get RETTYPE out of RETTYPE ROUTINE name (...) {...} without parsing ROUTINE and name? 10:56
(for the purpose of implicitly -->'ing it)
RETTYPE is properly part of the sig, but ther others aren't
audreyt RETTYPE can only be there if following declarator 10:57
properly speaking, it's the declarator's job to munge the Signature
not the Signature parser's
declarator means <my our> etc
aufrank so add <declarator>? <typeName>? to the beginning, or leave the munging out of Sig.pg? 10:58
10:58 kanru joined
audreyt I think it belongs to Declaration.pg 10:58
not Sig.pg per se.
aufrank fine by me 10:59
audreyt or rather, Declarator.pg
aufrank are [_]**{2..*} names allowed for variables?
audreyt I don't see why not. 11:00
1..* is sufficient.
aufrank is the ^ [s|tw]igil ever used in declaring types within a sig?
audreyt as a twigil, no 11:01
as a sigil... I don't think so. it's still the ::dwigil.
aufrank are :$^foo and :foo($^bar) allowed? 11:02
how about *@^baz ? ( a different question ) 11:03
audreyt I think none of them are allowed. 11:04
arguably
{ sub ($^x) { 3 } }
expands to
-> $x { sub ($x) { 3 } } 11:05
but it really does not make sense to construct a closure this way
so simply disallowing it makes more sense
pasteling "gaal" at 192.115.25.249 pasted "typing help needed" (33 lines, 1.3K) at sial.org/pbot/17116
aufrank and there's only ever one (non-unpacked) *@slurpy and one (non-unpacked) *%slurpy in a sig? Can you have both of those in a single sig? 11:06
11:06 penk left
gaal expects forall voodoo is required to fix this? 11:06
unrelatedly. couldn't haskell benefit from "mcase", where mcase aexp of ... is equivalent to do { exp <- aexp ; case exp of ... } 11:08
audreyt aexp >>= \exp -> case exp of 11:09
gaal: post a whole patch so I can build?
gaal sure, sec
audreyt aufrank: yes you can.
aufrank but just one of each, yes? 11:11
audreyt certainly 11:13
alternately, you can put multiple
gaal audreyt: perlcabal.org/~gaal/tmp/reduce.0.patch
audreyt but the ones after the first are all empty
I don't mind either interpretation; whatever is easier for you :) 11:14
aufrank k
audreyt the spec doesn't forbid multiples.
alternately, multiple ones _all_ receives the slurpy things.
aufrank it's really the difference between ? and *
audreyt that may not be too sane, though.
aufrank so not a big deal
and a final slurpy scalar treats everything it slurps as positional? 11:17
aufrank checks syn 11:18
audreyt gaal: it has nothing to do with forall 11:19
gaal braces for a silly error
aufrank biab & 11:22
audreyt gaal: getReduceFuncs can't be pure
either asks for context outside 11:23
or make it monadic
gaal eep, indeed (I wonder how the compilation didn't mention that?)
audreyt because you can supply an instance and make it work... 11:24
gaal but yeah, s/let// and s/=/<-/
audreyt also, (\a b -> f (a:b)) is somewhat bogus
because as specced, a and b are both Val
and you can't : them together
gaal I did have a hard time understanding how the foldMn version got away with that!
(but it does) 11:25
(also, isn't drop 1 list =:= tail list ?)
audreyt well because list2LoL gets [[V]] 11:26
I mean, it returns [[V]]
so the type "b" is [B]
not B
gaal uh huh. but scanM (sneakily) has the same signature as foldM; how come it doesn't work there then?
audreyt because you got the sig wrong 11:27
you had
scanM :: (Val -> Val -> Eval Val) -> Val -> [Val] -> Eval Val
but it's (sneakily)
scanM :: (Val -> [Val] -> Eval Val) -> Val -> [[Val]] -> Eval Val
er I mean
gaal (ah re: drop 1, no it isn't tail, because drop 1 [] is [], not an error) 11:29
audreyt scanM :: (a -> b -> Eval a) -> a -> [b] -> Eval a
gaal it's scanM :: (Val -> VList -> Eval Val) -> Val -> [VList] -> Eval Val 11:30
except that that doesn't work :)
audreyt no, as it's poly morphic
you were indeed right
scanM :: (Val -> b -> Eval Val) -> Val -> [b] -> Eval Val
works
gaal tries...
audreyt s/b/a/
the rest is left as exercise :)
dinner, bbiab & 11:31
gaal :)
bon mousse.
audreyt perlcabal.org/~audreyt/tmp/corrected.diff 11:33
gaal thanks, exercising
11:34 froh-doh joined 11:39 penk joined
gaal well, it complies and almost works, but the start and end values need to be kludged a bit. will do soon - thanks audreyt :) 11:41
bbiab &
11:45 reZo joined 11:47 reZo is now known as phpError 12:03 kolibrie joined 12:08 kane_ joined 12:09 trampel joined 12:10 jsiracusa joined 12:13 Eljin joined 12:17 hcchien joined 12:19 kane_ is now known as kane-xs 12:20 Eljin left 12:21 Khisanth joined 12:40 neoesque joined 12:41 iblechbot joined 12:59 Qiang joined 13:08 cdpruden joined
Juerd tnx.nl/php.jpg 13:12
13:19 Limbic_Region joined
cdpruden Juerd++ 13:23
13:29 hcchien joined 13:32 FurnaceBoy joined
scw clkao: :) 13:37
clkao scw: now get to work! ;) 13:38
scw clkao: I was thinking if I should use compile_p6grammar.pl or lreq
but lreq not yet support <-["]> :(
clkao make it so! 13:39
scw then what copy should I modify? misc/pX/Common/lreq or the one under Pugs-Compiler-Rule ? 13:40
13:43 chris2 joined 14:02 particle_ joined 14:05 hexmode joined 14:06 fglock joined
fglock scw: ping 14:07
14:11 lichtkind joined 14:13 justatheory joined
fglock scw: it's better to use compile_p6grammar.pl instead of lrep 14:15
lichtkind suddenly i installed pugs, font have to use evalbot :) 14:18
svnbot6 r10246 | gaal++ | * Reduce metaoperator in list context gives a scan. Tests included.
r10246 | gaal++ | TimToady++. Chained ops scanning doesn't; consulting with p6-l.
14:45 jserv-- joined 14:49 aufrank joined
aufrank hey all 14:52
14:57 fglock left
lichtkind hallo aufrank 15:02
15:06 vel joined
svnbot6 r10247 | scw++ | Pugs::Compiler::Rule 15:16
r10247 | scw++ | * character classes in Pugs::Emitter::Rule::Perl5
aufrank has TGE syntax solidified since www.parrotcode.org/docs/compiler_tools.html ? 15:28
audreyt more-or-less, yes 15:31
mostly it's just PIR. 15:32
see eg ./languages/pheme/lib/past2post.tg in the parrot tree
I'm trying to get the PIR part replaced with the MiniPerl6 scw's working on 15:33
in the interest of portable-productions-across-implementations
stevan heya audreyt 15:34
aufrank I get the feeling I should go scan through MiniPerl6
stevan audreyt: so do you think YAPC::NA will be an appropriate time to talk ObjectSpace stuff again?
audreyt surely 15:35
stevan cool
audreyt in particular the p5 objspace 15:36
stevan audreyt: I have been very out of the loop lately, what is the current state of p5 related stuff?
Pugs::Compiler::Rule is still written in p5 I assume 15:37
or rather p5 bootstrapping p6 15:38
audreyt take a look at misc/pX/Common/
the p5 stuff currently is centred around parsers 15:39
or rather, around the rule sublangauge
stevan yes
audreyt fglock has a SoC proposal to start working out the p5 objspace
stevan that is why I figured it was safe to drop back for a while :)
audreyt implementing primitive types in S06
but until that part of work starts, nothing much has happend in p5-as-a-target 15:40
<- just released Filter::Simple::Compile and Filter::Macro (C-style, not Perl6-style macros) to CPAN 15:41
scw new blog entry posted :)
audreyt dozens of .pmc modules lying around in $job code now ;)
stevan cool
audreyt scw++
nnunley Hrm.
stevan audreyt: so the .pmc thing is working really well then 15:42
very nice
scw my connection to blogger is extremely slow now :(
audreyt stevan: right. in $job I used a locally hacked version of Perl6::Subs
stevan nothingmuch and I have been talking about Moose compiled to .pmc
audreyt based on Filter::Simple::Compile
15:42 neoesque joined
stevan oh,.. very nice 15:42
audreyt so I get to write p6 method/subs and avoid startup overhead
stevan yeah
particle_ .pmc? 15:43
audreyt particle_: strange technology developed to allow mixing v6 code in v5 programs, among other uses
cf. search.cpan.org/dist/Module-Compile/
particle_ just happens to have the same extension as a parrot magic cracker... 15:44
audreyt slides: www.kwiki.org/Sporx/pmc.xul
yeah, but .pmc-for-perl5 was here first :)
(1999, as compared to parrot's 2001)
also, parrot doesn't install .pmc files, so it's probably safe.
particle_ just for a second thought you were using parrot for perl5 stuff 15:45
thx for the links
audreyt there is Inline::Parrot as well in the pugs tree... but not actively using it (yet)
scw: yeah, somehow blogger is really slow from hinet here as well 15:46
TimToady seems to me that patrick's P6 bootstrap might turn into MiniPerl6...
scw audreyt: I'm on TANet now...
audreyt TimToady: the goal is certainly to get pmichaud's minip6 to work as rule-production-language in parrot land...
TimToady nevermind, I'm inside out this morning... 15:47
audreyt mmm Klein Wall 15:48
TimToady It's that spot where the surface of my brain intersects itself that I'm having problems with... 15:50
audreyt TimToady: did you look at the Ropes structures? does it fit with your brain w.r.t Str? 15:55
svnbot6 r10248 | audreyt++ | * Pugs.Lexer: ruleWs -- proper dynamic <ws> support for
r10248 | audreyt++ | cases where we can't infer what <ws> means statically
r10248 | audreyt++ | from rules input.
TimToady link? 15:56
audreyt www.cs.ubc.ca/local/reading/proceed...spe986.pdf
basically, a Rope is a tree of concats of fragments 15:57
allowing sharing of fragments
TimToady sounds about right.
audreyt basically, O(1) concat with no memcpy. 15:58
(which is what we don't get currently in any of the 4 runtimes)
TimToady now just get the OS to mmap one. :) 15:59
audreyt well, a one-fragment is just a degenerate case :)
PerlJam audreyt: Are you going to implement Str as Ropes?
TimToady nod
audreyt PerlJam: probably not me. probably some lambdafolk.
(as Data.Rope on top of Data.ByteString)
16:00 pmurias joined
pmurias hi 16:00
audreyt just wondering if we can profitably reuse that work
seems to be yes to me, based on a cursory glance
pmurias: hi!
TimToady I'm wondering if @foo ~~ /.../ can view @foo as a funny kind of rope. 16:01
audreyt (or, more likely, Data.Rope on top of Data.EncodedString on top of Data.ByteString)
TimToady: well, it lacks position-index information
PerlJam interesting
audreyt so need to do a O(n) prescan
we'll call it study()! ;)
_why hey audrey 16:02
when you get a minute, i need to reconcile your syck stuff
TimToady seems like something would want to be done lazily in case it's never needed.
PerlJam audreyt: don't they keep the length information around? you could get position from that when needed.
audreyt _why: here's a minute 16:03
_why is it not too late there?
audreyt it is, but a minute is not much
PerlJam: yeah, but @foo[1] may need to stringify 16:04
@foo itself is sparse; it may contain nonstrings
but as TimToady said, this can be done in an as-needed basis
_why your syck binding has two alterations (aside from some very common patches i've seen in other bindings)
both are to the emitter
audreyt right. one is for mark_node to distinguish seen from unseen 16:05
_why i'm looking a YAML::Syck code, so maybe these were done by gaal or someone else?
tell me more about that one
svnbot6 r10249 | pmurias++ | a regex-based parser for vtable.h
audreyt 0.44? it's all by me I think
the C part
_why yes yes that's the one
audreyt ok. in Perl5 we have this strange idea that an object containing another object doesn't really do that 16:06
_why bizzxare, go on
audreyt it contains a "reference" which points to the real object
the reference is like a typed tag (of the 6 or so builtin types) to indicate the target object layout 16:07
as well as (potentially) the class
so in mark_emitter, there is always a level of indirection 16:08
see perl_common.h line 54
if (SvROK(sv)) perl_syck_mark_emitter(e, SvRV(sv));
this says "if this value on hand is a reference, mark the referenced item" 16:09
_why mmmnn
audreyt however, a reference can refer to itself.
my $x; $x = \$x;
is the canonical example
_why i follow
audreyt so that naive logic loops
16:10 ludan joined
audreyt to break the loop I need to know if it's previously marked somehow 16:10
I can use emitter_xtra and hold another map
ludan hi
audreyt or I can abuse the ->markers map.
I chose the abusive way. :)
end of explanation :)
_why ohhh
audreyt ludan: hi!
TimToady ludan: welcome, at least by default... :) 16:11
16:11 neoesque joined
_why the other patch concerns using quote2 styling for literals 16:12
i guess this is for compat with YAML.pm?
audreyt yeah.
though I remember i overloaded the meaning of scalar_fold 16:13
because I don't use it myself
so I made it mean scalar_utf8
_why yeah :)
audreyt if there is a real scalar_utf8, well, I'll use that :)
also, the json pre/post processor may be of interest to you
_why it is 16:14
audreyt in perl_syck.h
though if you have a native impl, I can throw that away as well :)
_why and lib/JSON and all that
working on it
audreyt yay :)
_why if i can get a real resolution from oren and ingy, i'll use that 16:15
audreyt k
ingy hello
audreyt ingy is right here :)
_why but in the meantime i'll probably use what you've got
ingy :D
_why that scared me
don't do that
HI INGY
ingy I broke my right arm
_why holy shit
ingy al
mo
st
_why bravo! 16:16
ingy on some good drugs
I can't really move the arm, but gugod gave me a nice external keyboard to use
audreyt ?eval [ [+] 1..10 ] 16:17
evalbot_10249 [55,]
_why wow this is almost like chatting with steven hawking
ingy what's up with yaml?
_why yeah that's a good question 16:18
audreyt ?eval [ [+] 1..10 ] 16:19
evalbot_10249 [1, 3, 6, 10, 15, 21, 28, 36, 45, 55]
audreyt that's more like it.
gaal++
xinming ?eval [ [+] 1..10 ]
evalbot_10249 [1, 3, 6, 10, 15, 21, 28, 36, 45, 55]
_why this is a magical place, for sure
audreyt ?eval [ [~] 'a'..'f' ] 16:20
evalbot_10249 ["a", "ab", "abc", "abcd", "abcde", "abcdef"]
audreyt TimToady++ # shiny new scanner
TimToady blush 16:21
audreyt TimToady: so, do we get that into S03? :)
aufrank gaal++ # show the r-l toggle with .reverse!
gaal meows
audreyt ?eva [ [+] reverse 1..10 ] 16:22
?eval [ [+] reverse 1..10 ]
evalbot_10249 [10, 19, 27, 34, 40, 45, 49, 52, 54, 55]
xinming ...
aufrank but that's doesn't show scanl vs scanr, does it?
particle_ so, S03 says, "[+&]() # +^0 (-1 on 2's complement machine)" 16:23
gaal ?eval list [+] 1..10 # also works
evalbot_10249 (1, 3, 6, 10, 15, 21, 28, 36, 45, 55)
particle_ but afaik we're doing ieee, which is sign-magnitude, so will never be 2's complement
audreyt aufrank: it does.
aufrank: to get True scanr, use two reverses 16:24
?eval [ [+] reverse 1..10 ].reverse
evalbot_10249 "55 45 25 94 54 04 43 72 91 01"
audreyt er, not _that_ reverse...
aufrank ?heh
audreyt ?eval list reverse [+] reverse 1..10
ingy haha
evalbot_10249 (55, 54, 52, 49, 45, 40, 34, 27, 19, 10)
audreyt better.
16:24 xinming joined
gaal hmm, should there not be an 'is zero' trait for operators? 16:25
audreyt no, it's just MMD
or rather, zero-arg multi dispatch
aufrank is impressed
gaal how do you do non-1 reduces with alternate zeros?
Juerd array.reverse stringifies the array? 16:26
audreyt no. it's a bug.
gaal fold (+) -42 ls
audreyt (due to confusion of Call vs Apply)
xinming wonders what the differences between `my method` and `submethod` are...
anyone here would clarify it for me please? :-/ 16:27
audreyt xinming: submethods are visible to outside world
submethod means "noninherited methods"
gaal C<< &infix:<+> but is assoc >> possible, C<< &infix:<+> but is zero(-42) >> not
xinming hmm, so, for `my method`, It can only be invoked in self.meth form? 16:28
gaal so the Grand Reduction Unification (3d) isn't complete
audreyt self!meth.
gaal: say what?
my &infix:<moose> ::= &infix<+> but assoc('right'); 16:29
my multi sub infix:<moose> () { -42 }
xinming hmm, Ok, thanks, sorry, I must go for learning driving for month... I'll read synopsis next month more carefully...
bye all
audreyt have fun :)
gaal: I suppose this works... if it doesn't, it can be made to work
gaal audreyt: okat, the fist form would probably be ok for on-the-fly reduces with the spelled-out version, &reduce 16:30
but the second won't, would it?
audreyt the second line adds a variant to infix moose 16:31
try again
gaal &foldr ::= -> &op, @l { reduce &op but assoc('right'), @l }
audreyt my multi sub infix:<moose> () { -42 }
TimToady MMD comprises much wonderfulness...
audreyt my multi sub infix:<moose> is assoc('right') ($, $) {...} 16:32
particle_ croons, "i'll take manhattan..."
gaal ok, what I'm asking is how do you write a (non ugly) fold with an overriden zero?
TimToady MMD: "Maybe Manhattan Dwims" 16:33
particle_ :)
audreyt gaal: a generic foldr or just for [+]?
gaal generic
no, ignore the r in foldr 16:34
audreyt so it takes $new_zero, &op, @l
like that?
gaal yes
audreyt well, trivially:
multi myfold ($v, &) { $v }
gaal Haskell has 3 dimensions: scan/fold, right/left, assumed nonzero/parameter zero
audreyt multi myfold ($, &o, *@l) { reduce &op, @l } 16:35
there you go
s/op/o/
gaal [] unifies them, and adds Logical Associativity<tm>
aufrank audreyt: what are bare sigils in Signatures?
audreyt aufrank: they have internal and external names as empty
aufrank: meaning, it is only checked for type and never really bound into variables 16:36
gaal but how do I say, I want to perform this fold on some arbitrary op, but with this arbitrary zero?
aufrank so a slightly picker *
audreyt yup
aufrank s/picker/pickier/
audreyt myfold 42, &infix:<+>, @list;
gaal: like that?
gaal yes but I said nonugly. :-)
aufrank how can I make a rule matchthe null string between $ and put it into $<external>? 16:37
gaal I guess "can't win 'em all"
aufrank s/$/$,/
audreyt [+] @list // 42;
[+] (@list // 42);
actually, that should be || here 16:38
PerlJam aufrank: Why do you want to match the null string and store it somewhere? 16:39
gaal looks pretty good, but would that work where @list = (undef) ?
audreyt but || would confer context
so
[+] *(@list || 42)
is the Really Works version
gaal: that's fine
aufrank oh, if I just don't match a name but the parse succeeds, the external name will be empty 16:40
audreyt ?eval my @list = 1..10; [+] *(@list || 42)
evalbot_10249 55
audreyt ?eval my @list = (undef); [+] *(@list || 42)
evalbot_10249 undef
audreyt ?eval my @list = (); [+] *(@list || 42)
evalbot_10249 42
gaal yay!
aufrank so it's { ... [<sigil> <name>?,]? ... } # or something like that
audreyt aufrank: I s'ppose so
TimToady where I presume "non-ugly" is everything but the variable must be punctuation. :) 16:41
audreyt looks like that :)
[+](*(@_||42))
[+](*(@_||$_))
TimToady I know some people who would call that cork-brained...
aufrank are you so surprised? there's been undisguised APL admiration on p6l today!
audreyt indeed
clearly it's Advanced PerL 16:42
TimToady yes, well, Mark's college honors project was to write an APL interpreter in PDP-11 assembly...
sufficiently advanced perl...
audreyt ...is indistinguishable from APL
TimToady now it's SAP as well...
gaal TimToady: well, seeing as Haskell has eight fold functions (and I haven't even gotten to monadic versions), I think we did pull a high order on them, even at the cost of a few punctuation chars :-) 16:43
TimToady I don't mind beating Haskell occasionally. Perl has to be good for something...
audreyt TimToady++
ludan TimToady: hi :) 16:44
TimToady by now that should be a "rehi" :)
ludan yep :P
TimToady what is your interest in P6? 16:45
ludan just lurking to undestand a bit more, and after i wanna start with a functional language like haskell 16:46
TimToady P6 is a functional language like haskell, sorta... :)
ludan TimToady: i'm not an active part of P6 :(
TimToady don't let audreyt hear you say that, or you'll have a commit bit before you know it... 16:47
ludan ok :)
aufrank sub foo (*@bar, *%baz, *@qux, *$qux) { ... } # how many slurpies can you have in a single sig, and what combinations are legal? 16:48
TimToady syntactically or semantically?
16:48 fglock joined
aufrank syntactically 16:48
TimToady I suspect the syntax doesn't care how many you put. 16:49
If it chokes, it'd be in semantic analysis somewhere.
audreyt ludan: danieleludovici at email.it ?
ludan audreyt: nooooooooo :)
TimToady oops, too late...
now you're done for...
ludan audreyt: i'm not joking, is not my email :)
aufrank TimToady: ok, audreyt said something similar this morning. I buy it. 16:50
TimToady so you only believe it if it comes from a guy, not a gal? :P
aufrank I forget things after too many hours of being awake. There's been sleep since then. 16:51
audreyt ludan: too bad :) not ludan at linux.it either?
ludan audreyt: no no no :
:)
TimToady I believe ludan is coming in from .nl somewhere...
aufrank also, I only asked about *@foo combining with *%bar... putting all of them in there confused me again
ludan i'm just studying in the netherlands
TimToady It's certainly allowed to mix one *@foo with one *%bar. 16:52
aufrank yes 16:53
sub foo ($bar, *@) {...} # ignores everything after the first scalar?
(are there unnamed slurpies in general?) 16:54
TimToady oughta work.
audreyt foo(1, baz=>3) would die.
need to say
sub foo ($bar, *@, *%)
to catch that
aufrank man, whatever
sub foo ($bar, *) { ... }
audreyt I guess
sub foo ($bar, *)
yeah, whatever, never mind :) 16:55
hm. 16:56
TimToady hr.
aufrank sub foo (:$, :foo$, :($bar)) # all syntactically ok?)
audreyt sub foo (*, *)
wouldn't make sense then
sub foo ($, $) 16:57
need that to confer arity
TimToady yep
aufrank audreyt: any whatever automatically puts you into the Inf parse, but so does any arity > 1, right?
audreyt right, though I'm not talking about parse
TimToady aufrank: I don't think :foo$ works 16:58
audreyt aufrank: syntactically, I'm not sure about :foo$
the other two is fine, though maybe semantically questionable
aufrank something something great minds something something
TimToady same gutter, I believe...
audreyt . o O (redundant array of inexpensive designers) 16:59
TimToady I resemble that remark.
SamB hehe
audreyt @Larry is RAID;
TimToady I don't find them so R
audreyt maybe because we are still at level 1 17:00
particle_ perhaps level 0
TimToady you lose your R when one of them goes sour, and with designers that happens frequently...
audreyt 1+0 maybe.
TimToady though I haven't noticed that expensive designers are any more reliable... 17:01
aufrank mumble arc 17:02
SamB well, expensive designers don't conk out on you as much because they need food
TimToady you sure you don't mean expansive designers? 17:03
they have more reserves...
SamB no, I mean, expensive designers can afford to buy food 17:04
TimToady expansive designers could afford to have not bought food...
SamB true!
lichtkind TimToady how did you knew that perl is the right name? (have to name my prject right now) 17:05
TimToady well, it wasn't the right name in Japan. :)
szbalint Oh, care to tell a bit more about that? :) 17:06
TimToady but mostly I knew that people irrationally like a work with positive connotations.
aufrank and so the Revolution resulted in Wii
TimToady szbalint: r and l are difficult in Japanese...
s/work/word/ 17:07
I have a bad name too...
particle_ levorution is still better than wii
szbalint I see.
TimToady I had to learn to answer to "Rally Woeru" 17:08
aufrank catchy
TimToady and Perl comes out "Paaru".
but that's okay. I think the Japanese and English are about equal in mangling each other's phonological systems... 17:09
SamB still wants to vandalize the boxes to say Revolution instead of Wii
TimToady English speakers can't tell the difference between dwarf trees and jumping out of airplanes... 17:10
aufrank phonologists love japanese vowels and prosody
TimToady especially pitch accent, yes.
'course, there's about 26 different pitch accent systems in Japan... 17:11
but at least everyone recognizes Tokyo accent.
sorry, you pushed my phonology button... 17:12
szbalint My girlfriend always said that I should stay away from linguists. That might be a bit hard as she's slowly becoming one. :)
TimToady It used to be I couldn't spell lingrist, and now I are one... 17:13
fglock scw: lrep is under a rewrite - compile_p6grammar.pl has less features, but it is more up-to-date
lichtkind thanks TimToady but had you a moment where you did know thats it ore just you feel there is no better
TimToady ore wa, ringurisuto... 17:14
17:15 cratylus joined
TimToady though, actually, they have their own word for linguist: č؀čŖžå­¦č€…, gengogakusha. 17:16
which, oddly, I can never remember...
aufrank rule default { = $/<default> := <expression( terminates => ',' ) } # valid rule, even with in-place redefinition of $/<default> ? 17:17
gaal @facs := [*] 1..*; &fac = { @facs[$^n] } 17:18
# I want memoization so that this works. :-)
# stolen cold from a nice Haskell page 17:19
aufrank what's the trait that means is memoized?
gaal 'is memoized' :)
aufrank I thought there was a sexier name
TimToady is cached
aufrank that's the ticket
TimToady it has a certain cachet
aufrank re: previous q on rules, does Match object binding need to be set off in curlies or anything, or can it happen unmarked in a rule? 17:20
TimToady I don't see anything wrong with what you wrote, though $/<default> can be shortened to $<default>. 17:21
gaal wonders if there'll be a programmatic extensible way of manipulating sub traits, so the memoization can get its many optional arguments set 17:22
TimToady I think, though, that the expression terminates on a precedence, not a token.
aufrank points at audreyt
she did it!
gaal bbiab
audreyt what did I do? 17:23
aufrank TT: (terminates => <some_level>) #?
TimToady Gave aufrank a token of your appreciation.
aufrank audreyt: you clued me in to the terminates argument to <expression>, and suggested ',' as the way to end defaults
(which was very helpful and kind of you to do!) 17:24
audreyt oh. that. :)
if -e {...}
surely the { is a token not a prec?
TimToady in discussions with pmichaud we determined that in general you want to terminate on a set of possible tokens, where that in part be specified by precedence.
s/in part/can in part/
audreyt that works, but sometimes you still want specific tokens 17:25
TimToady yes, "set of possible tokens".
but sets can be specified either by enumeration or by rule.
audreyt that worksforme
TimToady or, in this case, perhaps both. 17:26
aufrank we actually can't have just ',' here, because the default could be the last thing in the sig so it would terminate on ')'
TimToady right. 17:27
though that would also fall under the precedence rule, if you consider ) to be infinitely loose.
that is, in terms of surreal precedence, the levels inside the parens are a different order of infinity... 17:28
but most other languages would just say "it's just a term, use parens if you want ($a + $b)..." 17:29
In fact, some languages would just use parens without the =.
aufrank hrm 17:30
audreyt good night. have fun :)
aufrank thanks audreyt. sleep good.
audreyt waves
TimToady thx!
audreyt np hth hand :)
aufrank sub foo (%hash = :a<b>, :c<d>) { ... }# or 17:32
sub foo (%hash = (:a<b>, :c<d>)) {...} 17:33
?
(inclusive or)
theorbtwo grumbles over Parse::Binary's lack of documentation.
aufrank ?eval \) ~~ / <[,)]> / 17:36
evalbot_10249 Error: unexpected ")" expecting term
aufrank ?eval \) ~~ / <[,\)]> /
evalbot_10249 Error: unexpected ")" expecting term
aufrank # what's wrong there?
nm, duh
_why does anyone know what's up with ghc on freebsd6? 17:37
aufrank s/)/')'/
audreyt theorbtwo: try search.cpan.org/~simon/Parse-Binary...tive-1.00/
theorbtwo: lathos does a much better job than me; Parse::Binary is stagnated because $job now uses haskell 17:38
aufrank _why: off-topic: do I remember that you were going to put the atom syndication format into YAML at one point?
audreyt _why: I'm on fbsd6 with a working ghc... do you need a binary package?
_why yeah, !okay/news
i'm getting a linker error with libm.so.3 17:39
the lists say use compat5x
uhhh no such thing
aufrank _why: if that was an asnswer to me, I didn't grok it =\
_why aufrank: redhanded.hobix.com/index.yaml
audreyt _why: /usr/ports/misc/compat5x
_why mmmnn okay i will try! 17:40
theorbtwo audreyt: You should put out a new version with POD that says that, probably. 17:41
Er, not that you don't already have enough to do... 17:42
aufrank _why: thx
audreyt theorbtwo: do that for me?
you are now comaint on Parse::Binary.
theorbtwo Blast, now I have another module I need to put out a new version of... 17:43
17:43 hexmode joined
audreyt I still have... 30 or so 17:43
that I need to put out new versions of
so, load balancing :) 17:44
aufrank TimToady: are there named character classes in PGE that map to Ps and Pe?
audreyt theorbtwo++ # thanks in advance
17:44 DaGo joined
_why aufrank: here ya go -> yaml.kwiki.org/index.cgi?OkayNews 17:44
aufrank excellent 17:45
particle_ aufrank: what are Ps and Pe? 17:47
aufrank unicode character classes corresponding to opening and closing punctuation
punctuation -start and -end, from my understanding
particle_ ah, gotcha 17:48
audreyt _why: if perchance after you installed compat5x it still buildfails, a bindist is at perlcabal.org/~audreyt/tmp/ghc-6.4.2.tbz 17:50
_why: you can pkg_add -f it
g'nite for real &
lichtkind night 17:51
_why thanks audreyt, it's movin along now 17:52
glad for yr patches as well!
17:53 bernhard joined
TimToady aufrank: I don't know if PGE has that much Unicode support yet. I would guess not, given parrot's overall Unicode squeamishness... 17:54
aufrank so <alpha> is probably just <[A..Za..z]> ? 17:56
TimToady dunno 17:57
particle_ aufrank, there are pge tests for character classes 17:58
aufrank in parrot/compilers/PGE/t ? 17:59
particle_ looking fro the right one
t/compilers/pge/p6regex/...
builtins.t
TimToady ?eval "ƅ" ~~ / <alpha> / 18:00
evalbot_10249 *** unexpected "\\" expecting comment or "." at -e line 15, column 20
particle_ and they're defined in include/parrot/cclass.h 18:01
TimToady ?!!?!!
anyway, it doesn't match Latin-1 in my pugs. 18:02
aufrank yeah, I tried a couple of those and just put in a dumb character class for common delimiters
TimToady 'course, the real trick is to find the corresponding close delimiter. 18:04
aufrank a trick which I don't attempt =\
TimToady so you'll match (...] ?
aufrank unfortunately yes 18:05
aufrank waves the work in progress flag 18:06
TimToady it's okay for a bootstrap. 18:08
must commute to work... 18:09
biab &
svnbot6 r10250 | aufrank++ | made token/rule names consistent
r10250 | aufrank++ | support for arity declarations using unnamed params
r10250 | aufrank++ | support for trailing : in MMD
r10250 | aufrank++ | update to invocant detection rule from audreyt++
r10250 | aufrank++ | removed lots of outstanding questions, TimToady++ audreyt++
r10250 | aufrank++ | tokenized some previously hard-coded symbols
r10251 | aufrank++ | more permissive slurpy parsing 18:18
clkao oh, scanning reduce 18:23
aufrank clkao: have you looked at Signature.pg? I know you were one of the people considering starting in on it. Extra eyes appreciated.
clkao slightly, but there were many changes and i didn't follow 18:25
aufrank clkao: would smaller more frequent commits be easier to follow? I'm just getting used to working in a public repo. 18:29
18:45 trym joined, fglock joined
fglock aufrank: I did some experiments with Ps/Pe in perl5 - you can find (most) closing delimiters with 18:47
perl -e 'use charnames ":full"; use Unicode::CharName 'uname'; $s="{"; $n = uname(ord($s)); $n =~ s/LEFT/RIGHT/; print $s,"...",chr(charnames::vianame($n)),"\n"; '
pmurias how can i addres files relative to the location of the source file 18:48
clkao aufrank: sure, but the problem is i've been busy
aufrank clkao: no problem at all :) 18:54
fglock++ 18:55
very useful. Thank you. 18:58
19:01 kanru joined
fglock aufrank: np 19:03
TimToady note, however, that many of the ps/pe pairs are not described in terms of left/right 19:08
aufrank uh oh 19:09
TimToady see wall.org/~larry/pspe
you also want to look at lib/unicore/BidiMirroring.txt from a recent P5 distribution. 19:12
aufrank sees
aufrank looks
TimToady and as long as you're there, you could just extract the Ps/Pe from UnicodeData.txt 19:13
At least the BidiMirroring gives you the opposite character, unlike the Ps/Pe data.
aufrank ok, I guess I admit this is hard :) 19:14
TimToady I suspect for now we just hardwire it all...
pmichaud_ <TimToady> aufrank: I don't know if PGE has that much Unicode support yet. I would guess not, given parrot's overall Unicode squeamishness... 19:16
aufrank hi patrick
pmichaud_ actually, PGE has as much Unicode support as Parrot does, which is a fair amount
i.e., <alpha> really does match more than just A..Za..z
TimToady hmm, I wonder why my pugs case didn't work then... 19:17
I'm using pge embedded.
pmichaud_ but yes, for the time being I would just hardcode the open/close brackets
aufrank pmichaud_: nice to IRC-meet you :)
pmichaud_ TimToady: well, you have to compile parrot w/ICU, methinks
TimToady I just do whatever it does by default. 19:18
pmichaud_ linux? what distro?
TimToady I still tend to think that depending on ICU is a mistake.
19:18 Shillo joined
TimToady Fedora Core 5. 19:18
Shillo Woof!@
:)
TimToady but I build from repository every day
pmichaud_ you probably need to install the libicu and libicu-devel packages (or something like that)
they aren't installed by default. Then parrot will find them on the rebuild
particle_ icu is external for a reason... waiting replacment 19:19
pmichaud_ aufrank: very nice to meet you as well; are you the one working on Signature.pg and friends?
aufrank notices that [particle_ and pmichaud_] and [_why and ingy] are just as confusable as aufrank and audreyt 19:20
yeah, I got started on that
pmichaud_ TimToady: I agree that depending on icu is likely a mistake; I think it's just waiting for someone with the tuits to replace it with something else
TimToady well, I already have those installed...
Shillo Hey, a fun question: Is it allowed to define both foo() and infix:<foo>() ? 19:21
And if yes, what does foo foo foo mean? :)
TimToady should myconfig mention icu anywhere?
pmichaud_ TimToady: checking 19:22
TimToady yes, and it means exactly what you think it should mean.
theorbtwo TimToady: Is that foo(foo(foo())), or foo() foo foo() ?
chip Debian calls the packages "libicu#" and "libicu#-dev", where # is "34" in unstable
pmichaud_
.oO( "foo foo foo" means we're short a few bars. )
theorbtwo thinks we should all go down to the pub. 19:23
Shillo :) :) :)
TimToady theorbtwo has a good point
Shillo Yep, just what theorbtwo asked. I don't see which one should have the preference.
19:23 leo joined
pmichaud_ myconfig does not appear to reflect the availability of icu 19:24
TimToady It would depend on whether you defined normal foo as () or not.
if sub foo () {...}, then foo() foo foo().
otherwise foo(foo(foo()))
I was thinking of the former when you said "both foo() and infix:<foo>()". 19:25
Shillo Uhhmmm...
Just used () to indicate that I meant the function.
TimToady but of course that'd have to be infix:<foo>($x,$y) if you're talking about the signature.
Shillo No, just about function-ness. :) 19:26
TimToady yes, I just misread.
pmichaud_ TimToady: when running "perl Configure.pl", there should be a line like Determining whether ICU is installed...................................yes.
TimToady but anything that is looking for an argument is just going to see the ordinary sub next as a term, not an infix foo.
pmichaud_ if so, then parrot gets built with ICU. At any rate, PGE lets parrot handle any unicode-ness 19:27
TimToady have to prefer shift over reduce there.
Shillo Hmm, that seems kinda dangerous. Refactor the code so that foo stops expecting the argument, and everything still magically parses. Just bombs badly.
s/parses/parses and runs/ 19:28
Of course, 'don't do that sort of overloading' applies. ;)
TimToady It says ICU is installed when I configure 19:29
theorbtwo Also, don't do that kind of refactoring, unless you're sure you've caught all the callers.
TimToady I wonder if pugs is forcing some patterns into pcre...
you'd think / <alpha> / would certainly go to pge.
pmichaud_ is pge failing to catch an <alpha>?
aufrank ?eval 'œ' ~~ / <alpha> / 19:30
evalbot_10251 Match.new( ok => bool::true, from => 0, to => 1, str => "S", sub_pos => (), sub_named => { "alpha" => Match.new( ok => bool::true, from => 0, to => 1, str => "S", sub_pos => (), sub_named => {} ) } )
Shillo theorbtwo: Well, signatures should normally catch those. This just seems like a rather pathological exception.
aufrank huh. TimToady, what was your char that didn't match?
TimToady pugs> "ƅ" ~~ / <alpha> /
Match.new( ok => bool::false, from => 0, to => 0, str => "", sub_pos => (), sub_named => {}
)
theorbtwo wonders what tat str 'S' is. 19:31
pmichaud_ what character is the '?' I'm seeing?
aufrank mine or TT's ?
pmichaud_ (unicode codepoint)
TimToady The one that looks like Anstrom
pmichaud_ I saw aufrank's and saw that it worked
19:32 elmex joined
pmichaud_ okay, lemme try on my sys 19:32
TimToady U+00C5
pmichaud_ ah, excellent. that helps
TimToady ƅ 00C5 LATIN CAPITAL LETTER A WITH RING ABOVE 19:33
theorbtwo Possibly an issue because it's latin-1? (oe isn't, apparently -- U+153)
aufrank pmichaud_: worked for some value of worked. The str value of the match object didn't contain the correct character.
or is that right? 19:34
pmichaud_ aufrank: I'll try with your char also
TimToady: 00c5 matches <alpha> on my system
TimToady also fails: ě 011B LATIN SMALL LETTER E WITH CARON
pmichaud_ [pmichaud@orange trunk]$ ./parrot x.pir 19:35
"VAR1" => PMC 'PGE::Match' => "\x{c5}" @ 0 { <alpha> => PMC 'PGE::Match' => "\x{c5}" @ 0
}
and so does U+153
I suspect a translation problem between pugs and embedded parrot
aufrank doesn't embed
TimToady hmm, could be. 19:36
aufrank but evalbot probably does?
TimToady I don't think evalbot embeds.
pmichaud_ if evalbot (or emb parrot) doesn't send the string to parrot as a unicode string, then it probably gets treated as ascii 19:37
TimToady ?eval "ě" ~~ / <alpha> /
evalbot_10251 Match.new( ok => bool::false, from => 0, to => 0, str => "", sub_pos => (), sub_named => {} )
pmichaud_ ?eval "\xc5" ~~ / <alpha> /
evalbot_10251 Match.new( ok => bool::false, from => 0, to => 0, str => "", sub_pos => (), sub_named => {} )
TimToady probably something along those lines.
given that audreyt was going to propose that Haskell treat unicode like Perl does, there could well be a weakness there. 19:39
19:39 fglock joined
pmichaud_ maybe I'll do rules support in perl6.pbc before tackling subs 19:39
or at least get some form of C<~~> to work 19:40
that would make it easier to test such things :-)
TimToady yep. and maybe you could even bootstrap a little...
so you don't have to write everything in pir... 19:41
pmichaud_ pir's not so bad, really. :-)
TimToady given a choice between rules and pir, I think I know which I'd rather write in...
pmichaud_ well, notice that I don't use pir to actually write the grammar. :-)
TimToady ?eval "ě" ~~ / . / 19:43
evalbot_10251 Match.new( ok => bool::true, from => 0, to => 1, str => "\ESC", sub_pos => (), sub_named => {} )
TimToady interesting, in my pugs it also works, but prints str => "\283"
pmichaud_ looks octal 19:44
TimToady Can't be.
pmichaud_ I guess not.
pmichaud_ spies the '8'
however
283 octal == 303 octal == c5 hex :-) 19:45
TimToady but it's the e-wedge, not the Anstrom. I guess I picked badly...
pmichaud_ okay, I guess that decides it -- I need to implement ~~ 19:46
:-)
theorbtwo Theory: It's getting utf8 encoded, and maching <alpha> anyway, because the first byte of it is a latin-1 alpha char. 19:47
TimToady I wasn't using <alpha> the last one.
I was using .
?eval "ě" ~~ / .. / 19:48
pmichaud_ oh, an even better theory. What's the unicode point for e-wedge? I bet its first byte ends up being 9b
evalbot_10251 Match.new( ok => bool::false, from => 0, to => 0, str => "", sub_pos => (), sub_named => {} )
pmichaud_ nope, can't be
hmmmm
TimToady ?eval "ěě" ~~ / .. /
evalbot_10251 Match.new( ok => bool::true, from => 0, to => 2, str => "\ESC\ESC", sub_pos => (), sub_named => {} )
theorbtwo Yeah, that doesn't really make sense for the ESC thing...
pmichaud_ parrot strips high-bits when reading as ascii
theorbtwo ?eval "ěě".length
evalbot_10251 Error: No such method in class Str: "&length"
TimToady I suspect something between non-embedded pge and pugs there... 19:49
since my embedded works.
theorbtwo ?eval "ěě".graphemes
evalbot_10251 Error: No such method in class Str: "&graphemes"
theorbtwo ?eval "ěě".chars
evalbot_10251 2
theorbtwo Thank you.
?eval "ě" ~~ /../ 19:50
evalbot_10251 Match.new( ok => bool::false, from => 0, to => 0, str => "", sub_pos => (), sub_named => {} )
TimToady but this means we can pass unicode to pge and get it matched correctly.
theorbtwo Good.
TimToady so why doesn't this work 19:51
?eval "ě" ~~ /<alpha>/
evalbot_10251 Match.new( ok => bool::false, from => 0, to => 0, str => "", sub_pos => (), sub_named => {} )
TimToady ?eval "ě" ~~ /<+alpha>/
evalbot_10251 *** Cannot parse PGE: <+alpha> *** Error: Null PMC access in invoke() Match.new( ok => bool::false, from => 0, to => 0, str => "", sub_pos => (), sub_named => {} )
pmichaud_ (sanity test)
?eval 'a' ~~ /<alpha>/
evalbot_10251 Match.new( ok => bool::true, from => 0, to => 1, str => "a", sub_pos => (), sub_named => { "alpha" => Match.new( ok => bool::true, from => 0, to => 1, str => "a", sub_pos => (), sub_named => {} ) } )
pmichaud_ (pge doesn't understand <+rule> yet) 19:52
?eval "\x41" ~~ /<alpha>/
evalbot_10251 Match.new( ok => bool::true, from => 0, to => 1, str => "A", sub_pos => (), sub_named => { "alpha" => Match.new( ok => bool::true, from => 0, to => 1, str => "A", sub_pos => (), sub_named => {} ) } )
pmichaud_ ?eval "\xc5" ~~ /<alpha>/
evalbot_10251 Match.new( ok => bool::false, from => 0, to => 0, str => "", sub_pos => (), sub_named => {} )
pmichaud_ aha
TimToady oho
pmichaud_ "\xc5" isn't unicode or utf8 19:53
TimToady ehe
pmichaud_ ?eval "\x00\xc5" ~~ /<alpha>/
evalbot_10251 Match.new( ok => bool::false, from => 0, to => 0, str => "", sub_pos => (), sub_named => {} )
TimToady ?eval "\xc5" ~~ /^ . $/
evalbot_10251 Match.new( ok => bool::false, from => 0, to => 0, str => "", sub_pos => (), sub_named => {} )
pmichaud_ ?eval "\xc5" ~~ / .. /
evalbot_10251 Match.new( ok => bool::true, from => 0, to => 2, str => "\197", sub_pos => (), sub_named => {} ) 19:54
TimToady heh
particle_ hah
pmichaud_ ?eval "\x153" ~~ / .. / 19:55
evalbot_10251 Match.new( ok => bool::true, from => 0, to => 2, str => "\339", sub_pos => (), sub_named => {} ) 19:56
pmichaud_ very bizarre
TimToady ?eval "\x153" ~~ / ^ .. $ /
evalbot_10251 Match.new( ok => bool::true, from => 0, to => 2, str => "\339", sub_pos => (), sub_named => {} )
pmichaud_ for some reason pge is getting the string as bytes or ascii instead of utf8 or unicode
i.e., parrot is reporting the string as two characters 19:57
then when it comes back, whatever translates back is treating it as unicode
TimToady the cathedral and the bizarre
the question is, which is which... 19:58
pmichaud_ well, lemme see if I can get ~~ into perl6.pbc 19:59
TimToady thanks, I'll glare at six impossible lines of Haskell after lunch. 20:02
20:09 fglock joined
svnbot6 r10252 | fglock++ | RuleInline.pl - minor updates 20:09
20:13 Limbic_Region joined 20:31 egesch joined
svnbot6 r10253 | fglock++ | PCR - initial implementation of :ratchet, using inlined code 20:45
r10254 | pmurias++ | contains the special type PMC_T which i failed to get to work 20:54
pmurias good night 21:00
svnbot6 r10255 | fglock++ | PCR - t/09-ratchet.t - added failing tests 21:01
21:10 fglock left 21:23 Keith joined 21:30 larsen joined 21:31 FurnaceBoy_ joined, aeon_ joined 21:44 Odin-LAP joined 21:52 Shillo left
cratylus exit 22:31
22:37 mako132_ joined 22:52 mako132_ joined 23:09 phpError joined 23:12 sockstat joined 23:17 Craftyone joined 23:21 OutThere is now known as SubStack 23:22 weinig_ joined 23:24 Odin-LAP joined 23:25 AtomicStack joined 23:27 Quell joined 23:35 mako132_ joined 23:36 Ymmv joined

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!