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.
spinclad as eighter fly flies fly, fly flies fly eighternal 00:17
nm: that sounds like good sympathy -- share pain, not suffering 00:26
01:03 mako132_ joined 01:06 azuroth joined
azuroth I need food! 01:08
szbalint Food for the mind, you've came to the right place. 01:10
TimToady wolverian: all object attributes are supposed to work like variables, so "temp undefine $*ARGS.irs" or some such ought to work. 01:13
wolverian TimToady, yeah. what I really wanted, I guess, was a tempified copy of the object for the duration of the method call chain. 01:44
I'm not sure if that's very sane.
01:53 Quell joined 02:03 frederico joined 02:08 neoesque joined
spinclad { temp $foo; $foo.leg.mangle } # was it all just a bad dream? was it all just copyOnWrite? what is the extent of temporization? 02:14
02:16 integral_ joined
spinclad # the spatial scope? 02:16
# as it were
02:18 DesreveR joined
spinclad By S06, this seems to depend on $foo .TEMP method, and by default, preserves as much structure as $saved = $foo; ... $foo = $saved; , which again depends on $foo. 02:36
s[$foo][$foo's] 02:37
Note: S03 refers to S04 for temporization ['Note thae C<temp> and C<let> ... See S04 for more details.'] AFAICS this should refer to S06 instead. 02:43
s/thae/that/
02:53 integral joined 02:57 Khisanth joined 03:01 weinig is now known as weinig|sleep 03:06 macroron__ joined 03:14 FurnaceBoy_ joined 03:15 trym joined
aufrank ?eval rule plain_a { [a]* }; 'aaa' ~~ / <plain_a> /; 03:19
?eval rule assert_a { <[a]>* }; 'aaa' ~~ / <assert_a> /; 03:20
?eval say "1"
aufrank misses evalbot
03:21 kanru joined 03:28 svnbot6 joined 03:29 integral_ joined, evalbot_10232 joined
audreyt ?eval rule plain_a { [a]* }; 'aaa' ~~ / <plain_a> /; 03:30
evalbot_10232 Match.new( ok => bool::true, from => 0, to => 3, str => "aaa", sub_pos => (), sub_named => { "plain_a" => Match.new( ok => bool::true, from => 0, to => 3, str => "aaa", sub_pos => (), sub_named => {} ) } )
audreyt ?eval rule assert_a { <[a]>* }; 'aaa' ~~ / <assert_a> /;
evalbot_10232 Match.new( ok => bool::true, from => 0, to => 3, str => "aaa", sub_pos => (), sub_named => { "assert_a" => Match.new( ok => bool::true, from => 0, to => 3, str => "aaa", sub_pos => (), sub_named => {} ) } )
audreyt ?eval say "1"
evalbot_10232 OUTPUT[1 ] bool::true
obra 'morning, audreyt 03:31
audreyt greetings. 03:32
obra How's things?
audreyt just woke up. thinking about lighttpd packaging and general deployment strategy 03:33
clkao . o O (runapp)
there's a very noisy bird outside since 3am 03:34
obra audreyt: does this mean you're shaving the jifty packaging spec yak?
audreyt potentially. still want ot get the declarative-action yak done first today 03:35
obra *nod* very cool
audreyt there is a deadline some 26 hours from now for all this to work on win32 :)
so I'll... use whatever way most expedient.
obra Yow. understood.
audreyt (but the basic jifty was already ported a couple days ago in my checkin) 03:36
obra Yep. we've only found one small bug
File::Path chokes when you ask it to mkdir something existing 03:37
audreyt put a -e test before make_path? 03:39
obra -e or -d? 03:40
I'd gone with -d.
Oh. and actually, mason, deep inside somwhere does a ..unless (-e _)
which breaks with our loc
audreyt oh. right. I custom-patch that by hand 03:41
obra custom patch mason?
aufrank hi audreyt! hi evalbot_10232! 03:42
audreyt no, rewrite HTML:::Mason::Compiler::ToObject::compile_to_file
obra oh.
audreyt *H::M::C::TO::ctf = sub { ... };
obra "ew"
audreyt granted, not a good idea.
obra ponders *H::M::C::TO::_ = sub { } 03:43
audreyt notes that *ANY::_ is just *::_ 03:44
i.e. the parser has a special rule
obra oh god. I didn't realize. 03:45
obra cries
Perl6 will make this all better? right? ;)
FurnaceBoy_ pats obra 03:46
audreyt right :) actually Perl5.10 already does
via lexical _
aufrank basic rules question (the one I was trying to figure out with evalbot): what's the difference between matching an asserted character class using an assertion and matching a raw character class using an assertion? 03:49
audreyt [a] is just the char "a" 03:52
aufrank I know
audreyt [a-z] is the string "a-z"
<[a-z]> is a char between a and z
aufrank right
oh 03:53
audreyt did I unask your question? :)
aufrank <[a..z]> === <[a-z]> ?
audreyt oh wait a minute 03:54
a-z now means 'a', or '-', or 'z'
a..z is correct. sorry.
I was confused by the :Perl5 example in S05.
aufrank rule plain_alpha { [alpha]* }; rule assert_alpha { <[alpha]>* }; 'abc' ~~ / <plain_alpha> / ; 'abc' ~~ / <asserted_alpha> / # what's the difference ? 03:56
I think it's something about what goes into $/, but I don't know what 03:57
audreyt the first one doesn't match. 03:58
or rather, it matches empty.
?eval 'abc' ~~ /[alpha]/
evalbot_10232 Match.new( ok => bool::false, from => 0, to => 0, str => "", sub_pos => (), sub_named => {} )
audreyt ?eval 'abc' ~~ /[alpha]*/ 03:59
evalbot_10232 Match.new( ok => bool::true, from => 0, to => 0, str => "", sub_pos => (), sub_named => {} )
audreyt square brackets is not the perl5 (?:...)
so it's like (?:alpha)
s/not/now/
aufrank oh 04:00
non capturing group
got it 04:01
04:01 kanru joined
aufrank so <[...]> is enumerated or named character class, period. 04:01
aufrank wishes things sank in faster =\
audreyt just need more gravity 04:02
aufrank I'd better start digging
04:28 eccles joined
eccles hello 04:33
04:38 kanru joined
aufrank hello 04:38
audreyt: should the signature parser work like the Moose example in S06 line 1050, where it returns a new object of class Param based on the contents of the match? 04:44
and then an object of class Signature based on the Param objects?
audreyt sec 04:45
aufrank rule positional { / <sigil> <name> / { return Param.new( :sigil<$/<sigil>>, :name:<$/<name>>) } } # naive example 04:47
ok, that was wrong in a bunch of ways
but hopefully the right parts got the point across
audreyt yo 04:49
aufrank hi hi
audreyt I think just get the parse tree working is a good 1st step
aufrank ok
audreyt the "production" part is as simple as
rule positional { <sigil> <name> : { return Param(*()) } } 04:50
aufrank ok, that makes sense
audreyt (which translates to Param.new(sigil=>$<sigil>, name=>$<name>)
)
aufrank right
audreyt which is why we brought in *() notation :) 04:51
aufrank so is it my job or the production job to know that postfix:<?> sets the optional bit to true?
audreyt the production job. and production will be part of rule 04:52
so you'll eventually need that too
see ruleFormalParam
line 380 src/Pugs/Parser.hs
that's how it'd eventually look like 04:53
(conceptually not syntactically)
aufrank this helps 04:54
thanks
audreyt np :)
aufrank do you want me to add fixes so that the requirement/optional markers are postfix?
or will that cause too much breakage>
?
audreyt they already are 04:55
see line 386
aufrank -- $foo is required --> !$foo
audreyt it's after line 385
it's canonicalized internal representation
not parser
aufrank ok
audreyt see line 391
it's just so we can apply a foldr on it 04:56
aufrank ok, I get that :)
buu Wow 05:19
Your names are impressively hard to differentiate
aufrank yeah, that has come up today
05:19 aufrank is now known as aufrank_not_audr
aufrank_not_audr any better? 05:19
=P 05:20
05:20 aufrank_not_audr is now known as aufrank 06:01 ghenry joined
spinclad <auDREYT> auFRANK: much 06:15
:] 06:16
06:16 aufrank is now known as auFRANK
auFRANK fine by me 06:16
spinclad if you like 06:17
auFRANK what's a good rule for a valid key of a hash?
spinclad a string, unless the hash has another :shape
is that what you mean? 06:18
auFRANK ok, now say that in rule-speak :P
06:18 neoesque joined
auFRANK token key { [ <alpha> | \d | _ ]+ } is too restrictive 06:18
but is 06:19
spinclad depends who's parsing it, i imagine
auFRANK token key { <.>* } too pernissive>
?
06:19 cmarcelo joined
spinclad i think it just must not be confusible with whatever closes its parsing 06:19
auFRANK good call 06:20
spinclad :key %hash<key> %hash{'{'}
the last form should allow any string
auFRANK I'm only dealing with :key right now 06:21
spinclad there, i'm not sure you can make much wider than other var names 06:22
auFRANK good news to me
spinclad :key(1) etc 06:23
:key, ... :key:key2
auFRANK there's an exception, I think
I'm pretty sure :1(2) is allowed
but $1 is not allowed as a variable name, right?
audreyt :1(2) means something else 06:24
?eval :8(123) 06:25
evalbot_10232 83
spinclad :2(1) parses 1 in binary; i'm not sure how that falls out of, or away from, adverbs/pairs generally
audreyt the :name() and :digit() forms are separate forms
spinclad (like audreyt said)
audreyt you can't use the :digit form to passin named param
auFRANK good good
can you put traits on unpacked parameters? 06:26
( [ $foo, *@bar] is rw, $baz )
spinclad ( $pair [ $head, *@tail ] is rw ) ... i imagine so 06:28
(if i got that syntax right)
auFRANK ok, here goes nothin'
spinclad but in $pair ... is rw, perhaps the $pair is what's rw
but i think your form should work 06:29
svnbot6 r10233 | aufrank++ | initial attempt at rules for parsing signatures 06:32
r10233 | aufrank++ | only attempted single dispatch signatures so far
r10233 | aufrank++ | no invocant support, limited type parameter support
r10233 | aufrank++ | almost definitely need help with unpacked params
audreyt woot! auFRANK++ 06:33
auFRANK it's very possible that I am hopelessly off the mark
spinclad auFRANK++
audreyt (and I wouldn't mind if you re/nick back to lower-case :))
06:33 auFRANK is now known as aufrank
aufrank phew, that's more like it 06:33
audreyt: I don't think I quite grok the magic of <%hash> forms 06:34
audreyt actually that looks Just Right
aufrank so if they are useful, they're not in this version
audreyt oh, there's no need to grok them for sigs
they are for e.g. <%infix>
allowing user-defined infix operator
aufrank do you think that unpacked params take types and traits? 06:35
sub chase ( Dog [$fido, *@doggies] are vicious, $cat ) {...} 06:36
spinclad i seem to recall that the unpacking form is generally another signature, so yes, types, constraints, traits, pretty much everything i think 06:38
audreyt yeah. 06:39
spinclad (Dog @pack [ $fido is nice, *@others are bad ] ) 06:40
aufrank spinclad: are you think that sub would be called as chase(@pack1, @pack2) ?
my understanding was that unpacking happens in place, so you're describing two different array paremeters as opposed to unpacking the first one 06:41
s/think/thinking/
spinclad that would be (Dog @pack, Dog [ $fido is nice, *@others are bad ] ) 06:42
aufrank oh
wow, ok, need to think about that :)
spinclad i was intending one @pack, with aliases on its head and tail
audreyt (bbiab) 06:43
spinclad sorry, must get up tomorrow ... to bed now & 06:44
aufrank gnight
thanks for the help
spinclad glad to see this getting real, though 06:45
glad to contribute
svnbot6 r10234 | aufrank++ | s/packed/unpacked/ 06:56
r10234 | aufrank++ | more notes about broken stuff
r10234 | aufrank++ | attempted to handle observations from spinclad++
r10234 | aufrank++ | (@foo [$head, *@tail]) # unpacks by aliasing
r10234 | aufrank++ | (@foo, [$head, *@tail]) # unpacks second param in place
07:16 kattana_ joined 07:24 elmex joined 07:35 iblechbot joined
svnbot6 r10235 | aufrank++ | described access to the match object for some Signature object attributes 07:41
r10235 | aufrank++ | removed explicit whitespace in rules, they're supposed to DWIM with :w
r10235 | aufrank++ | attempted to fix up type parameter handling, probably closer now
r10235 | aufrank++ | described some more rough spots
07:44 cmarcelo left
aufrank ok, that's it for me for the night 07:54
08:20 SvenDowideit2 joined 08:28 lisppaste3 joined 09:03 larsen joined 09:26 drbean_ is now known as drbean 09:51 neoesque joined 09:59 iblechbot joined 10:37 larsen_ joined
meppl guten morgen 10:54
azuroth evening 10:55
meppl hello azuroth 10:57
azuroth hi meppl, what's up?
meppl its all fine 10:58
11:03 larsen joined 11:04 kanru joined
gaal how do I tell svn to remove a file from source control, without deleting my copy of it in the working directory? 11:08
I hope it isn't mv ; svn rm ; svn ci ; mv
11:40 FurnaceBoy joined
gaal say, why are statement-ending blocks desireable? To make C<if> and friends consistent with everything else? 11:42
11:53 larsen_ joined 12:32 macroron joined 12:44 jsiracusa joined 12:48 larsen joined 12:52 Limbic_Region joined 13:16 xinming joined 13:17 larsen_ joined 13:19 weinig|sleep is now known as weinig
TimToady gaal: yes, though rather than saying they're more consistent, I look at it more like, "the user doesn't have to worry about whether this block belongs to a built-in or not." 13:45
13:50 justatheory joined
theorbtwo ...and thus you can write a grope function that really does look like grep. 13:55
14:02 Southen_ joined
nothingmuch gaal: ping 14:02
pasteling "nothingmuch" at 84.108.53.74 pasted "YAPC iternary" (6 lines, 335B) at sial.org/pbot/17083 14:03
aufrank gaal: with svk there is a -K switch on rm that means "keep local copy" 14:18
I imagine there's something like that in svn 14:19
FurnaceBoy actually, I don't think there is 14:22
furthermore, even if you rm url:// remote and then 'svn up', it blows away the local file, rather than leaving it there unmanaged 14:23
FurnaceBoy can't remember what would happen with local changes though 14:24
FurnaceBoy should test it
14:43 fishb0t joined
fishb0t 'lo all: If I have questions about parrot install - better to ask here, or #parrot on irc.perl? 14:44
aufrank I've had luck with #parrot for that 14:48
fishb0t thnx, I'll try that. 14:53
gaal <nothingmuch aufrank FurnaceBoy> >>.ack 14:54
gaal writes a gripe function 14:55
szbalint :)) 14:57
aufrank TimToady: I finally checked a first pass at Signature parsing, if you want to check it out 15:09
src/Pugs/Parser/Signature.pg
svnbot6 r10236 | aufrank++ | some fixes on accessing match objects 15:35
r10236 | aufrank++ | clarified my ideas about aliasing matches
15:44 FurnaceBoy_ joined 16:21 Ymmv joined 16:26 jsiracusa joined 16:34 integral_ is now known as integral 17:00 justatheory joined
gaal { my $foo = 42 ; my $foo = 54 } # error? warning? 17:10
17:30 chris2 joined 17:32 elmex joined 17:33 m__ joined
szbalint why, gaal? 17:41
18:07 larsen joined 18:21 larsen_ joined 18:42 penk joined 18:43 penk joined 18:44 penk joined 18:50 Khisanth joined 18:55 marmic joined 18:56 ma|2duk joined 18:58 ilogger2 joined 19:26 DaGo joined 19:39 ruoso joined, aufrank joined 19:51 aufrank left 19:56 penk joined 19:57 penk joined 20:11 aufrank joined 20:33 ludan joined
ludan hi 20:33
20:36 Quell joined 20:38 Quell joined 20:40 DaGo joined
svnbot6 r10237 | wolverian++ | typo fix 20:52
20:52 aufrank joined 21:29 kanru2 joined
aufrank should sub foo (@a [$b, @c [$d, *@e]], *@f) {...} be allowed? 21:34
can you unpack a parameter that is already being unpacked?
azuroth looks sane to me 21:41
21:44 Raftman joined
wolverian can I use stub declarations to declare the signature of a set of multis? 21:51
s,stub declarations,a stub declaration,
azuroth like `multi foo(Int $x, Int $y), (Str $x, Num $y), ($x, $y, $z);`? interesting 21:53
21:53 xerox joined
xerox audreyt: ping? 21:54
Or anybody. 21:55
wolverian azuroth, actually, I was thinking along the lines of 'multi foo (Foo, Bar --> Baz) {...} multi foo ([], $bar) { ... } multi foo (@foo, $bar) { ... }'
xerox Anybody knows how does CPAN work on the security side?
szbalint depends which side of security. 21:57
xerox szbalint: mind to come to #haskell-overflow a second? 21:58
szbalint sure, if I can help it will turn out in a sec. :) 21:59
xerox Thank you
azuroth wolverian: that's a lot more normal. I don't know whether it's specced, but I'd also like it to be
wolverian good, someone agrees with me :) 22:01
btw, I also hope we have foldl, not just foldl1 (reduce).. 22:04
azuroth I would also kinda like a compile-time { ... }, if I were going to use it like that though 22:05
like... sub die_sub { die; die; die; }; macro infix:<...!> { $?ROUTINE := &die_sub # would this work? }; CHECK { for @subs -> $s { if $s == &die_sub && { die "this sub sucks" } } } 22:10
or something... I wonder whether it's doable
aufrank audreyt told me the other day about tiebreaking sigs for multis 22:11
so you can do `sub foo (Int $x, Str $y : Str $z Set $q) {...}`
s/sub/multi sub/
wolverian is that missing a ,? 22:12
aufrank that allows declaring more than set of invocants for the multi within the same signature
yes, there should be a , between $z and Set
and actually, I think it needs to be either a multi method or a multi submethod 22:16
because plain sub doesn't have an invocant? 22:17
azuroth hmm, maybe they aren't really invocants, just look like them? 22:21
aufrank I think they are actual invocants when they are split by colons like that 22:23
but I can't find this tiebreaking syntax anywhere in the syns
the closest I see is from S06 22:24
Multimethod and multisub invocants are specified at the start of the
parameter list, with a colon terminating the list of invocants:
multi sub handle_event ($window, $event: $mode) {...} # two invocants
multi method set_name ($self, $name: $nick) {...} # two invocants
audreyt or TimToady: is the tiebreaking inv syntax for multi sigs documented somewhere in the syns? If so, where? 22:27
ludan audreyt: are you living in the netherlands? 22:29
azuroth is off to TAFE
bye all!
23:19 SamB joined 23:20 avar joined 23:23 justatheory joined 23:26 pdcawley_ joined
meppl gute nacht 23:33
23:37 putter joined
putter abs.t has constructs like is(ref abs($_), ref $_, "got") 23:37
is this valid? 23:38
why does is it not parse like is(ref(...))?
wolverian ref has a signature that takes one scalar? 23:39
dvorak what wolverian said
putter my impression is pugs considers arity, but a spec p6 parse doesn't.
I'm wondering if i'm mistaken about the latter 23:40
or whether abs.t is just non-spec
wolverian I hope p6 does.
putter non-trivial interaction with optional args. 23:41
wolverian mm.
audreyt p6 does consider 0, 1, and Inf 23:47
where 1 may be optional 23:48
and that's all it's allowed to care
i.e. as far as parser is concerned, (), ($), ($?), (*@)
are the only 4 variants
TimToady: strangely, this is implied, never documented. should we document it somewhere? S02? 23:50
putter so the parser consults the scope to determine from proto signatures whether an ident should be considered term, prefix, or... greedy something? 23:51
audreyt in PGE it's called "prelist" 23:52
for "prefix listop" 23:53
in Pugs/Hs currently it's a toplevel "apply sub" rule
outside normal precedence parsing
putter fine name. still sitting here contemplating the consequences for the parser
nod 23:54
audreyt essentially that makes all listops the same prec level 23:55
otoh, not clear what "tighter" listops means.
23:55 pdcawley_ joined
aufrank audreyt: do you know if tiebreakers are documented in a syn? 23:58
putter so when I compile a file, its parse is dependent on the arity of subs defined in my ::* of the moment? 23:59

Powered by Cro and the Raku® Programming Language.

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

Thank you!