Think twice before running "make install" for Pugs | moritz.faui2k3.org/irclog/ | pugscode.org | sial.org/pbot/perl6 | ?eval [~] <m oo se> | We do Haskell, too | > reverse (show (scanl (*) 1 [1..] !! 4)) | "Perl 6 Today" video from YAPC::Asia: xrl.us/v6op
Set by agentzh on 6 May 2007.
Aankhen`` @seen agentzh 00:02
lambdabot I saw agentzh leaving #perl6 1d 18h 1m 5s ago, and .
Aankhen`` Am I correct in thinking that $obj.meth, $obj.?meth, $obj.*meth and $obj.+meth only affect the candidate list, which {call,next}{same,with} use? The interaction between the two is confusing me. :-S 00:06
Oh. Um. Methods won't magically call their superclass methods, will they, now? Silly me. 00:08
...no, still confused. 00:09
00:18 weinigLap joined 00:19 ozo joined 00:25 stevan_ joined 00:33 guinex joined, stef_ left 00:38 buetow joined, lisppaste3 joined 00:39 stef_ joined
TimToady Aankhen``: you delegate to the next candidate in the list using nextsame() 00:46
but if called with .* or .+ you don't need to do that because it happens automatically.
Aankhen`` Oh, okay.
What happens if it's called with .* or .+ and you use nextsame, though?
TimToady presumably the .* dispatcher is aware of that. 00:47
since it's basically just doing nextsame on your behalf
Aankhen`` Okay.
00:47 weinigLap joined
TimToady and presumably the nextsame call itself is what iterates the next list item 00:48
Aankhen`` ?eval 1.WHAT 00:55
00:55 evalbot_r16308 is now known as evalbot_r16314
evalbot_r16314 ::Int 00:55
Aankhen`` ?eval 1.HOW
evalbot_r16314 ^Int
Aankhen`` ?eval 1.HOW.methods
evalbot_r16314 ["","LIST","ITEM","bless"]
Patterner ?eval 1.VALUE
evalbot_r16314 Error: No such method in class Int: &VALUE
Aankhen`` ?eval class Foo { }; Foo.new.HOW.methods 00:56
evalbot_r16314 Error: No such method in class Class: &methods
Aankhen`` Bah.
?eval class Foo { }; Foo.new.HOW
evalbot_r16314 \Class.new(("attrs" => {}), ("does" => ()), ("is" => ("Object",)), ("name" => "Foo"))
Patterner ?eval 1.methods 00:57
evalbot_r16314 Error: No such method in class Int: &methods
TimToady we're all waiting for 6.28
00:59 ofer joined 01:11 lampus joined
Aankhen`` ?eval 1... 01:15
evalbot_r16314 pugs: out of memory (requested 1048576 bytes)
Aankhen`` There's this line in porting_howto: (?=foo) -> <?before foo>
The ? there is just to keep the <before> from being in the result object? 01:16
TimToady correct
Aankhen`` Kewl.
TimToady and lazy Range objects aren't really implemented yet either
Aankhen`` Yeh, I figured. 01:17
I just like to check up every now and then. :-P
01:19 Southen joined
Aankhen`` Huh, this doesn't seem right (again from porting_howto): \G -> <( .pos == $oldpos )> 01:20
Should that be <{ ... }> instead?
IIRC from S05, <( ... )> means that the result object only contains whatever's inside the parentheses.. 01:21
s/.$//
?eval say "\X[263a]" 01:22
evalbot_r16314 OUTPUT[X[263a]␤] Bool::True
Aankhen`` Poop.
TimToady should be <?{...}> or it uses the result as a subrule 01:23
Aankhen`` Ah, right, my bad.
Aankhen`` fixes.
svnbot6 r16315 | Aankhen++ | * docs/other/porting_howto: fixed \G translation 01:25
Aankhen`` Hmm, I wish I knew what I was thinking when I wrote this: /$1/ -> my $old1 = $1; /$old1/ 01:27
Or at least that I remembered writing it.
01:27 edenc_ joined
Aankhen`` Wouldn't work anyway, come to think of it. 01:27
01:31 japhb joined
svnbot6 r16316 | Aankhen++ | * docs/other/porting_howto: fixed (hopefully) `/$1/ -> ...` line. 01:31
01:34 weinigLap joined 01:47 ggoebel joined 02:10 torz joined 02:14 nekokak_ joined, mj41_ joined 02:15 mj41_ is now known as mj41, lampus joined, stef_ joined, spinclad joined, BooK joined, diakopter joined, PerlJam joined, wolverian joined, mr_ank joined, svnbot6 joined, zamolxes joined, fridim joined, lumi joined, Casan joined, ingy joined, TimToady joined, nekokak joined, pjcj joined, statico joined, wilx joined, clkao joined 02:22 zperl joined
zperl ?eval say 'Good morning, Pugs' 02:26
02:26 evalbot_r16314 is now known as evalbot_r16316
evalbot_r16316 OUTPUT[Good morning, Pugs␤] Bool::True 02:26
02:26 SubStack joined 02:28 weinigLap_ joined 02:29 weinigLap joined 02:36 mako132_ joined 02:37 dolmans joined 02:39 nipotaway is now known as nipotan
avar mm, yummy p6 sux thread on perlmonks:) 02:55
perlmonks.org/?node_id=614624 # If anyone knows the answer to the latest Q .. :) 02:57
lambdabot Title: What's wrong with Perl 6?
03:06 foof joined
PerlJam avar: That thread started days ago. 03:10
03:10 bonesss joined
avar ya, I know 03:10
PerlJam avar: Are you *sure* there are no Perl 6 modules on CPAN? 03:12
Aankhen`` Perl6::Junction? Perl6::Gather? Perl6::Perldoc?
(Those are the ones I remember off-hand.)
PerlJam But you've hit on an important point: CPAN needs to grow to accomodate Perl 6 and it hasn't really.
Aankhen`` Or did you mean modules written in Perl 6? 03:13
foof Perl6::Form
avar those are perl 5 modules that emulate p6
foof ... speaking of which, how do you format numbers with commas but no decimal points?
Aankhen`` So you did mean modules written in Perl 6.
avar yup, I thought that was apperenty, apperently not:)
Aankhen`` Well, for one, search.cpan.org would have to figure out how to distinguish between Perl 5's POD and Perl 6's Pod. 03:14
Er.
avar you could write it in p5 pod
Aankhen`` Sorry, s/search\.cpan\.org/CPAN as a whole/.
avar or you could just write it in p6 pod and have it look sucky
PerlJam avar: your objections are much better than w-ber's though. l6::Form
Aankhen`` :-)
PerlJam oops. stray mouse click.
perlmonks.org/?node_id=615819 03:15
lambdabot Title: Re: What's wrong with Perl 6?
Aankhen`` Forcing everyone to document Perl 6 modules using Perl 5 POD is a really bad idea going forwards.
avar true, but that's just search.cpan.org lameness, I'm more interested in adding an index of p6 modules and having something to install them 03:17
Aankhen`` Right.
I was just throwing up roadblocks. :-)
avar oh noes!:) 03:18
Aankhen`` Heh, one of the messages in that thread cites an article from 2003.
03:20 amnesiac joined
foof I was thinking a format like {],]]]} where the brackets to the right of the , are right brackets would indicate the , as a comma separator rather than a decimal separator, but then you couldn't right justify after a decimal separator. 03:21
Alternately, {],]]],]]]} could only mean , is a comma separator, but then you can't have a smaller width. 03:22
03:31 drbean joined
foof Actually, am I mistaken or is there no way to format numbers without putting them in a fixed-width field? 03:35
03:51 wolverian joined, diakopter joined, Casan joined, clkao joined, statico joined, ingy joined 03:52 lumi joined
Aankhen`` avar: One problem with Perl 6 modules on CPAN is… how do you differentiate Perl 5 modules from their Perl 6 ports 03:53
?
03:53 wilx joined
avar with magic 03:53
Aankhen`` Ah, glad we cleared that up. I feel a lot better now. 03:54
avar I don't know, how is this specced in the synopsis?
in any case nothing will break if they're not indexed by the normal index files
Aankhen`` I don't seem to recall reading any specs about this. :-S
03:54 stef_ joined
Aankhen`` Oh, sure. 03:54
03:56 lampus joined, spinclad joined, BooK joined, PerlJam joined, mr_ank joined, svnbot6 joined, zamolxes joined, fridim joined, TimToady joined, pjcj joined, BooK_ joined 03:57 PerlPilot joined 03:59 zamolxes_ joined, TimToady joined, pjcj_ joined, lampus_ joined 04:00 spinclad_ joined 04:02 mr_ank_ joined 04:07 nperez joined
foof "(define\n (fold kons knil ls)\n (define (loop ls acc) (if (null? ls) acc (loop (cdr ls) (kons (car ls) acc)))\n\n (if (null? ls) acc (loop (cdr ls) (kons (car ls) acc))))\n (loop ls knil))" 04:22
oops, wrong window 04:23
amnesiac what the...
foof sorry, my screen(1) installation is broken and I'm struggling with a bunch of scattered xterms 04:25
amnesiac what is kons and knil? user defined cons and nil symbols?
foof yeah, basically - it's a generic foldl operation 04:26
if you pass it cons and nil it's effectively a list-copy 04:27
... I mean reverse
amnesiac I see 04:29
04:48 zamolxes_ is now known as zamolxes 05:00 stef_ joined 05:05 ddwagnz joined
Aankhen`` (?(c)t|f) [ c :: t | f ] # is this accurate? I interpret the pattern on the right to mean c, with no backtracking allowed, followed by either t or f. :-S 05:05
05:05 BooK joined
Aankhen`` s/no backtracking allowed/backtracking causing the group to fail/ 05:06
I would have thought it's more like [ <{ (c) ?? t !! f }> ] 05:08
Speaking of which, from S05: "A leading { indicates code that produces a regex to be interpolated into the pattern at that point as a subrule". So String objects are automatically turned into Regex objects, with Regex objects being treated as subrules? 05:10
svnbot6 r16317 | Darren_Duncan++ | ext/QDRDBMS/ : updated some terminology in Language.pod, and some parameters in Operators.pm 05:27
05:32 devogon joined, SubStack joined 06:00 hirschnase joined 06:06 weinigLap joined
TimToady Aankhen``: yes, <{ $x }> will compile $x into Regex if it isn't already one, hopefully caching the Regex so it doesn't have to recompile if the string doesn't change. 06:08
though it occurs to me that <$( $x )> would do the same 06:09
Aankhen`` Neat. 06:11
TimToady on [ c :: t | f ] the :: binds tighter than the |
should probably be || nowadays though
Aankhen`` Okay.
TimToady if c fails it goes to f, if t fails it doesn't
Aankhen`` Ohh. 06:12
Hmm.
Ah.
TimToady so it really is if/then/else in regexland
Aankhen`` Makes sense.
TimToady and stacks vertically nicely, though that's hard to show on irc
Aankhen`` Should have known better than to doubt such an elegant construct. ^_^
TimToady was sort of an aha moment while writing A5
Aankhen`` So basically, the only change it needs is s/\|/||/ to make sure the sequence is preserved, aye? 06:13
TimToady yes, though arguably use of | just means the c's are tested in longest token order 06:14
06:14 rfordinal joined
TimToady since :: explicitly terminates the longest token match 06:14
in addition to controlling the backtrack
dduncan question: did there seem to be any merit to my suggestion of fundamentally using $ for all variables and changing @% etc to a subset of former uses? ... or did it sound without merit? ... or did I just miscommunicate the idea? 06:16
there were no replies, so I'm not sure if people didn't understand what I said, or if it was simply dismissed as not wasting time to reply to 06:17
svnbot6 r16318 | Aankhen++ | docs/other/porting_howto:
r16318 | Aankhen++ | * slight update to (?(c)t|f) translation.
dduncan (it was on p6l)
TimToady it seems to me to be akin to the original Highlander Variable proposal, and therefore suffer the same flaws mentioned in A2. 06:19
dduncan I'm not sure what you refer to exactly, but I don't think what I was proposing is so extreme. 06:21
it was just the @ and % sigils I suggested changing to @$ and %$, but any other sigils can stay the same 06:22
TimToady that would not be at all acceptable to the typical Perl 5 programmer.
06:23 Averell joined
TimToady and you have to explain why &$ is exempt 06:23
06:23 jisom joined
dduncan a question then, based on the way things currently are ... 06:23
TimToady I'd just rather keep those namespaces separate
dduncan I agree with keeping routines separate
now, ...
what sorts of sigils should each of these have: Set, Mapping, Capture, Signature, Pair? 06:24
I know you can use : or | etc in declaring ...
but afaik those aren't kept as sigils in variables of those types
TimToady none of those occur frequently enough to deserve anything beyond $ 06:25
dduncan Seq likewise?
so are you saying that Array and Hash get their own for huffmanizing?
TimToady that's more like an Array
dduncan they are used a lot
TimToady mostly anonymously
dduncan so would a Seq variable have a @ or $ ? 06:26
TimToady Seq is immutable, so it's not a variable... :P 06:27
dduncan let me rephrase that ...
TimToady Seq is not used much at all in the t/ director
y
dduncan would a variable defined to just hold Seq values have a @ or $ ?
TimToady you can bind a Seq to either $ or %
@ I mean 06:28
dduncan so then, Seq is the same as Array, in that it can bind to either $ or @ ?
TimToady basically
dduncan would then Mapping bind to $ or % as a Hash does?
TimToady presumably 06:29
and Code can bind to $ or &
dduncan how about Set and Bag etc ... I would imagine either % or $ as well?
would a Pair bind to a % , or just a $ ? 06:30
06:30 baest_ is now known as baest
TimToady Indeed, the synopses discuss modelling them on hashes with automatic values 06:30
I imagine you could bind a Pair to be a tiny hash
dduncan it is with the mutable versions, so I suppose the immutable versions could have the same sigils
okay ... 06:31
TimToady yeah, we don't really distinguish mutable vs immutable variables syntactically
they just complain if you try to vary them...
dduncan part of the reason I raised the whole issue is that it seemed not so good that just Array and Hash have those extra sigils just for them, and other container types didn't ... but from what I'm reading above, it doesn't sound like my concerns in this regard are valid ... the other collection types can bind to @ or % too 06:32
TimToady and presumably the compiler can sometimes intuit more than that
or both
dduncan still, that takes care of a large part of my concern, so thanks for clearing this up
TimToady sure
dduncan do the Synopsis cover all the details we just discussed, or do you mind updating them so? 06:33
TimToady array and hash sigils are a semantic huffmanization explored by earlier Perl with pretty good results
the * sigil was a mistake though...
dduncan sure
TimToady uh, I think you could derive everything I said, but it's perhaps not always stated in so many words 06:34
dduncan on a separate matter, I'm happy to see the returns->as rename occurred in the Synopsis as we discussed last week
sure, but sometimes it helps to have some things explicitly stated
TimToady up to a point, after which you're just repeating yourself... :) 06:35
dduncan eg, explicitly state what are all the built-in types that could typically bind to each of the sigils
so eg if you said %foo in a signature, you can have an idea what range of built-in types you may get
(and $ is like a maximal type sigil in that *anything* can bind to it, afaik) 06:36
TimToady I'll think about where best to put that--the original camel structure didn't really have a chapter for types...
dduncan sure 06:37
but afaik, the synopsis have somewhat evolved away from the Camel book already
TimToady so probably somewhere in S02
dduncan sure
on a different matter, possibly previously discussed ...
given that afaik it is best practice when declaring parameter types or testing argument types that .does() is used, so that valid arguments for the parameter are ones that .does() the declared parameter type ... 06:39
and given that we often have mutable types saying they do immutable types, eg Array does Seq ...
I get the impression that if I declare a parameter of type Seq, I could validly be handed an array? 06:40
06:40 Psyche^ joined
TimToady I suppose so. 'Course, all parameters are considered readonly by default anyway... 06:40
dduncan but the reason I asked for a Seq is that I want the value I am given to be immutable, so eg if I then assign it to one of my private attributes, then subsequent modification of a passed array argument won't affect my attribute 06:41
afaik, the read-only thing just prevents me inside my routine from changing it, but doesn't prevent the above scenario
or that is, read-only prevents me from assigning to the parameter 06:42
TimToady yeah, maybe we need an "is snap" or some such for a readonly snapshot
or COW or something
dduncan and so then if the original is immutable, then the process doesn't make a copy, and if the argument is mutable, then it does make a snapshot
TimToady have the same problem if we allow people to use a class as a role
we have to take a snapshot of the current state of the class and make an immutable role from it 06:43
if objects have some kind of versioning builtin, then it could just refer to the particular state in time
dduncan still, if we can get that kind of protection, then it would be very useful to me, as I won't have to clone the argument manually in case it was mutable, to protect my internals
TimToady might work with STM somehow 06:44
dduncan perhaps, though I thought STM was more for short-term atomicity
still, thanks for anything you can do to design a fix for this matter, as I see it as a very real issue that would affect many peole 06:45
TimToady well, it's about proving lack of contradiction in some update order, and it just feels like "not updated" is one variety of assertion about the update.
dduncan is it worth my bringing up this matter on p6l, or is my just telling you now enough?
TimToady but maybe it's more of a COW issue underneath, so you only have to take a snapshot if someone has already claimed the mutability of the object. 06:46
dduncan copy-on-write sounds reasonable ...
for that matter, while I mentioned Arrays, I see this issue potentially affecting every data type 06:47
TimToady might be worth discussing on p6l; I'm likely to get distracted, and it's partly the implementors that have to figure out what it really means underneath
dduncan for example, if we have a random object that does Str or Int or Bool, they could be mutable too
TimToady sure, might even be worth some very short sugar
dduncan and for those it may not even be possible to work around outside the language, short of eg $copy = "$original" which forces a new Str to be made 06:48
and that is inefficient
okay, I'll write up a quick thing about this on p6l now ... 06:49
TimToady maybe all objects have a .snap method to go with .bless and .clone
dduncan I may also just be lazy and copy this thread into it as the meat of the post, so they know what each of us said
TimToady would be okay too
or just link to log 06:50
dduncan but .snap would be called lazily, citing COW
TimToady well, it's a noop on immutable types
dduncan I'll do both, assuming this thread is short, unless we go on for awhile
yes
TimToady would only need COW on mutables
dduncan yes
TimToady well, I need to go to bed, my eyes are crossing
dduncan sure ... and it's midnight here 06:51
thanks for the chat
TimToady your welcome, night
zzz &
*you're
06:57 Psyche^ is now known as Patterner 07:05 drupek12 joined 07:15 franck__ joined
dduncan and the p6l post is done 08:04
and I'm gone 08:05
dduncan zzz
Tene dduncan: seeya
08:05 iblechbot joined 08:21 bernhard joined 08:31 japhb joined 08:43 Jedai joined 08:50 pipping joined 08:55 andara joined 09:13 riffraff joined 09:20 ludan joined
riffraff hi 09:21
09:27 nwc10 left
moritz hi riffraff ;) 09:29
09:33 dduncan left 09:38 buetow joined, polettix joined 09:48 ruoso joined 09:53 the_dormant joined 09:55 IllvilJa joined 10:13 hirschnase joined, lde joined 10:23 sapir joined
sapir is there a page somewhere that describes how far away Perl 6 is from a full working implementation? 10:25
10:27 chris2 joined
xinming svn.pugscode.org/pugs/docs/01Overview.html 10:38
lambdabot Title: Pugs Apocryphon 1
xinming spinclad_: svn.pugscode.org/pugs/docs/01Overview.html 10:39
lambdabot Title: Pugs Apocryphon 1
xinming spinclad_: sorry, wrong completion.
sapir: the url is for you.
sapir thanks
10:40 spinclad_ is now known as spinclad
sapir does that mean Pugs doesn't have classes or rules yet? 10:42
mj41 h, Mono:DLR www.oreillynet.com/xml/blog/2007/05...guage.html 10:44
lambdabot Title: [Mono:DLR] Hello, Dynamic Language Runtime-enabled World! - O&#039;Reilly XML Bl ..., tinyurl.com/yqexqp
10:59 edenc joined 11:06 mjk joined, the_dormant joined 11:14 rindolf joined 11:32 kanru joined 11:42 nipotan is now known as nipotaway 11:44 rho joined 11:47 jisom_ joined 11:51 lichtkind joined
lichtkind ?eval gather for (1,1,1,2,3) { state $previous = take $_; next if $_ === $previous; $previous = take $_ } 11:52
11:52 evalbot_r16316 is now known as evalbot_r16318
evalbot_r16318 (1, 1, 1, 1, 2, 3) 11:52
lichtkind hahahaha
like i thought the example in the syn 6is bad
see under dev.perl.org/perl6/doc/design/syn/S..._statement 11:53
lambdabot Title: Synopsis 4: Blocks and Statements - perl6:, tinyurl.com/o9tb7
11:56 pipping joined 11:59 DarkWolf84 joined 12:00 cernd joined 12:20 the_dormant joined
svnbot6 r16319 | rhr++ | [examples/unitsdat-grammar.pl] Handle nested parens correctly (maybe). 12:22
r16319 | rhr++ | Recursive grammars make my head hurt...
12:23 jerrylee joined 12:30 turrepurre joined 12:33 baest joined 12:53 carl_ joined, carl_ left 13:00 DarkWolf84 joined 13:02 lampus_ is now known as lampus
svnbot6 r16320 | rhr++ | [examples/unitsdat-grammar.pl] Big refactoring, it actually makes sense now. 13:14
13:17 ludan joined
lichtkind ?eval gather { take 1, 2; take 5 } 13:17
13:17 evalbot_r16318 is now known as evalbot_r16319
evalbot_r16319 (1, 2, 5) 13:17
lichtkind ?eval gather { take [1,1] ; take 5 }
evalbot_r16319 ([1, 1], 5)
lichtkind thanks 13:18
13:22 gnuvince left 13:23 the_dormant_ joined 13:27 [particle] joined
lichtkind ?eval do { 'as' } 13:28
evalbot_r16319 "as"
lichtkind ?eval gather { 'as' }
evalbot_r16319 ()
lichtkind ?eval gather { take 'cdf'; 'as' }
evalbot_r16319 ("cdf",)
lichtkind ?eval @maler = gather { 13:30
take 'Caspar David Friedrich';
take 'Paul Cezanne', 'Claude Monet';
'Alfred Sisley';
evalbot_r16319 Error: ␤Unexpected end of input␤expecting "use", statements or "}"
lichtkind }
?eval @maler = gather {take 'Caspar David Friedrich';take 'Paul Cezanne', 'Claude Monet';'Alfred Sisley';} 13:31
evalbot_r16319 ["Caspar David Friedrich", "Paul Cezanne", "Claude Monet"]
13:33 PerlPilot is now known as PerlJam
svnbot6 r16321 | rhr++ | [examples/unitsdat-grammar.pl] Add support for prefixes and plurals in definitions. 13:33
lichtkind ?eval do { leave 2 } 13:40
evalbot_r16319 (2,)
lichtkind ?eval do { leave 2; return 1 }
evalbot_r16319 (2,)
lichtkind ?eval do { return 1 }
evalbot_r16319 Error: Cannot cast from VUndef to Pugs.AST.Internals.VCode (VCode)
lichtkind ?eval sub r { return 1 } r(); 13:41
evalbot_r16319 1
lichtkind ?eval sub r { leave 2; return 1 } r();
evalbot_r16319 (no output)
svnbot6 r16322 | rhr++ | [examples/unitsdat-grammar.pl] Was allowing prefixes in the wrong place. 13:44
13:54 stevan_ joined 13:56 pack|pizza joined 14:06 dvorak joined 14:09 hatsuseno joined 14:20 fridim joined 14:31 pipping left 14:40 drupek12 joined 14:53 elmex joined 15:02 goban joined
goban when writing perl should i actually use $_ and such? seems this could conflict with other things (say its in a subroutine) 15:02
diakopter ?eval use $_ and such 15:03
15:03 evalbot_r16319 is now known as evalbot_r16322
evalbot_r16322 Error: ␤Unexpected "$_"␤expecting "=", use version or use package 15:03
15:03 dunstabulos joined
TimToady goban: it is true that $_ can only mean one thing at a time 15:04
15:04 amnesiac joined
TimToady so I wouldn't use it for a scratch variable unless you really want it to be the current topic 15:05
goban dont all those global variable side effects cause confusion?
(new to perl) 15:06
TimToady $_ isn't a global in p6, it's a lexical
so topics are well scoped
goban ah cool, the job im applying for is using perl5 still though, but p6 sounds really cool 15:07
TimToady yes, we're fixing a lot of those kinds of problems that p5 *does* have
but p5 is still an okay language 15:08
hopefully p6 will get the same result while requiring less self-discipline
we've worked hard at reattaching various previously global concepts to the correct scope or object in p6 15:09
so nearly all the global special variables are gone
goban will normal perl 6 let me use p5, haskell, and java modules like pugs? 15:10
TimToady if the underlying engine does; the intent is to run p6 on any of a number of engines
certainly the parrot engine is aiming to run multiple engines
s/engines/languages
though I doubt they're thinking of Haskell at the moment... 15:11
the current Haskell engine gives access to p5
15:11 rindolf joined
TimToady but nobody's working on the Java connection at the moment 15:11
once we have the front end done, it shouldn't be too difficult to retarget the back end to any engine 15:12
dunstabulos how is p6 on parrot versus p6 on pugs?
TimToady including the JVM
goban dunstabulos, i think pugs is farthur along at the moment
TimToady it's difficult to compare, because parrot's p6 is happening bottom up
while pugs's is happening top down 15:13
we're hoping they'll meet in the middle and help each other
goban parot is in p5?
TimToady right now both implementations are trying to converge on the standard grammar in pugs/src/perl6/Perl-6.0.0-STD.pm
dunstabulos is looking forward to playing with the grammar stuff 15:14
TimToady usually just referred to here as "STD"
parrot is written in C
there is a p5 implementation called "v6" by fglock
[particle] hurls parrotcode.org
lambdabot Title: Parrot Virtual Machine - parrotcode
goban maybe ill start a java implementation just for fun :) 15:15
TimToady would you like a pugs commit bit? you can basically put anything under that repository
the parrot folks are a little more disciplined :) 15:16
jerrylee so far I cannot compile code on pugs 15:17
can pugs be compiled? 15:18
i mean perl code on pugs
TimToady well, that's kinda what's happening with the prelude, but you have to write in a subset of Perl 6 for that currently
can't use "return" for instance
eventually the AST will have emitters for any back end you like 15:19
fglock gets a twinkle in his eye when he thinks about a C backend
but that would be hard, since you'd have to invent continuations and GC and everything parrot and haskell already provide 15:20
as it is, targeting the JVM will be difficult if you want efficient continuations 15:21
jerrylee hmm 15:22
TimToady for the most part we've tried to define p6 such that explicit continuations aren't necessary
but they would certainly make it easier to write a regex engine that backtracks into subrules correctly 15:23
15:24 rashakil joined 15:27 justatheory joined 15:29 andara left
jerrylee when will pugs become p6? 15:36
[particle] never--there is no single official perl 6 implementation 15:38
perl 6 has an official spec, and an official test suite 15:39
goban yeah thats one thing i hate about java (no continuations)
[particle] once any implementation passes the entire test suite, it's perl 6
jerrylee wow
so pugs could be p6 someday 15:40
[particle] it could be a p6, yes
jerrylee target date?
TimToady might even get there first, but parrot is likely to be faster
[particle] christmas
jerrylee really? this year?
TimToady year is unspecified
[particle] who said anything about year? ;)
TimToady after p6 is out, every day will seem like christmas... :) 15:41
jerrylee ha
goban maybe that should be in topic :) seems a common q
jerrylee really it is 15:42
looks pugs and parrot are both going to be perl6 15:43
TimToady the more the merrier 15:44
jerrylee :) 15:45
TimToady once the dam bursts, you're going to see a flooding algorithm at it's best 15:47
jerrylee so compiled on pugs can also integrate with other languages? like in parrot?
15:47 the_dormant joined
TimToady if the compiler is in Perl 6, the compiler is platform agnostic 15:48
only the emitters have to know what platform they're targeting
so you could run the compiler on parrot and emit Haskell, maybe. :) 15:49
or vice versa...
jerrylee it's really like a dream, man 15:50
TimToady we'd like to wake up someday and have it be real though... 15:51
jerrylee things are happening 15:53
TimToady well, all of us here are very stupidly persistent, so there's hope 15:54
dunstabulos is the perl6 test suite believed to be complete?
(believed because you never know if a test suite has complete coverage) 15:55
TimToady it's incomplete
I suspect we really want to have about twice as many tests as we have currently 15:56
15:57 the_dormant_ joined
[particle] maybe even more tests... but we need a complete spec before the test suite is complete 15:59
dunstabulos: have you seen the specs with tests linked to them? 16:00
spec.pugscode.org/
lambdabot Title: Official Perl 6 Documentation
jerrylee is really sleepy now 16:01
good night guys
see you
[particle] for example: perlcabal.org/syn/S03.html#Operator_precedence 16:03
lambdabot Title: S03
[particle] you can show or hide the tests related to that section of the spec
dunstabulos [particle]: nifty 16:04
[particle] it's a great tool to investigate spec coverage
16:04 zarchne joined 16:33 weinigLap joined 16:34 the_dormant joined 16:37 kanru joined
lichtkind hallo TimToady the snippet in dev.perl.org/perl6/doc/design/syn/S..._statement seems inkorrekt 16:41
lambdabot Title: Synopsis 4: Blocks and Statements - perl6:, tinyurl.com/o9tb7
lichtkind i think your intend was to filder unique values 16:42
PerlJam lichtkind: what do you think is incorrect about it? 16:44
TimToady just looks buggy to me 16:45
lichtkind PerlJam: it does nothing
i tested it one hour ago 16:46
TimToady with a buggy implementation :)
PerlJam What is === again?
That's not identity equality is it?
lichtkind yes
TimToady: but the "buggy implementation" behaved liek expected 16:47
PerlJam perhaps === should be == then
TimToady the bug may be that pugs doesn't always consider 1 === 1
lichtkind so maybe i misunderstood something
TimToady value types should always have an identity of their value, not their memory address 16:48
[particle] 1 is a singleton
rhr ?eval my @uniq = gather for (1, 1, 2, 2, 3, 3) { state $previous = take $_; next if $_ === $previous; say $previous; $previous = take $_; }
evalbot_r16322 OUTPUT[␤␤␤␤␤␤] [1, 1, 1, 2, 2, 3, 3]
[particle] or, more appropriately, highlander :)
rhr state doesn't work
TimToady take doesn't return its value yet
rhr or that :) 16:49
TimToady ?eval gather { say take 42 }
evalbot_r16322 OUTPUT[␤] (42,)
TimToady yes, take is not up to spec yet
so, buggy implementation, not buggy spec. :) 16:50
feel free to fix it. :D 16:51
16:51 integral joined, the_dormant_ joined
lichtkind TimToady: so take always takes only one value when running through a block and overwrites the old if i take another? 16:51
TimToady huh? 16:52
take pushes its arguments onto the gather's list, always
it also happens to return the argements, according to spec
s/e/u
are you confused by the fact that the pseudoassign to state declaration happens only once? 16:54
the second take is never reached on the first time through the loop (assuming take worked right) 16:56
lichtkind no that one i understood something else confuses me 16:57
so far i understand this implementation takes every value anyway so no filtering here 16:58
there should be no take in the first line of the block i think 16:59
i try to rewrite it like i think it should behave 17:00
TimToady the first take is guarded by being in a pseudo assign, so it only happens first time through 17:01
lichtkind ah it gets warmer 17:02
thanks now i understand
17:04 chris2 joined 17:05 franck__ joined
lichtkind TimToady+ 17:06
TimToady++
17:07 Psyche^ joined, edenc joined
rindolf Hi lichtkind 17:14
lichtkind: what's up?
lichtkind rindolf: perl6wise or general?
hi
rindolf? hm perl6wise im currently writing part 2 of an huge article which was very sucessfull amongst my friends all liked it even they know nearly nothing about perl :) 17:16
rindolf: but im currently dont feeling that terribly well
rindolf lichtkind: "I'm" 17:18
lichtkind: not "im".
perlbot: im
perlbot im means "instant message". If you mean "I'm" or "I am", please say that
rindolf lichtkind: are you sick?
17:19 polettix left
lichtkind rindolf do you prefer the gnostic answer? :) the normal would be slightly in between 17:19
rindolf lichtkind: what's "gnostic"? 17:20
lichtkind: agnostic?
17:20 rodi joined
lichtkind rindolf: gnostic is an old knoledgy tradition it actually means knowledgy agnotic means not to know (god) 17:21
rindolf lichtkind: knowledgy?
lichtkind rindolf: gnostics say that all illnises are illusions
knowledge 17:22
rindolf lichtkind: can you please use some "." or ","?
lichtkind k
rindolf lichtkind: thanks.
lichtkind: I don't believe all illnesses are illusions.
17:23 Psyche^ is now known as Patterner
lichtkind rindolf: i mostly agree to gnostic standpoints, even if they sound sometimes radical and can easily misunderstood 17:23
rindolf: but thats an long story. you are the australian wiki maintainer? yes i dindt do there anything since month 17:24
rindolf lichtkind: I'm not *the* perl.net.au maintainer. 17:26
lichtkind: but I'm editing content there.
lichtkind rindolf: yeah but you seemed to be one of the australian community core member 17:27
rindolf lichtkind: I'm not Australian.
lichtkind: I'm Israeli. 17:28
lichtkind: but I have some Aussie Perler friends.
lichtkind ah thatswhy my misinterpretation
rindolf lichtkind: but I found myself contributing to this wiki.
lichtkind: and also to perlmeme.org
lichtkind: perlmeme.org/ 17:29
lambdabot Title: Welcome to perlmeme.org
lichtkind rindolf: in last weeks i mostly wrote for wikipedia , damians article and some core improvements and now im on the german perl6 tutorial
rindolf lichtkind: the German wikipedia?
lichtkind rindolf: cool i recognized perlmeme since it started i made the link in our linklist there 17:30
rindolf: of course
rindolf lichtkind: oh.
lichtkind: yes, your English is probably not good enough to edit en.wikipedia.org
lichtkind rindolf: most time of the day it is :), im just sleepy now 17:31
rindolf lichtkind: oh OK.
lichtkind rindolf: i even started effort to synchronize between en and de perl main article. even juerd said that my english improved :) 17:32
rindolf lichtkind: oh OK.
lichtkind: why are you sleepy now? What time is it where you're at?
lichtkind rindolf: its 19:37 but it feels like 23:00 to me :) 17:33
rindolf lichtkind: oh. Why?
lichtkind maybe because last night i played till the next dawn
rindolf: if your israeli so you have to be in the pugs team right? :) 17:35
17:36 the_dormant joined, toshinori joined
rindolf lichtkind: I'm not really in the pugs team. 17:40
lichtkind: I know gaal and nothingmuch.
lichtkind: I just like to hang out in this channel.
lichtkind rindolf: yes its very friendly here :) 17:41
offby1 not enough free beer, in my opinion. 17:42
rindolf gives a keg of beer to offby1 17:44
offby1 that's more like it! Thanks.
rindolf offby1: it's online beer!
TimToady eep, it's turning into the chatterbox...
offby1 I notice it's lager, too, and it's warm ... :-| 17:45
rindolf TimToady: what's up?
TimToady trying to figure out how the summer will work
[particle] same way it always does here... days will get shorter, temps will get warmer 17:46
offby1 yup 17:47
I was going to make a similarly flip comment.
17:51 rlb3_work left 17:57 SCalimlim joined 18:10 weinigLap_ joined 18:11 the_dormant_ joined 18:13 justatheory joined 18:16 jisom joined, pack|pizza left 18:24 SubStack joined 18:51 GeJ joined
avar @tell fglock search.cpan.org/~avar/re-engine-PCR..._01/PCR.pm 18:53
lambdabot Consider it noted.
18:59 fglock joined 19:06 prism joined
fglock avar: looking 19:07
lambdabot fglock: You have 1 new message. '/msg lambdabot @messages' to read it.
jrockway @messages 19:08
lambdabot You don't have any new messages.
avar fglock: b.t.w. that segfault is my fault:)
19:09 bonesss joined
fglock avar++ 19:09
avar fglock: I also made a release of ::Plugin that works with blead currently: search.cpan.org/~avar/re-engine-Plu...Plugin.pod 19:13
lambdabot Title: re::engine::Plugin - API to write custom regex engines - search.cpan.org
avar lotsa docs and testcases too
You might be interested:)
fglock hmm - my cpan mirror doesn't know about it yet 19:15
avar noes ohes 19:16
sial.org/pbot/24942 19:17
lambdabot Title: Paste #24942 from Someone at 85.197.228.236
fglock avar: I'm now reading about non-perl backends, in order to plan a more portable 6-in-6
Perl 6 in Perl 5 can cheat... 19:18
avar cool
Have you looked at sbcl?
or other CL backends..
fglock no - looking now; I'm reading about Groovy
avar Okey, I won't bother you then:) 19:19
fglock avar: no, please - what's special about sbcl? 19:20
avar It's a common lisp runtime that compiles to machine code 19:21
19:21 svnbot7 joined
avar In general emitting common lisp would give access to around 20-30 VM's and compilers, including a gcc frontend:) 19:21
fglock avar: ok, it is now in my list :) 19:23
avar Learning CL is on my todo after the current project of hacking perl core / re::eng stuff:) 19:24
sial.org/pbot/24943 # urgh 19:25
lambdabot Title: Paste #24943 from Someone at 85.197.228.236
fglock I've done a lot of lisp ... but that was well before I learned Perl 4 19:27
re re::engine::PCR - is it ok if the PCR runtime needs plain p5 regex? 19:29
for character classes, for example
avar yup, the replacement of engines in p5.10 is lexical, so it's just as all right as it needing strict in some scope:) 19:30
19:37 dduncan joined
fglock avar: re Paste #24943 - I've never seen that; just in case, PCR Makefile.PL would clean up the cache for you 19:38
but may be a 5.9.5 bug 19:39
19:41 polettix joined 19:48 justatheory joined
avar fglock: Do you mainly think of the filecache as something to cache a regex in a loop? 19:52
fglock no, it is a persistent compiler cache; kind of make "pmc" files for each regex 19:53
if you run the same program twice, the second time it starts faster 19:54
19:55 the_dormant joined
fglock I wonder how sbcl compares to ghc in runtime speed 20:01
avar it's probably faster
well, depending on what of course.. 20:02
the speed also depends a lot on how it can optimize the code
if you dump the assembly of a function like (abs) it'll be somewhat slow, but if you say that you're only going to use fixnum it's as fast as it can be on x86 20:03
fglock found it: shootout.alioth.debian.org/debian/b...;lang2=ghc 20:04
lambdabot Title: Lisp SBCL benchmarks | Debian : AMD&#8482; Sempron&#8482; Computer Language Benc ..., tinyurl.com/23zmhu
avar shows what I know about ghc:) 20:06
DB<8> p $p_directory 20:18
/tmp/FileCache/v6-rules/2/f/8/
DB<9> do { my ($s) = $p_directory =~ /^(.*)$/; print "m => '$s'" }
m => ''
mm, wrong channel:)
20:20 rhandom joined
meppl good night 20:25
20:32 the_dormant_ joined, pbuetow joined
audreyt ok, I'm ready to push the pad-refactoring local branch, to be followed by the MO branch. the only problems is that it's 2 months of work and 723 commits... 20:41
20:41 ozo joined
audreyt ponders whether it's really a good idea to simply svk push --verbatim to replay the 723 commits. 20:41
[particle] holy cow, that's a backlog! 20:42
audreyt it also tells me that the next time, choose smaller bite-sizes for local branching
xinming audreyt: It's a good idea to see if there is confliction. ;-) 20:43
audreyt I guess I'll just review the commit log by hand and write a really, really long summary
[particle] what if you push it to a branch?
audreyt xinming: I pull+merge every week or so
[particle]: it's ready for trunk.
PerlJam audreyt: no guts no glory. :-)
audreyt this is more about log-comprehensibility than anything else...
xinming audreyt: Then, I have an idea... Just dump the log with svk log first.
[particle] audreyt: it'd give you a way to keep all the history, but the log would only have one merge entry 20:44
dduncan send it in and we'll deal with it
[particle] *trunk log
xinming audreyt: and then, merge with svn
audreyt [particle]: point taken
xinming svn co the local copy, and merge one by one.
dduncan by the way, have any of you tried the Git version control system?
xinming dduncan: Yes, But a bit hard to use for me.
Tene has tried git and is quite fond of it.
PerlJam dduncan: I used it a bit briefly
dduncan its possible that if we were using that rather than Svn, this matter may not be such a problem 20:45
PerlJam dduncan: merlyn is enamored of it.
[particle] there's no native windows client
only *blech* cygwin
dduncan I'm relatively new to it, but thanks to mugwump I am now using it for my Perl 5 projects, and I like what it brings
xinming Tene: I don't like it... As there will be hash collision in it's files. though, I'll never happen in our lives maybe.
dduncan one big thing that would help us here is that commits aren't linear
audreyt dduncan: true, but a switch to darcs or git just to get one merge going is... strange :)
Tene [particle]: nor is there a native beos client, nor an OS/2 client. ;) 20:46
dduncan yes, of course, we would switch to it just because its generally more useful, if that's true
audreyt I guess I'll just look at the log and summarize a bit without losing too much info and do it in one commit.
avar I use git for perl-core, but I don't really know it:)
avar knows, pull, update, restore, diff
dduncan I only know a minimum of it, but it seems to work
Tene audreyt: perhaps just a log of "Magic! Don't look behind the curtain! Worship me!"
audreyt heh :) 20:47
[particle] 6.28RC1 :)
PerlJam audreyt: you could output the entire log since you forked to a file and then just use that as the log message for the one commit :)
audreyt PerlJam: that's what I mean, with tiny cleanups
doing that now
dduncan either way, I eagerly await the 6.28 goodness
[particle] wonders if there's a commit log max length
PerlJam oh, "look at the log and summarize a bit" sounded quite different in my ears :) 20:48
xinming is waiting for audreyt's commits, and preparing to pulll... O_O 20:49
dduncan what x said
PerlJam I think we're all waiting to pull now that we know what's coming.
dduncan It's Christmas in July, er, May! 20:50
20:51 the_dormant joined
xinming audreyt is mother Christmas.... ;-) 20:53
fglock audreyt: hey :) 20:55
xinming hmm, no offending meaning... I just googled, Father Christmas is the man who gives people presents. So I don't know if Mother Christmas will be the nature form to tell me idea.
audreyt fglock: hey :)
xinming: I think it's known as "Santa Claus"
fglock feather christmas? 20:56
Patterner Hogfather.
PerlJam Note from Santa Claus: Christmas is suspended until Perl 6 arrives. 20:57
xinming Santa Claus's wife: He's right.
PerlJam (They've started using pugs to automate Santa's workshop, but it's not quite functional enough)
20:58 ggoebel joined
audreyt hm, it's 5am and I think summarize is impossible... change of plan: simply replay everything and summarize later ;) 21:00
21:00 lampus joined
Tene alternate plan: replay and let someone else summarize for you. 21:00
audreyt whilst I sleep? indeed, I've been away for too long that I forgot the gnomes :) 21:01
audreyt hacks svk to make it not re-post the merge-from-trunk logs
xinming good idea. Oh, my love, my darling, I've hungered for your commits a long, lonely time... :-) 21:02
21:02 weinigLap joined, ozo joined
xinming don't get me wrong, I am really happy to see audreyt is here again. 21:03
[particle] somebody disable the svnbot before the channel is flooded 21:04
dduncan audreyt, as much as I'd like to see your commits yesterday, I suggest you may want to sleep now and do your commit after you get up, so that something unpleasant doesn't happen due to sleep clouds
audreyt then discovers someone known as mndrix++ has already hacked that into svk ;) 21:05
dduncan: yes, but svnbot runs
so it shouldn't be _too_ bad...
(the svnbot7 was me testing from the pad branch)
and the relatively more fragile code from the MO branch, as well as more shaky speed optimizations, I don't plan to commit today...
japhb I can't wait for this merge ... my stack of 3D code porting was blocked on pre-merge pugs problems .... 21:06
dduncan I mean, you are wanting to get a few things set up just right for this one commit ... sleep can help you not forget something you want to include
audreyt true, but we can always revert 21:07
japhb dduncan: with great respect to audreyt++, I'm more afraid that life will happen for her by tomorrow morning. :-)
audreyt runs "env SVKRESOLVE=y svk push --verbatim"
dduncan okay then 21:08
svnbot6 r16323 | audreyt++ | * initial checkin for pad refactoring
r16325 | audreyt++ | * HTML::Entites - Fix infinite loop due to incorrect multi dispatch use: 21:09
r16325 | audreyt++ | # incorrect - "is rw" only distributes for *@x, not @x;
r16325 | audreyt++ | # the annotation would enable @x.push instead.
r16325 | audreyt++ | sub f (@x is rw) { $_++ for @x }
[particle] skipped a rev?
audreyt mergedown is skipped automagically
japhb [particle]: might be the merge from trunk
svnbot6 r16326 | audreyt++ | * temp checkin for pad refactoring; COMPILING pad thought out.
audreyt i.e. I pulled from trunk between those
[particle] aha
dduncan so we're getting it as a hundred commits then?
japhb dduncan: that's the --verbatim part, I believe
audreyt or more, yes.
dduncan okay
[particle] with realtime comments from the gallery :) 21:10
svnbot6 r16327 | audreyt++ | * more preliminary hacks.
audreyt reason I did it this way is that "svk log t/something.t" can display relevant logs
instead of utter mess of unrelated thing(s)
japhb audreyt: anyway to let us know when all of it is done, without you having to stay awake?
svnbot6 r16328 | audreyt++ | * Done with the parser part.
audreyt (the true reason however is that summarization to make it non-utter-mess is beyond my current brainpower)
[particle] japhb: sure, she could provide the text of the last commit log 21:11
audreyt japhb: sure, the last commit would be say that :)
*saying
dduncan and besides, why should we lose any details ... the verbose is better methinks
svnbot6 r16329 | audreyt++ | * syntax checks now!
japhb audreyt: the per-file svk log being relevent is actually very key for me, I appreciate that, even if it wasn't your core reason.
[particle] audreyt: you should just pretend you're in a different TZ... it's 2pm here, you should be wide awake :) 21:12
audreyt the reason against doing this is that since it's a local branch, my log messages are less formal than what I normally expect myself to give when committing to trunk, but oh well :)
svnbot6 r16330 | audreyt++ | * more whack at pad refactoring; lift freshness to VCode
r16330 | audreyt++ | level and invent the LexPads abstraction.
dduncan why that same number twice?
21:12 prism joined
audreyt two lines? 21:12
[particle] two line commit log
dduncan oh right
svnbot6 r16332 | audreyt++ | * Merged various HsSyck tweaks, including proper base64 routines
r16332 | audreyt++ | to separate Buf from Str later:
r16332 | audreyt++ | Double-quoted strings in YAML output no longer contains continuation
r16332 | audreyt++ | lines, which confused YAML.pm terribly.
r16332 | audreyt++ | Reported by: Nelson Elhage
japhb audreyt: I can guarantee that noone here is going to disrespect you for informality in pushing something we've all been salivating over for weeks .... 21:13
audreyt it's just half of it though, the less sexy, iron-out-long-known-bugs-and-sync-up-to-spec branch
svnbot6 r16333 | audreyt++ | * More pad refactoring.
r16334 | audreyt++ | * Now compiles again whee.
dduncan whee 21:14
audreyt I think I can get the other one in next week, but (again) that's only if life doesn't interfere
svnbot6 r16335 | audreyt++ | * another fix.
japhb audreyt: does this branch include any performance changes?
dduncan is "the other one" the last major thing needed for 6.28?
audreyt japhb: havn't benched. should be ~equiv or slightly faster/slower
japhb audreyt: I was going to do before & after performance tests
audreyt japhb: the one key optimization, O(1) pad lookup, is in the other ("MO") branch 21:15
svnbot6 r16336 | audreyt++ | * Remove unused imports.
r16336 | audreyt++ | (The YAML compression stuff should probably move over to HsSyck too.)
r16337 | audreyt++ | * compiled again.
audreyt otoh, closur entering and cloning is much faster in the new model
otoh, parsing could be slower
japhb audreyt: d'oh! That's the one I needed most, I think .... but closure enter might save me, not sure
audreyt honest answer: I don't know. I don't feel significant changes
svnbot6 r16338 | audreyt++ | * Pad refactoring now runs - snapshot the tree before we redo LexPad
audreyt japhb: yeah, I think I can pry O(1) out and commit it separately first 21:16
svnbot6 r16339 | audreyt++ | * Things actually run now, yay. Next step is to convert
r16339 | audreyt++ | Params over to the new convention...
audreyt but not today, at least not today morning :)
audreyt ponders a "svk log --edit" feature to retroactively edit logs 21:17
svnbot6 r16340 | audreyt++ | * Params unified with pad entries.
r16341 | audreyt++ | * Tonights work
japhb audreyt: no problem, thanks for considering pulling O(1) pad out for me, that's very nice of you!
audreyt clkao: thoughts? # "svk log --edit" to delete embarrasing stuff from local branch log
supposedly it'd display the "svk log" output in an editor, let you edit it, and then parse the log and the do revpropsets 21:18
21:18 drupek12 joined
svnbot6 r16343 | audreyt++ | * Loading recursive TVar finally works. 21:19
r16344 | audreyt++ | * Finally, circular loading of TVars. 21:20
r16345 | audreyt++ | * Repaired accessor - tests finally working.
r16346 | audreyt++ | * Interim commit before we can solve the disapperating-tvar bug. 21:21
r16347 | audreyt++ | * Another checkpoint; nested pads now work.
r16348 | audreyt++ | * This is the first moose that seems to work...
r16349 | audreyt++ | * Go back to Inline style, as now Prelude compilation seems to be working.
r16350 | audreyt++ | * Yet another speedup hack. 21:22
japhb Is there anyway to get svk info to recurse when you do it in a checkout of a local copy of a mirror? I'd like to see the rev of the actual source repo, but svk info in a WC of a local copy of a mirror only shows revs back to the mirror ...
audreyt japhb: "Merged From" displays the chain iirc 21:23
the last lines of an svk info
svnbot6 r16352 | audreyt++ | * Reclose global pad too.
japhb Hmmm, maybe because I'm running an old svk
1.08 (Debian testing)
audreyt *nod* svk1 is distant memory to me... 21:24
svnbot6 r16353 | audreyt++ | * "make ghci" now works GHC 6.7-trunk (which bundles the filepath package.)
japhb Yay! According to packages.qa.debian.org/s/svk.html it looks like svk 2 is only one day away from moving from unstable to testing! 21:26
lambdabot Title: Overview of svk source package
21:26 SubStack joined
audreyt nice! 21:27
what does "Not considered" mean?
(in that debian page)
svnbot6 r16355 | audreyt++ | [07-end-blocks.t] Explicitly refer to $package::var instead
r16355 | audreyt++ | of using "our $package_aliased_as_lex_var" as S04 mandates
r16355 | audreyt++ | that only package vars are safe to use in END blocks.
r16356 | audreyt++ | * DrIFT.YAML: Snapshot of the currently-seems-to-be-sane 21:28
r16356 | audreyt++ | treatment of cyclic YAML compression.
r16357 | audreyt++ | * Use "=begin END" to save some AST space as the END POD
r16357 | audreyt++ | block is not saved into the stream.
r16358 | audreyt++ | * Pugs.Parser: The Main::MAIN sub now remembers its package;
r16358 | audreyt++ | also END blocks should close over its compiling pad as outer pad.
audreyt with the pad branch in, it should be generally possible to take any .pm file and generate a .pm.yml without losing closure information 21:29
svnbot6 r16359 | audreyt++ | * The cached "subLexical" field is no more.
r16360 | audreyt++ | * Regen instances.
audreyt same goes for .t.yml
so that's another route to optimize things, perhaps -- putter did some work on that in Prelude.pm, so maybe reviving it could be considered for 6.28.0, if investigated in time 21:30
svnbot6 r16361 | audreyt++ | * Reclose pads at the end-of-compilation stage (before saving
r16361 | audreyt++ | it into serialized AST), not at beginning-of-runtime stage.
xinming audreyt++
svnbot6 r16362 | audreyt++ | * Include outer pads in prettified code blocks.
r16363 | audreyt++ | * named_2_positional.t: unTODO
japhb audreyt: I think the "not considered" means that svk and the packages it depends on are all too young. There's a 10-day waiting period for people in unstable to find nasty bugs before packages are allowed to move
audreyt also, we'd need to port over Moose.pm tests to t/oo/ so that I can commit the MO branch in good conscience
(pugs is supposed to pass the same object model tests and carry the same API, differing only in surface syntax.) 21:31
svnbot6 r16364 | audreyt++ | * Add placeholder variables into pad as soon as the first one is seen.
audreyt any takers? :)
Tene Where are Moose.pm tests? 21:32
svnbot6 r16365 | audreyt++ | fix an outdated comment
r16366 | audreyt++ | * Remove debug statement.
audreyt search.cpan.org/src/STEVAN/Moose-0.21/t/
lambdabot Title: Index of /src/STEVAN/Moose-0.21/t
svnbot6 r16367 | audreyt++ | * Pugs.Monads: Don't reclose subs that already started.
r16368 | audreyt++ | * Use the Code object's package and outer pads when evaluating
r16368 | audreyt++ | param default expressions.
audreyt warning: there are a lot of them. 21:33
stevan_'s project, that's to be expected :)
svnbot6 r16369 | audreyt++ | * dollar_underscore.t: Chase spec r14356:
r16369 | audreyt++ | Simplified -> binding not to autoalias $_ to first arg, per TheDamian++
r16370 | audreyt++ | * Separate "inline" calls with "displaced" callers, so caller() works more reliably. 21:35
xinming audreyt: Do we need to rewrite the moose test into perl6?
audreyt: If so, I think I can do this job.
audreyt xinming: we do, because the surface syntax is difference
of course if you want to somehow machine-translate it... :)
svnbot6 r16371 | audreyt++ | * Make nextwith/callwith work.
audreyt in any case, in the spirit of t/regex/from_perl6_rules/
I think we can put those into t/oo/from_moose/ or something like that 21:36
and retain the original test numbering/name filenames
svnbot6 r16372 | audreyt++ | * Pugs.Parser.*: New s_outerVars field to track intermediate
r16372 | audreyt++ | blocks crossed from an implicit OUTER reference; for each
r16372 | audreyt++ | compiling pad it tracks the set of vars that must not be
r16372 | audreyt++ | post-declared within it, lest an earlier reference changed.
r16372 | audreyt++ | (The algorithm is by courtesy of Isis Kang; add to AUTHORS.)
audreyt makes xreffing easier
makes some sense? 21:37
svnbot6 r16373 | audreyt++ | * Change the MPad data type to be a standalone datatype,
xinming audreyt: Ok, But I think I have to do this tomorrow evening.
svnbot6 r16373 | audreyt++ | instead of the possibly relocatable TVar.
xinming audreyt: yes,
audreyt: Is it hurry?
audreyt in any case it's a parallelizable job - someone can just commit them en masse as 001_recipe.t.orig or something 21:38
svnbot6 r16374 | audreyt++ | * Primitive blocks generates no closures.
[particle] thinks xinming doesn't understand the nature of open source... it's *never* a hurry
:)
audreyt and then anyone can port one of them and remove the .orig bit. I think.
xinming: it's *always* a hurry so do it asap! :) /j/k
svnbot6 r16375 | audreyt++ | * Reclose the MAIN code as well.
[particle] is suffering from buffering
xinming Ok, I'll do tomorrow, I will import the test later with files renamed.
audreyt xinming++ # or whoever_beats_xinming_to_it++ 21:39
svnbot6 r16376 | audreyt++ | * Correctly handle displaced calls into blocks
xinming [particle]: Yes, I know that, But at least, audreyt herself doesn't have much time, and I don't want to her to do something coolie work. ;-) 21:40
svnbot6 r16377 | audreyt++ | * Supply $_ to the MAIN closure if it wants that.
audreyt looks up "coolie" and realizes that 1)It's Chinese not English and 2)Its meaning has nothing to do with "cool" 21:42
svnbot6 r16378 | audreyt++ | * Fix trait blocks and their interactions.
21:42 mugwump joined
diakopter hands audreyt rose-colored glasses. 21:43
[particle] curses newegg.com's slow rma process
xinming audreyt: My dictionary database should be upgraded. >_<
svnbot6 r16379 | audreyt++ | * &?BLOCK and &?ROUTINE are now correctly (recursively, lexically)
r16379 | audreyt++ | bound, so a displaced blocks can still see its lexically contained
r16379 | audreyt++ | &?ROUTINE correctly.
r16380 | audreyt++ | * Do not recursively close over &?ROUTINE and &?BLOCK.
audreyt xinming: no, coolie is exactly the correct word to use here :) 21:44
it's just surprising :)
svnbot6 r16381 | audreyt++ | * Prelude: Erase the caller frames in &eval and &evalfile. 21:47
r16382 | audreyt++ | * $obj.^foo now means $obj.HOW.foo.
r16383 | audreyt++ | * Repair importing from perl5 by giving the ->$_{...}
r16383 | audreyt++ | primitive a default pad to hold $_.
r16384 | audreyt++ | * &map into a nullary block should still consume the list.
r16385 | audreyt++ | * gen_prelude.pl: Deal with "=begin END" in Prelude correctly.
r16386 | audreyt++ | * fp.pm: Use the modern export-multi style instead of global-operator style for exporting.
r16387 | audreyt++ | * fp - Now that the Inf term (?\226?\136?\158) is supported, test for it.
r16388 | audreyt++ | * Don't return from a pointy block.
r16389 | audreyt++ | * Test.pm - Eliminate an unneccessary double-closure in &try. 21:48
Tene so, import them as .t.orig ? 21:49
svnbot6 r16390 | audreyt++ | * Remove MessageBus as it no longer matches the perl5-IPC::PubSub interface
r16390 | audreyt++ | on CPAN; a backport from there would be better than taking this top-level
r16390 | audreyt++ | namespace.
r16391 | audreyt++ | * Stub code objects are ireelvant to parsing.
audreyt Tene: or .t.p5 or .t.moose or .t.legacy or .t.upstream or something that strikes your fancy :) 21:50
I like .orig because it's a well known extension that carries the "delete me soetime later" semantic meaning
svnbot6 r16392 | audreyt++ | * "my $x = 4" should simply generate "$x = 4" instead of
r16392 | audreyt++ | "($x = 4; $x)". This repairs PIR codegen.
xinming Tene: please don't...
Tene: I'm doing this now...
japhb audreyt: what does coolie mean?
xinming Tene: svk is pushing all the time...
Tene xinming: oh, I thought you said you'd do it tomorrow night. 21:51
audreyt japhb: coo means "suffering" and "lie" mean "labor"
Tene Nevermind, then. :)
xinming Tene: I will import this now, and probably do it tomorrow night.
audreyt (from Chinese, 苦力)
japhb audreyt: ah. "drudgery"
xinming Tene: But at least, currently, I am importing, to avoid collision. plesae don't push curently. 21:52
audreyt porting over all tests is not a job for a single person to do in one night, unless your night is stretchable and no life interferes, so I'd suggest parallelizable go-ahead-and-port-it for everyone interested, once the importing is done
xinming yes...
Tene xinming: will not push. :) 21:53
xinming but don't know why svk push takes so long...
stevan_ there are a good amount of tests which might not be relevant too
features in Moose which are not in Perl 6
[particle] moose.SKIP
stevan_ heya audreyt :)
[particle]: :)
they are not very graceful animals, i doubt they could skip well 21:54
audreyt stevan_: heya. maybe they should be in perl6 somehow *grin* but yes, would be good if you can help mark those ones
japhb Moose has features not in Perl 6? Like what?
Tene injures himself laughing at stevan_
stevan_ japhb: the before/around/after method modifiers i stole from CLOS
the augment/inner stuff I stole from BETA 21:55
the Role requires feature I took from Fortress (but this is in Perl 6 IIRC, just as method stubs)
amnesiac audreyt++ # Keynote tips
japhb stevan_: So these are things that might be elsewhere in the language, just not in the Perl 6 MOP right now?
audreyt before/around/after is just sugar 21:56
svnbot6 r16393 | audreyt++ | * Add svn props to 99problems tests.
audreyt can be ported easierly with callwith/nextwith
stevan_ japhb: Perl 6 will be flexible enough to add all of Moose without much trouble
[particle] m66se
svnbot6 r16394 | audreyt++ | * return_in_anonymous_subs - This test is obsolete per S04; maybe delete?
r16395 | audreyt++ | * coro.t: Correct the "state" test to avoid unintended cloning.
audreyt but I do find before/around/after easier to explain to an audience 21:57
Tene so perhaps perl6 moose will be flexible enough to skip gracefully?
svnbot6 r16396 | audreyt++ | * [t/run] - Correct misc. typoes.
stevan_ audreyt: is there any official hackathon planned for YAPC::EU?
audreyt stevan_: no, I've been too fried to do any logistics the past three months
[particle] no doubt japhb will have images of a moose skipping once he finishes porting his code to p6
stevan_ audreyt: ok,.. no worries, nuffin and I will be there with out hacking leaderhosen on 21:58
japhb [particle]: now there's an idea. I just need a moose model to play with ...
svnbot6 r16397 | audreyt++ | * if3.t: Make it report failure properly instead of dying
stevan_ s/out/our/
nothingmuch audreyt: i can set up crash space if you need it
oh wait, hackathon 21:59
i thought attendance
japhb "PIGGE: Your Moose, Digitized."
svnbot6 r16398 | audreyt++ | * if2.t: wrap the not-going-to-be-supported-in-this-release
r16398 | audreyt++ | feature (statement inside parens) inside and eval+todo.
r16399 | audreyt++ | * ternary.t: More eval()ify and comment about the true cause of failure.
[particle] perl 6: as powerful as a moose, as graceful as a skipping moose!
nothingmuch i can't crash $n hackers on my aunt ^_^
japhb [particle]: *chuckle* 22:00
svnbot6 r16400 | audreyt++ | * autoload.t - More obsoletification of tests
r16401 | audreyt++ | * operator_overloading.t - eval()ify the nasty tests.
r16402 | audreyt++ | * t/magicals/sub.t: Work around a parsing glitch, and
r16402 | audreyt++ | reverse a test because &?ROUTINE is now always defined
r16402 | audreyt++ | (the MAIN is a routine not a block.)
r16403 | audreyt++ | * macro_code_test.t - Make this a test, not a prose. 22:01
r16404 | audreyt++ | * caller.t - gotta have a plan.
r16405 | yiyihu++ | r16092@HomePc: xinming | 2007-04-18 02:10:47 +0800
r16405 | yiyihu++ | Replaced all './pugs' with $*EXECUTABLE_NAME in tests.
xinming ....
oops, It seems that I forgot to push to server what I did almost a month ago. >_< 22:02
stevan_ xinming: if you like, you can also grab the Moose tests from our SVN repo 22:03
22:03 weinigLap_ joined
stevan_ it is still pretty synced with CPAN right now, but you never know 22:03
22:04 riffraff joined
stevan_ xinming: code2.0beta.co.uk/moose/svn/ 22:04
lambdabot Title: Revision 2347: /
riffraff hi 22:05
svnbot6 r16406 | audreyt++ | * implicit_parameters.t: Contrary to what this test says,
r16406 | audreyt++ | {42} does not implicitly use $_ in its body. ;)
r16407 | audreyt++ | * t/02-test-pm/1-basic.t: Merge the same fix of returning
r16407 | audreyt++ | inside pointy blocks from ext/Test/.
riffraff audreyt is back in hacktion? 22:06
audreyt there's some minor regressions with this mergeback
before:
18919 test cases: 17895 ok, 1024 failed, 1987 todo, 715 skipped and 34 unexpectedly succeeded
er sorry, that's "after"
before is:
[particle] riffraff: this is a release of pent-up hacktion for the past few months
audreyt 18921 test cases: 18189 ok, 732 failed, 2054 todo, 719 skipped and 22 unexpectedly succeeded 22:07
but some of it is due to chasing specs
xinming audreyt: Is all change uploaded?
riffraff pent-up? 22:08
svnbot6 r16408 | yiyihu++ | r16515@HomePc: xinming | 2007-05-18 05:47:08 +0800
r16408 | yiyihu++ | Added tests from Moose-0.21 with all tests append .orig.
riffraff (sorry, not native speaker)
Tene riffraff: this is hacking accumulated over the past two months, all being uploaded now.
[particle] dictionary.reference.com/browse/pent%20up 22:09
lambdabot Title: pent up - Definitions from Dictionary.com
22:09 justatheory joined
svnbot6 r16409 | audreyt++ | * Data.Yaml.Syck: Misc code cleanups. 22:09
riffraff yeah I got the concept, thanks to both of you :)
[particle] dictionary.reference.com/browse/pent%20up
lambdabot Title: pent up - Definitions from Dictionary.com
[particle] repressed
riffraff nice expression, I love it when I learnstuff completely unrelated to the channels I'm in 22:10
audreyt xinming: no, they are not -- svk aborted the pushback following your commit 22:11
svnbot6 r16410 | audreyt++ | * DrIFT.YAML: Add "cleanSeen" to clean up the seen-recursive-nodes cache.
audreyt svk-- # not exactly sentient
[particle] hopes he doesn't get that comment on a particle-- someday
japhb heh
audreyt lol
Tene particle++ # quite sentient
diakopter karma [particle]
@karma [particle] 22:12
lambdabot [particle] has a karma of 3
Tene [particle]++ # also sentient
xinming audreyt: sorry.
svnbot6 r16411 | yiyihu++ | r16571@HomePc: xinming | 2007-05-18 06:11:02 +0800 22:13
r16411 | yiyihu++ | Added files from moose svn repository which is missing in 0.21
r16412 | audreyt++ | * Pugs.AST.Internals - Tidy up initial structs for the new VCode/Pad
r16412 | audreyt++ | layout for greater sanity.
audreyt xinming: no, svk needs to be sorry, not you :)
xinming++
xinming audreyt: and sorry again...
audreyt xinming: I'm committing the rest by hand, there's just a few of them
xinming audreyt: because the svk will be interrupted again. I didn't back log....
audreyt ponders how a sentient [particle]++ can be composed of nonsentient @particle++... 22:14
japhb audreyt is in the company of a great many sentients in pondering sentience 22:15
svnbot6 r16413 | audreyt++ | * Regen instances.
xinming audreyt: In my opinion, svk doesn't works well with directory deletion and creation some times. 22:16
svnbot6 r16414 | audreyt++ | * Pugs.AST.Pad: emptyPad is moved into .Internals.
r16415 | audreyt++ | * Misc. stylistic cleanups in Pugs.AST and Pugs.Eval.
audreyt xinming: indeed. 22:17
svnbot6 r16416 | audreyt++ | * Pugs.Parser: Each symbol is now exported exactly once -- the
r16416 | audreyt++ | old %EXPORT treatment of putting each multi variant into its
r16416 | audreyt++ | own array element is now replaced by simply exporting the
r16416 | audreyt++ | toplevel proto/multi sub.
r16417 | audreyt++ | * Pugs.Parser.*: More cleanups; fake a compPad so CHECK blocks 22:19
r16417 | audreyt++ | still thinks they are running in compile time, even though
r16417 | audreyt++ | nothing really is under compilation.
TimToady does this mean I'll have to recompile? 22:20
svnbot6 r16418 | audreyt++ | * Pugs.Monads: Finish up the enterSub/reclose* subsystem;
r16418 | audreyt++ | &?CALLER_CONTINUATION is disabled temporarily because
r16418 | audreyt++ | we're not at all sure if it's lexical or not -- this
r16418 | audreyt++ | perhaps needs a spec discussion.
r16419 | audreyt++ | * This concludes the pushback of the pads-rewrite branch.
audreyt TimToady: oh hey :)
TimToady: yes, if it compiles at all 22:21
(havn't really tried with GHC 6.6!)
TimToady well, it'll work out
svnbot6 r16420 | yiyihu++ | Renamed from_moose_0.21 to from_moose since we are sync'ed with moose repository
Tene What was the goal of the pad rewrite? Infrastructure to enable more development?
diakopter @karma audreyt
lambdabot audreyt has a karma of 779
audreyt Tene: several (I'm quite sleepy so not really coherent) 22:22
first, $_ &?ROUTINE etc were fake-lexicals in old pugs runtime -- you can't really close over them and they pollute dynamically, resulting in surprising bugs
TimToady commuting & # back in about 40 min 22:23
22:23 marmic joined
audreyt the premise of the MO branch is that a class is simply a closure factory, so things that can only be semi closed over must go 22:23
(cf. search.cpan.org/dist/Class-Closure/ if you havn't already for prior art by luqui)
lambdabot Title: Luke Palmer / Class-Closure - search.cpan.org
audreyt next, our YAML precompilation doesn't really work, because outer scopes only exist in runtime and not really in compile time in the old pugs 22:24
so only a very fragile subset of code can be put into Prelude.pm and Test.pm in order for them to be compield
and MO would require a great deal of OO+Closure code to be put into prelude 22:25
and we _really_ want to precompile them
pasteling "rhr" at 65.94.38.10 pasted "r16420 build failure w/ ghc 6.6" (7 lines, 310B) at sial.org/pbot/24950
audreyt so the fragile non-recursive-capable model of pads must go
Tene audreyt: thank you. 22:26
japhb audreyt++ # Yes, thank you for that explanation
audreyt there are many more reasons...
another is of performance; pad lookup is too expensive
and multi variant selection too, partly because of that
by locking up pads during compile time, we can emit code that looks up vars by offset 22:27
not by name
so instead of a large envLexical lookup it's now just two offsets (OUTER depth and slot idx), for (I hope obvious) wins
(where "now" means "not yet in trunk but soon")
22:27 ashleyb joined
japhb That's the O(1) pads, right? 22:28
audreyt yup
japhb It looks like svk push has finished, but I didn't see the "all done" message 22:29
xinming audreyt: I get the same error with Any not in scope.
audreyt 06:27 < svnbot6> r16419 | audreyt++ | * This concludes the pushback of the pads-rewrite branch.
perlbot What kind of idiot karmas himself? Your kind of idiot!
audreyt heh ;)
xinming audreyt: Do I need to upgrade ghc or wait for the fix?
audreyt xinming: fixing
japhb Oh duh, I misread that line (don't ask me how, it's clear now, but I looked at it, and saw something different) 22:30
svnbot6 r16421 | audreyt++ | * DrIFT.YAML: GHC 6.6 doesn't export the Any type, so we 22:31
r16421 | audreyt++ | disable the "seen/cleanSeen" cache in DrIFT.YAML now.
audreyt s/now/for now/
xinming making again.
japhb OK, if I have a WC of a local branch of a mirror of the pugs repo, how do I pull up to a specific pugs rev all the way through to my WC? 'svk sync -t REV //mirror/pugs' followed by 'svk up' in my WC? Does that DTRT WRT the local branch? 22:32
audreyt "svk pull" 22:33
rhr audreyt++ fix works
japhb But wouldn't that pull me up to HEAD? I want to pull to just before and just after your push, so I can compare before/after performance
audreyt oh.
svk sync -t rev, and then "svk up -m" 22:34
and then "svk pull"
22:35 ggoebel joined
japhb I'm not sure I understand the help text for 'svk up -m' ... does that mean that it does the //mirror/pugs -> //local/pugs sync before updating the WC? 22:36
audreyt yup
japhb OK, cool
audreyt ?eval 1+1 22:37
22:37 evalbot_r16322 is now known as evalbot_r16396
evalbot_r16396 2 22:37
22:39 h8crime joined
audreyt pugs compile on feather just finished 22:40
xinming 's computer is still doing compiling 22:41
rhr yay, it builds :)
audreyt now the stable (I hope!) half is in, I'll work on stabilizing the unstable part and commit piecemeal the parts that can be taken out first
with the ultimate goal of passing the (hopefully going to exist soon) moose tests
there's no large showstoppers really, so once it does, it's changelogging time 22:42
?eval 1+1
22:42 evalbot_r16396 is now known as evalbot_r16420
evalbot_r16420 2 22:42
audreyt but now, sleep :)
*wave* &
xinming audreyt: thanks
xinming goes to sleep also. 22:43
22:43 Jmax joined, eghjaytee joined 22:46 TimToady_ joined
rhr smoke time increased by ~2x here 23:03
"Useless use of a constant in void context" generates some impressively huge error messages 23:04
23:05 Aankhen`` joined 23:08 thepler joined 23:16 theorbtwo joined 23:25 fridim joined
japhb I'm getting an error message that simply says 'user error' right after pugs starts when trying one of my benchmarks. However, after that error message, it seems to continue working, but a while later goes into a CPU-chewing loop. This did not happen with the last rev before the big merge. 23:25
I've made clean and am now rebuilding to see if that matters 23:26
23:29 Limbic_Region joined
Limbic_Region woah, significant update with svn up - is audreyt back? 23:30
?seen audreyt
lambdabot audreyt is in #perl6 and #haskell. I last heard audreyt speak 48m 12s ago.
TimToady_ there's a 'user error' trying to precompile the prelude too, and that's after a realclean 23:31
Limbic_Region: yes, the pad refactor just went in
setting up for MO soon
Limbic_Region ok
23:31 buetow joined
TimToady_ not precompiling the prelude might explain the slower smoke 23:32
rhr aha, didn't notice that
Aankhen`` Whoo. 23:33
TimToady_ yeah, just running pugs gets the error
Aankhen`` Nice update indeed.
TimToady_ so it reloads the prelude, which doesn't seen to get the error again 23:34
*seem
"user error" is not the most helpful diagnostic I've ever seen... 23:35
rhr I get the "user error" but it seems to generate blib6/lib/Prelude.pm.yml ok
TimToady_ Oh, I was outside of /pugs when I got the user error
running in /pugs doesn't get it
Limbic_Region is not even going to attempt building without realclean given the extent of the rev
TimToady_ I'm recompiling on windows, up to 98 of 112
I also did a realclean first 23:36
(it'll reconfigure in any case)
Limbic_Region TimToady_ - what is your Win32 environment like - AS Perl with MinGW?
TimToady_ so apparently it's only compiling the prelude that produces 'user error'
cygwin with strawberry perl overlaid 23:37
Limbic_Region interesting
I assume that the cygwin portion has nothing to do with the pugs build?
TimToady_ so anyway, I don't think my "user error" theory explains the slowdown 23:38
I don't know if it's using anything from cygwin
it certainly isn't using cygwin perl, which doesn't work
(for compiling pugs) 23:39
Limbic_Region so in addition to a helpful error message, it is also slower? 23:40
japhb OK, confirmed, 'user error' disappears when running from within pugs WC. However, script eventually off into la-la land still happening. 23:41
rhr my smoke went from 6.5 to 12mins, YMMV
offby1 a long-lasting cigar!
TimToady_ heh, got a permission denied trying to pregen prelude on windows... 23:44
moritz I didn't even knew that windows has file permissions ;) 23:45
TimToady_ could be a cygwin thing
Jmax it has Object Permissions
Aankhen`` moritz: XP Pro (and, I think, most NT-based OSes) have this "advanced" security mode which is pretty intricate.
TimToady_ this is 2000 pro 23:46
dduncan has anyone completed a smoke of the new pugs?
TimToady_ but it's NT based
Aankhen`` Yup.
Jmax it's NTFS
TimToady_ I'm up to t/regex
Aankhen`` Ah, that's probably it.
dduncan okay
Jmax :)
Aankhen`` Jmax = smart. ;-)
dduncan I'm on "make" atm
23:46 Psyche^ joined, Psyche^ is now known as Patterner
moritz Aankhen``: I was just joking... I know that is has permissions, but most windows users really don't know about that 23:47
dduncan I recall some comment in the past about updating Yaml something or other ... is that because some recent version had a bug or because old ones just don't work ... I still have the version from last year
Aankhen`` moritz: Oh, sorry for spoiling the joke. :-(
moritz Aankhen``: my fault, I forgot the smiley ;-)
Aankhen`` Hehe.
Jmax dduncan: the only thing that comes to mind is incompatiblity with serializing blessed objects 23:48
Aankhen`` Nah, the IRC protocol is just missing a way to denote emotions.
How very primitive.
Aankhen`` ducks.
dduncan right now I only care if leaving things alone will prevent smokes from working
I haven't done a smoke for a few months
Jmax erm maybe that's YAML.pm and YAML/Syck.pm. I forgot this was #perl6 :< 23:49
TimToady_ smoke started on windows box okay
Tene ?eval ?eval my $f = 'foo'; $f ~~ m/f/
23:49 evalbot_r16420 is now known as evalbot_r16421
evalbot_r16421 Match.new(␤ ok => Bool::True, ␤ from => 0, ␤ to => 1, ␤ str => "f", ␤ sub_pos => (), ␤ sub_named => {}␤) 23:49
dduncan I'm going to do a smoke today, in any event
Aankhen`` Me too, once I'm done opening a thousand tabs in Firefox.
Should I be waiting for another update first, though, if stuff is broken or not working quite right at present? 23:50
dduncan if the build doesn't die and the smoke doesn't die or loop, that's good enough for me
pasteling "tene" at 24.2.71.129 pasted "pugs failure on matching" (4 lines, 1.4K) at sial.org/pbot/24952 23:51
TimToady_ rhr: is your box multi-cpu, and could it be using only one of them?
dduncan but now's certainly a good time for everyone to smoke, for obvious reasons
offby1 lights up 23:52
rhr TimToady_: yes, but it was using all of them (load was >4 when I checked)
offby1 don't mind if I do.
Limbic_Region TimToady - what if anything did you do about the permission denied issue with the prelude? 23:55
TimToady_ didn't do anything; seems to work 23:56
there's a Prelude.pm.yml with the right date in blib6/lib 23:57
so probably just carping about not being able to overwrite a file that isn't there, or some such
Tene Oh, I guess that's not a rules problem. I get the same error with "$m.say"
oh, seems to only happen in interactive pugs.

Powered by Cro and the Raku® Programming Language.

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

Thank you!