Pugs t-shirts www.cafepress.com/pugscode | Pugs 6.2.9 released | pugscode.org | pugs.kwiki.org | paste: sial.org/pbot/perl6 | www.geeksunite.net
Set by stevan on 15 August 2005.
stevan hey putter 01:01
01:33 sleepster is now known as typester
stevan autrijus: let me know when you commit Pugs.Class.* I am very interested to see the metamodel come to life in Haskell 02:05
stevan ponders porting the MM 2.0 tests to Pugs .... hmmm
geoffb rezzes in 05:00
Since I seem to be off schedule from . . . well, basically everyone right now . . .
rafl: what are the plans re: packaging parrot 0.3.0 for Debian?
autrijus: Will you be attempting 0.3.0 compat for 6.2.10 or 6.28.0? 05:01
stevan: I skimmed through MM 2.0, and I'm curious about stylistic choices:
1) Why so many routines forced into the P5 :: namespace? 05:02
2) Something about it feels . . . not sufficiently sugary. Is this intentional, for performance or stylistic reasons?
3) Are you planning to add more sugar in a later rev?
4) How much of the MM code exists merely to perform the bootstrapping, and how much is useful to the programmer after bootstrap? 05:03
Um . . . I think that's most of what I wanted to ask for now.
stevan hey geoffb 05:05
geoffb hey stevan 05:06
stevan to answer your questions ...
1) I was trying to avoid p5 namspaces alltogether, and the easist way to do that was to force the global :: namespace
2) the MM is not where the sugar is,.. the sugar is in the language itself 05:07
3) nope,.. no more sugar,.. thats autrijus's problem ;) 05:08
geoffb 1a) Why avoid P5 namespaces entirely? In other words, since in effect you chose one namespace to use everywhere, why not a different one?
2a) Sugar to make later MM .pl files cleaner, not sugar exposed to the user programmer 05:09
stevan 4) only genesis.pl is bootstrapping,.. the rest is building up the meta-objects, which (I hope) will be exposed for the user to manipulate
geoffb 4a) Fair enough.
stevan 1a) MM 2.0 was an attempt at a pure meta-model, not just a "lets see how this feels" prototype 05:10
I asked autrijus what I could expect to be available to me
he said "arrays, hashes, scalars and subs"
i did my best to build it out of just that,.. keeping as much of the "not just that" stuff in chaos.pl 05:11
also,... p5 namespaces/packages will not work the same was as p6 namespaces/packages
so I felt it best to avoid them 05:12
2a) ??
geoffb So am I correct to assume that MM 2 is optimized for implementation language portability?
stevan geoffb: yes
and theoretical correctness 05:13
based on the theory I know,. which I hope is correct :)
geoffb 2a.1) Um, clearly I'm not making any sense. :-) I guess I was trying to say that some of the files contain a lot of repeated very similar constructions; I expected to see some of that factored away.
I hope that makes more sense
heh
stevan I borrowed heavily from what I had read about CLOS and Smalltalk (including source code)
geoffb nod
stevan 2a.1) you mean like refactoring some common things into helper subs and such? 05:14
things like attribute access, and such
geoffb yeah, that's what I meant, exactly. I was wondering if it was intentional, or historical artifact, or todo, or what 05:15
stevan well, I didnt want to make any assumptions about sharing of code between various elements of the metamodel
geoffb . . . because some languages might not make that easy? 05:16
stevan exactly
geoffb ah-hah. OK. 05:17
stevan it might force the implementation to conform in a certain way
and besides,.. I am a fast typer,.. so I dont mind :)
geoffb The code was too regular not to have a reason for that, but I couldn't figure out what the reason was.
stevan agian,.. for "correctness" over anything else
and readability too
geoffb nod
stevan I suppose I saw that as an implementation descision,.. if you want/can do that,.. go for it 05:18
geoffb right, I think I understand now.
stevan other than that,.. no good reason really ;)
MM 1.0 was more about "what can I do to just make it work as specced" 05:19
geoffb So . . . what kind of performance are you expecting from your primitives (the above mentioned arrays, hashes, scalars, subs) . . . have you assumed roughly P5-level O() performance? 05:20
stevan MM 2.0 was more about "how should this really be designed, in order to make it work as specced"
geoffb: performance is something I have not even thought about until very recently
geoffb . . . the reason I ask is to know if a language that doesn't intrinsically have an array type with O(1) random access needs to have it emulated to get proper MM performance, for instance
stevan bascially when putter started adding it to PILRUN 05:21
geoffb nodnod
stevan to be honest,.. I haven't given it that depth of thought
geoffb nodnod.
stevan I know the way Smalltalk approaches it is .. 05:22
if it needs to be fast,.. it is a primative method
geoffb Perl 5's primitives can lull one into believing that hashes and arrays are intrinsically fast things . . . but that's only because of a lot of implementation effort on Larry's part
stevan and CLOS make a seperation between functional and procedural parts of the protocol,.. functional being (mostly) side-effect free,. and so easily memoizable 05:23
geoffb: yes, that is true
geoffb thats an interesting and useful distinction . . . (the CLOS one)
stevan yes, I only recently stumbled on that chapter :)
geoffb :-) 05:24
stevan hence the recent memoization of method dispatch in chaos.pl
it was about a 20% speed boost on my machine
geoffb nothing to sniff at. :-)
stevan geoffb: I guess I make an assumption that the implementation has a resonably fast Map (hash) and Array type 05:25
most do now adays
geoffb fair enough.
stevan its the compilers problem,. not mine ;)
geoffb is just poking at the design to see what goes "boing"
heh
stevan very much appreciates that 05:26
I am very interested to see what happens when autrijus starts hacking it into Haskell
the last time it exposed some issues in the design
geoffb ditto
nodnod 05:27
Is autrijus in flight?
stevan mostly that it was to dependent on Perl 5 OO
literally?
geoffb right, I vaguely remember that
stevan I have no idea
oh yeah,.. also did I mention I have obsessive compuslive personality disorder? 05:28
geoffb figuratively, I guess, as in "making his way home" as opposed to "doing a post ICFP hackathon" or "heading to another country not his own", or whatever
stevan not like wash your hands OCD
just I get obsessed with things,.. like MM purity
geoffb (which I have)
stevan I think he is heading to europe,... although I am not sure of that 05:29
EURO OSCON I think
geoffb I'm not like a movie character, but I definitely have (undiagnosed) OCD. I just accepted it, and went on with my life.
stevan scored higher on the ADD test then most tests he "took" in high school
geoffb heh
stevan and I take way too much pleasure in silly useless things 05:30
geoffb That, I think, is generally a good thing
stevan "All art is quite useless" - Oscar Wilde
geoffb is trying to decide what to do next with his awake time . . . 05:32
hmmm . . . catch up on $work, read online for a while, eat some good chocolate, . . . more than one of the above maybe
A little too burnt on massive Forth input to work on the interpreter right now (though from the last book, it's clear my interp needs a major overhaul) 05:33
stevan $work is no fun, chocolate is good,..
hey actually,.. that reminds me...
geoffb Yeah, but getting done with $work *is* fun . . . :-) 05:34
stevan I wanted to talk with you about an article idea I had while mowing my lawn
geoffb chuckles
stevan I actually threw it out to chromatic earlier tonight,.. no response yet.,.. last I heard he was on vacation
the title says it all 05:35
geoffb Geeks are so good at thinking during everyday activities.
stevan Perl for Programming Language Research
basically,. there seems to be a lot of activity lately around this area
geoffb That seems like a good idea to me . . . but it would either have to be high level or *long*
stevan you, nothingmuch with the Forth thing
nothingmuch's Blondie
I think it would end up being long 05:36
cause you just cant skim this one
geoffb From what I can tell, chromatic is religious about only doing O'Reilly mail during pure working hours, and he seems to only check it like once or twice a day in big passes
nodnod
Or at least, you shouldn't. :-)
stevan yeah I used the oreilly.com account,.. since it was work didnt want to bother him if he was on vaca 05:37
geoffb He may have gone off again, but I believe he was back as of a week ago
stevan ah,.. ok
geoffb Of course, he could be backlogged
stevan well we will see,.. I am no rush :)
it has always seemed to me though that Perl was at least as good as Smalltalk, LISP, etc 05:38
in it's own way of course
geoffb . . . especially if, like me, you write much slower than you type. So a few thousands words take many hours.
stevan I tend to blurt out a few hundred words,.. then refactor 05:39
a couple lawn movings later,.. and I am done :)
geoffb When I was in school, I was a one pass guy. Tanked for a while, then wrote straight from stream of consciousness. Now I always outline first for anything over about 100 words 05:40
heh
stevan went to art school,.. you dont really write many papers there
geoffb Oh? What was your focus? 05:41
stevan I dont think I ever wrote an actual "paper" in my life
Fine Art Painting,.. I did 2 years and decided to just learn to wait tables and save my parent some $$ 05:42
geoffb I tried to avoid them. I found only a couple teachers in my life who could actually mark up a paper worth a damn -- most of them just looked to see if you could spell and complete a sentence.
stevan ok, time for me to go to bed .... 05:43
good talking to you geoffb
geoffb My favorite language teacher had a policy that all papers should have approximately the same amount of red ink . . . because everyone can improve.
you too!
have a good sleep
stevan :D
*sigh* school,.. I never quite understood it 05:44
svnbot6 r7213 | Darren_Duncan++ | /modules/Rosetta-Incubator : Added all 15 initial core modules under lib/, though each is currently just a copy of the same template internally; Added t/00_Compile.t, which checks that all 15 would compile and are the right versions; Added usual svn props for all
geoffb heh
dduncan seen mugwump 05:54
jabbot dduncan: mugwump was seen 33 days 1 hours 37 minutes 54 seconds ago
rafl geoffb: I'm currently about to rework the parrot build system for that. Waiting for the comitter bit currently. 05:59
geoffb: Why?
geoffb curiosity, I guess. 06:01
Partly because I just built a new dev machine, and didn't feel like syncing parrot again. :-)
putter stevan: multi methods.. are external to mm2? 06:02
dduncan I'm having a SVN problem, regarding a project on a *different* server than Pugs, and I'm not sure if its at my end or the other end 06:03
geoffb putter, it will probably be a while, stevan just headed off to bed about 20 min ago
putter ah
thanks
geoffb dduncan, is it HTTP? You could just hit it from a browser and see if it's up.
putter, sure, np
dduncan everything works fine but committing 06:04
on commit, it says: The OPTIONS request returned invalid XML in the response: XML parse error at line 1: no element found
putter good night &
geoffb night, putter
dduncan, odd, that
Using svn client, or svk? 06:05
dduncan It's mugwump's server, and I tried asking him, but he's not around in this last half day
geoffb nodnod
dduncan the client is SVN 1.1.3, the server is SVN 1.1.4
same one I've been using for the last 2 months
geoffb wierd.
Sorry man, I've got nothin'.
dduncan thanks anyway
meppl guten morgen 06:16
brentdax waves.
geoffb morning, meppl
morning, brentdax
meppl good morning geoffb
dduncan geoffb, fyi, I did more testing and found that a different project on the same server would still commit, suggesting that the problem is specific to the first project, and not just the server or client 06:18
geoffb interesting 06:19
dduncan I left more messages for mugwump, and perhaps he can check for corruption on the server side
geoffb misconfiguration or corruption in the repo, maybe.
dduncan I had done a clean svn checkout between the first and second instances of the same problem
geoffb nodnod
dduncan anyway, I'm going to bed, goodnight 06:20
nothingmuch morning
geoffb gnight, dduncan 06:24
gmorning, nothingmuch
nothingmuch hi ho
geoffb what's on the agenda for today, nothingmuch? Blondie? 06:25
Khisanth Brunette! :) 06:26
geoffb Mmmm, Blondie . . . 06:27
geoffb searches for Best of Blondie in his collection
nothingmuch today - work 06:28
there's a holiday soon
i've been learning lots of type stuff
(though there is still no tapl =( )
i am getting a semi clear picture on Milner type inference
and I hope I can implement that WRT luqui's theory tuples 06:29
geoffb nodnod
nothingmuch and then i hope I can make it partial
geoffb Rosh Hashana? (forgive my spelling)
nothingmuch yep
spelling is good
i think it is, anyway
nothingmuch never really knows... there's so much of them this time of year 06:30
geoffb OOC, what is the work week in Israel?
nothingmuch sun-thu inclusive
geoffb 's guess was correct then
06:55 _jql is now known as jql
nothingmuch given $luqui.svn { when Broken { $nothingmuch.complain } } 07:08
svnbot6 r7214 | brentdax++ | Cipher API for cryptographic ciphers: 07:11
r7214 | brentdax++ | * Cipher base class, including docs
r7214 | brentdax++ | * Cipher::Caesar for shift ciphers (including rot13)
r7214 | brentdax++ | * Tests for Cipher::Caesar
r7214 | brentdax++ | To come: Cipher::Stream and Cipher::Block base classes, docs for Cipher::Caesar, more tests, real ciphers.
r7215 | luqui++ | Wrote intro to theory.pod. 07:36
nothingmuch pings luqui 07:47
07:56 erg_ is now known as erg
nothingmuch archive.cs.uu.nl/pub/RUU/CS/techrep...04-037.pdf 08:06
nothingmuch thinks that L::AG will be used to infer blondie's types 08:07
it's essentially a middle-order traversal of the tree
with some simple replacement
08:20 Aankh|Clone is now known as Aankhen```, Aankhen``` is now known as Aankhen``
nothingmuch yummy: āˆ€Ī±. āˆ€Ī². Ī± ā†’ Ī² 08:41
sili . 08:51
rafl geoffb: You asked about debianisation of parrot lately. Do you want to help me with that? 08:57
gaal brentdax: ping 09:00
svnbot6 r7216 | iblech++ | * Usual svn props. 11:31
r7216 | iblech++ | * util/svnlog2graph.pl: The scaling is adjusted automatically now.
r7216 | iblech++ | * STATUS: Minor updates WRT MM2.
r7217 | iblech++ | ext/Config-Tiny: 11:55
r7217 | iblech++ | * t/basic.t: s/ok !.../dies_ok {...}/
r7217 | iblech++ | * Config::Tiny: Removed dead code (a pugsbug was fixed).
r7217 | iblech++ | This means Config-Tiny's tests pass again.
masak is it possible to force list context in perl 6? 12:01
nothingmuch yes 12:02
* does that
masak thx, i'll try it
nothingmuch if it doesn't work post the code, maybe we can figure it out
nothingmuch wishes luqui would fix his svn already
i have a nice, shiney, failing tests for him to fix =) 12:03
pasteling "masak" at 130.238.83.176 pasted "I have a problem with context in the calculate_n() method" (73 lines, 2.3K) at sial.org/pbot/13409 12:07
masak * didn't work :(
help appreciated 12:08
nothingmuch masak: got a line number?
oh, calculate_n
masak yes, the reverse call, more specifically 12:09
nothingmuch reverse is from prelude, right?
masak i don't know
nothingmuch is it forcing list context on calculate_z? or is it returning a scalar?
okay
i don't have a working pugs ATM 12:10
byt try this:
@.n = $reversed_pattern.calculate_z();
say @.n;
and then @.n = reverse @.n;
masak oki, will try that
brb
nothingmuch so that we know for sure who is the guilty party
then we may want to fix revere, or alternatively, provide a new test case for pugs
masak ah! @.n is a one-element list all the time 12:12
svnbot6 r7218 | iblech++ | ext/Set:
r7218 | iblech++ | * Set::Hash: s/sub infix:<...> is export/multi *infix:<...>/
r7218 | iblech++ | * t/overload.t: Removed a now-obsolete pugsbug workaround.
r7218 | iblech++ | This means Set's tests pass again, and, as some modules depend on Set, they
r7218 | iblech++ | should work again as well.
nothingmuch okay, so it's time to debug calculate_z =) 12:13
masak yes
rafl How to declare class methods?
masak hm, "return @.z" seems like the right thing to me...
nothingmuch rafl: hmm.. good question 12:14
since overriding 'method new' seems to work, i guess you go with that 12:15
but ask stevan
masak: it does seem right
try return [@.z]
just to be sure
masak oki
still same problem 12:16
nothingmuch hmm =(
say want(); before the return
masak Scalar ((mkType "Scalar")) 12:17
nothingmuch okay
now, try this:.
@.n = reverse $reversed_pattern.calculate_z();
return @.n; 12:18
rafl nothingmuch: method methodname(Class $class: ...) {...} seems to do the job.
nothingmuch and see if it prints the same
masak nothingmuch: oki
nothingmuch rafl: hmm... sounds right
although $class might not necessarily be Class ;-)
rafl Err, Why?
nothingmuch you can override it
rafl What does ::?CLASS mean, btw?
nothingmuch ::?CLASS::Bar is Foo::Bar if $?CLASS eq "Foo" 12:19
rafl nothingmuch: Doesn't every class somehow inherit from 'Class'?
nothingmuch it's something you can put in a namespace directly
rafl: no... but i'm guessing for sanity every class at least does Class ;-)
i'm just being silly though, don't listen to me
rafl Aye. :-)
nothingmuch (btw, in which case methodname (Class $class: ...) will still work 12:20
masak nothingmuch: the want() is now "List (Any)" but the end result is still a scalar "0 0 0 4 0 0 2 0 0 0 0 21"
nothingmuch masak: style question - why do you say ./index_of_first_mismatch(substr(...\n<lots of whitespace>) and not ./index_of_first_mismatch(\n<only 4 more spaces> ? 12:21
hmm
okay
that's a good thing
say "@.z" in there, and make sure it's right
oh wait
sorry
@.z has 21 elems, right?
okay... say @.n and +@.n and make sure they're OK
masak nothingmuch: i wanted to vertically align the arguments
nothingmuch and say want in calculate_n
masak but i see they have become misaligned anyway 12:22
nothingmuch masak: they're still vertically aligned with a \n, but it's more space efficient IMHO
masak nothingmuch: ok, i'll try that
nothingmuch but nevermind, it's not really an issue
wait
svnbot6 r7219 | iblech++ | Recurrence:
r7219 | iblech++ | * Set::Symbols: Again, s/sub infix:<...> is export/multi *infix:<...>/
r7219 | iblech++ | This means Recurrence's and DateTime-Set's tests pass again.
nothingmuch what is @.z?
who do 0, 0, 0, 4, 0, 0, 2, etc belong to? @.z? or something else? 12:23
masak nothingmuch: an array of integers, and yes, the numbers belong to it
the methods are calculating information about a pattern later to be used for searching
nothingmuch aye 12:24
okay, so isn't it the last element of @.z that's being screwed up?
@.z[$k] = ./index_of_first_mismatch(.... # bad ctx?
masak "say +@.z" from within calculate_z() prints 12 12:25
but "say +@.n" from within calculate_n() prints 0
nothingmuch hmm
odd
try doing the reverse manually?
masak hmm :( 12:26
nothingmuch it's not that hard:
masak what do you mean about # bad ctx, btw?
nothingmuch i mean that maybe that assignment has a bad value
hmm
masak but calculate_z() works ok
nothingmuch yeah, i know that now =) 12:27
anywho
masak trying reverse manually :/
nothingmuch my @new; for @.n -> $x { unshift @new, $x }; say "@new"; @.n = @new; say "@.n"; return @new;
masak thx 12:28
i'll do it inline, though
@.n = (); for $reversed_pattern.z -> $x { unshift @.n, $x } 12:29
that worked 12:30
nothingmuch okay
lets find reverse
masak yes, let's :>
i want to have a quiet chat with reverse
nothingmuch it's not in the prelude 12:31
Prim.hs
masak ok
a few weeks ago i finally understood the subtleties of reverse in p5
now i'm being bitten by what seems like reverse bugs in p6... 12:32
nothingmuch hehe
masak nothingmuch: i don't read haskell very well
nothingmuch this is no subtlety... it's a bug =)
masak: add some tests for reverse if you like
i'll try to crack it
masak ok, i'll try and make a minimal test case
nothingmuch t/builtins/lists/reverse.t 12:33
coral .g "focused LEDs"
nothingmuch i'm guessing it's not imposing the right context
masak ?eval my $a = (12, 34); reverse $a
evalbot_7219 '43 21'
coral oh, i'm sorry, ww
masak is this right?
nothingmuch aha 12:34
that stringifies
masak yes
nothingmuch ?eval my $a = (12, 34); list reverse $a
evalbot_7219 ([12, 34])
nothingmuch phoeey
masak :)
it's congruent with the annoying behaviour in my program
nothingmuch ?eval sub foo { my $a = (12, 34); return my @n = reverse $a }; my @m = foo(); @m; 12:35
evalbot_7219 Error: No compatible subroutine found: "&my"
nothingmuch ?eval sub foo { my $a = (12, 34); my @n; return @n = reverse $a }; my @m = foo(); @m;
evalbot_7219 [[12, 34]]
nothingmuch blech
masak indeed
nothingmuch what was actually the problem?
masak that reverse stringifies
nothingmuch i mean, what was the last thing we tried before it succeeded?
masak hand-code reverse
nothingmuch no, one step earlier 12:36
masak uh... (backlogging)
nothingmuch i didn't see the code, so i may have lost synch
masak printing the numifications of the @.z and @.n values
which were 12 and 0, respectively 12:37
svnbot6 r7220 | iblech++ | * t/builtins/math/rounders.t: skip all tests undef PIL2JS and PIL-Run (no eval). 12:38
r7220 | iblech++ | * ext/MIME-Base64/t/Base64.t: Temporarily skip all tests so smoke loops
r7220 | iblech++ | have a chance of finishing again; it seems to be a Parrot-related issue
r7220 | iblech++ | anyway.
r7220 | iblech++ | * Pugs.Parser: Extremely minor cosmetic fix.
r7220 | iblech++ | * Added the standard shebang line to tests.
masak i think we did $reversed_pattern.calculate_z(); @.n = reverse $reversed_pattern.z; return @.n; 12:39
?eval my @z = (12, 34); my @n = reverse @z; say @n 12:40
evalbot_7220 3412 bool::true
masak this works fine
nothingmuch ?eval class { Foo { } 12:41
evalbot_7220 Error: unexpected "{" expecting term postfix, operator, ":", ",", postfix conditional, postfix loop, postfix iteration, ";", "}" or end of input
masak maybe it's the access method...
nothingmuch ?eval class Foo { has @.n; method foo { (1, 2, 3) }; method bar { return @.n = reverse ./foo } }
evalbot_7220 undef
nothingmuch ?eval class Foo { has @.n; method foo { (1, 2, 3) }; method bar { return @.n = reverse ./foo } } Foo.new.bar;
evalbot_7220 Error: No compatible subroutine found: "&foo"
nothingmuch ?eval class Foo { has @.n; method foo () { (1, 2, 3) }; method bar { return @.n = reverse ./foo } } Foo.new.bar;
evalbot_7220 Error: No compatible subroutine found: "&foo"
nothingmuch ?eval class Foo { has @.n; method foo () { (1, 2, 3) }; method bar { return @.n = reverse $?SELF.foo } } Foo.new.bar; 12:42
evalbot_7220 \'3 2 1'
nothingmuch ?eval class Foo { has @.n; method foo () { (1, 2, 3) }; method bar { return @.n = reverse $?SELF.foo } } my @n = Foo.new.bar; say @n; say +@n;
evalbot_7220 3 2 1 1 bool::true
nothingmuch aha
masak ?eval class Foo { has @.z; } my $f = Foo.new; $f.z = (12, 34); my @n = reverse $f.z; @n
evalbot_7220 [[12, 34]]
nothingmuch it makes one concatenated thingy
masak yes
nothingmuch okay, i'm adding my test case
masak nothingmuch++ 12:43
nothingmuch windows is making my head hurt
masak i know, it tends to do that
?eval class Foo { has @.z = (12, 34); } my $f = Foo.new; my @n = reverse $f.z; @n 12:44
evalbot_7220 Error: unexpected "=" expecting trait, "handles", ";" or "}"
masak ah
?eval class Foo { has @.z; } my $f = Foo.new(:z(12, 34)); my @n = reverse $f.z; @n 12:45
evalbot_7220 [34, 12]
masak ah-ha!
oh :) that's actually correct...
nothingmuch =)
i think the problem is the cxt passing
masak ok
i suspected that too, but i don't seem to find it and you already have, so... 12:46
besides, my program works now
thx
nothingmuch isn't 12:48
my @n = do {
(1, 2, 3)
}
say @n;
supposed to work?
i remember an S or an A saying that }\s*?\n does not require a ;
masak me too, vaguely 12:49
but add one for clarity anyway :)
nothingmuch it's a syntax error in pugs 12:50
added another test 12:51
svnbot6 r7221 | nothingmuch++ | r8773@syeeda: nothingmuch | 2005-10-02 14:45:15 +0200 12:54
r7221 | nothingmuch++ | maskak++ 's reverse bug
r7222 | nothingmuch++ | log want()s all around masak++ 's bug, and test that they are all List (Any)
rafl How to declare private class variables? 12:55
As well as public ones.
gaal rafl: class Foo; has $:a_public_var; has $.a_public_var; 12:58
masak gaal: you mean $:a_private_var; 12:59
rafl gaal: Thought that's the way to define instance variables. 13:01
masak rafl: ah, use my and our for class variables 13:02
rafl Ah, OK.
masak see S12
nothingmuch no prefix needed, btw 13:03
gaal oh, *class* vars, oops, right.
obviously what i need is to finish this coffee.
masak my generates access methods, our doesn't
...i think 13:04
nothingmuch is going to ask his boss to not do windows work
it's really frustrating
every time I think i finally managed to simplify things so they are not a big conglomerate of special cases, it turns out there is yet another catch 13:05
rafl just started porting XML::SAX, if anyonce cares. 13:08
masak rafl: will you use it for something afterwards? 13:10
rafl masak: XML-RPC would be nice to have. Of course there are lots of other things to use an XML parser for. Jabber/XMPP for example.. :-) 13:11
Hm, how to call class methods? ::Foo::Bar.meth() doesn't work for me. 13:13
masak rafl: xslt, too :) 13:14
rafl masak: Of course!
masak rafl: that seems to be the reasonable way... isn't that how new is called?
rafl I have class XML::SAX-0.01; method load_parsers(Class $class: Str $?dir, Str $?file) { ... }; and do ::XML::SAX.load_parsers(); 13:15
Hm, something about optional arguments seems to be broken. 13:22
XML::SAX.load_parsers("foo", "bar") works. 13:23
rafl wonders how File::Spec can work. It uses quite similar code. 13:24
Is there already a test for class methods with optional arguments? If not, where should it go? t/oo/methods/class.t ? 13:38
nothingmuch arrives at home 13:51
luqui! wake up! 14:00
rafl Hm. File::Spec.splitpath("/foo/bar") returns <obj:Module>. Why? 14:01
nothingmuch shouldd it really be done as class methods?
this interface issue has always bothered me about File::Spec
i don't understand what's so bad about calling File::Spec::splitpath
but it does hinder the user's ability to import, and makes the code yucky 14:02
rafl Hm, use File::Spec <splitpath>; splitpath("foo") works.
Calling File::Spec::spitpath doesn't. 14:03
nothingmuch how is it documented to work/ 14:04
rafl It's not documented how to call it without importing the subs. 14:06
nothingmuch hmm 14:07
beh. File::Spec is a horrible interface =(
it should have a .new
that is, File::Spec is a factory
and it creates a File::Spec::Unix, or File::Spec::Win32 based on $?OS 14:08
'use File::Spec' should export functions that call these methods on a singleton
based on $?OS
rafl You're welcome to rework it. :-)
nothingmuch anyway, File::Spec::Unix etc is sub yadda 14:09
so File::Spc::splitpath *should* work
you know what? no need for objects
rafl No compatible subroutine found: "&File::Spec::splitpath"
nothingmuch the current version is actually good enough
oh wait 14:10
stevan rafl: File::Spec was one of the early modules ported,.. and it is probably suffering from neglect
nothingmuch that's because of the horrible hack at the top
stevan have you ran the test suite? that should spotlight issues
nothingmuch rafl: okay, one idea is to hack File::Spec so that it loads the right one to itself and to it's caller 14:11
try to paste the if($?OS eq "MSWin32") part under File::Spec-0.0.1 too
rafl Well, simply importing everything I need is enough for me, currently. 14:16
nothingmuch the issue is that File::Spec doesn't really have any functions
it just delegates to the module that it loads 14:17
so that importing goes to it instead of to File::Spec
but it doesn't contain any functions
rafl Ah. 14:18
?eval class Foo { my constant $:bar = 42; method bar { $:bar } } Foo.new.bar.say 14:19
evalbot_7222 bool::true
rafl Shoudn't that say '42'?
nothingmuch ?eval class Foo { my constant $:bar = 42; method bar { $:bar } } Foo.new.bar 14:20
evalbot_7222 \undef
nothingmuch that's what's confusing you... it printed the result of the say
?eval class Foo { my $:bar = 42; method bar { $:bar } } Foo.new.bar 14:21
evalbot_7222 \undef
rafl Well, \undef is false nevertheless, isn't it?
nothingmuch odd
?eval class Foo { my $:bar = 42; method bar { $:bar = 42; $:bar } } Foo.new.bar
evalbot_7222 \42
nothingmuch hmm... i'm guessing the first assignment never happens
?eval class Foo { my $:bar = do { say "evaling"; 42 }; method bar { $:bar } } Foo.new.bar
evalbot_7222 evaling \undef
nothingmuch phhoey 14:22
lets see if it's the closure that's not sharing
?eval class Foo { my $:bar; method bar { $:bar } method set { $:bar = 42; $?SELF }} Foo.new.set.bar
evalbot_7222 Error: No compatible subroutine found: "&bar"
nothingmuch ?eval class Foo { my $:bar; method bar { $:bar } method set { $:bar = 42; $?SELF }} my $x = Foo.new; $x.set; $x.bar;
evalbot_7222 \42 14:23
nothingmuch this seems like a bug
i'm guessing the first $:bar container goes out of scope some how
rafl Could you please write a test for it? 14:26
svnbot6 r7223 | stevan++ | Perl6::MetaModel 2.0 - 14:31
r7223 | stevan++ | * some work on Synopsis 12.5
r7223 | stevan++ | * added lib6/ directory and began a *very rough* translation of the
r7223 | stevan++ | meta-model to Perl 6 code
rafl Does someone wants to port Config::Tiny to real OO and maybe to add read_string and read_fh methods? 14:41
nothingmuch you completely misunderstood Config::Tiny! 14:42
it's purpose is to be small
so that it loads extra fast
it doesn't need any features
that stuff will just add "bloat"
rafl OK, could someone please write Config::Bloated? :-) 14:43
nothingmuch you should ask Damian to port Config::Std 14:44
rafl nothingmuch: The original Config::Tiny has a read_string method, btw.
Ah, iblech already did it: tpe.freepan.org/repos/adamk/Config-...ig/Tiny.pm
Why isn't that version used for pugs?
nothingmuch hmm 14:45
we forgot about freepan
;-)
ingy: ping
rafl I want to what methods can be called on Arrays, but I can't find it in the Prelude. Where should I look for that? 15:10
I have an array of hashes: @a = ({Name => 'foo'}) and I want to add another hash as the value of a new hash key to the last hash in the array. How to do that? @array[-1]<Foo><key> = $val? 15:19
It should look like ({Name => 'foo', Foo => {key => $val}) afterwards. 15:20
nothingmuch rafl: Prim.hs has some stuff 15:45
e.g. List::elems
i suspect this is handled very dynamically
as for the hash assignment - that sounds right 15:46
rafl nothingmuch: OK, how to do that if I wan't interpolation in the assignment? @array[-1]{Features}{$0} = {$1} seems to try to execute &Features. 15:47
nothingmuch either << >> or the french variant 15:48
or use {"Features"}
{ no longer auto bareword }
@array[-1]<Features>{$0} 15:49
but that's ugly ;-)
i would use @array[-1]{'Features'}{$0} or @array[-1]<Features><<$0>>
but with french quotes, since I like unicode
rafl nothingmuch: That gives me [{('Features' => {.}), ('Name' => 'foo')}]. $0 and $1 are correctly set.
nothingmuch odd 15:50
what about @array[-1]<Features> = { $0 => $1 } ?
?eval my @array = { Name => 'foo' }; @array[-1]<Features><key> = "va";
evalbot_7223 Error: Can't modify constant item: VUndef
nothingmuch ?eval my @array = { Name => 'foo' }; @array[-1]<Features><key> = "val"; @array;
evalbot_7223 Error: Can't modify constant item: VUndef
nothingmuch ?eval my @array = { Name => 'foo' }; @array[-1]<Features>= { key = "val"; @array; 15:51
evalbot_7223 Error: unexpected end of input expecting end of input, ";", statements or "}"
nothingmuch ?eval my @array = { Name => 'foo' }; @array[-1]<Features>= { key => "val"}; @array;
evalbot_7223 Error: Can't modify constant item: VUndef
nothingmuch bah
?eval my @array = ({ Name => 'foo' }); @array[-1]<Features> = { key => "val"}; @array;
evalbot_7223 Error: Can't modify constant item: VUndef
nothingmuch ?eval my @array = ({ Name => 'foo' }); @array[-1]{'Features'} = { key => "val"}; @array;
evalbot_7223 Error: Can't modify constant item: VUndef
nothingmuch ?eval my @array; @array[0] = { Name => 'foo' }; @array[-1]{'Features'} = { key => "val"}; @array;
evalbot_7223 [{('Features' => {('key' => 'val')}), ('Name' => 'foo')}]
nothingmuch ?eval my @array; @array[0] = { Name => 'foo' }; @array[-1]<Features><key> = "val"; @array; 15:52
evalbot_7223 [{('Features' => {('key' => 'val')}), ('Name' => 'foo')}]
nothingmuch ?eval my @array; @array[0] = { Name => 'foo' }; my $x = "key"; @array[-1]<Features>{$key} = "val"; @array;
evalbot_7223 Error: Undeclared variable: "$key"
nothingmuch ?eval my @array; @array[0] = { Name => 'foo' }; my $x = "key"; @array[-1]<Features>{$x} = "val"; @array;
evalbot_7223 [{('Features' => {('key' => 'val')}), ('Name' => 'foo')}]
nothingmuch Works For Meā„¢
(after many tries, admittedly ;-) 15:53
rafl nothingmuch: Hm, maybe you wan't to take a look at the code: pastebin.com/380603 - My code is nearly identical and doesn't work though. 15:54
nothingmuch line? 15:55
rafl nothingmuch: 48
nothingmuch can't think of anything =( 15:56
rafl Running XML::SAX.load_parsers.perl.say gives me that: \Class.new(('known_parsers' => [{('Features' => {.}), ('Name' => MkMatch {matchOk = True, matchFrom = 1, matchTo = 19, matchStr = "XML::SAX::PurePerl", matchSubPos = [], matchSubNamed = {}})}, {('Features' => {.}), ('Name' => MkMatch {matchOk = True, matchFrom = 1, matchTo = 16, matchStr = "XML::SAX::Expat", matchSubPos = [], matchSubNamed = {}})}, {('Features' => {.}), ('Name' => MkMatch {matc 15:57
nothingmuch loves builtin Data::Dumper =) 15:58
rafl Well, it could be a bit more pretty, though.
nothingmuch i bet autrijus could be temped into using AGs to write a pretty printer ;-) 15:59
rafl I doubt that would fix my problem. ;-) 16:01
nothingmuch beh 16:03
?eval {.}
evalbot_7223 Error: unexpected "." expecting ";", statements or "}"
nothingmuch that sucks
it's not even .perl ;-)
rafl :-( 16:06
nothingmuch rafl: do you think you could reproduce that? 16:08
copy XML::SAX into a test
take away code till it stops working?
rafl nothingmuch: I'll try.
nothingmuch: Where should that test go?
nothingmuch pugsbugs 16:09
rafl nothingmuch: Can you exaplain that error: *** Can't modify constant item: VUndef 16:20
at nested_hash.t line 28, column 33-59
nested_hash.t line 58, column 4-41
nested_hash.t line 58, column 1-42
pastebin.com/380632
nothingmuch i'm guessing it's an out of bounds assignment 16:21
perhaps @config.push didn't work as expected
since [-1] returned a constant undef instead of a container (since there is no elem) it gave up
rafl @config.perl.say before the failing assignment gives me: [(('Name' => MkMatch {matchOk = True, matchFrom = 1, matchTo = 19, matchStr = "XML::SAX::PurePerl", matchSubPos = [], matchSubNamed = {}}))] 16:22
nothingmuch anyway, i need to go to dinner
blech
rafl Bye.
nothingmuch doesn't know what to say
rafl Oh, nein.. ich hab' nur noch zwei Blaettchen und kann bis Dienstag nicht einkaufen gehen. Duh! 16:25
nothingmuch summons luqui once more 17:21
it's already noon over there, right? 17:22
ingy nothingmuch: eh? 17:33
nothingmuch nothing... /me was just noting that freepan needs more P.R.
how's that patch application coming along?
ingy ?
nothingmuch Test::Base ;-) 17:34
ingy sorry... where is the patch?
nothingmuch i forgot again
i'll regenerate
against .43
ingy mail to [email@hidden.address]
nothingmuch you get to integrate, since it took you so long
okay
ingy okay 17:35
*kisses*&
nothingmuch sent 17:40
18:28 typester is now known as typeout
Aankhen`` goes to sleep. 18:35
nothingmuch ciao
Aankhen`` G'night.
geoffb rafl, sure, I've always wanted to learn more about Debian packaging. What should I read to get (more) up to speed? 19:29
19:41 typeout is now known as typester
wolverian geoffb: www.debian.org/doc/manuals/maint-gu...ex.en.html 19:57
geoffb wolverian, anything else, or will it all be pointed to from there? 20:03
wolverian I'm not sure, I'm in the same process as you are - learning. :)
geoffb (and thanks for the link)
ah, OK
rafl, anything else to look for besides what wolverian just linked?
wolverian (looking into updating the debian irssi-snapshot from SVN)
geoffb (for when he gets back :-) 20:04
wolverian (which has dismally failed thus far.. :)
geoffb ah
geoffb would like to debianize all of his code, and just host a package pool
but it
's always been a low-priority task
FEK
svnbot6 r7224 | stevan++ | Perl6::MetaModel 2.0 - 20:33
r7224 | stevan++ | * added implementation of singleton methods using AUTOLOAD and Roles
r7224 | stevan++ | to be honest, I really dont like it, however the more I read about
r7224 | stevan++ | singleton methods, the more I think they need to actually be integrated
r7224 | stevan++ | into the MM itself (it would make class-methods cleaner thats for sure)
wolverian geoffb: oh, you have code of your own? ;) 20:38
K where is the official PAR website? 20:43
revdiablo search.cpan.org? 20:45
obra par.perl.org, but it got blown away recently in a disk outage 20:52
K i noticed, it appeared to be a chinese blog
or something
obra That's autrijus' girlfriend.
Likely, it's the only thing still live on that box 20:53
nothingmuch oh luqui where art thou
does anybody know to diagnose svn server errors? 20:54
luqui's svn says: Berkeley DB error: PROPFIND request failed on '/svn/misc': Could not open the requested SVN filesystem
sounds like a runaway lock to me... either that or data corruption
rafl geoffb: What link? 21:26
geoffb: Fine. Let's talk about co-maintainership for pugs and parrot tommorow, OK? 21:27
geoffb rafl, OK 21:28
rafl, I've got a lot to learn, so you may need to do some teaching . . . .
rafl geoffb: That's fine with me. 21:29
geoffb wolverian, what, you're surprised? I do occasionally do *something* of value, besides take care of my kids. :-)
rafl, fair enough.
rafl geoffb: What did wolverian link? "22:04 < geoffb> rafl, anything else to look for besides what wolverian just linked?"
geoffb wolverian geoffb: www.debian.org/doc/manuals/maint-gu...ex.en.html
"Debian New Maintainers' Guide" 21:30
rafl geoffb: Well, as I last read the new-maint-guide it was kind of outdated and not very complete.
geoffb Oh, OK.
rafl geoffb: But it's a good starting point nevertheles.
revdiablo Heh, it's good to start out being out of date and incomplete?
geoffb Other places to look as well? I tend to dive deep for a couple days at a time on something, so it's good to have a queue of stuff to read 21:31
I'll backlog . . . 21:32
geoffb swaps back in "Watching Quantum Leap" task
rafl geoffb: Well, the policy is always good to read. The whole new maintainers corner on debian.org/devel as well. 21:36
geoffb: I'll send you some further links tomorrow. I'll take a bath now. Bye!
21:39 flgr__ is now known as flgr
geoffb thx, rafl . . . ttyl 21:56
wolverian geoffb: oh, I was just comparing you to myself. I'm far too lazy to be productive. :) 22:03
dudley_ rafl: ping 22:05
rafl dudley_: pong 22:25
dudley_ what are you planning to do with parrot's build system? 22:26
rafl dudley_: Well, changing it. Not sure how, yet. Currently it uses the MANIFEST to decide where files should be installed that's ought to change. 22:28
dudley_ I'm just curious becuase I was going to try to port parrot to VMS a long time ago, but the build system was too unixy for it to work at all 22:29
rafl dudley_: Well, what's the exactly the problem with VMS? We could write a new, more portable, build system together. It's ought to change anyway, because it doesn't support shared-lib builds and parallel builds very well. 22:30
sili i thought parrot was supposed to be portable?
dudley_ VMS doesn't have a 'make' that works
rafl sili: It's thought to be, but it isn't. 22:32
dudley_: What does it use then?
dudley_ there's a makelike utility called mms
sili there goes my dream of parrot on win3.11 22:33
rafl Well, at least a LSB compliant version would be great.
rafl goes to bed now. See you tomorrow.
dudley_ 'night
dduncan nighteynight 22:34

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!