-Ofun: xrl.us/hxhk | pugscode.org | pugs.kwiki.org | paste: paste.lisp.org/new/perl6 or sial.org/pbot/perl6
Set by apple-gunkies on 11 November 2005.
dduncan hmm, it seems that some update broke Test.skip() ... looking ... 00:28
okay, I see that I used it wrong ... fixing ... 00:42
fixed, commit pending ... it seems that Test or the parser was just more permissive before of a bad skip() call 00:47
er, actually I think I had done this correctly 00:54
mlh dduncan: just continuing a discussion the other day .. (sorry for disappearing) ...
dduncan and Test or the parser may be in error
mlh is there a switch to treat "hello" + 3 as an error? 00:55
dduncan I don't know, and I don't recall talking with you about it
mlh ok, thanks 00:56
dduncan or maybe you were the one that tried to ?eval that ...
and I said to use a ~ instead
to produce what you probably wanted
mlh yes, that was me 00:57
dduncan but I don't think we discussed treating that as an error
in any event, I know of no such switch 00:58
mlh indeed, but i meant to but got diverted
may i vote for one then :-) -- -a switch i mean -- because, for me at least, "string" + number would always be an error. 01:00
dduncan how about "3" + 4 ... is that an error? 01:03
or how about "3hello" + 4?
now that I think about it ... 01:04
?eval "hello" + 4
evalbot_8301 4.0
dduncan mlh, actually there is an error in Perl 5 (which Pugs doesn't seem to do yet) when you turn on warnings 01:06
it says: Argument "hello" isn't numeric in addition (+)
the program doesn't die, but you are alerted to this 01:07
... I don't know for sure, but Perl 5's use fatal may make that die?
mlh, are you there?
wolverian that's not an error, but a warning :)
dduncan yes it is, a warning
wolverian no, Fatal won't do that.
dduncan but how do I make pugs emit that warning?
wolverian larry talked about this sort of a thing on p6l in his latest post (or one of them anyway) 01:08
I doubt pugs has any warnings yet
dduncan what Larry said is that he didn't want use of undefs to be a fatal error, but rather a soft error, such as a warning provides
by default 01:09
that was his response to my "better undef handling" thread
he was citing hardware input/measurements as an example for where soft warnings are good, so a plane doesn't refuse to continue flying if some components break/misbehave 01:10
I'm not sure if that's the best analogy, ... 01:11
but I've conceded to letting things stay as they are
wolverian come up with a better one :)
dduncan my problem was that the analogy didn't seem to fit with the problem I was describing
I was talking about poorly written code, not airplane controls 01:12
regardless,
I've conceded that it is fine for the undef handling to remain as it is /how perl 5 does it
meaning non-fatal
at least we HAVE the warnings
dduncan away for dinner
mlh dduncan: thanks, and sorry again -- went out for lunch 01:23
i think larry's justification is fine for dynamic 'continue flying' cases, but if it can be detected statically ( which i _think_ is the intention for p6) then making it fatal would be a good option 01:35
clkao bwaha, 80.58 % 02:12
obra clkao++ # what was it? 02:14
clkao nothing, just increased the cpu limit for the processes 02:15
obra heh 02:18
dduncan Anyway, I noticed that some recent change to the parser or such had resulted in a "skip('foo')" being interpreted as 'foo'.skip() rather than Test.skip('foo');
obra which js backend does it use these days?
dduncan and that was a failure, since Str didn't have a skip() 02:19
clkao i use --run=jspm --perl5
so it has perl5 support
dduncan while adding another, first argument (1,), caused it to work, Test.pm defines skip() as a multi-sub, and the older form should have worked 02:20
obra jspm == Javascript.pm?
clkao ya. if you want to play with it, read the readme to find the fixed JavaScript.pm from my place 02:21
it actually has 80% coverage now!
obra nah. I'll crunch on my release 02:22
gaal clkao: a few days(?) ago you mentioned you saw a failure when you called "use" in eval not on column 0, but I couldn't reproduce it. was it in main pugs? can you point me at a demo failure? 05:47
stevan_ audreyt: ping 05:53
hello gaal
gaal hey stevan_ 05:57
audreyt: low prio, but could you look at r8298 and say if the new case where named args have Cxt annotations is desirable, or if indeed it's now the only form in which named args can come? (If the latter, I'll remove the old form to resist cruft accruing) 06:05
stevan_: WALK* :)
dduncan gaal, I can tell you that when I used the latest pugs in interactive mode and said to 'use NotExists', pugs died, rather than just displaying an error and waiting for the next input, like it does when you say '$undeclared = 3' 06:10
I would think that pugs shouldn't die
in this situation
may or may not be related to what clkao was talking about 06:11
gaal dduncan: yes, that's a bug in interactive pugs that should indeed be fixed.
it isn't what he's talking about though :)
dduncan on another matter ... 06:12
if all goes well, I should have a new working relational database implemented in perl within the next week
the pugs native version may even work too
then you won't have to link in perl 5 when making your first db using apps 06:13
gaal great!
dduncan however, I should emphasize this is a fully perl native self contained solution ... it is not something that talks to 'normal' databases 06:14
that will come later
it should also help show how perl 6's new constructs make such an implementation easier than perl 5 does 06:15
I think that gather/take would help with some manual-ish implemented ops like table joins and such, for example
gaal hmmm. I don't see why this interactive use is failing so hard. 06:20
audreyt: "use NoSuch" in interactive pugs fails. should Pugs.Prim.Eval's requireInc fail like op1 "die" does, with a shiftT and all that? 06:22
ahhh, it isn't just in interactive mode. you can'y catch use errors in a stringy eval on the command line, either. 06:23
gaal wonders how catchable errors at compile time should be encapsulated. and remembers he has to go to $work :( 06:24
bye folks &
avar "I'm reminded of the day my daughter came in, looked over my shoulder at some Perl 4 code, and said, 'What is that, swearing?" -Larry Wall 06:30
07:01 kakos` is now known as kakos
GeJ greetings all 07:12
07:42 DesreveR is now known as r0nny
audreyt stevan_: r8308 should fix the instantiate_container problem -- please sanity check 10:17
(bbl) 10:21
clkao audreyt: got laptop? 10:22
gaal: i will figure out use problem a bit later
audreyt clkao: no... gotta wait till tomorrow noon 10:31
sigh
kane_ waves at audreyt 10:33
clkao is taiwan turning like britian? it used to be fast! 10:34
audreyt clkao: asus's so-called "royal" club is incredibly popular
so popular that they placed cafe tables and chairs and free tea etc for people waiting to turn their laptop in 10:35
clkao ya. i went there and got things sorted immediately
XD
audreyt average waiting time is 2hrs last weekend I went
kane_: hey
clkao sucks
kane_ audreyt -- got a few mins to spare?
clkao seen the use jsperl5: stuff? want to make it sane?
jabbot clkao: I havn't seen the, clkao
audreyt kane_: yes I do
clkao: sure do
kane_ audreyt -- check /msg (or did you not register? :) 10:36
gaal hey :)
kane_ ah nm, here's fine too actually 10:37
it's p6 anyway -- audreyt, did you get a chance to think about the new proposed use syntax?
audreyt kane_: I didn't get msg
but I did register already
kane_ maybe it's me... we'll keep it here :)
audreyt so, what use syntax? 10:38
kane_ the one i showed you from the S22 draft like, a few weeks ago?
audreyt I thought I expressed that I'd like to use the pkgmgr information on the site
instead of forcing people writing code to specify packages?
kane_ yeah, and then i said some stuff, and you replied and then you wanted to think about it :) 10:39
audreyt ok... a few things
you assume Foo::Bar will be distributed as Foo-Bar instead of inside Foo
if it's inside Foo then one'd need to explicitly say 10:40
use Foo::Bar from Foo;
it strikes me as inconvenient... or did I understand it incorrectly?
kane_ heuristics can be applied, but essentially, yes
you could make 'use Foo::Bar' resolve as 'from Foo::Bar' and then as 'from Foo' as logical alternaties 10:41
but 'use Foo::Bar' from Zot is something one could never guess
audreyt so, in absense of pkgmgr info, I gree the fallback to Foo heuristics is okay 10:42
but we do have pkgmgr info, and it seems silly to discard it
so maybe allow an explicit form
and allow the pkgmgr to intervene if present?
kane_ well, there's a few hefty downsides by relying on the pkg manager
audreyt key here is that people often merge or split packages
kane_ like it doesn't know about homegronwn stuff, etc 10:43
audreyt but why? supposedly anything you can turn into a .jib can become a package
kane_ so that wouldn't work, also the META.info must be 100% correct or modules won't be found etc
audreyt and it should be possible to generate a .jib fromnothing but a package name and a module file
kane_ and then register it somewhere?
audreyt on your system, yes
kane_ that sounds like a lot more work
audreyt but Module::Install::all_from does exactly that 10:44
kane_ explain?
audreyt takes a .pm, figures out META.yml, writes it
so you can install it and register in local package conf
I mean, I don't think the system should stop working if pkgmgr is broken or nonexistent 10:45
kane_ but i dont see why we need all that extra work
audreyt so an explicit form to get stuff from pkg is clearly a win
but if I have a Foo-Bar that provides Foo-Util-Bar
I mean Foo::Util::Bar
kane_ according to perl naming policies, that's Evil btw ;)
audreyt mm. 10:46
it's sorta the norm though.
kane_ btw, the suggestion is to move the authority/version/etc requirements from the 'use' to the 'from' section
note that 'use Foo-(\d)-($authority)' is hard to resolve, and is actualyl something package wide, not module wide
audreyt yeah, I grok that part, that part is sane
kane_ so i feel very strongly about elminating that mpaping overhead, while leaving the package manager out of it 10:47
audreyt ok. so you make a few assumptions
kane_ and remember, in 90% of the cases, it'll do the right thing
audreyt 1. all modules in the package share the authority
kane_ by just saying 'use Foo::Bar'
audreyt 2. all modules in the package share the version of the package
3. packages are named such that it forms the common prefix for all modules in it. 10:48
kane_ 1 for sure, 2 most of the time
3 is an unspoken convention (not adhered to enough)
audreyt but if we are to ditch pkgmgr from the use line
then all(1,2,3) becomes neccessary
kane_ not necessarily 10:49
audreyt otherwise the author is forced to hardcode the package line
kane_ 1 is something i dont think can take much argument, so we'll take that as a given
audreyt why?
kane_ and the rules for resolution can be:
audreyt S11 thinks it's not a given
and thinks each class in the same file can have idfferent uahotiries
authorities
kane_ 'use Foo::Bar' to look for all Foo-Bar packages, then for Foo packages to lookfor lib/Foo/Bar.pm in them 10:50
version resolution can happen when the .pm is found
then the package version and authority become rather pointless
(re S11)
as they add nothing except that the combination should be unique 10:51
audreyt mm, that assumes the constraint that a single file can provide only one "use"
only one class, that is
(as in perl5)
kane_ yup
unless you want to compile /everything/ it's gonna be damn hard to change that 10:52
(or you have a way?)
audreyt yeah, it's the one you deemed too heavy -- namely a pkgmgr that maintains a map from strong names of classes to packages they are in.
I can argue both sides. your proposal is neat and enforces a clean state of expectations
so fits the p6 spirit in general 10:53
kane_ it's possible, but it imposes a crapload of overhead on the compiler and the perl interpreter
in KISS spirit i argue for 'from'
(and thanks :)
audreyt a pkgmgr (as adopted by .NET CLR, Sun Fortress, Haskell, etc) is much more suited to complex interdependent compiled language with static and dynamic linking semantics
kane_ s/perl interpreter/package manager/ of course
*nods* 10:54
audreyt which I suspect may become a part of perl6 as well.
kane_ we can change things though
here's a thought
audreyt we could. and they really don't clash
kane_ if you dont let the .jib code deal with this, but let it worry about putting things in the right spot, maintaining deps etc
audreyt it's complementary -- the lightweight one is always good as fallback
kane_ we can have install hooks keep caches etc that the interpreter can use 10:55
audreyt yes
kane_ and an explicit 'from' would dumb down the use statement to scan dirs
where a clever 'use' might consult a cache
audreyt yes.
kane_ that works for me.. also means we can work on the simple version, and add the complex one if need arises
audreyt so, tell me... why "from"?
why not 10:56
kane_ hmm?
didn't quite get that ;)
audreyt use Foo::Bar p6-Foo-Bar-(ANY)-(ANY);
or
kane_ so it reads like english
audreyt use Foo::BAR-p6-Foo-Bar-(ANY)-(ANY);
kane_ and it's a recognizable keyword
audreyt okay. another q 10:57
we used to have
use perl5:DBI
do you think it needs to become
use DBI from perl5;
use DBI from perl5-(ANY)-(ANY)-(ANY);
kane_ x:Y statements are fine, as long as you can clearly parse them
i think the use statement should translate internatlly to
yes, what you said
p5:DBI might just be shorthand 10:58
which is fine
audreyt use p6:Foo-Bar:1.13:cpan+kane:DBI;
kane_ they all work
as long as it's unambiguous, i dont care about the syntax 10:59
audreyt cool, I think we are clear :)
yup, just making sure it's really all part of pkg
never part of module
kane_ exactly
Foo::Bar version X still has meaning
but a different one than Foo::Bar from version X
(if you want it that way of course :)
audreyt I think per-module versioning and authority still has a meaning. 11:00
but it's okay to defer to package all the time for now
if we need a reverse index from modules to files-in-packages,we can add one later
KISS and all that.
glad we agreed :)
kane_ right.. we just get semantics here... i think we can define it as 'from: tells you where to find a file that might match your criteria, use: lets you check the file against those criteria' 11:01
audreyt so, what does it mean in a p5 context?
nodnod.
p5 context, as in multiversionning-for-perl5 as well as pugs's "use perl5:PKG:VER:AUTH:DBI;" line
kane_ we can make it strikingly similar to 'use only', but cleverer 11:02
but loading in 2 different versions of 1 pm at runtime and DTRT based on scoping is hard
if @INC is tie-able, it'll be a breeze
audreyt @INC is coderefable 11:03
kane_ not quite good enough
unshift @INC...
audreyt I think we can at least reject 2 diff versions when it arises
kane_ i fully plan to let .jib work against p5
with the 'only' like semantics of course
so you can use this PM instead of standard makefile.pl/build.pl if you wish 11:04
and training CPANPLUS to cope isn't too tricky either
audreyt you can tie inc.
kane_ right, so you can say 'use perl6::packages' and have it Just Work 11:05
audreyt I think it's easier to just override CORE::require 11:06
but timtowtdi
kane_ right
that's implementation
audreyt also we really need a sane p5 yaml binding to syck
I'll talk to ingy or maybe jfdi tomorrow 11:07
kane_ anyway, thanks for the few mins of clearing things up... it's clear for me what JIB needs to do, and i think it's clear for you on the pugs end too now :)
clkao tying inc was my plan
kane_ YAML.pm is working for me right now
audreyt yup, excellent
kane_ i've not found anything pathological that broke it in JIB code
audreyt kane_: it's not roundtripping to every other yaml implementation.
pugs included
so it clearly needs to go
kane_ right
gaal btw is there any reason to keep p5's require "filename-in-inc" semantics (as opposed to something like use Module)?
kane_ there's one 'gotcha' in the mixe though 11:08
audreyt gaal: require "foo.ini"
kane_ let me grab the right section
audreyt: we use this now: \[ ... ] # grouping
which serializes to a perlref
that's not portable to other languages
audreyt I see that
gaal audreyt: and that happens more often than require "Some/Module.pm" ? :-(
kane_ require-- # pure evil anyway 11:09
audreyt gaal: I think we just adopt a certain sane person's Module::Load
gaal I wonder if we can't make both work
audreyt for require() semantics
that sane person happens to be here, so feel free to chat :)
kane_ audreyt: if you have a suggestion for a token that serializes cross platform, i'm happy to hear about it
gaal: i'm yours when i hit the office.. i'm late as it is :)
kane_ office &
gaal haha! 11:10
audreyt kane_: nothing from topofmyhead now...
shower, bbiab
gaal kane++ # totally sane, indeed 11:15
audreyt decides to 1)sleep early and 2)go to asus and fetch laptop the first thing in the morning 11:26
g'nite :) & 11:27
gaal night audreyt, i hope we can moose tomorrow your-afternoon my-morning 11:28
araujo Good morning. 11:30
Ive got a kind of compromising question here :-) 11:33
pugs will be used to bootstrap a perl6 implementation right? 11:34
dada that's the plan AFAIK
but I don't think that's really compromising 11:35
araujo Ok, and, when this implementation gets done, will pugs continue under development?
Well, i was referring to that latest question :-)
gaal araujo: this is a while off, but it certainly may
dada oh, this last one seems a good question 11:36
compromising, even :-)
araujo hah, i warned you ;-)
integral pugs is just for fun, hopefully no one will care if pugs becomes obsolete :) 11:37
integral hopes it does
araujo I dont programm in perl, but i do in Haskell ... ive wanted to take a look at the project 11:38
gaal %implementations = ( c => { gcc => <free widely-available>, msdev => <visual-ide>, intel => <fast> }, haskell => { helium => <educational>, hugs => <interactive>, ghc => <production> ... })
araujo And one of the reasons i asked this... To know if you had any plan maintaining a perl implementation in haskell after the bootstrapping.
gaal so we can certainly have perl => { pugs => <fun experimental>, some_other => <efficient?>, ...} 11:39
integral "you" :)
araujo That's perl dialect which i don't understand, but i suppose it is a kind of hash :-) 11:40
gaal yeah
[ ("perl", [("pugs", ["fun", "experimental"]), ("some_other", ["efficient"])) ] 11:41
:-)
araujo :-)
gaal at some point it may certainly become "fun" to make pugs "efficient" though. :-) 11:42
araujo Nice.
dada my $pugs :fun :experimental; 11:43
avar ?eval [ ("perl", [("pugs", ["fun", "experimental"]), ("some_other", ["efficient"])) ].perl 11:45
11:45 evalbot_8301 is now known as evalbot_8307
evalbot_8307 Error: unexpected "[" expecting program 11:45
avar hah! 11:46
araujo So you guys are all lambda-camel? ;-) 11:47
dada ?eval hash( perl => { pugs => [<<fun experimental>>], some_other => [<<efficient>>] );
evalbot_8307 Error: unexpected "h" expecting program
dada ?eval my %perl = ( pugs => [<<fun experimental>>], some_other => [<<efficient>>] ); 11:48
evalbot_8307 {("pugs" => ["fun", "experimental"]), ("some_other" => ["efficient",])}
dada yep
araujo: not all. I'm just a camel
gaal and me, I'm back from a short away. beginner lambdahead. 11:55
kane-xs pokes gaal 11:56
gaal dada: I think this should also work w/o the explicit listref:
?eval my %h = ( key => <val1 val2> )
evalbot_8307 {("key" => ("val1", "val2")),}
gaal because in p6 => is a pair constructor, and its right operand is in scalar context, etc.
kane-xs: I like Module::Load! we should get its usage blessed by @Larry. 11:57
the keyword can stay C<require> for all I care 11:58
kane-xs gaal: by all means.. i use it where i can.. i'ts More Sane 12:00
and the code is trivial, as i'm sure you've seen
gaal yes 12:01
no reason not to implement it in the parser.
kane-xs sounds like a fun haskell excersize 12:02
kane-xs blesses gaal
gaal ruleRequireDecl = do { symbol "require" ; module <- choice [ ruleVarName, ruleUsePackage ] ..... } 12:04
kane-xs: go for it :)
kane-xs no no, i already pushed the monkey onto YOUR shoulder :)
gaal okay, in that case you do the p6-l work. :) 12:05
kane-xs is doing the p6 package manager work
EFULLOFMONKEYS :)
gaal okay okay, tonight when i get back from $work :) 12:08
oh, there already is a ruleRequireDeclaration 12:09
this is indeed very simple to do.
kane-xs gaal++ # porting man
gaal kane-xs++ # porting perldoc # ha 12:10
kane-xs i'll take the free karma :) 12:12
gaal kane-xs: your input on the module load thread is welcome. 13:04
r0nny re 13:18
kioto hi all 13:26
what is the state of perl6 ? 13:29
avar pre-super-alpha-the-spec-isn't-even-finished-yet 13:30
xinming audreyt: ping 13:41
gaal hmmm, require perl5:DBI should be made to work just as use perl5:DBI works no? So ruleRequireDeclaration should be refactored away 14:18
rulePerlVersion now assumes stringy, lexical comparison of version numbers. Is this correct? 14:25
should "require v6" (as opposed to "use v6") work? 14:27
avar muff 14:46
pugs has no printf() ? 14:47
gaal fumm
avar: hmm. there's a sprintf.
don't know why not a printf too. 14:48
avar pugs -e 'sprintf "%100000000000000", 1' 14:49
gaal ouch :) 14:51
trym someone needs to make mod_perl6 ;) 14:56
xinming trym: mod_parrot is enough 14:57
trym ooh.. didnt know it existed 14:58
xinming trym: It's a project which is already started
gaal avar: the problem is in the underlying haskell implementation.
trym xinming: sweet. how far along is it? 15:01
kane-xs gaal: i'm not on a mailing list for pugs -- if you got a link or so i'm hapy to reply 15:18
gaal kane-xs: www.nntp.perl.org/group/perl.perl6....uage/24337 15:26
kane-xs gaal: sane post.. i'll try and keep an eye on the thread :) 15:27
stevan_ audreyt++ # nice solution to the attribute issue :) 15:37
stevan_ removes it from the TODO list :) 15:38
15:40 stevan_ is now known as stevan
stevan is actually liking the mini-lang very much,.. who needs p6 ;) 15:41
clkao lol 15:42
stevan I am starting to think that assignment, looping and anything other than function calls are *highly* overrated 15:43
pdcawley The mini-lang? 15:45
stevan pdcawley: yes, it is part of the new Object Space core design
gaal passes stevan a Ī»
pdcawley really should take a look at the object space stuff...
How 'hooky' is it?
stevan Ī».(gaal) 15:46
clkao home.clkao.org/~clkao/tmp/
stevan hooky?
clkao test-good and bad-pil
pdcawley Well, one of the things I like about ruby is it's abundance of callback hooks for various things in the object model.
clkao the only difference is the first line being empty. 15:47
pdcawley Callbacks on the parent class whenever it's subclasses, on mixins whenever they're mixed in anywhere, on classes whenever they have modules mixed in...
stevan pdcawley: yup, p6 will have this (if I get my way that is ;)
pdcawley Good oh.
clkao then the variable gets into different namespace!
stevan pdcawley: the idea is to have a core set of runtime types,.. the smallest set we can manage
clkao test.pl in the same directory.
stevan then audreyt wrote a mini-language to manipulate those types 15:48
we are then building the metamodel with the mini-lang
pdcawley The various container types and basic object, class, metaclass?
clkao gaal: any idea?
stevan nope
pdcawley: lower level than that
pdcawley Okay.
stevan we make those things with the mini-language
gaal clkao: how are you invoking this, exactly? 15:49
pdcawley stevan: And then, presumably, only to the point that you can start implementing extra behaviour in Perl 6 itself?
stevan pdcawley: this is somewhat old, but it shows the basic structure/layers svn.openfoundry.org/pugs/perl5/Perl..._Space.jpg
pdcawley: yes, pretty much 15:50
although the mini-lang is exclusively within the runtime,.. and not exposed to the user-land yet
but I have thoughts on bridging that gap
pdcawley It's definitely worth taking a look at Squeak here... 15:51
stevan yes, I have been :)
pdcawley Heh.
stevan smalltalk & CLOS are my two favorite object models :)
pdcawley I've never managed to wrap my head around CLOS, mostly because I've never really tried.
stevan and I intend to shoehorn as much of both into P6 before I am done :)
pdcawley: I suggest the book "The Art of the MetaObject Protocol",.. it is execllent,.. even if you never end up programming in LISP 15:52
araujo Smalltalk rul3z
:-]
pdcawley It would probably be interesting to take a look at what the smalltalk VM provides; I don't have the book to hand though.
stevan there is a copy of the blue book online,.. I have the brown one right here next to me
users.ipa.net/~dwighth/smalltalk/bl...p_toc.html 15:53
pdcawley As you are possibly aware if you read my blog, I've been doing most of my programming in Ruby, which is a jolly nice language and all, but I'm feeling the gravitational pull of Smalltalk more and more...
stevan pdcawley: we are thinking of moving to ruby at $work as well ... at least until p6 is done :)
pdcawley Indeed. It's a much cleaner (and, importantly, concise) OO language than Perl 5, but I find it rather lacking compared to what I know is coming in Perl 6. 15:54
clkao gaal: ./pugs -Iext/Test/lib -CPIL
/tmp/test.pl
pdcawley Dammit, I *like* the autovivification magic that Sigils give me. 15:55
clkao compare the output for the /tmp/test.pl with and without the first blank line
stevan pdcawley: p6 is much more multi-paradigm-ish,.. and it has MI and MMD :)
gaal clkao: ahh, okay.. because with with normal pugs it did work. ok, looking.
clkao because foo gets declared into Test:: namespace in jsbackend
pdcawley I'm less than convinced of the desireability of MI in the face of Roles. But MMD doesn't suck.
clkao iblech suspected it's pil bug 15:56
araujo I also like pretty much Smalltalk , now you talk aboit OO, will p6 implement OOP?
clkao er, buggy pil
stevan pdcawley: quite true, however, with C3 MRO,.,. MI becomes much more sane
araujo: p6 will be all OO
pdcawley grins, "I have objects, I have first class functions/closures and I have continuations. If I need anything else I can roll my own paradigm"
araujo stevan, Interesting.
stevan araujo: just about everything you touch in p6 will, on some level, be an object 15:57
pdcawley: quite true :)
araujo stevan, something like... ruby?
stevan araujo: ruby++ if you will ;)
araujo I mean, at the conceptual level....
Cool.
pdcawley araujo: Yeah, but with a more perlish feel to it
araujo stevan, I used (use sometimes) programm in Smalltalk
Best OO language ive ever found out there 15:58
Though.. there doesn't exist a serious and well-integrated implementation for unix-like system :-(
Something very bad.
stevan araujo: I have never programmed in it much myself, I tend to sit back and appreciate it like the great work of art it is :)
araujo *free software* i meant.
It is. 15:59
stevan araujo: Squeak?
pdcawley points out that Squeak used just to have blocks that weren't closures and objects but it was implemented in itself, so someone managed to implement first class continuations in it by monkeying with the way the interpreter handled its control stack.
araujo I tried several times to look at ruby.. but i don't know... i didn't find my way around it.
stevan, you mean the implementation available?
pdcawley araujo: I found that if I think of it as 'perl with better objects' I tend to fall off it slightly, but if I think of it as 'watered down smalltalk' I get along with it much better. 16:00
stevan araujo: is there no Squeak implementation for your platform of choice?
araujo yes, Squeak is good, but it doesn't fit into unix :-(
stevan pdcawley: that is a good way to look at it actually :)
araujo One of the biggest advantages and, (ironically) disa-advanatages of Smalltalk, is the fact that it is a whole OS. Or at least it tries to emulate it.
pdcawley, really?, that sounds nice. 16:01
pdcawley Yeah, you can't easily write the kind of glue you write in Perl/Ruby in Squeak.
araujo: The ruby way is heavy on blocks and stuff.
araujo yeah :-(
pdcawley But less good on things like the smalltalk Collection suite.
araujo To develop in Squeak, means not going out from Squek.
Pretty much the same for most of the ST implementations. 16:02
pdcawley, I agree.
pdcawley You're pretty much limited to Array/Hash/Set <=> OrderedCollection/Dictionary/Set
araujo the collectors are very weak in ruby compared to ST.
pdcawley *everyone*'s collections are very weak compared to Smalltalks.
stevan pdcawley: AMEN to that ! 16:03
araujo hah
pdcawley notes that the new Perl 6 signatures mean it's going to be easier to write code that looks like smalltalk method selectors... 16:04
Which is nice.
stevan yes, quite true
stevan was always jealous of that
pdcawley $thing.at(1, :put<bar>)
araujo pdcawley, really? 16:05
oh
$thin is an object?
pdcawley Yeah.
araujo Interesting :-)
araujo didn't expect perl6 was taking this way
pdcawley You can now specify internal and external names for arguments.
stevan the method would look something like this: method at (Int $index, +$put) { ... } 16:06
pdcawley So internally you use your handy dandy type suggesting argument name, but you give it a meaningful name for the client context too.
stevan although my named arg syntax might be a little behind
araujo Im not involved with perl at all... i used it for some scripots yeras ago, but that was all. I programm in haskell, and now ive found interesting the project because of it, now i see that perl has changed very much in this new version.
pdcawley Stevan: You can also do (syntax rusty): method at (Int $index, :put<$an_object>)
If memory serves. 16:07
stevan pdcawley: your the summarizer,.. I just skim p6l mostly these days
pdcawley What blows my mind about the changes is how much of the language is unchanged, just with all sorts of extras if you want them.
There's going to be an awful lot of p5 scripts that come over with only cosmetic changes. 16:08
araujo I see.
pdcawley But idiomatic Perl 6 will probably look very different from current idiomatic Perl 5.
gaal clkao: this also happens with -CPugs
still don't know why it happens though.
araujo Well, the language specification is still being designed right? 16:09
pdcawley Yes, sort of.
An awful lot's been nailed down and at the 'dotting is and crossing ts' stage of the game.
Other stuff is still up for grabs. 16:10
araujo i see
pdcawley I don't think we'll be seeing any big syntactic changes between now and Perl 6.0.0, but stranger things have happened.
The promised macro system means that if anyone wants big syntactic changes they can have 'em, they just have to implement them first :) 16:11
araujo hah
stevan chears for LISP style macros 16:12
araujo a macro system, Ć  la lisp?
stevan araujo: that is currently the plan
araujo that'd be nice.
stevan very :)
pdcawley Well, the entire Perl 6 grammar's going to be available as a set of rules.
So you define a macro by adding/overriding rules in the Perl 6 grammar. 16:13
Or, if you feel like it, you can replace the Perl 6 grammar with something else entirely.
stevan clisp.p6grammer ;) 16:14
pdcawley use Compiler 'Scheme' :{(def ...)}
Ferinstance.
Actually, more like:
araujo pdcawley, that's cool.
pdcawley use Inline 'Scheme'
sub foo (...) { ... } 16:15
araujo So i suppose you could re-define things like operators, clauses etc?
pdcawley scheme (...)
Whatever you like.
As Larry says "All's fair if you predeclare"
clkao gaal: but how come it works properly :p 16:16
gaal clkao: the outputs for -CPugs are very different! I have no idea what's causing this :-( 16:26
clkao subName = "&Test::foo" 16:36
gaal yes, i mean what in the parser might trigger these different interpretations 16:39
stevan audreyt: ping 19:25
buu Did anyone fix the svnbot? 19:45
svnbot6 r8308 | stevan++ | PIL/Native/Bootstrap.pil 20:22
r8308 | stevan++ | - setting the attributes correctly based on audreyt++ solution to
r8308 | stevan++ | the instantiate container issue ...
r8308 | stevan++ | - some misc. comments
20:37 kakos is now known as kakos|away
svnbot6 r8309 | stevan++ | PIL/Native/Bootstrap.pil 23:02
r8309 | stevan++ | - simplified CREATE using -> {}.do_for()
r8309 | stevan++ | - added a &WALKCLASS based version of CREATE for
r8309 | stevan++ | when the dispatcher is working correctly
r8309 | stevan++ | - converted BUILDALL to use -> {}.do_for() as well
r8309 | stevan++ | - commented out the &WALKMETH version for now
r8309 | stevan++ | - cleaned up some of the Bootstrapping code
r8309 | stevan++ | - added SEE ALSO section in the POD to point to the
r8309 | stevan++ | p5 prototypes of the MetaModel and ObjectSpace
r8309 | stevan++ | t/pil/metamodel.t 23:03
r8309 | stevan++ | - adding some tests for creating new classes, and then
r8309 | stevan++ | creating instances of that class. This tests the new
r8309 | stevan++ | BUILDALL/BUILD functionality
23:03 stevan_ is now known as stevan
stevan is a karma whore ;) 23:04
audreyt: I would like to discuss the dispatcher with you when you return
stevan off to dinner (be back in ~2 hours) &
obra seen audreyt 23:07
jabbot obra: audreyt was seen 11 hours 40 minutes 47 seconds ago
Juerd audreyt: Is your journal no longer maintained? 23:09
Or am I looking in the wrong place?

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!