6.2.7 released! | pugscode.org/ <Overview Journal Logs> | pugs.kwiki.org | paste: sial.org/pbot/perl6
Set by autrijus on 13 June 2005.
wolverian nothingmuch: right. thanks. 00:01
("its")
mrborisguy opinion question for you guys... 00:49
I'm thinking about learning Haskell
but not really to do anything with it, I just think it'd be good to take a step into the functional languages to learn what it's about
would you say learning Haskell will improve my programming skills/logical skills when it comes to programming? 00:50
(I ask here, 'cause obviously nearly everybody here is still doing perl, but I know quite a few of you know quite abit about Haskell too) 00:51
mugwump so, of course 00:52
s/so,//
PerlJam mrborisguy: yes.
mrborisguy: What languages do you know?
Darren_Duncan large LKT design-based update committed, corres to large p5 update, ... which among other things should simplify getting it working in Perl 6 00:53
mainly in that regard, the test suite is more focused on cases that should succeed, rather than fail
mrborisguy PerlJam: starting in order of learning: QBasic, C, Java, perl5, a bit of perl6 (and probably a few others scattered in there, like a step into php and python) 00:54
svnbot6 r4657, Darren_Duncan++ | large Locale-KeyedText update that corresponds to an in-progress Perl 5 version large update : removed superfluous input filters on Msg/Trn.new() so that you can now set each property element to any defined value ; added input constraint that each Trn property must have at least 1 element; updated POD to reflect the change ; removed and added a lot of MsgProp and TrnProp tests so they focus mainly on success cases rather than failure
PerlJam mrborisguy: If you're just trying to broaden your knowledge base, I'd suggest picking up a little lisp/scheme too and maybe some prolog. And if you do haskell after lisp/scheme you'll have a tremendous sense of joy at what you have discovered. :) 00:56
Well, you may have that anyway, but you'll be happier that haskell is *not* lisp/scheme :-)
mrborisguy haha, okay
but those would be the best to start off with? 00:57
mugwump elisp
mugwump hides
PerlJam mrborisguy: lisp/scheem teaches you about functional programming and pain. Prolog teaches you about declarative programming and pain. :-) 01:02
mrborisguy: oh, and play with bc or postscript too (I found postscript kind of fun but some people don't quite think of it that way) 01:04
mrborisguy PerlJam: how much time do you think I have on my hands! ;) 01:05
PerlJam mrborisguy: a whole lifetime
mrborisguy PerlJam: ah... good point 01:06
PerlJam Also, playing with these other languages and learning to think about things differently will help you better understand and use perl (especially perl6) 01:07
Do you have a copy of HOP? That's a good intro to FP via perl.
mrborisguy High Order Perl? no, I don't
I hear alot of people mention it, I'm guessing it's a good read then? 01:08
PerlJam you could wait for MJD to make it available on his website (or even volunteer to help him make it available on his website) 01:09
yes, it's a good book (modulo the errata) 01:10
rjbs It's great. 01:12
mugwump mrborisguy: one interesting idea might be to learn OCaml and work through Types and Programming Languages
as all the examples in that book are in OCaml 01:13
mrborisguy does he plan putting it on the website soon, do you know?
I'm a bit strapped for cash, and the last couple of times I was in Barnes and Noble, i don't think it was there, otherwise I would have paged through it! 01:14
mugwump I heard mutterings about a PDF being around for it
But that might not be entirely endorsed/legitimate
Just an idea, anyway. Getting a book that covers various FP topics, and learning whatever language the book uses. 01:17
Khisanth meh actually tried two B&N :/ 01:20
PerlJam I seem to recall the MJD needed help converting the source documents into something more webable. 01:21
meppl gute nacht - good night 01:22
PerlJam Ah yes ... hop.perl.plover.com/announce/20 01:23
So send mjd a note that you'd like to help and you may get a pre-release version of what goes on the web. : 01:24
:-) even
Odin-LAP PostScript 'fun'?
Hm.
That's counter to almost every other description I've heard of it. :) 01:25
Khisanth PerlJam ~~ Masochist :) 01:26
geoffb I've known several people that enjoyed hand-coding PostScript, including a DJ who designed all his tape and CD covers in raw PS. 01:28
And yes, he was a freak. :-)
arcady I even know someone who wrote a postscript "virus" 01:29
PerlJam arcady: have you seen the postscript web server? 01:30
geoffb At least one friend used to have a ball upping the ante in display wars when running on NeXT boxen -- he would melt their displays along nice curves. :-)
Odin-LAP Doesn't there exist a post ... nevermind.
PerlJam: Do you have a link? :>
PerlJam unfortunately no, but I'm sure google does :) 01:31
Odin-LAP True, but I'm lazy. ;)
PerlJam Since there's no socket code in postscript it obviously has to run via inetd
Odin-LAP Aren't there network extensions?
I'm sure I heard of a server which ran on printers... :p 01:32
Hmm. 01:33
PerlJam anyway, postscript is fun to code. Not so much by hand but it's really easy to write a compiler that outputs postscript :-) 01:34
Odin-LAP Ahhhh. 01:35
That's your argument, then. :p
PerlJam Well, it's fun but tedious to code by hand. The tediousness kinda saps the fun. 01:36
Odin-LAP Any pointers as to how to learn it? :p 01:37
Odin-LAP has actually taken a look, once or twice.
PerlJam Odin-LAP: there are several full books on it available for free (in PDF format no less :) 01:38
Odin-LAP Hmm. 01:39
PerlJam Thinking In Postscript is probably a good start
Also, I believe Adobe has made the complete reference manual available for free too 01:40
(It's been a while since I've played with postscript obviously)
Odin-LAP :) 01:41
There's quite a bit of stuff from Adobe, actually. 01:42
PerlJam Yep ... partners.adobe.com/public/developer...refman.pdf
Odin-LAP It's just not very useful to learn from. :p
PerlJam Try "Thinking In Postscript" 01:45
Odin-LAP Yeah, will do. :)
PerlJam anyway ... time for me to go hang out with my kids a little bit :) 01:46
btw, did I mention here that my wife is pregnant again?
Found out yesterday that it's not just a third child we've having, but a third *and* a fourth. 01:47
Odin-LAP What fun.
PerlJam we'll see. Hopefully the twins turn out like their older siblings
anyway, gotta go.
g'night all
PerlJam &
03:47 Boogie__ is now known as Boogie
geoffb I'm just starting to read the core Haskell source for Pugs . . . and I came across the following in Prim.hs: 03:48
op2 "~~" = op2Match
op2 "!~" = op2Cmp vCastStr (/=)
That seems very wrong to me. Shouldn't !~ just be the opposite of ~~ ? 03:49
arcady I think so 03:51
scook0 try to come up with a simple test case to check 03:52
arcady ne is string inequality
gaal good catch - it can't be *just* the opposite though, because ~~ returns an "interesting" value. 03:53
so i'd expect you need something like (not . VBool) op2Match 03:54
arcady yeah, that's what I was going to say
had I known haskell syntax
gaal er, no; s/VBool/the thing that casts to vbool/
:-p
good morning, btw :) 03:55
geoffb good morning, gaal!
gaal try vCast 03:56
geoffb as above "I'm just starting to read the core Haskell source for Pugs . . ." -- it'll be a bit before I go hacking 03:57
:-)
gaal oh, i seem to recall that !~ is not an oversight, it was added so that certain tests can parse 03:58
before there was a full ~~
QtPlatypus BTW is there an introspective thing that gives you a list of methods that an object supports? 03:59
gaal (~~ was once just a test for equality)
arcady maybe a "methods" method
gaal Qt, not yet, and even when we have introspection, that's impossible to fully do in a language that has AUTOLOAD :)
QtPlatypus nods "I'm aware of the AUTOLOAD limitation" 04:00
gaal i think there's some symtable stuff already? in a pinch, you can ref the object, and grep { Code } its class' symtable 04:01
then walk its @INC etc
mugwump that will be at $object.ref.meta.methods
eventually :) 04:02
arcady that looks kind of icky
mugwump you can judge your age by the amount of pain you feel when coming across a new idea
It works like this. $object.ref returns the type of the object 04:03
arcady $object.meta.getmethods is mentioned in S12
mugwump ah, you're right 04:04
wolverian I wonder why it has 'get' in the name 04:05
mugwump with that awful .dispatcher
arcady $object.ref.meta.methods would give you the methods of the metaclass
the ones that you can call on the class object
mugwump so if $object is a MyClass, then $obj.ref =:= ::MyClass 04:06
arcady and if $object is some anonymous type, then $obj.ref is the only way you can actually get at its class object
mugwump you might want to look at ext/Perl-MetaModel/docs/perl6_meta_model.pod 04:11
I am expecting to get the .dispatcher anomaly addressed eventually. .ref() is all you need 04:12
.ref() returns a type, which is a dispatcher - something like a package in P5 land
it might not be a class at all
.meta() can be called on that object and returns the Class (or whatever) object that is the "master" for this class 04:13
s/class$/type/
geoffb Augh, finally: 04:16
pugs> not(0 !~ rx:Perl5/0/)
bool::false
pugs> not(0 ~~ rx:Perl5/0/)
bool::false
mugwump the proxy methods, like $object.meta, ::($foo).getmethods, etc are shortcuts which bely the structure of the metamodel, so I don't like using them :)
geoffb It turns out things tend to stringify in such a way that it takes a little effort to find a case where string match doesn't give the same answer as smart match 04:17
arcady come to think of it, $foo.ref.methods makes sense 04:20
it all depends on where you want to think the methods live 04:21
mugwump yes, .ref.meta.methods might reasonably only return the methods defined in that class 04:22
.ref.methods would probably return everything that the .ref.can()
svnbot6 r4658, geoffb++ | Test for bug: !~ is treated as string ne 04:29
geoffb absentmindedly does 'make un<tab>' in his shell and gets a list of matching make targets 04:33
oooooh, shiny
wolverian geoffb: what shell is that? 04:43
geoffb bash
wolverian what do I have to do to get that feature? :)
geoffb I'm guessing "enable bash_completion"
What distro are you running? 04:44
wolverian ubuntu and debian
mugwump oh yeah, that annoying feature 04:45
geoffb Assuming ubuntu shares Debian's bash config (one would hope), then give me a sec and I'll look it up
wolverian thanks!
mugwump: why is it annoying?
mugwump the one that always makes me think directories are empty when I can't tab complete the files in it, then realise that it's because of the `cd' at the beginning of the line 04:46
wolverian does 'cd'ing to a file make any sense? :)
geoffb wolverian, uncomment the 3-line bash_completion block in either /etc/bash.bashrc or ~/.bashrc 04:47
mugwump It's just the way I've grown used to the shell tbh. I'm sure if it had always behaved like that for me, it would be nice
wolverian geoffb: oh, that's easy. thanks!
mugwump: right. oh well.
geoffb wolverian, it doesn't, which is why the behavior is technically correct, but I've been annoyed by that just like mugwump. Some kind of out-of-band notice that the tab completion is being artificially limited would be nice
wolverian hmm. it doesn't work. 04:48
it works for 'ssh' though.
geoffb mugwump: you could always tune bash_completion to not do that for cd . . . .
wolverian I guess I'm missing the 'make' thing for the completion
geoffb wolverian, did you source your .bashrc?
wolverian yes
as I said, 'ssh <tab>' works
(shows remembered hosts) 04:49
mugwump sure, I could. Just like I could set a VIM option to turn off auto-indent. But what happens when I paste some code in there and VIM f*cks it up? I quit and start nvi :)
geoffb (I was behind in my responses)
wolverian mugwump: :paste
geoffb mugwump, chuckle
wolverian nice, perl -l<tab> 04:51
er, perl -<tab>
geoffb Heh, interesting 04:52
mugwump you could make it search through your command history for one liners when you: 04:53
perl -e '<tab>
geoffb Ouch: disk grinds when you do perl -M<tab> -- It seems to search for all available modules! 04:54
Yep, very interesting. perl -MTe<tab> gets me a nice chunk of Test:: and Text:: modules
Interesting . . . if (...) { ... } else { ... } parses faster than ... ?? ... :: ... 05:09
mugwump backtracking 05:10
geoffb The question is, what about the if form reduces backtracking . . . or conversely, what about ternary op increases it? 05:11
Could someone who knows makefiles please make Pugs stop spuriously building on 'make install' when it is unchanged? 05:13
link_time_frustration++ 05:14
Or, since I guess that is interpreted as karma:
link_time_frustration-- 05:15
QtPlatypus Seems with the PGE and rules its quite easy to create rules that suck up all the memory and Bus error. 05:21
05:41 _jql is now known as jql
geoffb I've managed to pull about 13% out of the parse time of Test.pm in my local tree, but it's been like pulling teeth the whole way 05:50
mugwump nice 05:51
by changing the source?
or hacking Parser.hs et al?
geoffb so far, by doing transforms of the source 05:53
I have a feeling that to get much more, I may have to understand Parser.hs even to know what Test.pm source transforms will help 05:57
Of course, there's the nifty fact that about half of the test scripts use nothing other than plan, ok, and is . . . so a Test::Minimal or suchlike would have a huge impact on make test time 05:58
mugwump Sounds like a job for an autoloader! :-D 05:59
geoffb If only . . . 06:04
If I knew how to autoload w/ "is export", I'd do that 06:05
pasteling "geoffb" at 69.110.115.185 pasted "Cumulative count of test files using various Test.pm functions" (20 lines, 691B) at sial.org/pbot/11074 06:08
sdtr443w How would I define an array of hashes and/or vice versa in Perl 6? 06:09
Or rather, declare?
mugwump my Array %HoA
geoffb As mentioned above, with just 3 functions, Test::Minimal would have 50% test file coverage; with half of Test.pm's functions, it would have met the 80/20 rule
mugwump my Hash @AoH
wolverian or my @AoH is Array of Hash; 06:10
geoffb It's somewhat silly that Test.pm includes functions that never get used at all, and we have to pay the parse cost for them over and over (removing those is *not* included in the 13% figure above) 06:11
sdtr443w in Pugs I just tried "my @array_of_hashes is Array of Hash;" and it didn't seem to work
geoffb well, it's getting late here. 06:12
mugwump sdtr443w: for now you can use my Hash @array_of_hashes
geoffb Time to trundle off to bed, I think
wolverian geoffb: can't we precompile Test.pm? 06:13
nothingmuch morning
wolverian has no idea about the compiling features pugs currently has
sdtr443w mugwump: What then, of an array of hashes of arrays?
wolverian my Hash of Array @AoHoA 06:14
mugwump which is also currently parsefail ;)
geoffb wolverian, that *might* be possible . . . but I don't recall pugs being able to compile just one module and use that, as opposed to an entire script
wolverian geoffb: right.
just a thought. :) 06:15
sdtr443w Hmm yeah Pugs wasn't reaching a say statement right after "Hash @array_of_hashes;" either so I'm thinking it's not enjoying that.
Aankhen`` sdtr443w >> The sigil determines the primary type if it's @ or %, and any type restrictions apply to the elements.
sdtr443w To be fair, I'm using 6.0.11 -- I think.
Aankhen`` ?eval my Hash @array_of_hashes;
wolverian wow. that is ancient.
evalbot6 undef
sdtr443w I tried building the new one and had some trouble with ghc coming up in interactive mode during the build
Aankhen`` ?eval my Hash @array_of_hashes; @array_of_hashes[0] = (foo => "bar", baz => "quux");
evalbot6 [('foo' => 'bar'), ('baz' => 'quux')]
wolverian sdtr443w: mail juerd for an account on feather.perl6.nl?
Aankhen`` ?eval my Hash @array_of_hashes; @array_of_hashes[0] = (foo => "bar", baz => "quux"); @array_of_hashes 06:16
evalbot6 [[('foo' => 'bar'), ('baz' => 'quux')]]
mugwump sdtr443w: you have ghc 6.4?
Aankhen`` ?eval my Hash @array_of_hashes; @array_of_hashes[0] = { foo => "bar", baz => "quux" }; @array_of_hashes
evalbot6 [{('baz' => 'quux'), ('foo' => 'bar')}]
geoffb is really going now
sleeping &
wolverian ?eval my Hash @aoh; @aoh[0] = "hi"; # should fail?
evalbot6 \'hi'
sdtr443w mugwump: Yes, that's how I got this version of Pugs to build... 06:17
Aankhen`` Shouldn't that be: ({ 'baz' => "quux", 'foo' => "bar" })
wolverian Aankhen``: yes. pugs doesn't do typing very well yet
QtPlatypus I don't think it does typing at all..
wolverian exactly. :)
mugwump sure it does
Aankhen`` wolverian >> If what Larry said about "my Dog $spot is just a statement" still holds true, the line you tried shouldn't fail.
mugwump MMD is basically type-based method dispatch 06:18
QtPlatypus ?eval my Int $anint = "BooWhoo";$anint;
evalbot6 \'BooWhoo'
Aankhen`` mugwump >> It handles the primary type of the variable fine, but if the variable is a container, it doesn't seem to be handling the types of its elements very well... 06:19
wolverian Aankhen``: the @aoh[0] = "hi" should fail, I think 06:20
Aankhen`` wolverian >> Depends on whether types are supposed to be enforced outside of signatures.
QtPlatypus And my example should fail as well.
Aankhen`` Although I guess what I quoted above is quite outdated.
QtPlatypus Aankhen``: Why declear then if you don't wish them to be enforced? 06:21
mugwump ?eval multi sub foo(Int $foo) { "int $foo" };multi sub foo(Str $foo) { "str $foo" }; "&foo(1), &foo('bar')"
sdtr443w Well ok presuming Pugs is having a little bit of trouble with AoHoA or anything like that, then how would I dereference that. Could it be something like my $final_scalar = @AoHoA[1]{"I like cake"}[3];
evalbot6 'int 1, str bar'
wolverian Aankhen``: yes, they are. saying my Foo @bar is a statement that only Foo-type things can be put into @bar. I don't see a way for an Str to be coerced into a Hash :)
Aankhen`` wolverian >> OK. :-)
QtPlatypus >> Dunno, read the relevant Apocalypse...
wolverian Synopses are the recommended reading now. 06:22
(they are actually updated.)
Aankhen`` I know.
wolverian (sometimes.)
Aankhen`` That's why I said my quote was outdated.
arcady I think the "my Dog $spot is just a statement" means that it doesn't actually make a new Dog
mugwump in theory my Hash $foo; $foo = "bar"; should be a *compile time* error
Aankhen`` RFC 218: my Dog $spot Is Just an Assertion
I expect that a declaration of the form:
my Dog $spot;
is merely an assertion that you will not use $spot inconsistently with it being a Dog.
wolverian isn't that RFC about not instantiating an object immediately? 06:23
Aankhen`` Guess that's pretty old. :-)
wolverian mugwump: should $foo = bar; when bar is known and well typed be a compile time error, if bar is incompatible with $foo?
mugwump yes, the error should be detected as early as possible if types are declared 06:24
wolverian I'm just wondering how that can happen when the sub can be swapped out at runtime.
Aankhen`` It's ye old question of: "should it be an error, or should I grab something to eat?" 06:25
mugwump well, if you're going to change method signatures incompatibly at runtime, you deserve any problems you get :) 06:26
wolverian Aankhen``: I want to have the cake and eat it too!
er. or something like that.
mugwump In fact that's an interesting case. In some cases it should be an exception to change the signature of a method that has already had that signature used to shortcut evaluation in any way
sdtr443w Ah well I have some more reading to do about regex and OOP. This stuff is all very overwhelming.
Aankhen`` wolverian >> Sorry, we're all out of cake... 06:27
I finished the last bit yesterday...
Would you like some soap instead?
mugwump The exception *could*, in theory, go back and compile the pieces of code that used its signature... 06:28
wolverian no, I _hate_ SOAP. I prefer REST. :)
mugwump which in that case would then give you an exception when you try to alter the method's signature
lumi_ There's blue ice cubes in the freezer
Aankhen`` Hmm.
Maybe we could work out something.
A compromise.
I'll, uh, give you SOAP, and you, uh, eat SOAP. 06:29
mugwump & # home...
QtPlatypus suggeses "use strict <types>"
wolverian I'm not sure if I see the compromise there...
Aankhen`` That's okay, you don't have to. :-D 06:30
That's the beauty of it. :-D
wolverian that only applies in a relationship.
Aankhen`` You just have to eat the SOAP -- even lazily, if you want! -- and I'll just point and laugh. :-D 06:31
QtPlatypus The Appocalype seems to be more permissive then I would like.
wolverian than 06:36
svnbot6 r4659, Aankhen++ | * libs/pugs/run.pod -- English tweaks. 06:54
nothingmuch goes to work 06:58
,
08:09 integral1 is now known as integral
wolverian hmm, nice. dan is still blogging a bit. 09:27
crysflame . 09:48
wolverian that's an interesting dot 09:52
chip I wonder what stopped him from working on the register allocator before he quit. 10:05
QtPlatypus ?eval "alphachars" ~~ m/<alpha>+/ 10:09
evalbot6 pugs: src/events.c:265: init_events_first: Assertion `rc == 0' failed.
QtPlatypus ?eval "alphachars" ~~ m/char/ 10:10
evalbot6 pugs: src/events.c:265: init_events_first: Assertion `rc == 0' failed.
wolverian chip: me too. 10:15
10:18 Steve_p_ is now known as Steve_p
QtPlatypus How do you say "Match alpha charicters" in perl6 rules. A05 says its <alpha> but PGE doesn't seem recognize it. 10:23
wolverian I don't think the builtin rules are implemented yet 10:28
mr_ank ?eval (1,1)+(1,1) 10:35
evalbot6 4.0
svnbot6 r4660, autrijus++ | * Emit.PIR - does not depend on Pugs.* anymore; Harrorth can just use it. 10:40
r4660, autrijus++ | * abstract syntax for preludePIR landed.
chip so a list in a scalar context is its length? I thought Perl 5 was the end of that 10:42
Juerd I understand that PIR generation is far from complete, but isn't the error message for pugs -C PIR -e'for(1..2){say}' not a little too verbose? 10:43
chip: Lists don't exist *at all* in scalar context. 10:44
chip: So the question I have to ask you is whether you mean "," in scalar context, or an array.
Doesn't matter much, because , in scalar context returns an array reference 10:45
An array in scalar context returns a reference to itself.
If you further specify scalar context, you get more interesting stuff out of it
String context will stringify (by default: join on whitespace) 10:46
Num context will evaluate to the number of arguments
Bool context will evaluate to "does it have elements?"
(Possibly by falling back to Num)
mr_ank ?eval ?() 10:47
evalbot6 bool::false
Juerd All the same is true for array references too, except for bare scalar context, because it is already a reference
mr_ank i don't know enough of perl 6 yet - i found it strange that it returned a 4.0 and not a 4
Juerd mr_ank: That's a pugs thing, not necessarily perl 6 10:48
mr_ank Why? Does it depend on an internal type? 10:49
(i.e., is it a bug we should write a test case for?) 10:55
wolverian ?eval ref 2+2 10:57
evalbot6 ::Int
wolverian ?eval 2+2
evalbot6 4
wolverian hmm.
?eval (1,1)+(1,1), ref (1,1)+(1,1)
evalbot6 (4.0, ::Num)
wolverian ?eval ref +(1,1)
evalbot6 ::Int
wolverian er. that looks weird. Int+Int -> Num? 10:58
Juerd doesn't understand the 4.
wolverian Juerd: why not?
Juerd Oh, I do.
wolverian okay.
Juerd The ones are misleading
(Damnit, use fantasy values!)
?eval (15,42)+(99,69)
evalbot6 4.0
mr_ank ?eval ()+1 10:59
Juerd Never ones, or 1, 2, 3, 4
evalbot6 1.0
Juerd 0, 1, 2 is bad too
wolverian Juerd: right. good point.
Juerd And, of course, this is better written as [15,42]+[99,69] anyway. 11:00
wolverian at least it confuses the hell out of perl5 coders.
well, they both do.
Juerd In fact, I wouldn't mind if Perl disallowed comma in scalar context.
It's great in void and list context
wolverian I like the fleeting List type :)
mr_ank (134,184,33)+314
?eval (134,184,33)+314
Juerd In scalar it's always better written as [LIST].
evalbot6 317.0 11:01
wolverian ?eval +[1,2,3] ~ "a"
evalbot6 '3a'
wolverian ?eval +[1,2,3]
Juerd 1, 2, 3 is misleading
evalbot6 3
wolverian yes. sorry.
Juerd Is the resulting '3' the number of elements, or the -1th element 11:02
?eval +[42, 15, 99]
evalbot6 3
wolverian ?eval [42,15,99]+[99,12,94]
evalbot6 6.0
wolverian okay, they work the same then.
Juerd Yes.
mr_ank yep
wolverian bogus Int+Int -> Num in both
Juerd Comma in scalar context returns an arrayref.
wolverian right.
Juerd And yes, pugs uses nums always iirc 11:03
wolverian in expressions you mean?
mr_ank I keep punching random bits in just to see what happens
wolverian (as in, results of them)
Juerd wolverian: Yes
I think it doesn't really know what ints are
wolverian okay. that explains it then :)
?eval +[49]
mr_ank thought so
wolverian er
evalbot6 1
wolverian ?eval ref +[49]
evalbot6 ::Int
wolverian it knows it up to that point, or at least pretends it does 11:04
Juerd Hm
mr_ank ?eval [34,134]
Juerd ?eval ref +1
evalbot6 [34, 134]
::Int
mr_ank hm
Juerd ?eval ~('x' xx 42) 11:05
mr_ank ?eval 33+7.0
evalbot6 'x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x'
40
Juerd ?eval ('x' xx 42)
evalbot6 ('x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x', 'x')
Juerd :)
?eval want
evalbot6 'Scalar (Any), LValue'
Juerd hm? not list context?
mr_ank weird
Juerd Then how can 'x' xx 42 be a list here?
mr_ank 33+7.0 --> no floating bit [list]+[list} --> floating? 11:06
Juerd mr_ank:
?eval 1+1
mr_ank ?eval 33.0+7.0
evalbot6 2
40
Juerd hm
o...kay.
Now I don't get it anymore.
?eval ref 1+1
evalbot6 ::Int
wolverian literals are optimised
apparently
?eval ref 3.0 11:07
evalbot6 ::Rat
mr_ank ?eval 3.1-0.1
wolverian well. at least in an expression!
evalbot6 3
Juerd ?eval ref eval("1") + eval("1")
evalbot6 ::Int
Juerd Rat?!~
s/~//
wolverian rational
Juerd Why not just Num?
does Rat, sure, but is?
wolverian I have no idea.
I don't know if pugs's numeric type system is specced at all
or perl6's, for that matter 11:08
?eval "1.0" + "4.0" 11:09
evalbot6 5.0
wolverian okay! 11:10
Juerd ?eval "1" + "4" 11:11
evalbot6 5.0
svnbot6 r4661, Aankhen++ | * added 'temp-ex*' and 'unlink-test*' to clean targets for Makefile.PL. 12:34
r4662, Aankhen++ | * lib/pugs/hack.pod -- small tweaks. 12:39
autrijus Juerd: pugs uses the scheme numeric tower. 12:43
Aankhen`` autrijus, got a minute? 12:45
svnbot6 r4663, andras++ | fake type methods
autrijus Aankhen``: yes 12:46
Aankhen`` svn.perl.org/perl6/pugs/trunk/docs/...rview.html
Could you look over the answer to "Can Pugs work with Perl 5..."?
autrijus sure
Aankhen`` (I modified it last night, just want to make sure that it's correct)
autrijus Aankhen``: they're fine. 12:47
thanks!
Aankhen`` OK.
NP. :-D
autrijus also please note that pugs can already interface hs-plugins
as well as dynload haskell modules
and eval_haskell
Aankhen`` Alright.
autrijus so the question below it
can be changed as well
Aankhen`` I was thinking of adding that, but I'm not very familiar with the Haskell side of it, so I held back.
Aankhen`` goes to update it. 12:48
hs-plugins basically allows Pugs to use Haskell libraries, right? 12:49
autrijus right.
Aankhen`` OK.
Limbic_Region wonders if there is some cosmic force that intentionally attempts to take a very focused thread off on tangents 12:50
Aankhen`` Can I expand "dynload" to "dynamically load" without losing accuracy?
dkf L~R: Entropy, baby! 12:51
Limbic_Region More likely - people want to push their own agenda and take any opportunity to do so even if the originator of the thread has made their intention to stay focused very clear
Aankhen`` &eval_haskell == &eval('...', :lang<haskell>)?
autrijus nod 12:53
Aankhen``: yes
Aankhen`` Committed. 12:54
s/ed/ing/
s/ing/ed/ now. :-) 12:55
Hmm... would it be appropriate to add "What is the relationship between Pugs and Parrot?" with the answer including the fact that Pugs can/will target Parrot? 12:56
svnbot6 r4664, Aankhen++ | * docs/01Overview.html -- updated "Can Pugs work with Haskell libraries?".
autrijus Aankhen``: yes. make it "can" 13:00
"already"
"and is improving steadily"
Aankhen`` OK.
I'll do it in a sec.
autrijus is working on getting svnbot.p6 compiled.
Aankhen`` Just tweaking the HTML.
autrijus it's so nice that it doesn't use OO :) 13:01
or rather, it uses OO without using "class" keyword
Aankhen`` Heh. 13:02
Shall I add the relationship question after "How can Pugs help the Perl 6 language design?"? 13:05
Limbic_Region Aankhen`` - 1. By helping flesh out corner cases 13:06
2. By exposing unspecced areas
svnbot6 r4665, Aankhen++ | * docs/01Overview.html -- HTML tweaks.
Aankhen`` Limbic_Region >> I'll add that, but what I'm asking is where to add "What is the relationship between Pugs and Parrot?".
Limbic_Region sorry - misunderstood 13:07
was helping answer the proposed new question
Aankhen`` No problem, those are good points to add. :-)
Limbic_Region++
I have to say, that file has some very strange formatting. 13:08
Limbic_Region well - the 3rd I am thinking about is hard to boil down
Aankhen`` I understand the 80 chars wrapping thing, but I don't understand the seemingly random 60 chars wrapping thing.
Limbic_Region but basically writing examples of "real" code helps decision makers make sane decisions
Aankhen`` Well, not 60, more like 70.
Hmm.
"By providing concrete examples for the language designers to use in the decision making process"? 13:09
autrijus it's Nvu.
Aankhen`` autrijus >> Ah, I see.
autrijus I wonder if we should dig out 01Overview.kwid/.pod
and just modify that instead.
that's prolly Good Idea (tm).
Aankhen`` Dig out as in find a buried file or convert this one? 13:10
Limbic_Region Aankhen`` I like how you phrased it - I might only add "real world"
Aankhen`` Limbic_Region >> A'right.
autrijus search.cpan.org/src/AUTRIJUS/Perl6-.../PA01.kwid
Aankhen``: buried file... you can trace a bit and replay the html changes (not many) to this 13:11
Aankhen`` Hmm.
autrijus and replace it, in the source tree, as 01Overview.kwid
I think.
Aankhen`` OK.
autrijus danke.
Limbic_Region autrijus - hit chip for us - he didn't journal up yesterday
;-)
Aankhen`` I shall try. :-)
autrijus chip: Hit!
autrijus hits chip! autrijus kicks chip! --More--
autrijus casts a "summon journal" at chip. You hear distant rumbling... --More-- 13:12
Aankhen`` LMAO. 13:13
Where can I find a brief listing of Kwid syntax?
Limbic_Region the masses demand blood err writing
Aankhen`` Just need something to refer to if I get confused.
Limbic_Region where's ingy
stevan violence at the hackathon!!!
in a street fight, who do you think would win? the dog with the squished up nose, or the bird with the colorful feathers? 13:14
Limbic_Region I think the dog eats the bird whole, the bird starts pecking away at the dog from the inside, and we end up with a mutant 13:15
stevan that sounds about right
throw an AST or 2 in there, and we got a hackathon
QtPlatypus Could someone confirm to me that it makes no sence for 0-nary subs to be defined as (post|pre)fix: ? 13:20
svnbot6 r4666, Aankhen++ | * docs/01Overview.html -- tiny language tweaks.
Limbic_Region I am not sure I am following QtPlatypus
autrijus Aankhen``: p6bible s26 13:21
Limbic_Region if you don't have any arguments - that doesn't mean you don't need to check something before doing what you are doing
QtPlatypus multi sub postfix:<silly> () {...}
Limbic_Region no - I know what you mean
I just don't understand why you think it doesn't make any sense
the arguments aren't the only thing a sub may need to check before doing work right? 13:22
or am I missing the boat here?
QtPlatypus Well isn't that going to act in exactly the same mannor as
multi sub silly () {...}
Limbic_Region I guess I need to read up on postfix: and prefix: then 13:23
woah - do I feel dumb 13:26
QtPlatypus - not at all what I was thinking they were - I thought you were talking about sub wrappers
Aankhen`` autrijus >> Ah, thanks. 13:27
Limbic_Region I agree, though obviously you shouldn't take just my word for it, that 0-ary subs don't make sense in (post|pre)fix:
QtPlatypus Limbic_Region: I'm constenstently feeling dumb.
Limbic_Region fwiw - I was thinking about PRE/POST/FIRST/LAST/NEXT/KEEP/UNDO stuff 13:29
QtPlatypus Arh
Limbic_Region where the PRE/POST ne prefix: postfix:
Aankhen`` Considering the whole TMTOWTDI thing, and how Perl offers choices of pattern matching grammar, documentation syntax, and even the language itself, with macros, Perl 6 seems to be less of a language and more of a meta-language. 13:30
s/Perl offers/Perl 6 offers/
q[merlyn] waits for Limbic_Region 13:37
gaal|work hola 13:50
lumi Hi Gaal 13:52
gaal|work hey lumi
was kinda swamped with $work lately, what's up? what's preludePIR i see in the svnlog? 13:53
Aankhen`` I've added PA01.kwid. Will work some more on it later (I've chosen a few sections to work on). Going out for a movie. & 13:54
svnbot6 r4667, Aankhen++ | * readded and updated docs/PA01.kwid
gaal|work sub "&say" in Emit.PIR is wrong for a Perl 6 &say, i think? is it meant to be a demo? 13:58
svnbot6 r4668, iblech++ | * Usual svn props added to examples/typefunctions/. 14:19
r4668, iblech++ | * Not so usual svn props (charset: iso-8859-15) added to t/pugsbugs/hyper_latin1.t.
r4668, iblech++ | * New test: macro postfix:<!> (t/macros/operator_macros.t).
r4669, iblech++ | * Added "*.hi" to the svn:ignore of src/Emit and added the usual svn props to
r4669, iblech++ | docs/PA01.kwid.
r4669, iblech++ | * Renamed examples/typefunctions to examples/unitfunctions.
r4669, iblech++ | * Fixed "!~" to mean "not(... ~~ ...)", instead of "ne".
r4669, iblech++ | * Moved now succeeding t/pugsbugs/inverted_smartmatch.t out of pugsbugs/.
Boogie_ iblech: ping 14:22
iblech Boogie_: pong
gaal|work iblech, iso-8859-15 is latin-9, not latin1 :) 14:23
Boogie_ iblech: why did you renamed typefunctions? it's not a problem, I just interested why do you think it's a better name?
iblech: maybe the right name is fakemethods, but I'm not sure 14:24
iblech Boogie_: Sure: Many folks here associate type with the types like Int or (map :: (a -> b) -> [a] -> [b]). Units is not ambiguous 14:25
gaal|work: Oh, yes, of course, fixing
svnbot6 r4670, iblech++ | Fixef svn:mime-type of t/pugsbugs/hyper_latin1.t, gaal++. 14:26
Boogie_ iblech: when I named it, I used type in this way. 5 is an Int, and 5.bytes is a function for the the Int. 10.minutes is the same with the type time (what is an Int) 14:27
iblech: but I know, that it's not the best name for it. :) 14:28
nothingmuch evening
Boogie_ iblech: fakemethods is maybe better, it looks like methods for the type Int, but just sample functions that can be called with Ints 14:29
iblech Boogie_: Yep, but, if I understand the code correctly, the key of your examples is not a function minuts which has type Int -> Time, but the unit conversions -- minutes to hours, etc. But I maybe I'
Boogie_: Yep, but, if I understand the code correctly, the key of your examples is not a function minuts which has type Int -> Time, but the unit conversions -- minutes to hours, etc. But I maybe I've complete misunderstood your code :)
Boogie_: Like fakemethods, too
eric256 i can't seem to find that unitfunctions directory or file.... 14:30
Boogie_ iblech: oh, now I got the meaning of "unit". :) I thought it the Pascal "unit" way, which as I remembers is like a module...
iblech ETOOMANYMEANINGS :) 14:31
Boogie_ iblech: My examples are about presenting, that you can extend types, and the subject of the examples are unit convesions. :)) Like you don't call the dir of hangman hangman, you call it games. Maybe not the best example why I gave that name for it. 14:33
yes, toooooo many meanings. :) 14:34
while just unit conversions inside, I think the name is ok, and better than the initial one. ;)
iblech Boogie_: Hm, then maybe we should rename it to fakemethods and add a non-unit example :) 14:35
eric256 where is the file Boogie_? i can't seem to find it at all
Boogie_ eric256: examples/unitfunctions 14:36
eric256 thats what i thought, but svn isn't bringing it to me...hmm
Boogie_ eric256: examples/typefunctions? 14:37
iblech: do you have any idea for a non-unit example? :)
eric256 neither where there... svn was choking on one file and stoping with what apparently was a fatal error.
iblech Boogie_: Hm... maybe "some_string".escape(:html)? 14:38
eric256 "some string".words;
Boogie_ iblech & eric256: good idea. do you have time for it? 14:39
iblech Boogie_: I'll do .escape(:html)
Boogie_ cool. :)
steveq How about Base64? 14:40
eric256 i will. but i'm off to work for now. ;) back in a bit
steveq is there/could there be a method to do that?
Boogie_ steveq: I like this idea, too. I think not yet a Perl6 module for it, but you can use the Perl5 module. 14:41
steveq yeah, I often have to load the MIME::Base64 module to do that encoding
kungfuftr hhhmmm p6 VMethods
Boogie_ I'm going back to work, too. ;) 14:42
kungfuftr: yes! 'virtual method' is the right expression for it. I wasn't able to remember for it.
kungfuftr Boogie_: yar, is what Template Toolkit calls it 14:43
svnbot6 r4671, iblech++ | Added escape example to unitfunctions/. 14:49
Boogie_ iblech: nice! 14:56
I love Perl6... :)
iblech :)
Ok, should we rename it to fakemethods of virtualmethods now?
or virtual_methods
Boogie_ virtual_methods looks fine, yes. 14:57
iblech Ok. Will you do the svn mv or should I? 14:58
kungfuftr vmethods 14:59
!!! 15:00
iblech I'm fine with that, too
kungfuftr =0) 15:01
iblech renames it to virtual_methods now 15:06
Boogie_ boogie thinks he was faster. :) 15:09
svnbot6 r4672, andras++ | rename unitfunctions to vmethods
Boogie_ anyway: thanks for the help 15:15
svnbot6 r4673, iblech++ | Added examples/output/vmethods/escape and modified example.t accordingly so it
r4673, iblech++ | tests escape.p6.
Boogie_ iblech: what does it mean: "Usual svn props added to examples/typefunctions/"? 15:16
iblech See docs/other/development_tips
steveq svn proplist foo
iblech These are Subversion properties so svn converts "\n" to the linebreak of your current OS
Boogie_ ah, ok. I'll check the doc. 15:17
autrijus nested expressions works! 15:20
now moving on to user subs and named params
chip nested expressions .. you mean like (1+2)*3?
autrijus chip: right 15:21
chip ok
autrijus sooner or later I'll need help on transcribing the entire Prim.hs into the prim table in Emit.PIR
chip darn it, everything is working but threading. this is soooo weird
autrijus i.e. having a one-to-one correspondence between our ops and parrot ones
I'll check in now if peopl want to take a look.
iblech++ # virtual methods is darned cool
10.minutes.ago == time()-600 15:22
10.minutes.until(x) == x-600
iblech Boogie_++ # They're Boogie_'s files :)
autrijus ok :)
iblech BTW, autrijus, are there any reasons that we don't create a src/parrot/prelude.pir, similar to Prelude.pm? (Just wondering) 15:24
chip OK, I'm really not liking this 10.iminutes thing
autrijus iblech: well, the prelude pir isn't written in pir 15:28
svnbot6 r4674, autrijus++ | * nested expressions and function calls (no `is lazy` support yet)
r4674, autrijus++ | * a much saner preludePIR syntax.
r4675, iblech++ | * t/pugsbugs/hyper_latin1.t -- Use \ddd notation so we can safe the file as
r4675, iblech++ | utf-8 (so we don't confuse editors and my add-the-usual-svn-props script).
r4675, iblech++ | * Added a new pugsbugs test: t/pugsbugs/assuming_and_mmd.t.
autrijus chip: it's not builtin
iblech: programs written in PIR, at this moment, are almost guaranteed to net you segfaults.
;)
so I write in the "PIR" data structure sublanguage in haskell 15:29
that implements a segfaultless, typed, well-formed subset of PIR.
iblech: so unlike Prelude.pm, the prelude pir is actually in haskell.
gaal|work autrijus, except as an optimization, why do we need to write pir ops in pir at all?
autrijus iblech: look at the preludePIR for examples
gaal|work oh
hmm
autrijus gaal|work: there's no optimisation involved.
iblech autrijus: Ah, makes sense :)
autrijus gaal|work: we need to have our Prims in a form parrot can understand. 15:30
that is just no avoiding it :)
just like Prim.hs
gaal|work problem is that they do less than the (current) prims
autrijus right, but the current prims is of no use to parrot
unless you statically link ghc or something
or always embed parrot and foreign export them
gaal|work say $*ERR: for example works in the main prim
autrijus both ways aren't happen
to be interesting enough
gaal|work right
autrijus hmmmmmmm
actually no. 15:31
there is a chance.
to just use Prim.hs in embedparrot case.
gaal|work okay, this is serious incentive to push more and more to Prelude.pm :)
autrijus that will nest us a faster runloop
but nothing more
and is harder to maintain
so perhaps not a good idea after all.
gaal|work *nod*
autrijus gaal|work: right. whatever in Purelude.pm is shared.
gaal|work: so Prims are truly Prims
and Prelude is Prelude.
(the apparent tautology notwithstanding) 15:32
gaal|work heh, except that *most* of what's there now at least relies on very specific semantics from the internals.
autrijus gaal|work: we can recreate them at pir level.
i.e. all Pugs::Internals::* should be carried to pir.
in the P::I namespace
gaal|work what was it that i saw in the channel logs a while ago? ETOOMANYLANGUAGES :-) 15:33
but i agree we do need this
autrijus gaal|work: perl6
Odin-LAP Hm.
chip FUCK FUCK FUCK 15:34
Odin-LAP What'd that be, teleported to p6 exceptions?
chip DISTCLEAN ME HARDER
gaal|work how do you write Pugs::Internals::openFile in perl6? what actually opens the file?
Odin-LAP chip: O_o
chip OK, parrot's distclean just DELETED MY TODO AND MY OTHER SIDE DOCUMENTS
Odin-LAP Ouch.
gaal|work chip, what fs?
"FAT has to be better for something" :-) 15:35
Limbic_Region chip - linux?
perlmonks.org/index.pl?node_id=106709 # restoring deleted files on linux
Limbic_Region was literally just reading that moments ago
gaal|work grep -500 "my data" /dev/hda1 ? 15:36
whaddya know, essentially yes. 15:37
in any case you want to remount ro asap 15:38
chip ext3
it's too late
crysflame ow smashy
chip autrijus: I think it's time to test your memeory 15:40
autrijus: whatever I said was on my TODO list is now toast
autrijus ok...
problem is you didn't tell me anything about your todo list.
autrijus tries to imagine 15:41
chip it was all the things you said you wanted. single-quoted strings in PIR, := vs. =, that sort of thing
gaal|work item #1: fix "make distclean" target
autrijus oh!
actually, those two, yes, but you said you'd like to fix the parsing too, like not relying on the [a;b;c] to parse different things
.namespace [P0]
nothingmuch autrijus: buahahaha
-- Forth CPS ;-)
execute ((Call word):exps) = execute ((body word) ++ exps)
Odin-LAP gaal|work: "What, the kid already fell down the well. No need to board it *now*."
autrijus chip: but for me... '' and := are the two big ones 15:42
nothingmuch: right... that _is_ CPS
nothingmuch: Emit.PIR is ready for you to target, btw.
nothingmuch: I'm still trying to get svnbot working
nothingmuch how much will it keep me from learning PIR, btw?
autrijus nothingmuch: you need to learn an idealised form of PIR :)
nothingmuch: just look at Emit.PIR.preludePIR 15:43
nothingmuch =)
autrijus I'll populate it with lots of other prims
to get svnbot and mandel working
nothingmuch okies ;-)
autrijus and then you can help me :D and target that.
nothingmuch anybody got a nice PIR tutorial?
well, you can target forth
;-)
autrijus nothingmuch: right... just give me the INFIX_LEFT_ANGLE_BRACKET_STAR_RIGHT_ANGLE_BRACKET word! 15:44
oh and I want MMD too
;)
chip nothingmuch: anyone who knows PIR wants to talk about it
s/anyone/no one/ # so much for that joke
nothingmuch dan had a wonderful blog topic some time ago.. something like 'forth love if honk then'
autrijus that's lovely.
nothingmuch: but in all seriousness, parrot/imcc/docs/ is not actively misleading.
Boogie_ gaal|work: ETOOMANYLANGUAGES means, that we totally misunderstood each other with iblech. He mean units like minutes, etc., I mean units like Pascal's modules. And ecetera, we did it with some words. :)
autrijus nothingmuch: and parrot/docs/compiler_faq is not all wrong either. 15:45
nothingmuch okay, i'll take a look
first I need to document this progress
autrijus cool. you r0x0rz.
right, that has priority
nothingmuch dances the r0x0rz dance
autrijus I count on your tutorial to teach other yapc hackathoners haskell
gaal|work homewardses & 15:46
nothingmuch hurrah
chip That's my old like
line
nothingmuch btw, does Interp and Eval belong in AST now?
chip YOU FORTH LOVE IF HONK THEN ... Forth bumper sticker
nothingmuch otherwise I need recursive modules
autrijus nothingmuch: you can factor them to AST, yes.
gaal|work Boogie_ - oh. i thought it was a babelfish fatigue error.
autrijus nothingmuch: or create a new Types and shuffle them there.
gaal|work anyway, i'm outta here, see y'all!
nothingmuch ciao gaal_ 15:47
autrijus gaal|work: ciao
chip YOU SPEAK YODA SPEAK LIKE IF FORTH LANGUAGE FAVORITE IS THEN
nothingmuch Aankhen``: heads up, editing work soon
autrijus chip: *lol*
SamB that doesn't look right to me... I thought there was some bracket usage in the words for conditionals...
chip parentheses are an infix plot
nothingmuch you speak yoda =
oh wait, I get it
damn, forth *IS* confusing 15:48
chip it's a mystery of the INNER CIRCLE!! GET IT?!?!
nothingmuch no, i missed that =/
chip links to timecube.com and wonders why the world has more than four time zones 15:49
Odin-LAP Forth. It's lisp, the other way around without all the parentheses? :D
SamB not exactly. 15:53
svnbot6 r4676, andras++ | vmethods: some example, sample outputs, etc.
r4677, andras++ | vmethods: extra comments
r4678, iblech++ | Added the usual svn props to andras++'s new examples/vmethods outputs and 15:58
r4678, iblech++ | updated example.t so the examples are actually run.
chip Lisp programmers work with grand concepts. Forth programmers work with a 4K stack. 16:02
Odin-LAP Hehe. 16:06
chip And if your brain is shaped like Chuck Moore's, that's OK with you. 16:08
geoffb Chuck Moore's brain apparently being shaped like a 16x16 square with pretty colors . . . . 17:03
Questions about last night's (my timezone) scrollback:
iblech: would you mind putting your svnprop-adding script in util/ ? 17:04
iblech geoffb: It's a two-line sh-script :)
geoffb: But yes, will do so in ten minutes or so 17:05
geoffb chip: Will parrot (or more accurately Perl6 on parrot) be doing the crazy crap that P5 does to try to downgrade numbers whenever you operate on them (trying to fit an NV into a UV or IV)?
That just can't be good for performance on modern branch-hating processors 17:06
iblech, thanks
putter autrijus: Could you give a one-line description of <-- .- <-& .& -->? Thanks. 17:12
geoffb sometimes svnbot6 is ripping fast, and sometimes it takes quite a while . . . though I suppose it could be receiving brain surgery from autrijus 17:16
svnbot6 r4679, geoffb++ | Added description entry for examples/vmethods to pugs::hack 17:17
geoffb speak of the devil
iblech geoffb: That's because the readline() of the socket to the IRC server is blocking. That means: As long as the bot doesn't receive something from IRC, it can't check for new revisions. 17:18
geoffb: You can force an immediate check for new revisions by /msg svnbot6 ?check (this will cause readline() to return and svnbot to check for new revs) 17:19
autrijus 17:22
svnbot6 r4680, iblech++ | Added util/add-svn-props.sh, per geoffb++'s request.
geoffb iblech, very interesting! 17:23
autrijus, you were trying to say something?
svnbot6 r4681, iblech++ | pugs::hack -- Filled in description of examples/p6explain. 17:24
iblech food &
geoffb does svk pull just to see that description
Aankhen`` Back! 17:27
autrijus >> Should I delete docs/01Overview.html?
Support for `where` would be really useful for cases like that escape vmethod example. :-) 17:29
Hrm... are any Pugs binaries available apart from the Windows binaries included in PXPerl? 17:33
gaal vmethods/README says that in a file called "bytes.p6", 1.megabyte is 1024. that can't be right :) 17:36
chip geoffb: You are SO looking at the small potatoes. Branch-hating CPUs will die of apoplexy when running Parrot. 17:41
svnbot6 r4682, gaal++ | fix README's idea of a megabyte
r4683, Aankhen++ | * PA01: updated bit about binaries.
geoffb chip: LOL
Well, fair enough
I only thought of it because of the early-morning discussion about Pugs numbers varying almost randomly between Int, Rat, and Num 17:42
chip Mm. Parrot numbers will convert either according to the PMC's whim or according to the compiler's exact request.
e.g. $I0 = $N0 or vice versa 17:43
geoffb And I remember just *retching* looking at the morphing code in perl . . . which I only discovered because that morphing code exposed some bugs in SDL_perl, which expected exactly NVs in a few cases, and did bad things when an NV "promoted" itself to IV
fair enough 17:44
chip well, SDL will be a bunch of specialized PMCs in Parrot, and I think it'll be happy 17:45
geoffb As long as it won't happen "behind my back", at least when writing raw PIR, I'm fine. And Perl 6 damn well better not try to change type under me when I specify a lowercase type . . . . 17:46
chip, yeah, I'm wondering how OpenGL will be supported for Parrot's SDL
Aankhen`` On the subject of wondering, I'm still wondering how a language can be parsed and compiled by a compiler written in itself... the mind boggles. :-) 17:52
arcady Aankhen``, ever heard of gcc? : ) 17:55
Aankhen`` Yes, and GHC... 17:56
I know that these things exist.
It's just one of those things that defies comprehension until you take a closer look. :-)
(perhaps a much closer look :-P)
Hmm... what backends are available for Pugs? 17:57
Oh, wait.
svnbot6 r4684, putter++ | Added some 0-ary primatives to Emit/PIR.hs.
Aankhen`` checks the svn log. 17:58
In case it's of any interest to anyone: I'm going out of town Friday morning, will be back on the 3rd of July, and cheerfully resume Pugs hacking in the next few days. :-) 17:59
iblech You'll be be away for approx. half a month? We'll miss you... 18:02
Aankhen`` iblech++ # flattery will get you everywhere ;-)
iblech But it doesn't work at my school's director (I wanted to attend APW and the hackaton, but as we write tests ATM, I had to stay...) 18:04
putter "Welcome back Aankhen``! Pugs achieved sentience last tuesday and..." 18:05
Aankhen`` Awww. :-(
putter >> I wouldn't be in the least surprised. ;-)
iblech >> I guess he's not quite as malleable as me... 18:06
cognominal what is the purpose of calling a directory "t e s t" (with spaces)? 18:13
arcady about the same as calling a program "oo, ick" 18:14
to annoy people
putter cognominal: breaking my find scripts?
cognominal I know how to use "find -print 0¨ and "xargs -0" but I don see the purpose
except to gratuitouly annoy unix people 18:15
PerlJam cognominal: What is the purpose of calling a directory "test" or "foo" or "lib"? Whoever named that one "t e s t" probably had a reason that made sense to them.
gaal it was to test filenames with spaces, actually.
PerlJam Off hand, I'd guess that they wanted it to stand out a little more on thier terminal. 18:16
svnbot6 r4685, putter++ | Added int registers to Emit/PIR.hs.
cognominal a fxg='find -type f -print0 | xargs -0 grep -n'
a fxgi='find -type f -print0 | xargs -0 grep -ni'
PerlJam See? gaal gave good reason
cognominal that should to the trick for me
gaal though arguably the tests can generate that directory.
(arguably they should. there's not much much to argue that they *can* :) 18:17
cognominal next is using control chars...
gaal heh 18:18
utf8
cognominal I am probably too old skool
gaal ^G^G^G
or, as screen would put it, "Wuff! Wuff!" 18:19
cognominal now, I have forgotten what I searched in the first place :( 18:20
svnbot6 r4686, Aankhen++ | docs/PA01.kwid: * thorough edit. 18:25
r4686, Aankhen++ | * added bit about writing tests under "I have learned some Perl 6..."
r4686, Aankhen++ | * added link to porting HOWTO under "I know Perl 5 but not Haskell..."
r4687, Aankhen++ | * PA01: updated "Which compiler backends...". 18:32
Darren_Duncan question: what is the proper way to declare constructor methods, namely new(), so they get invoked? 18:35
case in point, the classes in my Locale::KeyedText module
the new() used to be invoked a few weeks back, but now most of my current test failures seem to be that they aren't
its like some default constructor is being invoked instead
is what I'm doing out of date, or is Pugs broken? 18:36
Aankhen`` Could someone look over the "Which compiler backends..." question and verify the accuracy of the answer?
Darren_Duncan the version in the repository now (since yesterday) is the newest 18:37
Aankhen`` Darren_Duncan >> I'm not able to answer whether Pugs is broken, but in general, it's better to use C<submethod BUILD ($foo, $bar) { $.foo = $foo; .perform_some_initialisation; }> and so on.
Darren_Duncan the signiture is: method new( $class: Str $msg_key, Str ?%msg_vars ) returns Locale::KeyedText::Message {
Aankhen`` (atleast, that's my understanding) 18:38
svnbot6 r4688, putter++ | Added lit Double to Emit/PIR.hs.
Darren_Duncan so 'submethod' is what you invoke on a class name, whereas 'method' is invoked on an object?
will try something like that 18:39
Aankhen`` No, see, when you do C<my $foo = Foo.new()>, the C<new> method of class Foo will return an object, and then C<BUILDALL> will be called on that object, which in turn will call C<BUILD>.
So you already have an empty object, you just need to fill in the blanks.
Once again, that's my understanding of it. I have been known to be wrong. :-) 18:40
Take a look at the Tree and Set modules.
And HTTP::Headers.
Darren_Duncan will do
putter autrijus: ping? 18:41
Darren_Duncan I'm getting something ... 18:43
Aankhen`` Goodie. ^_^
svnbot6 r4689, Aankhen++ | * PA01: added "What is the relationship between Pugs and Parrot?".
cognominal are the tests backend agnostic?
gaal they're all interpreted, cognominal. 18:48
autrijus putter: pong
putter: please add comment / haddock for me
<--
svnbot6 r4690, autrijus++ | * Emit.PIR - move `&bool::*` into the "bool" namespace.
r4690, autrijus++ | * Lexical pads and assignments. This now works:
r4690, autrijus++ | my $x = $x = $x = 123; say $x;
autrijus is calling opcode 18:49
that returns a value
.-
is calling opcode without a rv
<-&
is calling a user-defined function
that returns a list of values
assigning to lhs
.&
is calling user-defined function
without a rv.
clkao autrijus: hey! dont fix bug just by half!
autrijus clkao: huh? 18:50
Aankhen`` autrijus >> Are you still busy?
clkao autrijus: svk ls -f doesn't have depotname part anymore
Aankhen`` I mean, are you still working on Emit.PIR?
(I need review for a modification in PA01 :-)
autrijus clkao: oh. sorry. 18:51
Aankhen``: yes.
Aankhen`` OK. 18:52
I shall ask another time, then. :-)
autrijus Aankhen``: mv it to 01Overview.kwid
when you have a bit time :)
Aankhen`` Alright.
cognominal how can one see the code emitted by pugs for a given backend?
putter autrijus: re haddoc, will do
autrijus Aankhen``: also the bootstrap plan needs to be revised in light of hw2005.tex
putter: danke
cognominal: ./pugs -CPIR -e 'say 123'
cat dump.ast
Aankhen`` cognominal >> `pugs -C backend foo.p6`, `cat dump.ast`?
autrijus ditto for -CPugs etc
Aankhen`` autrijus >> I will revise it tomorrow morning.
autrijus andynu: danke! 18:53
cognominal thx
Aankhen`` (I'd do it right now, but I'm guessing I'd have to actually grok hw2005.txt, which would require more mental power than I possess at the moment)
Also, should I delete 01Overview.html?
Aankhen`` goes to do something brainless. 18:54
UT2004, here I come...
&
autrijus Aankhen``: don't delete it just now... :) 18:55
svnbot6 r4691, Aankhen++ | * moved docs/PA01.kwid to docs/01Overview.kwid 18:59
putter autrijus: fyi, I'm getting a type error on the new/current PIR.hs. 19:10
autrijus !.
really. what's it?
pasteling "putter" at 66.30.119.55 pasted "type error" (16 lines, 672B) at sial.org/pbot/11096 19:11
autrijus it worksforme. you don't have local additions?
putter I'll try a clean build.
autrijus fwiw, my line 105 doesn't look like that. 19:13
so you mean want to resync.
my l105is
funC <- compile fun
putter Thanks. 19:14
autrijus np :)
thank _you_
gaal nice meta line.
autrijus purrs 19:15
autrijus trying to think how to express C-style loops.
say... if you don't have c-style loops and have to reinvent it in perl6
svnbot6 r4692, putter++ | Added comment to Emit/PIR.hs.
autrijus how will you do it?
goto label?
and if you don't have goto label?
autrijus ponders some more. 19:16
integral you have while loops?
autrijus currently I have nothing... all I have is function calls that get tail optimised automagically
but if you see a need for a fundamental while loop 19:17
I can add it too
basically I'm thinking in term of the primitive semantics
integral tail optimised function call == goto
autrijus you need letrec for that
oh wait. no, there's Y.
autrijus forgot about Y.
so maybe we Y everything?
gaal Y not. 19:18
autrijus but Y...
integral this is in PIR?
autrijus integral: this is in PugsAST
the conceptual Perl6 ast
integral there isn't a &?CURRENT_FUNCTION equivalent?
autrijus has nothing to do with parse tree
nothing to do with pir either.
raw perl 6 semantic ast
integral oh, ok, you're deciding on basic perl6 primitives, I get it
autrijus yup.
which is something nobody had thought about until now :p 19:19
(but I see it as essential to get retargetable compilation)
I should look at PyPy's tree again.
autrijus decides to look at what pypy considers as prim 19:21
putter autrijus: What would pop look like? This: sub "&pop" [tempPMC] [tempPMC <-- "pop" $ [tempPMC]] --> [tempPMC] not being it. Once there are a few more examples it will be easier for folks to add things. 19:29
autrijus putter: right. 19:30
putter: let me write pop etc
putter ok. Thanks. ;)
Darren_Duncan ?eval class Foo { submethod BUILD( Str $bar ) { say "bar is '$bar'" } } my $baz = Foo.new( 'hello' ); 19:35
evalbot6 Error: No compatible subroutine found: "&say"
autrijus my @a = (1,2,3); say (pop @a);
that should be a good example.
putter Thanks! 19:36
Darren_Duncan ?eval class Foo { submethod BUILD( Str $bar ) { say "bar is '$bar'"; } } my $baz = Foo.new( 'hello' );
evalbot6 Error: No compatible subroutine found: "&say"
Darren_Duncan what I get for output on my Pugs is: bar is '' 19:37
what is the problem here? 19:38
putter Re primatives, one might do a kernel language... (like Oz) 19:41
Boogie should something like this have to work? $num % all(2..$num/2) != 0; 19:47
svnbot6 r4693, andras++ | vmethods: add some math
autrijus putter: right... what's in Pugs.Compile.PIR as PAST
will end up as Pugs's kenrel lang
putter ah 19:48
autrijus so need to keep the nodes very narrow
Darren_Duncan ?eval my $z = 13; class Foo { submethod BUILD( Str $bar ) { $z = 26; } } my $baz = Foo.new( 'hello' ); say "z is '$z'"
evalbot6 Error: No compatible subroutine found: "&say" 19:49
autrijus in order for Eval to be fast and for retargetting other VMs.
(and Ms)
Darren_Duncan this is getting frustrating
autrijus Darren_Duncan: er. "say" is disabled in safe mode
Aankhen`` ?eval my $z = 13; class Foo { submethod BUILD( Str $bar ) { $z = 26; } } my $baz = Foo.new( 'hello' ); "z is '$z'"
evalbot6 'z is \'26\''
autrijus Darren_Duncan: so I think you'll be frustrated to no end.
Darren_Duncan how do we debug then?
autrijus Darren_Duncan: like Aankhen`` did 19:50
(or run it on feather)
Darren_Duncan uhuh
just a minute ...
?eval class Foo { submethod BUILD( Str $bar ) { "bar is '$bar'"; } } my $baz = Foo.new( 'hello' );
evalbot6 \{obj:Foo}
autrijus that sounds correct.
Boogie is there somebody can answer my junctions question? ;) 19:51
Aankhen`` Darren_Duncan >> C<BUILD> doesn't return the value...
autrijus ?eval my @out; class Foo { submethod BUILD( Str $bar ) { push @out, "bar is '$bar'"; } } my $baz = Foo.new( 'hello' ); @out
Darren_Duncan class Foo { submethod BUILD( $bon: Str $bar ) { "bar is '$bar'"; } } my $baz = Foo.new( 'hello' );
evalbot6 ['bar is \'\'']
autrijus ?eval my @out; class Foo { submethod BUILD( Str $bar ) { push @out, "bar is '$bar'"; } } my $baz = Foo.new( bar => 'hello' ); @out
evalbot6 ['bar is \'hello\'']
Aankhen`` Also, C<BUILD> expects named parameters.
autrijus I think it's named only.
Aankhen`` So, you need to do it like autrijus did.
autrijus all *%_ is.
Aankhen`` submethod (*%_) { ... }
submethod BUILD (*%_) { ... } 19:52
Darren_Duncan just a minute ...
PerlJam From whence does this idea that BUILD needs named parameters come?
Aankhen`` PerlJam >> One of the Apocalypses.
Hmm, according to S12, it could take regular positional parameters too... 19:53
autrijus oh?
bug!
PerlJam indeed
autrijus write test!
Aankhen`` OK.
autrijus I completely missed that paragraph.
PerlJam Aankhen``++
autrijus but that's kind of usual with S12.
;)
Darren_Duncan FYI, what I *really* want to do is wrap the built-in new() that checks its input arguments first and causes new() to return undef instead of an object if they are bad ... how do I do this?
Aankhen`` autrijus >> It's not specifically mentioned. 19:54
However, there's an example.
Submethods are for declaring infrastructural methods that shouldn't be inherited by subclasses, such as initializers:
submethod BUILD ($arg) {
autrijus Darren_Duncan: well you can do your own new and call to super new
Darren_Duncan I did that before (checked in) but it wasn't being executed
autrijus ?eval class Foo { sub new { 123} } Foo.new
evalbot6 123
Darren_Duncan see the Locale-KeyedText I have checked in now ... maybe you can tell me why new() isn't being executed 19:55
run test LKT_20_MsgProps.t
autrijus ?eval class Foo { method new { 123} } Foo.new
evalbot6 123
autrijus you mean _10 19:56
ok
Darren_Duncan yes _10
autrijus Locale::KeyedText.new_message should do undef
is that the idea?
Darren_Duncan the first call should, the rest shouuldn't 19:57
in the test file
autrijus ?eval class Foo { method new { return } } Foo.new
evalbot6 undef
Darren_Duncan maybe it has something to do with my class name having a :: in it
autrijus probably not
Darren_Duncan or that I say 'use' in a different file
the Locale::KeyedText::Message.new() gives a "can't find" error without the preceeding :: 19:58
svnbot6 r4694, Aankhen++ | * added bug test for positional parameters not being passed on to the BUILD submethod.
Darren_Duncan I put 'say' inside new_message and Message.new ... the first was printed, the second wasn't ... 19:59
suggesting that something like SUPER::new was being executed rather than my new() ...
since an object is always returned having empty properties
autrijus ?eval class Foo { method new { return } } ::Foo.new
evalbot6 undef
autrijus ?eval ::Foo.new(1); class Foo { method new ($x) { return } } 20:00
evalbot6 undef
Aankhen`` ?eval class Foo { method new { return "foo" } } ::Foo.new
evalbot6 'foo'
autrijus weird.
Darren_Duncan: hm, weird indeed.
Darren_Duncan did you run my test and can you see the problem?
autrijus I did run your test and I can't easily see the problem. 20:01
:-(
Darren_Duncan do the tests all say 'ok' then?
putter pugs: internal error: startTask: Can't create new task Please report this as a bug to [email@hidden.address]
Aankhen`` Darren_Duncan >> Where is the test?
Darren_Duncan see /ext/Locale-KeyedText/t/
autrijus Darren_Duncan: no, 12 failed 20:02
putter: er?
Aankhen`` Darren_Duncan >> I see only LKT_20_TrnProps.t, not MsgProps.t.
Darren_Duncan see 10_MsgProps
Aankhen`` OK.
Darren_Duncan I get 12 failed too, but I haven't been able to figure out why 20:03
Aankhen`` C:\Development\pugs\ext\Locale-KeyedText\t>pugs LKT_10_MsgProps.t
pugs: *** Can't locate t_LKT_Util.pm in @*INC (@*INC contains: t/lib ext/Locale-
KeyedText/t/lib C:\Development\Perl6\lib C:\Development\Perl6\lib C:\Development
\Perl6\site\lib C:\Development\Perl6\site\lib .).
Darren_Duncan I put a 'say' at the start of my new() and it doesn't print anything
autrijus Darren_Duncan: right, because your new() is not being dispatched to.
Darren_Duncan so you confirmed that
the question is why
autrijus Darren_Duncan: move the entire class to before the declaration and drop the ::?
yeah the question is why since it's not a mmd.
Darren_Duncan will try that 20:04
er, move what before what? 20:05
autrijus move the entire class
before the other class that calls its new
Darren_Duncan just a minute
thanks, that improved things it seems 20:06
autrijus cool! 20:07
putter autrijus: re "er?", I went to add a say() to evalbot which raises a "sorry, no"... but the following made ghc go boom:
Darren_Duncan still other problems, but that seems to have helped ... will know soon
please keep watching this channel 20:08
putter EVALBOT_CODE='42' EVALBOT_TMPFILE=./deleteme EVALBOT_PUGS=/.../pugs perl -w evalhelper.p5
autrijus Darren_Duncan: sure
Aankhen`` autrijus, in the meantime, could you look over something for me to verify its accuracy? :-) 20:10
autrijus Aankhen``: sure.. ?
Aankhen`` rt.openfoundry.org/Foundry/Project/...d?rev=4691
"What is the relationship between Pugs and Parrot?"
And "Which compiler backends are available? Will more be added?"
autrijus Aankhen``: ok... change Parrot AST to Parrot PIR. 20:11
Aankhen`` Ahh!
One sec.
autrijus because there's no such thing as Parrot AST :)
Aankhen`` Yeh, I was a little confused about that.
autrijus otherwise fine. 20:12
putter Ahem. Could someone on a non-x86_64 machine try the following:
EVALBOT_CODE='42' EVALBOT_TMPFILE=/tmp/deleteme EVALBOT_PUGS=`pwd`/pugs perl -w examples/network/evalbot/evalhelper.p5; cat /tmp/deleteme
Aankhen`` OK.
Er.
Sorry
putter Thanks. I get a ghc error.
Aankhen`` .
autrijus I'll brb. 20:13
Aankhen`` Um, I'm on Windows... should I still test?
Boogie I solved the prime question. But just another. Is :by() supported by Pugs? If yes, how should I use it? 20:14
Aankhen`` Apparently not. No BSD::Resource. :-P
putter Aankhen``: Probably not worth the... yes, trouble. Thanks anyway. 20:15
Aankhen`` Heh.
NP.
svnbot6 r4695, Aankhen++ | * docs/01Overview.kwid: s:g/Parrot AST/Parrot PIR/.
Aankhen`` Boogie >> It doesn't seem to be supported. 20:18
Reading A03, it seems to be an adverbial trait for the range operator, so you'd probably use it like so: 1..10 :by(2)
Boogie OK, thx. It was my opinion, too, but not worked. 20:19
Aankhen`` C<..>'s full name is C<< infix:<..>:(Int, Int) >>, so you can always use C<< infix:<..>($int, $other_int, :by($step)) >> to test. 20:20
Boogie Aankhen``, I think it's alien language to me... ;) 20:22
Aankhen`` Heh.
autrijus alien as in space aliens.
pasteling "Aankhen``" at 61.246.21.218 pasted "Testing :by" (7 lines, 226B) at sial.org/pbot/11104 20:23
Boogie repeat: it's chinese to me. :)
Aankhen`` Bah. :-P
putter Sigh. And ghc head isn't even building at the moment. Punt. 20:24
Aankhen`` I could try explaining if you want to understand. :-)
putter Hmm, unless there is a snapshot...
Boogie oh, not your paste code is "alien" to me. I check it.
Aankhen`` Ah.
I wish learning new languages was as easy as "push @brain, $lang". 20:25
autrijus you think CPU thinks it's easy? ;)
Aankhen`` Hmm, but then again, that wouldn't be any fun.
Hehe.
autrijus indeed.
Aankhen`` OK, I have a new wish. 20:26
autrijus For what do you wish? --More--
Boogie Ok, I understand it. So it seems, that it's not working yet. ;)
autrijus Boogie: that would be Totally Correct.
Aankhen`` I wish taking time out to do stuff was as easy as "push @day, $hours;".
autrijus Boogie: I don't even know how to parse it ;)
?eval 3 :by(3)
evalbot6 Error: unexpected ":" expecting operator, postfix conditional, postfix loop, postfix iteration, ";" or end of input
Aankhen`` It has to be used on a range. 20:27
Something like:
autrijus ?eval 3,:by(3)
evalbot6 (3, ('by' => 3))
Aankhen`` ?eval 1..10:by(3)
evalbot6 Error: unexpected ":" expecting digit, fraction, exponent, term postfix, operator, "does", "but", "is", "^..^", "cmp", "<=>", "^..", "..^", "..", postfix conditional, postfix loop, postfix iteration, ";" or end of input
autrijus right... what I'm not getting is
does all infix get that?
Aankhen`` Parse error...?
autrijus no
adverb
Boogie 1..10:by(3) is 1, 4, 7, 10
autrijus 100 * 200 :modulo(10)
Aankhen`` I didn't understand what you said. :-P
Oh, wait.
I think only range gets :by. 20:28
autrijus sure
but does all infix get potentially adverbs.
Aankhen`` Doesn't make sense to have 1+5:by(3).
autrijus is what I'm asking :)
Aankhen`` Ah.
Yes, AFAIK.
autrijus ok, that makes it significantly easier.
Aankhen`` Any operator could get an adverbial modifier.
But please wait.
I am checking the Synposes.
autrijus also, an adverbial? 20:29
what about multiple?
Boogie S09 has a short info about it.
Aankhen`` Ah.
As a unary prefix operator, you may also apply adverbs:
for *$handle :prompt('$ ') { say $_ + 1 }
I think all operators can get adverbial modifiers. 20:30
(if not all functions)
autrijus cool. write tests then :)
Boogie And it should work, too: 1...:by(2)
autrijus for userdefined too.
Aankhen`` I'll do that, just going to read some more to make sure I grok it.
autrijus Aankhen++
Darren_Duncan I committed what I have so far so we're all on the same page 20:35
autrijus yay. "1=2" is now a compile time error in the compile monad.
autrijus is happy that more and more things are catched/warned/strictness-controlled at compile level 20:36
integral autrijus++
Darren_Duncan I agree
autrijus integral: I found that GADTs partially obsoletes metacomputation. 20:37
because data types are now as good as functions.
so it can guide the dynam monad while being fully reified.
integral wow, that sounds good
autrijus integral: Pugs.Compile.PIR's PAST structure is that
it guides the underlying Trans monad
instance Translate (PAST a) a where
...
svnbot6 r4696, Darren_Duncan++ | Locale-KeyedText updates : moved 'Locale::KeyedText' class declaration so it is the last one rather than the first; this is so that Message.new() and Translator.new() actually get executed; also fixed a syntax error in Translator.new()
integral hmm, so is AST.Internals looking any better now? 20:38
20:38 rjbs_ is now known as rjbs
autrijus integral: it will, once I switch the current Eval 20:38
to eval PAST (the syntax tree) instead of Exp (the parse tree)
integral sounds good :-) autrijus++
Aankhen`` ?eval 1..10[0,1]
autrijus we can finally do away with this "evaluate the parse tree" mess!
evalbot6 (1, 2)
Aankhen`` ?eval 1...[0,1] 20:39
evalbot6 Error: unexpected "[" expecting operator, postfix conditional, postfix loop, postfix iteration, ";" or end of input
Aankhen`` ?eval (1...)[0,1]
evalbot6 [1, 2]
integral it's a pity all these cool ideas have a "hack lots and lots and lots of code" stage =)
autrijus integral: nah, you just have to spec them down.
it's a nice byproduct that our specification language, which is called Haskell, is also executable 20:40
;)
Aankhen`` ?eval my @foo = (1...); @foo[0,1]
evalbot6 pugs: out of memory (requested 1048576 bytes)
integral hmm, maybe, but you can't get away from the "convert everything in Eval.hs at once" thing
Aankhen`` Hmm.
How does that work?
integral Aankhen``: afaik it's because = for an array is eager 20:41
Aankhen`` Can I use lazy { (1...) }?
Darren_Duncan currently debugging a problem where an array arg seems to always contain exactly 1 element ...
affecting 20_TrnProps ...
Aankhen`` Bleh, that doesn't work.
Boogie oh_my_god: ?none(2..sqrt($value)).is_factor_of($value); :D
Aankhen`` ?eval 1...[0,1] 20:42
evalbot6 Error: unexpected "[" expecting operator, postfix conditional, postfix loop, postfix iteration, ";" or end of input
Aankhen`` ?eval (1...)[0,1]
evalbot6 [1, 2]
Aankhen`` That works, but assigning it to an array doesn't. :-(
autrijus integral: I can :) 20:45
integral: by having a PAST backdoor to Exp.
integral: so what's not compiled gets interpreted.
integral 8-) ok, that's scary :) 20:46
autrijus PRaw Exp 20:48
just like that :)
so have two runloops
I've got it all worked out :D
putter ghc bug report sent. pop task_stack. 20:50
svnbot6 r4697, andras++ | vmethods: more more math
Aankhen`` autrijus >> I will submit the test in the morning. 20:51
My mother's telling me to go sleep.
G'night.
autrijus Aankhen``: you rock. go sleep.
ntie!
arcady too bad "for 1... -> $i { last if $i > 10 }" doesn't work 20:58
putter I propose to add the following line to evalbot.p6 evalhelper(): 21:02
$code = 'sub say(*@a){die("Cant say() in evalbot")};' + $code;
But I can't really test it.
autrijus putter: just commit ahead then.
we'll fix when it's broken 21:03
putter Ok. Tnx.
Shillo Hullo, folks!
arcady except you don't use '+' for string concatenation!
autrijus right.
Shillo Hey, just occured to me to check whether I remember an apocalypse correctly...
What exactly is the left side of := supposed to be?
I think I remember that := really imitates parameter binding in a function call. 21:04
Or... ?
autrijus Shillo: that's right.
however it's poorly implemented at the moment.
Shillo nods. 'what I wanted to ask next.
(+$x, +$y) = (:x<foo>, :y<bar>) should parse? 21:05
autrijus yes, but currently not
Shillo noddles. 21:06
arcady well, it does parse... as something... something not very correct 21:07
Shillo looks into writing a test, just for fun....
chip what does that + mean? 21:12
svnbot6 r4698, putter++ | evalbot.p6 now says it cant say().
r4699, putter++ | evalbot.p6 tweak.
r4700, putter++ | evalbot.p6 another tweak. Sigh.
chip oh dear, that's 'named parameter' +?
?eval (+$x) := (:x<1>) 21:13
evalbot6 Error: Cannot bind this as lhs: Pos (MkPos "<eval>" 1 2 1 5) (App (Var "&prefix:+") Nothing [Var "$x"])
chip jeepers
Shillo chip: I think := in pugs currently just behaves like =
That's kinda bleh, I so liked [foo, bar, *baz] = list idiom that I added it in the one scripting language that was *fully* subjected to my design hackery. :) 21:14
chip: Er, I mean, just /parses/ like = 21:15
putter , in wait state for PIR.hs examples, ponders what to do next.
Aankhen``++ # perl6 as "meta-language" 21:17
autrijus: Is there any community you wish was more involved in pugs development? (so many languages, so little time) 21:19
svnbot6 r4701, Darren_Duncan++ | Locale-KeyedText update : changed all LKT_20_TrnProps.t tests that passed array ref args so they pass parenthesized lists instead; previously each Translator array argument was always getting a single argument that was an Array
autrijus putter: elaborate? 21:24
Shillo ($a, @rest) = (1,2,3) <-- works
autrijus is hunting down a rather weird bug in the Writer monad's "pass" function
Shillo ($a, *@rest) = (1,2,3) <-- works but @rest ends up empty. Bug?
nothingmuch fucking ISP 21:25
every 25 minutes the connection goes down
autrijus Shillo: slurpy star at assignment lhs is compile time error
ideally
that only makes sense in rvalue context.
but becaus pugs's Eval has no notion of compile time error
it ends up being a runtime bogusness.
Shillo Duh! I meant :=... testing again...
($a, *@rest) := (1,2,3) <-- parse error 21:26
($a, @rest) := (1,2,3) <-- works but @rest == (2)
Hm, I think the latter is okay, actually.
autrijus right. 21:27
Shillo Former?
autrijus former is borken. 21:28
Darren_Duncan based on what I've seen so far with such as @foo arguments, I'm thinking of making my Perl 6 code look more Perl 5 ish and use a $foo argument instead, which would hold an Array reference ... then invoking said function with ['a','b','c'] should do what I expect 21:32
and for the near future it is important that my Perl 5 and Perl 6 versions of the same modules can be used in the same way 21:33
so I'll probably undo my last commit as part of this change 21:34
anyway, have to leave comp for awhile and will attend to this later ... have a good day y'all!
fyi, in the meantime, the checked-in LKT_20_TrnProps.t currently passes all tests 21:35
autrijus woot
Darren_Duncan++
Darren_Duncan I'm still having problems with hashes though, which affects 10_MsgProps 21:36
20_TrnProps is arrays by contrast
Shillo commits the tests for the conversation I just had. :) 21:37
svnbot6 r4702, shillo++ | tests: lhs of := is a parameter list (tested keywords and the slurpy)
Darren_Duncan autrijus, I would appreciate it if you could improve Pugs so I can declare my classes in any order and calling new() would still work
autrijus Darren_Duncan: I am in fact working on that. 21:38
(compilation phase)
please be patient :)
21:40 khisanth_ is now known as Khisanth
autrijus putter: I checked in some more pir prims. 21:43
Darren_Duncan great ... anyway, back in while
autrijus Darren_Duncan: the thing is that currently there's only parse time and runtime
so any info (like classes) not available immediately in parse time
is lost.
Darren_Duncan yes
autrijus so another compiler phase will fix this.
Darren_Duncan okay
this probably goes without saying, but I would like the flexability I have in Perl 5 with most reasonable things 21:44
autrijus right.
me too.
<- heavily invested in perl5 :)
Darren_Duncan it also occurs to me that as I've continued hacking LKT for Perl 6, its looking more like perl 5 as I go along 21:45
autrijus nod.
Darren_Duncan so I'm not really using that many Perl-6-added features
autrijus because pugs is currently implementing very few p6 oo features.
that will also improve with time :)
Darren_Duncan question: how long are you online today?
how much longer that is
autrijus Darren_Duncan: not much... 21:46
<.5hr
I hope.
Darren_Duncan so I thought
Shillo autrijus: Hey, ye still in the neibourhood? :)
autrijus Shillo: you mean austria?
Darren_Duncan so I'll work on my own the rest of the day and discuss with you again tomorrow, for the most part
Shillo autrijus: Yep!
autrijus Darren_Duncan: cool. thanks
Shillo: yes
Shillo Heh... been doing 1 trip to Vianna/month till about half a year ago (project at work). 21:47
Darren_Duncan that said, I will try to confirm a fix for the new() thing while you are online now if it is committed soon ... otherwise tomorrow
just give the word ... 21:48
svnbot6 r4703, autrijus++ | * Sequentialised even more evaluation order. This now works:
r4703, autrijus++ | my $x = 9; say $x++; say $x
Shillo So it's neighbourhood. ;)
putter autrijus: re elaborate? Embedding languages in p6 is so easy, I had to start n of them. So I was trying to find a way to choose which to poke further. 21:51
autrijus: re pir, great. will putter. 21:52
Thanks, autrijus. 21:54
autrijus np :) 21:57
I'm feeling like Day 4 all over again.
autrijus just got substr() working
mugwump including lvalue ? 21:58
hi btw
autrijus hi mugwump.
no, rvalue only at the moment because parrot doesn't do lvalue substr natively.
mugwump on other subjects, I was wondering if there was some kind of model of PMCs that might be relevant to study for the MetaModel... 22:00
do you model PMCs at all or just target them? 22:01
autrijus just target them. 22:03
but we get to create our own PMC/ParrotObjects.
mugwump which source file is best to read? 22:07
Emit/PIR.hs?
autrijus mugwump: no, the PDDs 22:10
pdd15
mugwump oh cool 22:11
svnbot6 r4704, autrijus++ | * Add backdoors to raw Exp in PAST, and to raw Doc in PIR. 22:40
r4704, autrijus++ | This is to get the "compile if you can, interpret if you can't"
r4704, autrijus++ | semantic working for gradual transition of Eval.hs.
r4704, autrijus++ | * As an example, the old Parrot backend is now bridged for
r4704, autrijus++ | (Syn "loop") syntax; this means mandelbrot runs for free. ;)
r4704, autrijus++ | * For lexical names, generate a ".local" name instead of using
r4704, autrijus++ | fresh supply of registers. That will make lexical access
r4704, autrijus++ | potentially much faster (constant time) without going thru
r4704, autrijus++ | the fetch_lex code.
autrijus waves & 22:45
svnbot6 r4705, putter++ | Added abs() and prefix:- to Emit/PIR.hs. But - seems to be returning a ResizablePMCArray?!? 23:34
obra Where do the tests of synopses live these days?
putter Any parrot-familiar folk around? (I seek clarity on infix .MMD_foo) 23:54

Powered by Cro and the Raku® Programming Language.

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

Thank you!