6.2.7 released! | pugscode.org/ <Overview Journal Logs> | pugs.kwiki.org | paste: sial.org/pbot/perl6
Set by autrijus on 13 June 2005.
kelan ?eval 1+1 01:44
QtPlatypus Eval bot is dead kelan 01:45
kelan oh? 01:46
geoffb Re: the "usual SVN properties" -- I have another project that I have just moved to SVN, that should work on both *nix and Win32. Is there any reason not to just apply the same SVN props as we use here to basically every non-binary file? 02:34
mugwump sure, if you have mixed win32/unix devs and prefer utf-8 02:35
geoffb Meaning "sure, it's the right thing to do"?
mugwump Āŷĕ 02:36
geoffb OK, wheee
Now to update properties on 160 files . . . .
Darren_Duncan you should just use utf-8 everywhere regardless of platform 02:41
nothingmuch evening from the hackathon 06:23
nothingmuch_ hackathon networking has been restored 06:34
Darren_Duncan good evening to you too 06:44
checking journal ...
I noticed no svn commits since this morning or earlier 06:45
no journal ... maybe another time
good night
nothingmuch_ no internet up till now 07:10
autrijus already went to bed
which means the commit should start momentarily
07:10 nothingmuch_ is now known as nothingmuch
nothingmuch foo! 08:50
broquaint bar! 08:52
nothingmuch good night good people of the world 09:11
Juerd Is being able to init a hash with a hashref a bug? 09:26
It certainly ought to be, because a hashref is a valid hash key.
broquaint Is it not magically deref'ed in list context? 09:27
Juerd No
References may never deref in list context or generic scalar context 09:28
Dereffing can only happen with explicit use-as-array or use-as-hash
%hash = { 1 => 2 } should set %hash{ { 1 => 2 } } = undef 09:29
broquaint Fair enough. Sounds like a bug. 09:32
arcady I'm not so sure it's a bug 10:12
oh, wait, it is 10:15
%hash = [ 1, 2] also does the same thing
12:22 Nattfodd is now known as Odin, Odin is now known as Nattfodd
pdcawley ?eval sub foo { my $x = 1; return sub ($codestring) { eval $codestring } }; foo()("$x") 13:07
broquaint I think the evalbot is down atm. 13:08
pdcawley Yeah, I guessed that. Just tried it on feather instead. 13:09
broquaint This looks to work as expected: sub foo { my $x = 1; return sub ($codestring) { eval($codestring).perl.say } }; foo()(q[$x]) 13:10
Juerd What was expected?
broquaint To return the value of $x 13:11
Juerd I see.
I think it should by default not work
broquaint Why is that? 13:12
Juerd It kind of defeats the purpose of having a garbage collector.
broquaint ?
pdcawley I'd disagree, you know lexically that you've got an 'undecided' scope down there, so for that function you hang onto the entire lexical stack.
Juerd pdcawley: Because of the eval? Or always? 13:13
I can live with smart detection of things that eval
pdcawley Because of the eval.
Juerd But not if that is taken to extremes: calling another sub that evals, which could in theory try to access $CALLER::x
Although it will be weird to have code that starts working when you add eval ''; :) 13:14
I'm not sure what I think of this
The impact of deep analysis is huge, but it would be needed to avoid weirdness
pdcawley I'm not sure what GC has to do with anything tbh. 13:15
At least, not in that example.
Actually, no, tell a lie.
Juerd If variables are held onto, they cannot be destroyed.
pdcawley Until the instance of teh sub that's holding onto them is itself destroyed. 13:16
Juerd That is true, but that can take a very long time.
pdcawley And that's a problem to someone writing that kind of code because?
If you're writing that it's because it's precisely what you want. 13:17
Juerd Because my values are often 100 MB.
One little string eval used as a shortcut could severely fuck up memory usage for the kind of scripts I use.
I'm not sure I like the idea of big red flags with every single string eval, as in PHP.
pdcawley Juerd: But things like that are only going to be seriously long lived when you do &whatever := func_that_returns_a_func. 13:18
Remember, you don't start holding onto the stack until and unless you call foo.
Juerd Or put them in a hash for much later use. 13:19
Which I tend to do. I use anonymous subs all over the place. I like working with them.
pdcawley Anon subs with evals in them?
Juerd I admit that I don't usually combine these two things: string eval and storing closures
pdcawley And you're stashing them with side effects?
Juerd But I do use both, and often too, so it's bound to happen.
pdcawley How very impure of you.
Juerd Not always side effects 13:20
pdcawley Stashing them in a longlived hash smacks of side effects to me.
Juerd The problem is that Perl can't see that eval "$integer1 / $integer2" is in fact rather pure -- it has to account for $integer1 being a string that uses a variable.
pdcawley Yeah. Great isn't it? 13:21
Juerd No, not really.
pdcawley All htat means is that every sub in that sub's lexical chain (not dynamic) needs to close over everything.
Juerd Which may mean my 100 MB variable is closed over without any chance of it ever being needed. That's a terrible waste of memory. 13:22
pdcawley Then write your code more carefully.
Juerd In general, if extra memory can be used to gain some speed up, I'm all for it. But this is waste.
I don't always want to write code carefully. That is WHY I like Perl.
It lets me code carefully, when I want that. 13:23
But is still the useful fast tool if I don't.
I'm not in the "if you code sloppily, every bit of slowdown is your own fault" camp
pdcawley I'm not in the 'closing down useful possibilities in order to chase performance' camp either. 13:24
Juerd We don't have string eval to code carefully. It's there especially for the people who don't care much about purity, because they can sometimes cut a few corners and be home earlier. 13:25
pdcawley That's something the programmer should be doing, not the language.
Disagree, very strongly.
Juerd pdcawley: The problem with this is that analysis is expensive. You can't just look for eval, you also have to look for my_eval, that uses eval.
Which means that in practice, every single sub needs to have a flag that says wether it uses something that is flagged.
pdcawley But you know the lexical structure at compile time. This has nothing to do with the dynamic structure. 13:26
Juerd Which can result in numerous subs being tainted in the way that using it will cause the enclosing block to close over the entire lexical scope, which is a slowdown at very great distance.
This isn't a bug you could easily find even if you wanted to. 13:27
pdcawley What on earth do you mean 'using them'?
Juerd I think there should be a way to close over everything, but never this implicit.
pdcawley The only time it's going to crop up is when you *write* a function. 13:28
Juerd pdcawley: foo calls bar, which calls quux, which calls xyzzy, which calls Some::Module::blah, which calls Other::Module::whee, which calls YetAnother::Module::blergh, which evals a string.
pdcawley: Now foo is as eval, in that when it is used, perl must close over all variables in scope.
pdcawley: Because miles away, something evals, which could use $CALLER::CALLER::CALLER..::x
pdcawley And which has no effect on teh lexical state of bar, quux, xyzzy, Some::Module::blah, or Other::Module::whee. 13:29
Juerd Now, if I use foo in the closure that I return from a sub that also gets a huge variable, the variable is kept, even if it will never be necessary in YetAnotherModule::Blergh's eval
And all that because one module author of code I presumably neither wrote nor read cut a corner by using eval. 13:30
Good luck figuring out why your program performs so badly...
pdcawley To even begin to implement your straw man you have to solve the halting problem. That's not what I'm proposing.
Juerd I'm saying that if you want eval '$x' to work, you should somewhere draw the line, and that that line is CALLER. If you also want '$CALLER::x' to work, you lose a terrible amount of efficiency. 13:31
pdcawley Or you say "This is a language where we value that, turn it on for everything".
Juerd So the purity and predictability must be broken somewhere in order to avoid greater problems. 13:32
pdcawley Full speed ahead and damn the torpedoes.
Juerd pdcawley: I don't think enabling it for everything is an option even, if we want Perl to ever be recognised as a smart and efficient language.
So half-enabling and not-enabling are the things I'm personally trying to choose from, and I tend to choose not-enabling by default, with some option of half or fully enabling. 13:33
(If it were up to me, CALLER:: would be enterily illegal outside signatures) 13:34
pdcawley Argh. Actually, I used Perl as the pseudocode for something I'm arguing that Parrot must support; the CALLER thing should come fenced about with "Using this is as bad as using $& was in Perl 5 days!" warnings.
Juerd: I like it for weird introspective and macroy ideas.
Juerd I agree that CALLER is a problem. Still, it is specced, and we need to work around problems it creates. 13:35
pdcawley: Then if you need it, you can specify that in the signature :)
But never ever in any eval :)
pdcawley Illegal outside macros and signatures I could go with -- you can always write a macro tweaks the signature.
Juerd And CALLER::CALLER:: is just waaay to horrible to even think about, for me :)
pdcawley Right. Must. Write. Summary. 13:36
Juerd Have fun
svnbot6 r4917, iblech++ | Prelude::PIR -- Unbroke &sleep and &exit (was broken because of the recent 13:41
r4917, iblech++ | Pugs::Internals -> Perl6::Internals name change).
masak Juerd: I must admit that I'm on your side in this issue. I think the analysis required for it would be too hairy. 13:45
autrijus greetings lambdacamels! 13:47
pdcawley Hi autrijus.
autrijus we have network!
PerlJam greets autrijus
autrijus: who are "we"? :) 13:48
autrijus people who have network, of course
PerlJam patiently waits for someone to prod lwall into joining #perl6 13:49
autrijus lambdacamels, please check this for insanity
perlcabal.org/~autrijus/perl6-as-of-20050623.txt
before I rip the relevant parts of pugs out
(based on discussion with lwall last night -- but we didn't have network so I didn't get to upload it) 13:50
PerlJam autrijus: prefix:<=> is .shift ? That seems odd. 13:51
integral likes the bits he understands
autrijus PerlJam: it's a destructive update on an iterator
(in item context)
for =@array { ... } ; # @array is empty after this
integral Does the Any change mean that: sub foo (Any $a) { ... } no longer autothreads? 13:52
PerlJam And we use the slurpy variant when we don't want destruction?
autrijus er, no, slurpy context prefix:<=> is also destructive... 13:53
it's more like splice(@x, 0, -1)
but now splice also preserves generators
PerlJam So I have to somehow explicitly declare an Iterator (or just copy the thing) if I don't want destruction?
autrijus if you don't want destruction... drop the = :) 13:54
for @array { ... }
pdcawley I'm unsure about the my thing.
autrijus pdcawley: it's conforming to the spec
almost too literally. 13:55
PerlJam autrijus: oh duh. I was getting hung up.
pdcawley But it can be worked around I'm sure.
autrijus my $x; { say $x; $x += my($x) * $x if $x; }
here the first three $x (lexically) are the same 13:56
pdcawley was thinking of the classic sub is_odd { $^a == 1 || is_even($^a - 1) }; sub is_even {...} type thing.
autrijus and the last three $x are the same.
pdcawley Eew!
And bugger me but that's bad and evil code.
autrijus but the third $x is "erroneous"; we may or may not actually support it. if not supported, that'd give an error.
integral: right, sub foo ($a) now means 13:57
sub foo (Item $a)
and the top type Any is divided into Item and Junction
pdcawley has been thinking idly about dependency injection... I want to write UtilityClass.new in my 'client' code and have my application framework decide that, today, UtilityClass is actually implemented using ThisSpecificUtilityClass. 13:58
Which is tangential, and I'm supposed to be writing the summary.
integral keeps thinking of "Knit" when things like pdcawley's are mentioned
PerlJam I always think of summarys when pdcawleys are mentioned ;-) 13:59
Arathorn autrijus: in the my() example above, wouldn't the final $x be from the outer block's scope? 14:00
despite being physically placed at the end of the line, wouldn't the if-clause's condition be evaluated before the left-hand-side?
autrijus Arathorn: I'd expect that too
but larry sort of insisted on the strict lexicalness. 14:01
i.e. it doesn't have anything to do with eventual evaluation order
Arathorn but only the physical position in the file
that's diabolical, imo :\
autrijus it does make it easier to explain.
integral hmm, but what about: my $x = 5; { my $x = $x } ? Isn't a strict left-to-right reading mean that this doesn't work? 14:02
autrijus integral: yeah, you'd have to say $OUTER::x. that's in the spec
pdcawley Well put that man!
Ghod but that's ugly.
integral oh. That makes me sad. :-)
autrijus beginning of S04 14:03
but consider:
pdcawley my $x = 5; { $x ==> my $x } ?
autrijus pdcawley: amazingly that works
my $y = my($x) zor $x;
now zor's evaluation order is unspecified
maybe it evaluates the second argument first.
you don't know.
so that's the argument for strict lexicalness.
integral hmm, is the old rule of declaration only taking effect after the statement too confusing? 14:04
Arathorn are there any other pseudo-functions like my() whose behaviour would depend on their physical lexical position rather than their position in the parse-tree had they been a real function?
autrijus integral: it was my second alternative but larry struck that down as well... because following that rule you get unspecified intra-statement results 14:05
my $x if 0;
Arathorn: those are called scope declarators... currently in PIL there are only three special forms that is not function application:
assignment, binding, scoping
integral hmm, how does this new rule fix that?
PerlJam Are there any other languages that adhere to such strict lexicalness in declaration? 14:06
Arathorn right
autrijus PerlJam: actually, most languages do, where they are known as the let-forms
and they are bound strictly lexically without regard to the evaluation order also.
Khisanth hmm ugly but who in their right mind would write code like that? :) 14:07
pdcawley autrijus: Woo!
PIL sounds promising.
autrijus pdcawley: see src/Pugs/Compile.hs 14:08
line 30 to 51 contains the entire spec of pil (For this moment)
pdcawley I will. But not 'til I've finished the summary. 14:11
Arathorn hm, when you for @x -> $y {} # the $y is implicitly lexically scoped, right? 14:12
pdcawley BTW, do we have full continuations yet? I want to solve that sudoku.
Arathorn: It's a parameter in the block, and lexically scoped to that, yes.
Arathorn pdcawley: a sudoko solver using proper continuations would def be very nice evangelism example :)
right
autrijus pdcawley: let me see if i can compile it to PIR today 14:13
pdcawley Arathorn: It's a horribly inefficient brute force thing, but it should work.
Once the core is working I'll see about adding the shiny interface.
The code is gloriously declarative.
NonDeterministic programming makes me happy. 14:14
theorbtwo Greetings from the hackathon, BTW. 14:18
Juerd 15:51 * PerlJam patiently waits for someone to prod lwall into joining #perl6
PerlJam: TBH, I don't think that would be a great idea. Many ideas need to be discussed long before decision makers see them. 14:19
theorbtwo The connection here sucks, so don't expect it in the next three days.
stevan is shocked to see autrijus did not take his laptop into the shower 14:20
rjbs he did, but he's playing solitaire
theorbtwo Nah, it's on the table next to me.
ihb Juerd: "TBH"?
Khisanth to be honest 14:21
theorbtwo We managed to get through a lot of corneer cases last night, though.
ihb Khisanth: danke. 14:22
theorbtwo (Not that I actually understood many of them, but...)
Arathorn cana nyone point me to the right apoc/synopsis number for a discussion of how the -> bind operator thingy works?
Khisanth wonders how autrijus has been able to avoid electrocution with the laptop in the shower
Arathorn ('cos it don't seem to be in [AS]0[1-4])
theorbtwo It's defined to work just like sub, except when it doesn't/ 14:23
QtPlatypus I thought that -> was a sub
Arathorn ah, found it in S06
bingo - thanks :)
pdcawley my $a = 10; my $c = 30; my $b := $a; $a := $c; say $b; 14:24
What should that print out?
PerlJam 30 14:27
theorbtwo Yeah.
pdcawley Exactly. So Brent's wrong about 'alias' being spelled ':=' in perl 6.
svnbot6 r4918, iblech++ | Emit.PIR, Prelude::PIR -- &eval_parrot (but Parrot segfaults when trying to 14:30
r4918, iblech++ | invoke a sub compiled at runtime).
Juerd autrijus: How do levels work as in "0-level" for deref?
pdcawley Oh fsck! 132 more messages in p6l before I've done this summary. 14:31
autrijus Juerd: oh, it just means there's no autodeferencing anymore.
dot explicitly dereferences
Juerd autrijus: But there is for coercion, right?
theorbtwo (There's still autoENreferenceing on %.foo.)
autrijus bracketlikes eg $foo[0] is merely shorthand for $foo.[0]
Juerd ~$aref and ~@array
autrijus @foo.[0] auto-enreferences @foo into something.
Juerd That I know
autrijus so it'd be in effect the same as (\@foo).[0] 14:32
but if you have a reference to a number, you can't use it as a number.
Juerd postfix ops provide Array|Hash|Code context to their LHS
autrijus so basically we're back at p5 land.
Juerd Good!
autrijus theorbtwo was very happy too :) 14:33
Juerd Now, I do think there should be some postfix scalar dereffer, but that's for later :)
autrijus maybe we can hijack .val for that.
Juerd And I *still* think it can be just $, but people seem to be very opposed to that ;)
$scalarref$ :)
autrijus $Id$ $Date$
theorbtwo .val already does enough... $foo.deref sounds good.
Juerd So don't call your references that
PerlJam Juerd: $foo\ :-P 14:34
Juerd The exception we have to make for #! is bad enough already :)
Khisanth autrijus: so on that first one it would be $y.++?
Juerd PerlJam: I wish to reserve \ for infix operation.
theorbtwo We do? Isn't #! just a comment?
Juerd theorbtwo: Not if it's the first line 14:35
autrijus Khisanth: I don't know if it should work like that.
Juerd Shebang behaviour is emulated
It has to be for the -T exceptio
n
autrijus $$y++ # is the way I'd envision
Khisanth autrijus: I am kinda hoping it wouldn't :)
Juerd theorbtwo: Ooh! .deref 14:36
theorbtwo: A generic whateverref dereffer!
BUT
How it's not possible for it to be generic.
That's unfortunate, really.
theorbtwo isn't so sure of that. It'd... yeah, that's not possible.
Juerd So let's have syntax. Then the asymmetry isn't so blatantly obvious.
theorbtwo Which is unfornunate, because it means that we have to find a name where people don't expect it to be one. 14:37
PerlJam So is there going to be some syntax for the infinite deref?
Juerd perlcabal.org/~autrijus/perl6-as-of-20050623.txt ++ # I like much of it.
PerlJam: Yes, it can be postfix $ :)
theorbtwo There already is syntax, just not postfix syntax (for finate deref.) 14:38
Khisanth return is just another way to throw an exception? :)
Juerd And infinite is a problem :)
That takes too long.
theorbtwo Yep, Khis. A very unexceptional exception.
PerlJam theorbtwo: perhaps I should have said "is there some simple punctuation?" :-)
Juerd So... what is the current state of <->? 14:39
theorbtwo Juerd: I don't think anybody thinks it's useful enough.
PerlJam Juerd: it still only exists in your imagination :)
Juerd Grumble. 14:40
Khisanth the lightsaber operator? :)
theorbtwo So make one
Juerd Khisanth: No, that's ====================@##@####
hmm
Khisanth: No, that's 4====================@##@####
there.
theorbtwo: I want <-> to be the default for $_-blocks. 14:41
theorbtwo: rather than "-> $_ is rw"
Because that adds some weirdness to the defaulting thing
Khisanth ah a sith lightsaber at that!
Juerd Anywhere $_ is a default, adding $_ shouldn't break things.
theorbtwo We're trying to decide what to change it to right now.
Juerd "$_ is rw" as a default makes very little sense.
<-> solves this in a subtle way 14:42
theorbtwo -> ?$_ default $OUTER::_ seems to be the conclusion.
Juerd And makes possible my for %hash <-> $key, $value { ... }, which is still much easier to type than for %hash -> $key is rw, $value is rw { ... }
theorbtwo: Holy bovine. 14:43
theorbtwo OK, the conversation is getting aways from me.
Juerd There goes simplicity :(
PerlJam Juerd: I was going to say that <-> only seems really useful when you have multiple parameters that you want rw, but then theorbtwo's comment shocked me into silence. 14:44
Juerd PerlJam: It's even more useful there, and that's when it becomes tedious to write "is rw" instead
PerlJam Can we do for @foo -> ($a,$b,$c) is rw { ... } ?
Juerd PerlJam: I'm more concerned with the default of "-> $_ is rw", because I'm rather convinced the default should be "-> $_" or "<-> $_", not something involving an external property like "is rw" 14:45
PerlJam: Not currently.
PerlJam: I find for @foo <-> $a, $b, $c { ... } a hell of a lot easier on the eyes and fingers still.
Khisanth PerlJam: but wouldn't that mean the list and not the individual elements are rw?
Juerd The thing is that <-> is conceptually NOT "yet another way"
PerlJam for some reason <-> always makes me think of the "in", "out" and "inout" parameters in XS 14:46
Juerd It's like q() and q[].
Limbic_Region perlbot nopaste
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
Juerd PerlJam: I don't know what those are, but <-> does really mean things are bidirectional, and -> means they're unidirectional.
Khisanth -> readonly, <- write only, <-> read write? :) 14:47
pasteling "Limbic_Region" at 129.33.119.12 pasted "Could someone please run this code - and if it works, nopaste the results for me?" (76 lines, 2.7K) at sial.org/pbot/11293
Juerd Yes, except that write only has no practical use and can thus be skipped.
Especially because <- is qw(-, which is too dangerous a thing.
Limbic_Region the binary build Jonathan puts out is broke today (after I already wiped out my working copy of Pugs) - so I have no way of testing
but if someone wouldn't mind trying it for me I would be very appreciative
PerlJam Khisanth: or ... for @a -> $a,$b { ... } === for $a,$b <- @a { ... } :-)
Juerd <-> is acceptable because it makes one thing different, instead of an infinite number of things.
PerlJam: But that screams for $value <= $key too, which isn't going to work. 14:48
PerlJam Beware the hobgoblins of foolish symmetry ;) 14:49
Juerd Limbic_Region:
unexpected "{"
expecting operator, ",", ":", term postfix, postfix conditional, postfix loop, postfix iteration, ";" or "}"
at - line 33, column 17
chip "the boy who cried 'foolish consistency'" 14:50
Limbic_Region thanks Juerd - will investigate 14:51
PerlJam Limbic_Region: what's the top level when acting upon?
Limbic_Region: i.e., I see no given $result_1 { ... } 14:52
Limbic_Region PerlJam - I am still working on it - hard to debug without Pugs 14:53
Juerd 30 when 1 { return %ball{3} == %ball{10} ?? (5, 1) :: (3, -1); }
Not that it has to do with the problem at hand, but doesn't that need () around the comparison?
PerlJam Juerd: Why should it? 14:54
Juerd Oh, no
That was with = :)
Juerd always has a problem remembering ?:'s precedence
BASIC's iif() isn't so stupid, really :)
theorbtwo Juerd: you mean ??::... <g> 14:55
Juerd No
I used ?: to indicate that the problem is bigger than just Perl 6
theorbtwo Oh.
Limbic_Region btw - does adding in the given $result_1 { line make it work?
Juerd Great, I got a warning by SMS that the sun would be dangerously powerful today, and I read it way after the most dangerous times! 14:56
So much for avoiding skin cancer.
Limbic_Region I want to verify that it does prior to merging -1 and 1 into a single default case
PerlJam Limbic_Region: Can't you just use feather.perl6.nl to test this stuff?
Limbic_Region no - I can't get access to anything outside other than http and https 14:57
PerlJam Time to tunnel ssh over http ;)
Limbic_Region *shrug* - normally isn't a problem as I can download binary builds - but today's build is broke and I didn't bother to check that before wiping out a working one 14:58
PerlJam Limbic_Region: perhaps you should build a form interface to feather such that you can paste code snippets, have feather execute them and show you the results. 14:59
(assuming Juerd is okay with that)
Juerd If it's properly ulimited, sure 15:00
Limbic_Region well - I have to run to a meeting now anyway 15:01
Khisanth shouldn't line 9 in Limbic_Region's code result in a warning? or is that not implemented yet? :) 15:06
autrijus only compiler can give a warning -- pugs's main evaluator is not yet a compiler :) 15:07
Khisanth I just recall someone in #perl mentioning a -w for pugs yesterday 15:09
autrijus theorbtwo++ 15:29
svnbot6 r4919, autrijus++ | * make PIR backend respect @*INC.
r4920, theorbtwo++ | Change castV to fromVal for better monadic goodness.
r4920, theorbtwo++ | Oh, and it's ft, not FT, but m is already used, so it's still M.
geoffb Dammit people, why must all the good conversations happen when I'm asleep? 15:57
:-)
Good morning, all
16:01 nnunley__ is now known as nnunley
Limbic_Region geooffb - mind helping me troubleshoot something (I think you will find it a fun project) 16:06
geoffb sure, watcha got for me? 16:07
pasteling "Limbic_Region" at 129.33.119.12 pasted "If this runs without error - what are the results" (55 lines, 1.9K) at sial.org/pbot/11294 16:08
Limbic_Region it is a puzzle - I am presenting a solution in p6, but the challenge will to be to derive a solution using code (not your head)
geoffb (And I'll trade you -- I need people to test compiles of svn.openfoundry.org/sdlperl1/ , and let me know how it goes, since I only have compilers for Debian)
Limbic_Region well - you will have to wait until I get home then 16:09
geoffb running it . . . 16:10
Does it take a long time to run, normally?
oh wait, it dies
Limbic_Region dunno - never ran it 16:11
geoffb unexpected "{"
expecting operator, ",", ":", term postfix, postfix conditional, postfix loop, postfix iteration, ";" or "}"
at ./lr-thingie line 13, column 21
(that line is off by two, I added a #! and a blank line at the top)
Limbic_Region ahh - likely need parens around the infix operators
or maybe not
can we do /msg so that I don't flood the main channel with "try this" 16:12
integral geoffb: the Makefile.PL does nothing on OS X, not even complain.
geoffb oh sure, why not? :-)
integral, yeah, the makefile mess is something I gave off tackling until today -- yesterday I noticed that someone had created a Makefile.netbsd, and then never put an entry in Makefile.PL for it, so I added that myself, but I haven't done anything else to fix the mess. 16:13
It's definitely next up on the priority list . . . but if you'd like to contribute, I'd be VERY happy, since OS X is the one major thing SDL_Perl 1 didn't run on, I think. 16:14
integral doesn't even know if he's got SDL installed
theorbtwo Oh, not pug's makefile.pl. 16:15
geoffb theorbtwo, sorry for the confusion, no -- it just came up as a trade with L_R, and integral was his usual cool self. :-) 16:16
PerlJam pdcawley++ on transparent v. opaque references ... "I'll tell you what, let's swap places: you read the thread and write me a summary of it" 16:25
Limbic_Region should my $number = .1 * -1; # fail to parse 16:26
my $number = 0.1 * -1; # parses fine btw
PerlJam is catching up on his summaries
wolverian well, that'll disallow method names that are only digits, but oh well.. 16:27
PerlJam Limbic_Region: good question. I'd think it should parse fine.
wolverian: we're already there aren't we?
wolverian PerlJam: probably.
geoffb FWIW, it would annoy me to have to do 0.1 everywhere -- because I have a lot of code with values between 0 and 1, in arrays -- so that would actually be significant clutter for me 16:29
Limbic_Region well - it doesn't currently work in pugs so it is either a bug or needs a doc patch
autrijus it's a parsebug. tests welcome 16:30
Limbic_Region will wait until geoffb has finished helping me debug all the code so I can get them all at once 16:31
geoffb :-)
autrijus is still fighting with the fallout of theorbtwo's patch... 16:37
Limbic_Region thinks he has found another odd parse bug where the presence or absense of prior statement will make it work/break 16:57
autrijus - I don't think sial.org/pbot/11300 is small enough for a test case 16:58
but if you remove the line - say "balancing group A and B"; # it stops working 16:59
geoffb got tired of wondering what Sudoku was, and finally read www.nationmaster.com/encyclopedia/Sudoku . . . 17:40
pretty cool, that.
Khisanth latest checkout doesn't compile ... 17:43
theorbtwo Khisanth, autrijus is working on it. 17:44
Khisanth oh that is what he meant :) 17:45
theorbtwo Yup. 17:46
All my fault.
.oO(fmap!)
svnbot6 r4921, iblech++ | * Pugs.Eval -- Removed unneeded import. 17:52
r4921, iblech++ | * Test -- s/say/print/ -- now a failed test outputs "not ok n" again (was
r4921, iblech++ | "not \nok n").
r4921, iblech++ | * examples.t -- unTODOed a succeeding test.
Arathorn they're quite fun 18:04
geoffb finally catches up with backlog 18:12
geoffb praises the death of infinite auto-deref 18:13
nothingmuch wakes up
geoffb welcome to the land of the living, nothingmuch 18:14
obra hey luqui_. how's the hackathon? 18:30
luqui_ it's like a design meeting 18:31
but without a damian :-)
obra Heh.
luqui_ fun
obra Last night, Allison sold me on the Artistic 2.0. (Pointed out why most of my concerns weren't actually an issue) 18:32
18:32 khisanth_ is now known as Khisanth
Limbic_Region geoffb - see perlmonks.org/index.pl?node_id=469482 18:32
luqui_ knows nothing of legalness 18:33
Limbic_Region actually, everyone is welcome to try the challenge - sorry about the broken p6 code but it seems I have uncovered several bugs in pugs
obra luqui_: I'm a bit jealous ;)
luqui_ just barely skimmed over artistic 2.0
Limbic_Region does anyone else have a Win32 binary build of Pugs I can download?
luqui - did you get a chance to see my follow up email? 18:34
luqui_ limbic_region, yeah, but I forgot what you asked
and the net is sloooow here so I'd rather not look
Limbic_Region ok - will regurgitate here if you don't mind 18:35
while $ref() -> @array { ... } # doesn't currently work in Pugs
you indicated that if $ref() returned a list, it would become an array, and then bound to the block parameter 18:36
luqui_ yeah, we made that decision pretty quickly
Limbic_Region that's what I want to happen and I will return an empty list so that it knows when I am done
luqui_ and it's not implemented
Limbic_Region using undef won't work - must be empty list
geoffb sorry, l_r, was AFK. reading back ...
luqui_ but for such a thing, you're probably better off using an iterator 18:37
Limbic_Region it is an iterator
the $ref() returns a list until it is done at which point it returns an empty list
anyway - moving on
you said while condition -> ?$x {
luqui_ oh, but you're not using the perl iterator interface
Limbic_Region right 18:38
PerlJam oh yeah, that reminds me ... Limbic_Region++ for the iterators article
Limbic_Region p5 -> p6 with as little translation as possible
that last part while condition -> ?$x { confused me for a couple of reasons
#1 shouldn't that be ?@x and not ?$x
second - why use ? (assuming that means it is optional)
luqui_ I was just outlining our decision about while's topicalization nature
Limbic_Region ok 18:39
luqui_ it had little to do with your question, but your question triggered the decision
Darren_Duncan fyi, 4291 fails to compile ; problem is in src/Pugs/AST/Internals.hs:538:13
Limbic_Region gotcha
so back to my question then
when the bound array is empty - the while loop finishes?
luqui_ Darren, we're aware of that :-(
Darren_Duncan okay
luqui_ well... 18:40
your undef thing doesn't quite work I think
basically the loop is finished whenever the return evaluates false
Limbic_Region it shouldn't be undef
undef is a single element resulting in an array of 1 element and hence true
luqui_ in scalar context
Limbic_Region that's why I was explicit about returning an empty list
empty meaning 0 elements 18:41
luqui_ yeah, but the single element might not turn itself into an array ref in scalar context
Limbic_Region grrrr
if in my iterator I have a line like
luqui_ is asking
Limbic_Region return () if $finished;
I would expect that to know I want a list of 0 elements 18:42
nothingmuch_ "...that works?..." 18:43
=D
Limbic_Region which is different IMO from return undef if $finished or return if $finished;
PerlJam Limbic_Region: return [] if $finished; doesn't do it for you?
(what with the transparent reference semantics and all)
Limbic_Region PerlJam - I would think that would be non-intutive 18:44
geoffb is increasingly hating transparent (de/en)referencing
Limbic_Region you are returning an explicit list everywhere else but when you mean empty list you change to array ref
that's less dwimery
PerlJam "explicit list"? 18:45
Limbic_Region return (1, 2, 3);
PerlJam parens aren't list constructors.
Limbic_Region right
geoffb Plus, I can easily see people following the rule elsewhere and saying "that's a list containing one element, an empty arrayref"
luqui_ Limbic_Region, we think this will do the trick
return (,) if $finished
PerlJam Limbic_Region: yeah, I see your point though.
luqui_ (not sure that , needs to be there)
anyway
Limbic_Region hugs luqui_ and the rest of the hackathoners
luqui_ if you want to return a single element list that turns into an arrayref in scalar context, it's like this: 18:46
return (1,)
so return (0,) will evaluate true according to while
PerlJam why am I suddenly reminded of python?
luqui_ larry was too ;-)
PerlJam (because that's one of the pythonisms that bugs me the most! that's why)
luqui_ well, we've got to get through this issue somehow 18:47
Limbic_Region the return (); would be best but return (,) is acceptable
PerlJam right.
luqui_ return () might work
but return (1) does not
PerlJam I don't have any good answers
geoffb Fair enough . . . my requirement is basically that I have a single simple rule that will allow me to disambiguate () and {}, because otherwise I can't even imagine the bughunts trying to figure out what I really should be getting
Limbic_Region ok - so now about all the other bugs I found with my latest challenge
PerlJam return () is syntactically clear if you ask me.
Limbic_Region does anyone have a Win32 binary build of pugs available for download? 18:48
PerlJam luqui_: surely [1] works as (1,) ?
Limbic_Region would like to write tests for the things identified so far before he forgets them
geoffb Personally, while I really love most of Perl 6 so far . . . methinks Perl 5 did Just Fine (tm) with the semantics of return
luqui_ perljam, [1] is a listref even in list context
we want a flat list
basically, we want an X such that: 18:49
$x = X # listref
@x = X # no listref
18:49 luqui_ is now known as luqui
Limbic_Region believes luqui's time is better spent with the other people in the hackathon then answering his silly questions 18:50
PerlJam luqui: are you saying that if X were [1] that @x[0] would be [1] rather than 1 ?
luqui right
Limbic_Region now that I have a warm fuzzy that something sane will be done - I will stop whining
luqui :-) 18:51
luqui &
PerlJam I wonder what a hackathon sounds like.
Limbic_Region no one has a Win32 binary build available huh?
geoffb loud, I would think
luqui oh, jonathan was making those for a while 18:52
PerlJam geoffb: it's either loud or silent but always with keyboards clicking away :)
luqui I don't know much about it though
luqui fg
luqui &
Limbic_Region jonathan's build is broke ATM and hence the problem
autrijus Limbic_Region: I can build one for you but it'd take a very long while to upload it from the cell phone connection here 18:56
Limbic_Region heh 18:57
Limbic_Region hugs autrijus too for being willing to go the extra mile
I have written down the bugs well enough to hopefully be able to write tests later tonight
luqui hey autrijus, I can get parrot talking to pug 18:59
pugs
but I'm not sure how to get pugs's output into parrot
PerlJam what does a parrot say to a pug?
luqui eval?
autrijus luqui: I'm converging to minimize our location delta 19:01
Darren_Duncan eval? 1+1 19:05
?eval 1+1
Limbic_Region the evalbot's not on the channel ATM Darren_Duncan 19:08
svnbot6 r4922, autrijus++ | * clean up theorbtwo's vCast patch. I expect most tests
r4922, autrijus++ | to still break... :)
Limbic_Region but I guess you figured that out already
PerlJam Is it too transitioning to the hackathon? 19:09
Darren_Duncan yes, it hasn't been for several days now 19:14
r4922 compiles successfully for me 19:31
arcady pugs' ~~ doesn't do some of the things that it's supposed to 21:01
42 ~~ 40..45 returns false, where S04 says it should be true
since matching a number against a range tells if the number is in the range 21:02
theorbtwo arcady: Find a test, or write one? 21:03
arcady aha, there are in fact tests 21:04
I think what I really need to do is learn Haskell and fix all these things 21:06
if only I had the time for that...
autrijus using a pair variable to bind into named params finally works 21:18
Limbic_Region autrijus++ 21:19
svnbot6 r4923, autrijus++ | * After extensive discussion with lwall, the named binding semantics
r4923, autrijus++ | is now implemented as an inferencing phase before binding:
r4923, autrijus++ | sub foo ($x) { ... }
r4923, autrijus++ | sub bar { (x => 3) }
r4923, autrijus++ | sub baz returns Pair () { (x => 3) }
r4923, autrijus++ | foo(bar()); # This assigns (x=>3) to $x
r4923, autrijus++ | foo(baz()); # This assigns 3 to $x
r4923, autrijus++ | foo(*baz()); # This assigns (x=>3) to $x
r4923, autrijus++ | foo($y); # This assigns 3 to $x, too
Limbic_Region how is the timezone difference treating you autrijus?
autrijus i'm doing fine :) 21:20
Limbic_Region i'm currently building Pugs (at home) to write tests for the bugs geoffb helped me find earlier
how relavent is the patch theorbtwo applied 21:21
WRT bugs not really being bugs?
autrijus theoretically theorbtwo's patch only makes all castfails trappable. 21:23
practically though... it breaks almost everything. I'm still sorting thru it
geoffb autrijus, let us know when the coast is clear to do another pull/build and be able to assume life will be good 21:24
er, please
pmichaud how goes the hackathon?
Limbic_Region well - I know these bugs were b0rk prior to the patch so I will post em anyway
theorbtwo
.oO(I tried to warn him...)
 21:29
Limbic_Region ughh - /me checks revision history to see what rev # was theorbtwo's patch 21:32
autrijus sure
pmichaud: very well! got lots of corner cases specced and/or clarified
pmichaud excellent. I'm looking forward to arriving tomorrow. Anything the group wants me to work on between now and then?
Limbic_Region how do I (using svk) go back to an older revision? 21:33
4919 for instance
autrijus pmichaud: invokecc for PGE closures? :) 21:34
pmichaud oh yeah, I can probably have that done at some point in the hackathon, np
I'm curious to know more about parrot namespaces and how we expect perl 6 objects (esp. subroutines) to appear in those namespaces 21:35
also, anything we have on the ast representation 21:36
autrijus I can braindump to you tomorrow :) 21:38
as you can already compile Test.pm and other stuff to PIR
eg. Test::ok is just &ok in the Test namespace 21:39
Limbic_Region kicks svk for not listening to him
autrijus also luqui is looking on getting compreg working even for external parrot (by shelling out to pugs)
so in any case you should be able just to ask compreg to produce something for you that you can call.
I'll take care of compiling the closure in the Rule into a Closure object pmc.
(it's already done, actually) 21:40
pmichaud np
autrijus as for AST, the PIL nodes should be a decent start; we can work out how to map them into p6 space
luqui and I sort of think to represent them adequately we'd need parameterised classes, aka generics 21:41
pmichaud yeah, that's the piece I'm wondering about next
so being able to go through that over the weekend will be really helpful
autrijus sure... I look forward to it. when do you arrive tomorrow?
pmichaud I should make it to the hackathon around 3pm, depending on airline delays and traffic 21:42
Juerd has anyone experienced downtime of feather, or was it my home connection?
pmichaud my flight arrives toronto 1pm
Limbic_Region finds it very hard to believe that you can't checkout a specific revision with svk 22:00
svnbot6 r4924, autrijus++ | * If the parameter is expecting a Pair... don't namify it.
autrijus yow. 22:03
7532 test cases: 6916 ok, 616 failed, 942 todo, 432 skipped and 4 unexpectedly succeeded
Limbic_Region: huh? svk up -r
svk co -r
geoffb That test summary looks close to normalcy -- only a few percent still failing 22:05
svnbot6 r4925, autrijus++ | * test cleanups; add makefile.pl so smoke tests run
Limbic_Region autrijus - it doesn't work 22:09
Invalid filesystem revision number: No such revision 4919
mugwump svn switch -r 22:10
(another way for svn)
Limbic_Region they all say the same damn thing
Invalid filesystem revision number: No such revision 4919
mugwump are you connecting from the right space/time zone? 22:12
autrijus weird... ^W^Wsvk takes local rev numbers
not remote ones
so you can't do -r 4919 when 4919 is the pugs svn rev
Limbic_Region I even tried wiping out the entire pugs distribution, svk, and starting out fresh with an initial checkout of 4919 - it just looked at me funny and kept doing the same crappy things 22:13
autrijus you need to translate it to your local number using svk log or something
Limbic_Region tries something 22:14
this is just plain stupid 22:16
22:17 knewt__ is now known as knewt_
geoffb It should be easy to specify a revision relative to the repo being mirrored -- if not, there's a good complaint for clkao 22:20
Limbic_Region I am firmly under the belief that I am just too stupid to run a computer 22:24
I have tried it relative to local revisions
I have tried starting out fresh checking out a specific versions
I have tried everything I can think of
it refuses to do anything other than give me the latest
geoffb blinks 22:30
Limbic_Region, did you nuke the offending depot in between all these efforts?
(here I'm assuming pugs is the only thing you're mirroring in that depot . . . )
Limbic_Region geoffb - yes, in fact I uninstalled everything and rmdir /s /q anything that remained 22:31
geoffb wowsers 22:33
Limbic_Region and so I tried to download svn 1.2.0 and it seems all those links are broken
Limbic_Region thinks the universe really doesn't want him to be playing with Pugs tonight
Limbic_Region opts for svn 1.1.4 and hopes for the best 22:34
you have got to be kidding me
now I am getting a "Program too big to fit into memory" error 22:35
geoffb OK, not to disparage your system or anything, but that's fubar.
Hmmm . . . maybe SVK reacts badly to OOM issues? 22:36
Limbic_Region that doesn't make any sense
it works fine unless I ask for something other than the most recent directory 22:37
s/directory/revision/
geoffb and "Program too big to fit into memory" does?!?
Limbic_Region no - that's svn not svk
Limbic_Region gives up for tonight
it isn't meant to be
Juerd Too many things are not meant to be 22:38
geoffb Limbic_Region appears to be playing "boss who blows up everything vaguely technical when he walks in the door" today
Juerd Those things take us to be
Perl 6 is one of those things
Getting living beings in and out of space was too
mugwump dang, I'm getting that field `_crypt_struct' has incomplete type error building Pugs.Embed.Perl5 now 22:39
Juerd I find the parrot binary rather large. Is there a reason it has to be so big? 22:41
mugwump stripped it?
Juerd Oh, it's not stripped by make? 22:42
That explains it :)
Thanks, this helps much
mugwump Mine's 2Meg after stripping
Juerd That saves a lot of memory during runtime
And a second execution time! 22:44
Imagine what that does to testing.
mugwump you don't embed? :) 22:45
Juerd Hm?
svnbot6 r4926, luqui++ | Added pugsSlurp. 23:04
autrijus luqen.wikipedia.org/wiki/Afshar_experiment 23:10
svnbot6 r4927, autrijus++ | * unTODO succeeding tests.
Juerd Is there a test for %hash = { ... } yet? 23:17
svnbot6 r4928, autrijus++ | * after consulting with lwall and luqui, in light of the new
r4928, autrijus++ | semantic of bare blocks always taking a (?$_=$OUTER::_) as
r4928, autrijus++ | parameter, the "Parametric" type is now obsolete. Also gone
r4928, autrijus++ | with it is the "Bare" type -- all blocks are just Block now.
Juerd It works (at least worked earlier today), but shouldn't work.
Well, it should work, but using the hashref as the first key, not as the entire new hash.
autrijus right. that was one of the deref issues following 0-level 23:24
I'll fix... actually the entire deref semantics is now changed
I'm also trying (not very successfully) to release a version today before really ripping everything out
Darren_Duncan that is a good time to do it 23:32
keeps a more permanent record of a lot of older work that would seem to get destroyed 23:33
revdiablo heh, "older" 23:34
the ancient code that's 2 weeks old
geoffb Um, isn't that what version control is for, Darren_Duncan?
Darren_Duncan I said MORE permanent
geoffb (well, one of the uses, anyway)
heh
fair 'nuf
Darren_Duncan all the versions in cvs aren't on CPAN
I mean svn
revdiablo and not only record but recognition
Darren_Duncan for those that only see CPAN, the releases contain all the visible work 23:35
geoffb sigh
revdiablo people are sensitive butterflies
coral flap, flap 23:44
svnbot6 r4929, autrijus++ | * first cut of transliterating &pugs_eval into Emit.PIR
theorbtwo Change going in for better error messages on castfails. There's probably a better way to do it, but this way works...) 23:49

Powered by Cro and the Raku® Programming Language.

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

Thank you!