Check your feather email | spec.pugscode.org | paste: sial.org/pbot/perl6 | pugs.blogs.com | www.treehugger.com/files/th_images/paradigm.jpg [set by audreyt on 2006-08-29 05:26:57 -0700]
Set by Akwa|user on 14 September 2006.
00:11 weinig is now known as weinig|bbl 00:14 Qiang_ joined
svnbot6 r13516 | lwall++ | Answer to sqrt question. 00:19
00:21 nekokak joined 00:23 hikozaemon joined 00:30 rodi joined 00:31 vel joined 00:40 plural_ joined 00:50 Qiang joined 00:52 mjk joined 00:53 justatheory joined, pen1 is now known as penk
svnbot6 r13517 | markstos++ | S29_spec.t: Add smart link and improve comment. 01:01
r13518 | markstos++ | mv t/xx-uncategorized/S29_spec.t t/builtins/type_declarations.t 01:04
r13518 | markstos++ | (Part of my secret plan to make the test structure layout more closely match
r13518 | markstos++ | the spec layout)
r13519 | lwall++ | Tweaks to character types. Ling and Graf are dead.
r13520 | lwall++ | Fixed tests to go with new S29 types. 01:08
ajs_home Hmmm... there's an S29_spec.t? Perhaps I should be looking at that ;)
TimToady er, no, it's just the Char types. 01:09
01:20 mdiep joined
svnbot6 r13521 | markstos++ | [t/operators] - two smart links and a comment. 01:23
01:31 Qiang_ joined 01:33 mdiep joined
TreyHarris About an hour ago, I got a duplicate copy of markstos's message to perl6-users from last Saturday. Did anyone else? 01:36
01:36 tcliou joined
TimToady yes 01:37
TreyHarris odd. 01:38
TimToady yes 01:42
01:54 mako132_ joined
svnbot6 r13522 | markstos++ | Test.pm: The docs stated that "is_deeply" was a future future. 02:02
r13522 | markstos++ | Clarify that some version of it works now.
02:12 mako132_ joined
svnbot6 r13523 | markstos++ | Test.pm - POD is further enhanced with "=head3" for each available function, 02:13
r13523 | markstos++ | so they are easily browsable after pod2html conversion. See:
r13523 | markstos++ | feather.perl6.nl/~markstos/Test.pm.html
lambdabot Title: Test - Test support module for perl6, tinyurl.com/p2dor
02:14 Qiang joined
svnbot6 r13524 | markstos++ | Test.pm: clarify what I believe to be true: 02:31
r13524 | markstos++ | - eval_ok should be used if code might be parse, but has some overhead.
r13524 | markstos++ | - die_ok is a better choice if the code parses, but might die.
r13524 | markstos++ | Without this clarification, it was confusing what the difference was.
r13525 | markstos++ | Test.pm: Improve docs for :todo and :depends. 02:47
r13526 | markstos++ | classify.t: new tests for newly spec'ed "classify". 02:55
03:10 Qiang_ joined 03:12 frederico joined
svnbot6 r13527 | lwall++ | 1st whack at exporting methods. 03:46
r13527 | lwall++ | Some bogus package names are turning into export tags.
r13527 | lwall++ | Added generalized pick method.
r13527 | lwall++ | Added signatures as sort criteria
TimToady I'm not done with that file, so don't expect complete consistency... 03:49
'course, it won't be completely consistent after I'm done with it either... 03:58
Eidolos Completeness or consistency. Choose one, maybe. :) 04:56
Teratogen thus spake Godel 05:00
svnbot6 r13528 | ajs++ | Per Larry answer, moved radical glyph to API doc with :UNI export tagging and cleaned up API doc a bit 05:04
r13529 | trey++ | Deleting test pm files used by old use_import_not_found.t, which was 05:19
r13529 | trey++ | deleted in r13468.
05:19 BooK_ joined 05:21 kyrbe joined
TreyHarris ?eval my $a = (1,2,3); my ($b) = [,] @$a; $b 05:30
05:30 evalbot_r13492 is now known as evalbot_r13529
evalbot_r13529 [1, 2, 3] 05:30
TreyHarris is that right?
TimToady no, should just end up with 1 05:34
in fact [,] should generally be a noop on @ in list context. 05:35
the Capture generated by [,] is supposed to reinterpolate. 05:36
now if there weren't parens around $b, that'd be different, I think.
since [,] in scalar context would just return the Capture. 05:37
TreyHarris right
?eval my $a = (1,2,3); my ($b) = @$a; $b
evalbot_r13529 \1
TreyHarris that works fine
svnbot6 r13530 | trey++ | [t/operators/splat.t] 05:38
r13530 | trey++ | Added failing test for flattened dereffed arrayref in list context.
TreyHarris though i guess 'splat' is no longer a good name for that test file... 05:39
05:46 f0rth joined
TreyHarris "Lazy of Pair"? Lazy is a type? 05:58
just noticed that in docs/Perl6/Spec/Functions.pod 05:59
TimToady probably just means a List 06:08
since List is already presumably lazy 06:09
but maybe we'll end up calling them all Feeds anyway... 06:11
TreyHarris "multi Lazy Array::reverse (@values:)" should be written "multi Array::reverse (@values:) is lazy", no? 06:12
TimToady that's bogus syntax several ways. 06:13
TreyHarris which one?
oh... yes. 06:14
i just cut and pasted, and edited and pasted
TimToady our Lazy multi method reverse (@values:)
plus an "is export" so we can say reverse(@foo).
I've been trying to get rid of all the bogus Array:: thingies too. 06:15
it's just in the List package, so anthing that does the List role and can bind to @values can be reversed. 06:16
TreyHarris couldn't reverse of an actual array be optimized over reverse of a list potentially? 06:17
TimToady maybe, maybe not.
It's probably a lot faster to reverse 1..100000 than 1,2,3,4,....100000 06:18
TreyHarris but i guess you don't have to specify in the docs any multis of subclasses you do specify... that should be transparent. (unless you want to play games with .wrap or .assuming or other Code methods...) 06:19
TimToady since you can just turn it into 100000..1:by(-1)
TreyHarris right... 06:20
TimToady these are interfaces, which is why I started putting in mumbojumbo about roles/classes.
reverse probably comes from the Ordered role, in fact...
or some such
TreyHarris right... multis of subtypes of types you do specify...
err... wait, is a subclass of Foo also a subtype of Foo? subtype has special meaning 06:21
06:21 justatheory joined
TreyHarris which lets you warp your type constraints in various dimensions... but inheritance is one of the dimensions already provided for i hope? 06:21
TimToady one would hope to get a bit of polymorphism out of our MMD mechanism... 06:22
TreyHarris sure... i was just asking if it's valid to use the word "subtype" to include a class that does or isa something else 06:23
TimToady I have no idea what "subtype" would mean in the context of P6. 06:26
TreyHarris well, S12 uses the word 'subtype' a dozenish times 06:27
TimToady probably means a dozen different things there. :)
TreyHarris and says "A subtype is not a subclass." which is why i was searching for a different word to encompass "any type that is natuarally encompassed by another type", so for example Int is encompassed by Num, and 4 is encompassed by Int... 06:28
liskov-like
06:44 iblechbot joined 06:46 mdiep joined 07:14 marmic joined 07:26 zakharyas joined 07:38 kane-xs joined 07:46 mdiep joined 08:03 pdcawley joined 08:07 ludan joined
ludan howdy 08:07
08:16 jferrero joined 08:17 justatheory joined
mugwump a subtype might not be a subclass, but isn't a subclass a subtype? 08:19
08:24 integral_ joined 08:26 scw joined
TreyHarris mugwump: that's what I was thinking, yes 08:27
TreyHarris heads off to bed & 08:28
08:30 integral_ is now known as integral 08:41 vytautas joined 08:55 azr|el joined 09:10 avarab joined 09:15 elmex joined
lumi How much slower is pugs today than perl? 09:16
Juerd Depends on what you're doing :) 09:17
lumi A lot of string processing
09:39 avarab is now known as avar 10:00 zeriod joined 10:02 Odin-LAP joined
obra 10:15
nothingmuch
buu
nothingmuch ack, /nick timmy didn't work
10:37 rodi joined 10:47 ruoso joined, buubot joined 10:52 norageek2 joined 11:03 buetow joined 11:09 chris2 joined 11:25 mdiep joined 11:26 ofer__ joined, ofer__ is now known as ofer1 11:31 pdcawley joined 11:40 Limbic_Region joined 11:52 norageek2 joined 11:56 weinig joined
Limbic_Region ?seen audreyt 12:01
lambdabot audreyt is in #perl6, #ghc and #haskell. I last heard audreyt speak 10h 35m 7s ago.
Limbic_Region audreyt ping
12:09 pdcawley joined 12:10 ayrnieu joined 12:21 masak joined 12:31 iblechbot joined 12:36 kane-xs_ joined 12:46 zeriod joined 12:59 KingDiamond joined
svnbot6 r13531 | malon++ | eval.t - test fatal C<use> statements wrapped in C<eval> 13:20
r13531 | malon++ | * Test.pm itself is broken by the Pugs bug that this test documents,
r13531 | malon++ | so you won't even see "Looks like you planned 6 tests but only ran 5."
13:21 mdiep joined 13:26 crem joined, cjeris joined 13:27 dakkar joined 13:28 dakkar joined 13:31 dakkar joined 13:37 vel joined
svnbot6 r13532 | malon++ | [ext/re] 13:43
r13532 | malon++ | * C<use re 'PGE'> will globally (not lexically) switch the regex engine
r13532 | malon++ | in the middle of a program. It simply resets the environment variable.
r13532 | malon++ | * Requested by audreyt++ as a supplement to r13513.
13:46 fglock joined 13:50 pdcawley joined 13:51 vel joined 13:52 vel joined 13:54 frederico joined 14:10 mdiep joined 14:25 pdcawley joined
ajs_ TreyHarris: Lazy is a type that Larry put into S29, and I ran with. Lazy of Pair is something that I'm not sure abut, but seemed to be the only way to get what I needed. 14:33
The 'our Lazy of Pair multi method classify ( @values: Matcher $test )' is meant to indicate that classify must return a Lazy list comprised of Pairs. If we take as written that Larry's Lazy type is, in fact, a container akin to Array, then it works fine. 14:35
14:40 pdcawley joined 14:55 lollan joined
TreyHarris ajs_: ok... I guess I'm just confused about "Lazy of Pair" versus "Lazy Array". the first makes Lazy sound as if it's a container, the second not so much 15:30
and what is the answer to what Larry said last night? what's the difference between Lazy and List? 15:31
15:36 justatheory joined 15:54 tcliou left 16:00 mauke_ joined 16:03 rlb3 joined 16:10 Psyche^ joined 16:15 weinig is now known as weinig|away 16:18 mauke_ is now known as mauke 16:23 Psyche^ is now known as Patterner 16:31 Daveman joined
TimToady according to src/Pugs/Types.hs, it would appear that the distinction between Lazy and List is that a List can be either Lazy or Eager. 16:34
ajs_ TreyHarris: I'm not aware of Larry saying something last night about Lazy and/or List 16:35
TimToady no, I just said it this morning. :) 16:36
Juerd Heh 16:38
TimToady however, it's not clear to me whether we really need an Eager type, or whether we can define eager lists operationally.
ajs_ sigh... I've been cattering here for how long without realizing that you were Larry?!
chattering even
TimToady np 16:39
rodi damn, and I could have been Larry to ajs_.
TimToady I guess if we need an Eager type it's for signatures.
so many missed opportunities in life...
rodi err... if I didn't ask dumb questions on p6l, anyway. 16:40
ajs_ So... given that, is my ad hoc "Lazy of Pair" reasonable, or should that be something else?
TimToady I'm not sure we need a DumbQuestion type either. :)
ajs_ rodi: you will always be my $larry...
TimToady it's probably okay, unless some kind of weird contra/covariance thing makes it not general enough. 16:41
ajs_ Ok, I'll leave it alone for now, then.
TimToady but I think naming a lazy list as Lazy probably buys us documentation at least.
that's my feelers too
ajs_ Your lack of whitespace DWIMery made me feel much better about |, BTW 16:42
TimToady good.
I'm only partly done with my recent whackage on S29.
ajs_ Ok, I'll keep my eyes on the logs. Do you have a feeling one way or the other about the audreyt-requested ongoing API/S29 split? 16:43
TimToady there are still a bunch of Foo:: things in there that should probably be implicitly in the role/class that provides/exports them.
no, that's been mostly off my radar. 16:44
which means I'm probably stomping all over someone's carefull crafted S29 extraction code. :)
ajs_ It's not that big a deal. Just removing things that aren't really language features to the API docs. Methods, constants, etc. 16:45
Juerd TimToady: Defining a type that is both an array and a string goes like class Foo { does Array; does Str; ... }, right?
TimToady but this should really be defined in Perl files, not Pod files.
Juerd, yes, in theory, and assuming we properly role-ify the basic types. 16:46
ajs_ Absolutely. As soon as pugs can handle the full spectrum of what's in these API docs, I think they should become the POD associated with the appropriate code, not stand-alone.
justatheory coughs
TimToady development-driven specification, just like test-driven development. :) 16:47
ajs_ heh 16:48
TimToady just run the whole waterfall uphill.
someday we might even have some requirements.
justatheory www.waterfall2006.com/
lambdabot Title: Waterfall 2006 - International Conference on Sequential Development
ajs_ "You've always known a good waterfall-based process is the right way to develop software projects." -- I have?! 16:50
TimToady speaking of waterfalls, I'm off to the shower... 16:51
biab &
ajs_ laters 16:52
Juerd TimToady: Given this syntax, how would you promise that the members of said Array are of MyType? 16:56
17:01 lanny joined 17:03 araujo joined
svnbot6 r13533 | fglock++ | t/xx-uncategorized/rule_recursive.t - s/rule/regex/ to enable backtracking 17:06
kolibrie to find every match for a regex called 'talk', should this work? while (my $match = $content ~~ m:c /<talk>/) { say $match } 17:07
is there a better way?
(is there a way which currently works?) 17:08
kolibrie is working on his 'Practical Extraction with Perl 6' talk for Saturday 17:09
fglock kolibrie: until Saturday we can find a way that works 17:10
kolibrie fglock: awesome
fglock: should I put that snippet in my slide for now?
for reference, I am building my talk here: feather.perl6.nl/~kolibrie/extraction.xul 17:11
lambdabot tinyurl.com/rxtmp
fglock kolibrie: with :global / :overlap / :exhaustive you can capture to an array directly 17:12
but :c is easier to implement
17:13 penk joined
kolibrie fglock: so this should work? my @match = $content ~~ m:g /<talk>/; 17:13
but really, I would prefer one match object at a time for this 17:14
[particle] kolibrie: change your while to for
17:15 streakceur joined
[particle] ...then you can use m:g 17:15
fglock bbiab &
kolibrie [particle]: for $content ~~ m:g /<talk>/ -> $match { say $match }; 17:16
17:16 penk joined
[particle] whoa, that's scary when you write it :) 17:16
17:17 penk joined
kolibrie and I think/hope that is lazily generated 17:17
[particle] same thoughts/hopes here
lanny ?eval srand(3);
17:18 evalbot_r13529 is now known as evalbot_r13533
evalbot_r13533 Error: No compatible subroutine found: "&srand" 17:18
lanny Oh good. Broken here too.
Juerd srand shouldn't work :) 17:20
lanny Really? Something to do with Haskell core? 17:21
Juerd Nah, it reeks of perl4ness :P
IMO randomness with user defined seeds should be OO. 17:22
So that you can have several sequences at the same time.
Maybe $*RAND can be the standard object, and $*RAND = Rand.new; could re-init or something
lanny I agree. But specs are specs. That's actually how I've implemented. Figured there would be a $*RandomGenerator having a default for rand()
I guess I'll add to prelude then and see what happens. 17:23
ajs_ rand/srand are a nod to POSIX. It is assumed that a real random library will be provided which can track multiple seeds, plug in PRNGs, provide the best local source of true randomness, etc. 17:41
Once it is, rand/srand will almost certainly be font-ends to same. 17:42
lanny I know. On the other hand having a basic rand() to just select among different things without having to OO it all is nice.
my @random = sort { int(rand 3) -1 }, @sorted;
kolibrie first example up, comments welcome: feather.perl6.nl/~kolibrie/extraction.xul 17:43
lambdabot tinyurl.com/rxtmp
ajs_ lanny: You know that that isn't guaranteed to finish, right?
lanny If you know @sorted is finite then I hope it finishes.
ajs_ no, it depends on what kind of sort you are doing. 17:44
And if you are memoizing
lanny It is memoizing because that's a KeyExtractor. hmm.
my @random = sort { $^a; $^b; int(rand 3) - 1 }, @sorted; 17:45
Ugly but now it's a Comparator and would work.
[particle] kolibrie: looks good so far 17:46
lanny Although first code works in Pugs now since need subsets working to put sort() into place.
ajs_ $test.(1,2) still has a random result. I think bubble sort chokes on that, but of course, Perl 6 is not likely to resort to a bubble sort ;) 17:47
Still, it's safer to just use a true shuffle (and guaranteed to be optimal) 17:48
lanny I've got a spec implementatin of sort() using mergesort.
kolibrie [particle]: thanks
lanny I know. N*logN just to shuffle is overkill.
[particle] there's nothnig wrong with bubble sort if you apply it correctly :P
17:48 BooK_ joined
ajs_ [particle], the problem with that is that applying it correctly means knowning things about your data up-front (e.g. is it already nearly sorted). 17:49
lanny Even nearly sorted you get N^2 comparisons. Although if you've got lots of time it does sort. 17:50
[particle] that's true for any algorithm. you need to know your data in order to optimize
svnbot6 r13534 | fglock++ | v6 - implemented m:c/rule/ - still needs work
ajs_ Point is that by definition, a sort assumes that any two elements have a static relationship. If you break that assumption, all bets are off, and you should not be shocked to find that a local implementation hoses you.
fglock re 13534: there is a problem related to cacheing - it only works the first time you run it 17:51
ajs_ [particle]: Correct. bubble just has a narrower domain than most. I assume we both know this, and we're just demonstrating that we do ;)
fglock kolibrie: I'm testing with: perl -Ilib -e 'use v6-alpha' - ' my $content = "xyz"; while $content ~~ m:c/<alpha>/ { say $/ } '
lanny On the flip side of knowing your data you can also know your implementation and do things you probably shouldn't.
But yes. I'm in total agreement, ajs_.
ajs_ fglock: did you hear my comment about v6 and "my A $x" turning into "my A; $x;"? 17:52
fglock ajs_: yes - v6.pm is currently too forgiving with syntax errors 17:53
kolibrie fglock: that looks like a good test
ajs_ ?... errors?
lambdabot Not enough arguments to @.
fglock ajs_: oh, you mean A is a Type
ajs_ Right
fglock ajs_: it's a grammar error then 17:54
ajs_ ok. just wanted to make sure you knew. I tried to hunt it down, but got a bit lost
v6 is a tiny bit of code ;-)
17:55 eden_c joined
fglock ajs_: the place to start is Grammar::Term.pm, line 299 - 'my' => ... 17:56
ajs_: tiny as in compact?
17:56 rodi joined
ajs_ That was sarcasm. I found it difficult to find the error would be the more useful way to state it 17:57
TreyHarris < Juerd> TimToady: Given this syntax, how would you promise that the members of said Array are of MyType? 18:01
Juerd: I'm not TimToady, but isn't "Array of MyType" what you want?
Juerd TreyHarris: Can that be used with "does"? 18:04
TimToady I presume he's referring back to the "does Array; does Str". In which case Array of Mytype ought to turn into "does Array[:of(MyType)]".
Juerd class Foo { does Array of MyType; }
Ahh
TimToady just generic instantiation.
fglock kolibrie: that example works with v6.pm r13535
Juerd Thanks
This clears that up 18:05
svnbot6 r13535 | fglock++ | v6 - $/ is initialized to undef at runtime
TreyHarris TimToady: yes, we probably don't need an Eager type. But I think some way (even a no-op eager role so you can say C<is eager>) would be useful for documentation. "Note: foo will expand the list, so do not pass it an infinite (or very long) list" is easy to get lost, it'd be nice if eagerness could be documented in the signature
TimToady and if you just say "does Array" it presumably autoinstantiates with Any or something.
TreyHarris How does Array[:of(MyType)] differ from 'Array of MyType' or 'Array[MyType]'? 18:06
TimToady TreyHarris: that's the main reason for keeping it there, though I suppose we could just have "is eager" as a trait.
Juerd TimToady: I found while I was writing some pseudocode that types provide great documentation. I just found myself looking for a way to express this :)
TimToady: In that respect, I think I'm fully convinced about types now :)
kolibrie fglock: awesome, thank you
TimToady the last assume :$of is the first positional parameter.
*assumes 18:07
kolibrie with .xul slideshows, how to I specify a specific slide in the URL? 18:08
TimToady and since that depends on how the role write writes the role, I thought I'd be more specific and use a named arg.
TreyHarris TimToady: right.
TimToady arguably it should be :of(:(MyType)) or :of(:(-->MyType)) though.
TreyHarris I don't care about Bertrand Meyer's invective, I like taxomania. I think an empty 'eager' role would be fine
TimToady unless we make types morph into sigs, which is likely. 18:09
lanny :(-->MyType) isn't allowed at the moment. Can't have --> in :() literals.
And I think as a signature :of(:(MyType)) is better. 18:10
TreyHarris Array of Employee is so nice though... I hope it will DWIM regardless of what it desugars into
Hash of Potatoes... mmmm
TimToady Array of Choices
rodi Array of Sun... ahhhh
Juerd I was thinking of Hash of Web::Argument
Or HTTP::Argument. Haven't made up my mind yet.
[particle] String of Puns 18:11
rodi LOL
Juerd In any case, I want to get rid of the hideous param(), except for compatibility with modules.
Not even for compatibility with people wanting param().
(Unfortunately, there's no way to enforce *that*)
TimToady make param a keyword. :) 18:12
Juerd TimToady: How would that help? :)
svnbot6 r13536 | fglock++ | v6 - updated TODO, ChangeLog
Juerd The Perl 5 modules wouldn't realise it's a keyword?
TimToady get Nick to sneak it into 5.8.9 18:13
kolibrie isn't sure if slide 16 is correct, particularly if <( )> is used, does $/ have just that captured section in the { return } block
Juerd Hehe
TreyHarris back to the earlier exchange between TimToady and ajs_... shouldn't global multis be documented together, wherever they originate from? (I realize in the general case that means you have to generate the documentation dynamically, but at least for the core API...)
TimToady $$/ has just that section. $/ still has the whole thing. 18:15
kolibrie TimToady: ah, thank you
fglock: you may have also noticed that I'm using .+? in rules (which currently isn't implemented in v6.pm) 18:16
TimToady TreyHarris: yes, but I wish we could find a way to do that without using bogus declaration syntax like Array::push. 18:17
fglock kolibrie: yes, I was just going to tell you that
nor <(...)>
kolibrie the troubles of writing a talk about bleeding edge technology 18:18
TimToady there are a lot of obsolete talks down in docs/talks :) 18:19
ajs_ TreyHarris: global multis should be documented together, that is a fair statement.
TreyHarris TimToady: does the =item *have* to be parseable as a valid signature and/or generated from the source file? if not, just add (defined in Array)
fglock kolibrie: otoh, things can be implemented on demand :)
TimToady that's what I've been doing at the top of each section.
"the following are declared in the Num role/class." 18:20
kolibrie fglock: I don't want to be demanding, but it sure would be nice to have these implemented :)
Juerd TimToady: Can a class export subs?
TimToady but for the global listing, that's inside out
TreyHarris TimToady: right, but for multis whose implementations are shared between several role/classes, put them together but say which role/class it comes from after each sig
ajs_ Yeah, and the API documents are just placeholders that will be wrapped up into those roles/classes when they are written
Juerd TimToady: Or does that require a module and a class of the same name?
ajs_ TreyHarris: That is exactly what we do 18:21
in S29
TimToady A class is a module so can export subs.
Juerd Great
TimToady Nowadays you can even export methods
TreyHarris Juerd: 'is export' is attached to methods in S29
TimToady which turn into multis.
Juerd Export... methods :)
18:21 mako132 joined
Juerd That takes away my explanation of roles! :) 18:21
TimToady push(@array: @list) vs push(@array, @list)
Juerd has so far explained roles in terms of exportation
TreyHarris ajs_: it is exactly what you do? what's an example, I'm looking at S29 now 18:22
TimToady our Int multi method push ( @array: *@values ) is export 18:23
ajs_ We do that for everything. Everything that is global is listed there. There are still some methods and other junk that need to be moved out into API docs, but basically, what's in S29 are "Functions", as the title suggests. The global ones that are exported
Juerd TimToady: Can "is export" be in the block, like in class Foo { is Bar; }?
ajs_ We don't add in "is export" typically, as that is assumed to be a part of anything in S29. 18:24
fglock kolibrie: 'token metasyntax {}' in Pugs::Grammar::Rule2 works with nested xml-like markup, maybe it helps
TreyHarris ajs_: sorry, i mean what's an example of multis defined in different places being listed together but with each sig's origin mentioned?
ajs_ Which we should state explicitly
Juerd Part of "anything"?
TreyHarris ajs_: part of any sub, surely, but not part of any method?
ajs_ TreyHarris: Every function mentions its package explicitly in the signature
TimToady Hash vs Array reverse
ajs_ TreyHarris: Methods in S29 are an anachronism, and are being removed. They are not "Functions", which is what the document is about. 18:25
TreyHarris oh, wild. so our 'Int multi method push ( @array: *@values ) is export' is defined in Int? I would have thought they'd be defined in Array
TimToady I'm going exactly the opposite direction and installing methods into S29
no, that's the return type
not the method's class
TreyHarris ajs_: ==TimToady. I'm very confused.
ajs_ TimToady: we should talk... 18:26
;)
TreyHarris ok. I'll let you two talk :-) my point was just that all multi subs/methods called <push> should be listed together, but they should say where the package they come from, as separate from type and not resorting to the incorrect multi Package::push syntax. 18:27
TimToady audreyt and I cooked up the method is export thing some time ago to get rid of the mutual fallback model. 18:28
ajs_ audreyt's request, which spawned all of this, and which I mentioned way back when on the list was this: (paraphrase) remove the methods and constants from S29 because they aren't actually part of the Functions list that S29 is setting out to update the reader on. Put them in API documents that define the classes and roles used.
Juerd repeatedly reads "TreyHarris" as "TimToady" and vice versa.
Apparently my brain only sees "T[a-z]+[A-Z][a-z]+"
TreyHarris Juerd: good for me, perhaps, not so good for TimToady ;-)
ajs_ TimToady: Ok, so what's a method that is export?
TimToady push
Juerd Which in Perl 6 regex would be... OMG... T<[a-z]>+<[A-Z]><[a-z]>+ 18:29
ajs_ Sorry, was asking what it means
Juerd Do we have <lower> and <upper>? :)
TreyHarris commutes &
ajs_ does it mean "this method also defines a function of the same name" or does it mean something more clever?
Juerd TreyHarris: Smart, to commute in the background! Then you can still be active here! :)
TimToady It's a multi that happens to wrap a method, for things that want to muddy the waters, such as abs($x) vs $x.abs 18:30
Juerd If a method can be exported as a multi - can a class import from itself, to get the multis?
[particle] juerd: yes, we have <upper> and <lower>, as well as <alpha> <alnum> <ident> <digit> <xdigit> ... 18:31
Juerd And why do I keep using " - "...
Juerd uses it as a larger comma.
TimToady presumably it can just get the multis as globals like anyone else.
Juerd [particle]: Good
[particle] suggests =>
Juerd T<lower>+<upper><lower>+
Much better. 18:32
[particle] i'm going to have trouble converting my fat comma usage to p6
ajs_ Ok, that sounds reasonable, and justifies methods being in S29, IMHO. However, there is perhaps a level of detail at which it makes sense to talk about classess and roles, not "functions". It is at that level of fuzzy detail that I propose to remove things from S29 into API docs. One such example being constants. Is that fair?
Juerd Can we write the + inside the <> without breaking it?
[particle] juerd: not as currently specced
Juerd [particle]: I'm already having trouble reading Perl 5 code with fat arrow in non-pair context.
Juerd has not really used Perl 5 for a while.
ajs_ Juerd => what do you mean by that? 18:33
Juerd My brain is almost constantly in Perl 6 mode.
TimToady do you mean + as in **{1..*}?
Juerd ajs_: That I interpret that as :Juerd<"what do you mean by that?")
[particle] <lower>+ =?= <lower+> # i say no
Juerd TimToady: Yes, though in practice I'm going to munge Perl so that it uses something other than **{}
TimToady the + in <lower+upper> would be a different +
Juerd Ah, good point. 18:34
ajs_ TimToady: The other option, if you feel that EVERYTHING should be in S29 is to just re-merge the API docs and go back to working on just S29, but you should stake out some territory in there so that I can work around you.
TimToady well, one big problem is that there really is almost no such thing as a function anymore 18:35
lanny Dang. Thought so. I do not understand how the prelude works.
TimToady it's all multis and methods
ajs_ So, we say that S29 is "all core-module-default-exported multis and methods" 18:36
?
TimToady something like that
in fact, I put a remark remarkably like that near the top of S29 18:37
Juerd TimToady: Why have singles anyway?
ajs_ Ok, so any method not exported as a multi or any multi which is not exported (especially :UNI aliases) will go into the API docs?
TimToady singles are mostly for hiding all multies of the same short name.
and to give the user a way of future proofing their program against future multi additions. 18:38
Juerd Hm
TimToady proto fills the same role if the user wants to write their own future-proof set of multis. 18:40
ajs_ hmm 18:41
TimToady you leave out the proto if you just want to intermingle with any future multis
ajs_ multi sounds increasingly like the useful default. If it wasn't required in many cases to clarify the function definition, I would say that we should toss it and make it the default. 18:42
TimToady multi depends crucially on types, and types are optional
ajs_ right 18:43
TimToady where are you updating S29? 18:44
In the pugs repo?
TimToady docs/Perl6/Spec/Functions.pod
Also, singles can easily be inlined. 18:46
ajs_ Good, thanks.
TimToady I just wish I didn't keep typing mutli 18:47
too much music going through my corpus collasum to send typing signals through, i guess. 18:48
svnbot6 r13537 | fglock++ | Pugs-Compiler-Rule - added <(...)> to the grammar - "capture_as_result"
TimToady or however you spell that... 18:49
fglock callosum :)
ajs_ OK, so another topic: macros. I keep arm-waving around them because it's not clear to me exactly how they'll manipulate ASTs, and at what level of granularity. Are we far enough down the road that examples could be added that demonstrate this, or do we need a more complete implementation?
18:49 bernhard joined
fglock ajs_: we don't have a standard/portable AST yest 18:50
yet
TimToady he, according to google there's really no agreement on how to spell it. 18:51
it finds lots of results for various spellings
kolibrie fglock: you are suggesting using token metasyntax in place of .+?
ajs_ fglock: Well, that's kind of what I'm getting at. If macros are only going to touch ASTs at a very high level, then we can just work with very general subrule names as AST fields. However, if macros will be able to / expected to portably manipulate everything, then there's a lot more work there. 18:52
[particle] callosum is latin for tough 18:53
fglock kolibrie: that's in the case you want to parse xml; '.+?' is easier to write
TimToady basically, all the implementors of Perl 6 have to converge on an AST definition. I am not one of them. :)
kolibrie fglock: I'll go for easier to write for this talk, I think
[particle] fglock: pmichaud's working on ast as we speak
TimToady so one side of my brain thinks it's tough, while the other side of my brain thinks it's huge 18:54
fglock [particle]: yes - gaal has been working on it too
[particle] fglock: ah, i thought you were. i'll get them in touch with each other. 18:55
TimToady: one side of your brain is pretty small, then
fglock [particle]: I am, but I'm writing just enough to have v6.pm working - it can be changed 18:56
TimToady it == corpus c[ao]l+os+e?um
ajs_ Would it be out of line for me to write up an API doc for macro to define what it is that macros will expect of the various ASTs at a high level?
TimToady it wouldn't be out of line to take a first whack at it, I wouldn't think. 18:58
ajs_ TimToady: As long as your corpora cavernosa aren't doing the thinking, you're fine.
fglock ajs_, [particle]: see docs/notes/p6ast_draft - this was written very early in the v6.pm development, and it was not updated - but some ideas may still work 18:59
ajs_ fglock: I'll have a look
TimToady ads_: I'm not sure it makes a vas deferens 19:00
[particle] fglock: thanks
ajs_ ouch
19:10 vel joined 19:15 eden_c left
svnbot6 r13538 | fglock++ | Pugs-Compiler-Rule - implemented <(...)> in Token/Rule 19:19
19:23 rlb3 joined 19:26 buubot joined
svnbot6 r13539 | fglock++ | Pugs-Compiler-Rule - implemented <(...)> in Regex 19:34
19:36 fglock joined
svnbot6 r13540 | fglock++ | Pugs-Compiler-Rule - ChangeLog update 19:36
kolibrie fglock: do you have any tips/code samples for converting my match object into a YAML structure with: rulename: captured_value 19:57
19:59 Limbic_Region joined
kolibrie fglock: I think I want some kind of emitter, like Pugs::Emitter::Rule::Perl5, but in Perl 6, and shorter 20:02
fglock kolibrie: the .perl method makes a dump, but this is not what you want 20:03
kolibrie correct, I just want a very small part of the match object
fglock $$/.perl should work, but it is currently a parser error in v6.pm 20:07
svnbot6 r13541 | fglock++ | Pugs-Compiler-Rule - version 0.18 in cpan
fglock this works, but I'm not sure it is correct: perl -Ilib -e 'use v6-alpha' - ' my $content = "xyz"; $content ~~ /<alpha> { return [42] } /; say $/().perl '
kolibrie I'd like to coerce that to pretty yaml (rather than .perl) 20:09
pugs builtin .yaml is also too verbose
fglock would you hack .yaml into the v6 runtime?
look for 'sub perl' into Pugs::Runtime::Perl6 - it's plain perl5 20:11
glasser can somebody remind me what the status quo for ghc on intel mac is?
(preferably darwinports based?)
Limbic_Region ?seen audreyt
lambdabot audreyt is in #perl6, #ghc and #haskell. I last heard audreyt speak 18h 44m 59s ago.
kolibrie well, perhaps, but I don't understand the pugs output structure for .yaml, and don't really want that structure
fglock kolibrie: rev 13542: perl -e 'use v6-alpha' - ' my $content = "xyz"; $content ~~ / { return [42] } /; say $$/.perl ' 20:15
svnbot6 r13542 | fglock++ | v6.pm - implemented $$/
kolibrie fglock: ah, very nice 20:16
kolibrie svk pulls
20:16 cognominal joined
svnbot6 r13543 | fglock++ | v6.pm - ChangeLog update 20:20
fglock kolibrie: btw, if you update PCR, this will work now too: my $content = "xyz"; $content ~~ / x <(y)> z /; say $$/.perl 20:21
20:21 _bernhard joined
kolibrie fglock: oh, thank you! 20:22
fglock kolibrie: np
kolibrie fglock: is v6.pm the one in perl5/Pugs-Compiler-Perl6/lib? 20:25
fglock kolibrie: yes
kolibrie ok
fglock which yaml module should v6.pm use for .yaml() and eval( :lang<yaml> ) ? 20:27
Limbic_Region FYI all - smoke server is currently giving a 500 error
20:27 [mago] joined
Limbic_Region and now it is back - nevermind 20:27
kolibrie fglock: audrey usually uses her YAML::Syck module
I think
fglock k 20:28
bbiab &
20:29 Odin-LAP joined 20:30 elmex joined
rodi Is it true that anything that could be expressed by a rule could also be expressed by a token? 20:36
pasteling "kolibrie" at 66.239.158.2 pasted "fglock: /me is having technical difficulties (v6.pm)" (17 lines, 673B) at sial.org/pbot/19855 20:38
kolibrie fglock: but I'm more worried about my talk than my v6.pm installation
TimToady rodi: yes, you'd just have to be explicit about whitespace
rodi TimToady: thanks, I think it all just came clear. Both, in turn, could theoretically be expressed as regexen (w/out using backtracking)? 20:39
[particle] rodi: token == regex :ratchet ; rule = regex :ratchet :sigspace ; so token = rule :sigspace 20:40
err. rule = token :sigspace 20:41
fglock kolibrie: perl Makefile.PL should fix this (old date in the cache)
rodi Man, S05 is dense. I see it now that I /'ed for ratchet.
kolibrie fglock: then make and install? 20:42
fglock: it skipped all files, claiming they were unchanged 20:45
fglock kolibrie: just running Makefile.PL might be enough, it calls $cache->Clear
maybe 'make clean' too 20:46
that will delete old '.pmc' files
kolibrie fglock: hmm, didn't seem to do anything 20:49
fglock kolibrie: please make install both Pugs-Compiler-Rule and Pugs-Compiler-Perl6 - it seems to be using incompatible versions of runtime/emitter 20:51
kolibrie fglock: /me re-installs Pugs-Compiler-Perl6 20:53
rodi my make smoke (r13538) is hanging indefinitely on t/xx-uncategorized/rule_recursive.t 20:54
mmm, looks like it is recursing forever on test 9: ok "x" ~~ /<r>$/, '"x" ~~ /<r>$/ matched'; 20:59
fglock rodi: yes, I've seen it
rodi fglock: good stuff. I think recursing on a recursion test is a step forward from just failing :-D 21:02
fglock rodi: :) - it's an error in the internal calling parameters, needs some more debugging 21:04
kolibrie fglock: ha! my v6.pm is working again. Thank you!
fglock kolibrie: :) 21:05
kolibrie I added some handwaving to my talk (slide 21). If anyone can help me show how to convert my $match object into the YAML on the following slide, that would be great 21:06
fglock kolibrie: I'm working on it 21:07
kolibrie fglock: you're awesome!
kolibrie thought that maybe some other people would like credits in my talk, too 21:08
21:13 frederico joined
svnbot6 r13544 | fglock++ | v6 - added .yaml() method 21:21
r13544 | fglock++ | perl -Ilib -e 'use v6-alpha' - ' my $content = 123; say $content.yaml '
r13544 | fglock++ | --- 123
21:31 [mago] joined, frederico joined, elmex joined, Odin-LAP joined, fglock joined, buubot joined, rlb3 joined, mako132 joined, BooK_ joined, penk joined, streakceur joined, araujo joined, Patterner joined, mauke joined, justatheory joined, mdiep joined, dakkar joined, cjeris joined, crem joined, KingDiamond joined, zeriod joined, kane-xs_ joined, iblechbot joined, masak joined, ayrnieu joined, weinig|away joined, avar joined, azr|el joined, scw joined, ludan joined, kane-xs joined, marmic joined, f0rth joined, Qiang_ joined, plural joined, xerox joined, ajs_ joined, idiotmax joined, ruz joined, lisppaste3 joined, stevan joined, webmind joined, gugod joined, baest joined, mtve joined, ofer0 joined, Eidolos joined, nothingmuch joined, wilx joined, jdv79 joined, theorbtwo joined, glasser joined, lampus joined, miyagawa_ joined, thepler joined, lumi joined, TimToady joined, cookys joined, szabgab joined, amv joined, norageek joined, [particle] joined, Khisanth joined, c6rbon joined, gaal joined, revdiablo joined, PerlJam joined, pjcj joined, xinming_ joined, evalbot_r13533 joined, buu joined, autark joined, xern joined, frankg joined, mollmerx joined, LCamel joined, dvorak joined, spoop joined, Yappo joined, rafl joined, Teratogen joined, silug joined, perlbot joined, prism joined, drbean joined, rafl_ joined, spinclad joined, nine joined, hcchien joined, cj joined, kolibrie joined, orafu joined, statico joined, jabbot joined, Southen joined, meppl joined, gantrixx joined, kalven joined, svnbot6 joined, Grrrr joined, ajs_home joined, nnunley joined, takanori_ joined, nperez joined, ingy joined, qmole joined, TreyHarris joined, Maddingue joined, notsri joined, kcwu joined, pmurias joined, broquaint joined, clkao joined, obra joined, wolverian joined, audreyt joined, jiing joined, cmarcelo joined, rgs joined, leo joined, chip joined, pmichaud joined, Juerd joined, Shabble joined 21:34 Limbic_Region joined
svnbot6 r13546 | fglock++ | Pugs-Compiler-Rule - added support for $/.yaml 21:34
fglock $$/.yaml is not going to work for a while - it needs more work at the v6.pm runtime side 21:35
svnbot6 r13547 | fglock++ | v6.pm - ChangeLog update; Makefile.PL recommends 'YAML::Syck' 21:40
TreyHarris glasser: re your earlier query on ghc for darwin/intel: what was your understanding of the prior status? i know something of status, but it's a few weeks old 21:47
glasser TreyHarris: Oh, I'm quite out of date 21:48
mostly just knew that "port install ghc" doesn't work :) 21:49
TreyHarris glasser: see cvs.haskell.org/trac/ghc/wiki/X86OSXGhc
lambdabot Title: X86OSXGhc - GHC - Trac, tinyurl.com/za42p
TreyHarris if you search for "audreytang" on that page, that one should work for you, it does for others
(I will find out tonight, just got an intel mac) 21:50
glasser right, i knew she had done something
TreyHarris er... maybe tomorrow, knowing how long ghc can take to build from source
glasser Oh cool, there's a build from last week too
TreyHarris no readline...\ 21:51
which is a must for pugs, IMHO... but i'm a pushover for arrow keys ;-)
glasser well, won't readline install with dports?
TreyHarris glasser: hm. maybe. getting readline support into ghc is a black art AFAICT 21:52
the distribs usually figure themselves out, but the nightly builds never do
glasser but 6.6 is almost out right? 21:53
svnbot6 r13548 | fglock++ | v6.pm - fixed '$$/.yaml'
TreyHarris glasser: i don't know. i'm sure they'd be able to tell you something better on #haskell 21:56
21:57 ruoso joined
fglock kolibrie: the "magic hand waving" part needs coercing Match into Str 22:04
kolibrie: otherwise, something like this should work:
perl -Ilib -e 'use v6-alpha' - ' my $content = "xyz abc"; $content ~~ / <ident> \s <ident> { return $<ident> } /; say $$/.yaml '
kolibrie: it currently outputs 2 Match objects, instead of 2 plain strings 22:05
22:11 Aankhen`` joined 22:24 mako132_ joined, beppu_ joined 22:31 beppu_ joined 22:53 Aankh|Clone joined
TreyHarris Juerd: I was thinking about what TimToady wrote to you, and it seems to me that "@items X~X ','" or [email@hidden.address] *would* be the most obvious way to write such a thing... i guess you'd have to write '"{@items.join(',')}," to get it to work even for zero items... 22:57
22:59 buetow joined 23:02 cjeris left 23:03 Aankh|Clone joined 23:10 SubStack joined
TimToady maybe we need @items.unchop(",") or something. :) 23:12
23:12 Aankhen`` joined
TimToady j 23:13
oops
Juerd TreyHarris: I don't remember discussing joining... 23:14
Ah, hadn't read that reply yet 23:15
TimToady: Normally, code generators do join. That is, mine do. Because in regexes (?:foo|bar|) hurts, and I'm afraid of similar hurting elsewhere :) 23:16
If [()] is an empty slice, that's perfect. It just means putting parens around the list, and that's easier than special casing.
I generally join on ",", because you need to join anyway. 23:17
TreyHarris or ", " if you want to print more prettily...
Juerd s/I/I expect I will/
TreyHarris: I don't need generated code to be pretty. I don't look at it except for debugging.
Juerd has this weird habit of generating code and eval'ing it, because it's very often the easiest and fastest way to do something. This includes runtime :) 23:18
TreyHarris ah... i always pipe my generated code through Perl::Tidy, but I'm a little anal retentive in such things :-)
Juerd I'm not spending CPU cycles on tidy'ing if the code is thrown away.
The point of translating something to Perl code is efficiency in the first place.
TimToady I'm just saying maybe we're missing something in the method syntax. with join(',', @x, '') it's easy in P5, but .join(',') doesn't make it so easy 23:19
TreyHarris ohhhh. yes, i rarely do that. if i'm generating code, it's getting written to, and some sysadmin somewhere is going to look at it when things are in flames, so i want it to be at least somewhat legible
TimToady (@x,'').join(',')
TreyHarris TimToady: that's not bad.
Juerd We have ~ now, that joins things too
TreyHarris ?eval my @x = 1..3; (@x,'').join(", ")
23:20 evalbot_r13533 is now known as evalbot_r13548
evalbot_r13548 "1 2 3, " 23:20
Juerd ~map { "$_," }, @x
?eval my @x = 1..3; [@x,''].join(', ')
evalbot_r13548 "1, 2, 3, "
Juerd ?eval my @x = 1..3; ~map { "$_," }, @x 23:21
evalbot_r13548 "1, 2, 3,"
TreyHarris ?eval my @x = 1..3; ~@x.map:{ "$_," }, @x
d'oh!
evalbot_r13548 ("1, 2, 3,", [1, 2, 3])
Juerd Heh.
Hm, @x isn't passed to map?
Alternatively, Hm, @x isn't flattened? 23:22
TreyHarris i've done that fairly frequently. i start typing a methodized form that was a builtin in perl 5, and before i get to the end i've forgotten that i'm using a method
TimToady should be. though perhaps .map is ignoring additional args?
Juerd TimToady: Then I'd still expect some flattening
TreyHarris ?eval my @x = 1..3; ~(@x,@x).map:{ "$_," }
evalbot_r13548 "1 2 3, 1 2 3," 23:23
TimToady looks more like :{...} is still be analyzed as an adverb rather than as : {...}
TreyHarris ?eval my @x = 1..3; ~{ "$_," }.map(@x, @x)
evalbot_r13548 "1, 2, 3, 1, 2, 3,"
TreyHarris ahh.
TimToady ?eval my @x = 1..3; ~@x.map: { "$_," }, @x
evalbot_r13548 Error: Cannot cast from VList [VCode (MkCode {isMulti = False, subName = "<anon>", subType = SubBlock, subEnv = Just (MkEnv {envContext = CxtSlurpy (mkType "Any"), envLValue = False, envLexical = MkPad (padToList [("@x",[(<ref:0xb79781ac>,<ref:0xb797819c>)]),("$_",[(<ref:0xb78e3c18>,<ref:0xb78e3c08>)]),("$?1",[(<ref:0xb7996e0c>,<ref:0xb7996dfc>)]),("$__evalbot_print",[(<ref:0xb6d30c44>,<ref:0xb6d30c34>)])]), envImplicit = {"$_":=()}, envGlobal = <
TreyHarris whoops :-)
Juerd BTW, While writing up my Web toolkit pseudocode, I really loved Perl 6 much again 23:24
A few lines of code. So consise, so well suited to express thoughts in.
23:24 polettix joined
TimToady looks like the compiler catches the fact that .map only wants one arg if you do : {...} 23:24
TreyHarris ?eval my @x = 1..3; ~@x.map: { "$_," }
TimToady it seems to have a nice texture
evalbot_r13548 "1, 2, 3," 23:25
TimToady the words and the non-words seem to play off each other pretty well
Juerd TimToady: Indeed. I can really appreciate the syntax
Though lines tend to get longer. I need to revisit my line wrapping style. 23:26
TreyHarris TimToady: you do have to slide into the gestalt, though. i've noticed that some folks on here seem to be fighting every new thing, instead of seeing the normalization that makes it all so beautiful. braces are closures. no, really, they are. yes, even there. :-)
Juerd And my 80 character limitation.
TreyHarris shudders.
Juerd TreyHarris: Curlies encloding code scares some people even in Perl 5 :) 23:27
TreyHarris Juerd: I've been wrapping every test I'm modifying anyway to 76 columns ;-)
TimToady I try to stay within 80 columns too.
Juerd See recent p5p re ${code; code;}
TimToady: I find it harder with Perl 6.
TimToady 'cuz the methods cascade?
Juerd Mostly for declarations.
TimToady ah 23:28
Juerd I find it very beautiful when the has'es line up without vertical whitespace.
But that's hard if you want to "handles" too.
TimToady I expect we'll see a lot of linebreaks right before the declared word, as in C.
TreyHarris Oh, dear. Declarations have to all be on one line?
TimToady nah.
Juerd TreyHarris: No
TimToady our Lazy multi method 23:29
foo (...)
Juerd has Some::Type $.foobarbaz handles <foo bar baz quux xyzzy but_most method_names are_actually much_longer>;
That really needs the handles thing on a different line, but that breaks a structure that I like so much.
TimToady I'm sure we'll see a lot of vertical <...> things.
Juerd Splitting it to another statement is bad too, because then I'd have to repeat the attribute name 23:30
polettix I know it's a bit unpolite to jump in the middle of a conversation, but I was wondering if the "~" in "~@x.map: { ... }" is really that necessary (from a linguistic point of view)
pasteling "TreyHarris" at 207.171.180.101 pasted "line-broken sigs" (16 lines, 530B) at sial.org/pbot/19858
Juerd If I put the <> list, or part of it, on another line, the horizontal lineup for $.foobarbaz and handles no longer makes sense.
And things start to look like a mess.
TimToady handles q:w:to/END/
Juerd TimToady: I don't mind the <>, multiline. 23:31
TreyHarris I wrote the above a bit ago, though as of yesterday's discussion I realize my parameterized types are now no good
and I guess * needs to be written Whatever in a type declaration....
Juerd will just treat <> like () and {}
has Some::Type $.foobarbaz handles <
TimToady there are times we probably want to distinguish * from Whatever.
Juerd foo bar baz
quux xyzzy but_most
...
>;
TimToady q:w:to/END/ 23:32
foo
bar
baz
END
Juerd prefers <...> then, TimToady
TimToady I like to see the ; on the first line.
Juerd I'm apathetic about ; placement, as long as it terminates things rather than prefix them :) 23:33
TimToady speaking of .join(';')...
Juerd used to care, until he started to indent () like: 23:34
TreyHarris polettix: they would be different. without the initial ~, you'd get:
Juerd foo.bar.baz(
$arg1,
$arg2,
);
TreyHarris ?eval my @x = 1..3; @x.map: { "$_," }
evalbot_r13548 ("1,", "2,", "3,")
polettix aaaaah-s
TreyHarris the ~ concats them together into a single string.
TimToady ?eval my @x = 1..3; @x.map: { "$_," }.join 23:35
evalbot_r13548 "1,2,3,"
TreyHarris Juerd: yes, that's what I was going to say about my paste, I've started to do that too. line up vertically rather than cascading like we usually see in perl 5
polettix but how are those ":" make our golf better than ruby?
23:35 Aankh|Clone joined
TreyHarris hmm... when would ~listy be different from listy.join? i can't think of a case right off, but they are rather different operations, so maybe one could construct one... 23:36
Juerd TimToady: I still want symmetry between split and join, i.e. that join defaults to whitespace. 23:37
TimToady the : means you don't have to use () on your args if there are any.
TreyHarris polettix: hyperops are way easier than .reduce { |x| .... } ;-)
Juerd And ~list joining on null string makes more sense to me anyway
TimToady Juerd: check Functions and you'll see it currently defaults to ' ' 23:38
Juerd TimToady: Ah, cool.
TimToady pugs is behind the times.
TreyHarris s/hyper/meta/
Juerd ++ :)
polettix eheh, good then
Juerd TimToady: Then perhaps it's useful to make ~ join on empty string?
TimToady [~] does that. 23:39
Juerd needs serious tuits, and to reread docs.
TimToady: Good point.
But why would ~ join on space?
TreyHarris ?eval my @a = 1..3; "{~@a}::{[~] @a}"
evalbot_r13548 "1 2 3::123"
TimToady for consistency with all the other stringified arrays.
Juerd Oh, right, interpolation does this too. I tend to forget about stringification of arrays. 23:40
Juerd uses it in golf only :)
TreyHarris Juerd: yes, of course, i could have written 23:41
?eval my @a = 1..3; "@a::{[~] @a}"
evalbot_r13548 "\@a::123"
Juerd TreyHarris: Compare: ~list.blah list.join.blah
TreyHarris ?eval my @a = 1..3; "@a[]::{[~] @a}"
evalbot_r13548 "1 2 3::123"
TreyHarris but i was trying to make the ~ thing explicit
Juerd TreyHarris: The former says "stringify whatever comes", the latter says "stringification is just a step in the process"
TreyHarris: I wasn't referring to your code examples re interpolation, but to Larry's point about stringification. 23:42
TreyHarris Juerd: yep... i got that. but i was thinking about a single token like C<listy>
Juerd TreyHarris: For single tokens it's mostly a matter of expression.
TreyHarris i guess listy could override stringification but not join().
er, whatever listy returns rather. 23:43
Juerd I'll use ~ and + much to mean "scalar".
That is, "item".
TreyHarris *nod* yes. self-documentation is good.
Juerd Whereas join is explicit concatenation.
Something could return something different from a joined string in scalar context! 23:44
TimToady @foo.cat for @foo.join('') and @foo.cat(',') for @foo.join(',') ~ ',' maybe
lambdabot Unknown command, try @list
Juerd TimToady: I think map suffices
TimToady by that argument we get scheme.
Juerd [~] @foo »~ "," 23:45
And by this argument? I wonder what language has all this ;)
TreyHarris Juerd: oh, speaking of self documentation, were you the one who said you didn't understand my point about self-documentation in captures? I was basically just saying that if I have ($lots, $of, :$args, :$in, :$my, *@signature) and i wanted to use a .call later, i'd have to repeat .call(\($lots, $of, :$args...)), and it would be tempting to just change my sig to (\$args) so I can just .call($args)
Juerd TreyHarris: Not that I can recall. 23:46
23:46 awwaiid joined
TreyHarris can't remember whom... 23:47
ah, ok. somebody was saying they didn't understand why i said that having to repeat all your args to get a capture of your args could lead to a loss of self-documentation.
ok, time to go home and set up the new mac. hopefully i'll be doing a smoke-compile-smoke cycle in less than an hour by this time tomorrow... 23:48
TimToady that's why we have |args that now just takes a snapshot en passant.
Juerd TimToady: Ah, so $ and @ can be infix now?
Yay, larger alphabet :)
TreyHarris TimToady: yep, it's a good compromise, like ($myself:).
Juerd OTOH, 3|4 was nicely tight. 23:49
TimToady still is.
TreyHarris is not stating that he is a good compromise....
TimToady that | is where an operator is expected.
Juerd Ah
foo|4 then :P
TimToady no ambiguity
Juerd But that's bad style anyway.
TimToady see my reply to ajs in p6l on that. 23:50
23:50 weinig|away is now known as weinig
Juerd Will do soon. I'm lagging behind on p6l. 23:50
Have been spending some time on p6u.
TimToady noticed that. :)
TreyHarris would |4 be legal for \4?
TimToady no.
that'd be like confusing C's * with & 23:51
Juerd Web programming is how I originally started using Perl. I'm very passionate about it, partly because Perl 6 enables some things that I have wanted for a long time.
TimToady |4 is no different from 4
TreyHarris TimToady: I haven't finished reading your most recent synopsis changes introducing |, sorry
Juerd It took a while for me to realise that roles are exactly what I was looking for. Even when I asked people about why you'd use roles, I didn't get it yet. 23:52
TimToady it's something we're all feeling our way into.
Juerd And roles combined with types just being classes... Splendid.
TimToady well, the proof will be in the pudding, as they say. 23:53
Juerd Re that - is .[] a method or an op?
TimToady We'll just have to see how it all works out.
TreyHarris i thought the whole point of autoboxing is that you *could* ignore the difference between boxing and unboxing. but @(thing) "derefs" and then immediately array-ifies thing, so why wouldn't |(4) "de-sig" and then immediately capture-fy 4?
TimToady but | autointerpolates into its surroundings immediately like @, only more so. So you just end up with a 4 23:54
TreyHarris i mean, @(4) gives you (4,), not...
ahhh
yes. 23:55
TreyHarris smacks his head
Juerd goes to bed
TimToady post hoc...
Juerd 1:55am, alarm clock is set for 7:30 :(
TreyHarris TimToady: thanks. I'll go read the syn changes and write some tests and then it'll all make sense to me
ergo propter hoc
TimToady seems like Juerd goes to bed when you smack your head. 23:56
TreyHarris *because* i smack my head. post hoc ergo propter hoc.
Juerd So that must be why I'm going to bed. Right :)
TimToady *smack*
23:56 jferrero joined
Juerd Funny is that I only yesterday first saw "post hoc ergo propter hoc". I looked it up on Wikipedia. I knew the fallacy, not its name. 23:56
Good night! 23:57
TimToady Good hoc!
TreyHarris videor habe potentum supra Juerdu... 23:59

Powered by Cro and the Raku® Programming Language.

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

Thank you!