»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg p6eval perl6: ... | irclog: irc.pugscode.org/ | UTF-8 is our friend! | Rakudo Star Released!
Set by moderator on 11 August 2010.
00:00 Psyche^ joined
jnthn Time for some rest. o/ 00:04
00:04 macroron joined 00:05 thowe joined
thowe ls 00:05
00:24 ajs joined
ajs I have a module with "multi sub infix:<+>(Foo $lhs, Foo $rhs) is export { ... }" in it. But when I use that module and then try to add two Foos, I get "Can't take numeric value for object of type Foo"... am I doing that wrong? 00:27
TimToady perhaps you need to define how to convert a numeric value to a Foo 00:29
sorear: no, I didn't see the question, but the existence of such a question is sufficient for me to see a problem. :) 00:31
sorear STD.pm6 doesn't implement matchification very well 00:34
TimToady attaching action rules to derived grammars is pretty easy when they're named, and pretty hard when they're anonymos
sorear so I cheat using $M->isa('STD::Regex') etc
TimToady *mous
sorear the most important colliding names are nibbler, LIST, and backslash 00:35
TimToady well, at least those are named subgrammars
the main trickiness is for the reduction to know the pedigree
sorear how does that help me?
AFAICT my code only works because of a bug in Cursor 00:36
TimToady so insted of calling, say Actions::nibbler it can call Actions::Regex::nibbler
sorear _MATCHIFY should be reblessing to Match and erasing the grammar pedigree
but Cursor doesn't have a Match yet
TimToady well, MATCHIFY may be misnamed 00:39
at most, it creates a Cursor that can act a bit like a Match object 00:40
pmichaud TimToady: when you get a chance, I didn't quite follow your answer to my earlier @_ question (more) 00:41
in particular, I'm wondering if @_ should allow modification of its arguments, as in p5
00:42 _jaldhar joined
pmichaud (assuming the arguments are lvalues, of course) 00:42
TimToady I'm saying (if that's the most efficient way to do it), it should allow it if you can get away with it, but the compiler is allowed to complain if it notices 00:43
00:44 _jaldhar joined
TimToady sorear: real matchification is done by a wrapper such as m// or .match 00:45
pmichaud the "compiler is allowed" part is the part I don't understand. Some p6 implementations would be able to disallow argument modification through @_, while others could allow it? That seems very inconsistent
TimToady that's what "erroneous" means
the compiler may complain but is not required to
because it may be too difficult to determine 00:46
pmichaud okay, so the answer I guess I"m hearing is that @_ in p6 doesn't have a way to act like @_ in p5?
00:46 tom_tsuruhara joined
TimToady but they're *supposed* to use "is rw" or "is parcel" to give explicit instruction 00:46
00:46 simcop2387 joined
pmichaud okay, so sub xyz(*@_ is rw) {...} is the normal way to do it? 00:47
TimToady that would require all the args to be rw, parcel would allow some of them to be, and not complain on @[42] = 3
pmichaud is that specific to slurpies, or is that true for positional parameters in general? 00:48
sub xyz(@abc is rw) { ... } # how does this differ from sub xyz(@abc) { ... } ?
TimToady well, enforcement levels may be different between positionals and slurpies, but the intent is that, without rw, @abc.push should be either an error or erroneous 00:50
it seems more worthwhile enforcing that for positionals, somehow
pmichaud so, .push, .pop, etc. need to check the "readonly" nature of the invocant? 00:51
TimToady maybe readonly proxies is one of those things you can turn off for efficiency after you're sure you're okay
pmichaud so far we haven't had any methods that restrict manipulation of the invocant
TimToady the invocant is probably born rw
well, maybe not 00:52
pmichaud that's what gets kinda of weird.
TimToady doesn't matter for $item really
pmichaud sub abc($a) { $a.push(5); }; my @xyz = 1,2,3; abc(@foo); # error?
TimToady but array methods are rare enough that we could require @array is rw: @args, I suppose
pmichaud er, @array
er, @xyz # !!!
sub abc($a) { $a.push(5); }; my @xyz = 1,2,3; abc(@xyz); # error?
TimToady readonliness doesn't go down normal method calls, I suspect, though I can argue it the other way too 00:54
pmichaud right, so if readonliness doesn't go down normal method calls, what makes @a.push different?
TimToady the proxy builder knows it's building an array proxy, maybe? 00:55
("knows" being static knowledge here) 00:56
pmichaud so far we haven't had our "proxies" intercepting method calls
or changing the meaning of them
TimToady people are going to expect readonlyness to apply to the whole composite container, though 00:57
pmichaud if @abc is readonly and denies .push, I'd expect that $a would be readonly and deny .push also 00:58
TimToady that's a very reasonable position
pmichaud so then we'd say that .push requires its invocant to be "rw" ?
TimToady it may well work better to have a capabilities wrapper than a constraints wrapper 00:59
the abstract model is that every mutable type is just a wrapper for the "current" immutable value 01:00
of course, arrays and hashes want to cheat on this
as do strings from time to time 01:01
pmichaud going back to an earlier example, if we have
TimToady but there's some kind of purity about, once you strip off the rw wrapper, you can't put it back on again
pmichaud sub abc(@xyz) { @xyz[0] = 4; }; my @def = 1,2,3,4; abc(@def);
does readonly-ness also extend to the elements in the container, as well as the container itself? 01:02
TimToady the slots are part of the container, so yes 01:06
pmichaud okay.
TimToady for the direct assignment to a slot, not for methods on that slot
pmichaud well, I suspect that postcircumfix:<[ ]> on a readonly array would need to return a readonly proxy to its slot 01:07
TimToady sorry 'bout the delay, the circuit breaker in my kitchen blew, taking down the WiFi
pmichaud or, more precisely, it returns the value and not the slot
sorear remember to take $foo[12] into account
pmichaud (the readonly value)
sorear: yes, got it.
sorear: internally, Rakudo treats those the same righ tnow. 01:08
okay, those clues help a lot. thanks. 01:10
another related question:
my @a; say @a.defined; # false?
TimToady after we split out Nil from (), I expect it to be false 01:11
it may be fairly artificial
depending primarily on whether you've most recently said @a = Nil vs @a = ()
where = Nil represents un-initialization 01:12
pmichaud is splitting Nil from () the likely result? Rakudo had it that way last year (was declared incorrect) and now it's a bit disturbing to be going back to that
TimToady well, I'd let you say "I told you so" :)
pmichaud that would help. :)
what does Nil become, if no longer () ? 01:13
Perhaps Nil is really just Parcel ?
TimToady the Nil eqv () notion, while seemingly simpler, seems to produce heartburn
pmichaud i.e., the Parcel type object?
TimToady yes, perhaps it's an alias for that
pmichaud that way it still has the aspects of a Parcel, but is undefined 01:14
TimToady but that has problems too
I'm still not quite sure that Nil is a type
unless it's "bottom"
pmichaud I did like turning it into a constant a few weeks ago :)
TimToady I think it's a singleton constant that represents the absence of anything meaningful
pmichaud I don't have any particular heartburn over making it into a "bottom type".
Or a special constent, sort of like EMPTY (do we still need EMPTY?) 01:15
TimToady perhaps even our "doesn't exist" value
I think we got EMPTY to go away, but Nil is very like it
except that Nil just interpolates nothing, like ()
pmichaud possibly related: should ().item result in an empty Seq ? 01:16
TimToady and Nil doesn't terminate any outer list like EMPTY did
pmichaud currently (1,2,3).item becomes a Seq
TimToady well, I'd really like to make .item into a no-op, but maybe
pmichaud and (1,@abc,3).item is a Seq 01:17
(so that we get the flattening behavior of parcels)
TimToady whereas Nil.item is probably Failure
or Uninit, or something
pmichaud Failure seems good
TimToady though .item is not the same as .[9] 01:18
.[0] rather
pmichaud right
anyway, if ().item is undefined somehow, then Nil could perhaps remain as ()
(just brainstorming) 01:19
TimToady well, I thought about just defining () to be undefined, but it seems problematic
sorear Wait, we got rid of EMPTY!?
TimToady because the concept of 0 is fairly well defined
pmichaud I suspect that the way to make something undefined would be to use "undefine", as opposed to assigning Nil to it
sorear still has EMPTYT
pmichaud sorear: I haven't needed EMPTY in Rakudo since the list/iterator refactor
sorear: if there's a case where we still need it, it can remain spec 01:20
but I'd need the use case :)
TimToady rakudo doesn't need EMPTY in the same sense that STD never needs an end of list marker 01:21
the list just runs out if it can't return another cursor
sorear I seem to have copied Rakudo's list code wrong then
pmichaud correct.
TimToady cursors are supposed to be immutable, at least in theory
(yeah, STD cheats on that) 01:22
sorear niecza's iterators have $.value and $.next, and the .value === EMPTY for the end of list
TimToady why not just no .next?
pmichaud there may be some fragments of EMPTY floating around in Rakudo, but those are fossils
Rakudo's iterators just keep track of whatever they reified to and return that
sorear TimToady: because then it's impossible to create a lazy empty list
pmichaud (which may or may not include a $.next iterator)
sorear yeah my iterators are pretty broken 01:23
$it.validate unless $it.valid; $it.value
TimToady for an empty list you don't have any cursors to reify in the first place 01:24
(list being a mutable wrapper around a set of reified values and imaginary feeds)
sorear I mean more like 01:25
my @arr = lines "slow.com/empty-file";
# Should return immediately 01:26
s/=/:=/
TimToady yes, the bind should just steal the iterators, more or less 01:27
pmichaud that's what currently happens in Rakudo
(depending on the degree of laziness you're looking for here... it would likely open the connection, but not read any lines until needed by @a) 01:28
(but that may change with lazy IO objects, too)
TimToady in fact, arguably, stealable iterators is the thing that distinguishes a list from an array, which must clone iterators to be multi-iterable
pmichaud s/lazy IO objects/lazy IO opens/
TimToady: so, you're currently leaning towards "Nil as special" as opposed to "() can be undefined"? 01:30
what problems do you envision with () as undefined, ooc?
sorear TimToady: I am having trouble reconciling your talk of stripping layers of rw-ness with my $a = 2; my $b ::= $a; $a = 1; ok $b == 1 01:32
Unless the test is wrong
(Which would actually be a great outcome for me, I think(
pmichaud ...is that a current test?
I'd argue that test is wrong.
no, wait 01:33
I'd argue the test is correct.
(sorry, got my negatives backwards)
TimToady you don't want () to be undefined, or my @foo = function // die; can never return () as a defined new value for @foo 01:35
er, function() :) 01:36
which, if () is defined, can return () to return nothing, or just say 'return;' to indicate that it doesn't expect its return value to be used at all 01:37
assuming return; defaults to return Nil
pmichaud okay. Now that you mention it, this argument sounds strangely familiar.... :) 01:38
whiteknight if I use postcircumfix:<[ ]> as an lvalue, rakudo assumes I'm returning a container to assign into? 01:41
pmichaud whiteknight: it depends more on what you're using postcircumfix:<[ ]> on
arrays already have containers for each element, so they return those directly 01:42
whiteknight pmichaud: I'm writing a matrix type for Rakudo that wraps PMCs I've written for the parrot-linear-algebra project
TimToady it assumes it should return something that can create a container to assign to via WHENCE if the element doesn't exist yet
pmichaud other types return values that may or may not be containers, depending on the base type
anyway, lvalue-ness isn't built into the postcircumfix:<[ ]> operator itself. 01:43
sorear whiteknight: container lvalues in Parrot are the reason I gave up on blizkost. You have been warned.
whiteknight what I really want is to be able to translate $foo[1, 2] = $bar into something like &set_at(1, 2, $bar)
sorear you'll need to use Proxy for that 01:44
whiteknight since I don't have a reference to a cell in the matrix, it's just a buffer
pmichaud whiteknight: well, that doesn't really fit with p6 syntax, since $foo[1,2] would normally be a slice
(returning the second and third elements of $foo)
sorear has Proxy been reimplemented in Rakudo since branches/list?
pmichaud sorear: no; we also haven't needed Proxy. Basically anything that supports STORE as a method can be a proxy, though. 01:45
sorear (as pmichaud points out the correct p6syntax is @foo[1; 2] = $bar
(
))
pmichaud (and rakudo doesn't support the slicel syntax yet)
whiteknight pmichaud: what then is the syntax for multidimensional array indexing?
sorear pmichaud: we haven't *needed* Proxy but it's still useful to have
whiteknight $foo[1][2]?
pmichaud whiteknight: or $foo[1;2] as sorear++ noted above
whiteknight ah, ok.
pmichaud but again, nyi in Rakudo. 01:46
01:46 ___damog joined
pmichaud you could potentially get $foo[1][2] to work, but then $foo[1] needs to return something that can be subsequently indexed as [2] 01:46
whiteknight it doesn't, in this case
pmichaud sorear: since Proxy wasn't/isn't officially "spec" I didn't feel comfortable leaving it in yet. 01:47
(yes, Proxy is mentioned in S06, but I consider that more "fossil" than "official" given its age)
whiteknight okay, so the best thing I could do is create my own proxy object for my matrix types? Or, should I just disallow postcircumfix:<[ ]> because it's too complicate in this case?
pmichaud whiteknight: yes, one of those should work. In rakudo, the current mechanism to provide .[ ] access is to overload the "at_pos" method. 01:48
whiteknight ok
pmichaud it *is* possible to completely overload postcircumfix:<[ ]>, and then you could get $foo[1,2] to work, but it would ultimately clash with other P6 forms of aggregate access 01:49
I suspect it also wouldn't be too difficult to get a basic form of slicels implemented in Rakudo. 01:50
whiteknight the syntax to use for it is not a concern for me. I'm mostly just concerned with the mechanism for using postcircumfix:<[ ]> as an lvalue in one of my matrices
pmichaud right now postcircumfix:<[ ]> simply returns the container directly (for arrays) 01:52
and then that can be assigned to
(since containers are normally lvalues already)
sorear I am now fairly convinced that ::= should strip the old container entirely off 01:53
so changing the container doesn't affect the binding
pmichaud I don't think I'd have a problem with that. 01:55
I've just been following whatever is requested by the spectests :-)
02:05 simcop2387 joined
dalek book: 6151b71 | (Nikolai Prokoschenko)++ | (7 files): 02:11
book: * Introduce new book cover by Sebastian Riedel
book: * Include CC license in the back cover
02:12 dalek joined 02:13 kst joined 02:15 forrest joined
tylercurtis rakudo: my $a; say $a.perl; my @b; say @b.perl; my @b := $a; say $a.perl; say @b.perl; say $a =:= @b; 02:17
p6eval rakudo 9e1bcd: OUTPUT«===SORRY!===␤Redeclaration of symbol @b at line 22, near " := $a; sa"␤»
tylercurtis rakudo: my $a; say $a.perl; my @b; say @b.perl; @b := $a; say $a.perl; say @b.perl; say $a =:= @b;
p6eval rakudo 9e1bcd: OUTPUT«Any␤[]␤Any␤Any␤0␤»
tylercurtis Should that be allowed> 02:18
?
sorear Yes 02:26
Any can be used as an array container
That's how @foo[0][0] works
tylercurtis Ah, right. 02:27
pmichaud I suspect @b := $a should be disallowed 02:29
since Any !~~ Positional
so that's likely a rakudo bug.
(known)
rakudo: my @b := 1; 02:30
p6eval rakudo 9e1bcd: ( no output )
pmichaud rakudo: my @b := 1; say @b;
p6eval rakudo 9e1bcd: OUTPUT«1␤»
pmichaud yeah, bug.
02:37 treed joined
treed Hey guys. 02:37
I don't suppose there's an API reference for Perl6 yet? Something that documents the methods for each class? 02:38
I found Class.HOW.methods, but it doesn't seem like it restricts to the class I actually asked about, and doesn't tell you what each thing actually does.
sorear Try .^methods(:local) 02:46
treed Aha, that helps. 02:48
But still doesn't tell me what they each do. 02:49
I'm actually looking for a way to either reverse a string or to at least access the individual chars so that I can reverse them myself.
02:51 ashleydev joined
plobsing rakudo: "abc".flip 02:51
p6eval rakudo 9e1bcd: ( no output )
plobsing rakudo: "abc".flip.say 02:52
p6eval rakudo 9e1bcd: OUTPUT«cba␤»
02:53 kst joined
treed huh. That's not listed in methods. 02:57
plobsing it's under Cool 02:58
treed What's Cool? 02:59
pmichaud perlgeek.de/blog-en/perl-6/cool.html 03:01
plobsing sort of a catch-all for perl-ish behaviour. also a really hard typename to google
03:15 drbean joined
Tene treed: if you still want a list of chars, use split 03:17
rakudo: say "abcde".split('').perl 03:18
p6eval rakudo 9e1bcd: OUTPUT«("a", "b", "c", "d", "e")␤»
pmichaud or .comb
rakudo: say "abcde".comb(/./).perl
p6eval rakudo 9e1bcd: OUTPUT«("a", "b", "c", "d", "e")␤»
03:19 risou joined
sorear TimToady: I suppose you've heard the news that package Foo { } is legal perl5 syntax now? 03:20
treed Thanks, guys. 03:24
03:35 Holy_Cow joined 03:37 kst joined
ashleyde1 theme 03:48
sorear From now on I'm going to call simple Perl 6 stuff like Int, Num, Str perinative types. 04:02
04:31 redicaps joined
sorear How antisocial would it be for me to overide _REDUCE in Niecza::Grammar? 04:35
04:35 kst joined
pragma_ nazi::grammar 04:38
04:38 ashleyde1 joined 04:42 ashleyde1 joined 04:50 MattMan joined 04:57 cono joined 05:04 tom_tsuruhara joined 05:08 drbean joined
dalek ecza: 14138d4 | sorear++ | Niecza/Actions.pm:
Try harder to return nonzero from the compiler after crashing (...)
 05:09
ecza: 4f89f0b | sorear++ | (2 files):
Use inline code expansions for quantifiers
05:18 envi^home joined 05:36 ingy joined 05:37 Axius joined 05:40 Axius joined 05:43 kst joined
ingy star: say "FOO".lc 05:44
p6eval star 2010.07: OUTPUT«foo␤»
05:59 Axius joined
pugssvn r32000 | pmichaud++ | [t/spec]: *..* is no longer a valid Range, per S02. 06:01
dalek kudo: 8455499 | pmichaud++ | (17 files):
Merge branch 'master' into range
 06:06
kudo: e8f2c6b | pmichaud++ | src/core/Range (2 files):
A few more Range updates for Whatever, .from, .to, .bounds.
kudo: 3f202d5 | pmichaud++ | src/core/ (3 files):
More range improvements. All spectests pass.
kudo: 6be4293 | pmichaud++ | src/core/Range.pm:
No longer need separate checks for +/- Inf in Range.
06:19 whiteknight joined
whiteknight what are the current best practices for making a new module for rakudo? 06:25
06:26 MattMan joined
whiteknight is there a current module builder tool? 06:26
diakopter there are modules? 06:28
whiteknight i dont know what they are called 06:30
sorear modules.perl6.org?
whiteknight is proto the current package manager? 06:31
sorear it's either proto, pls, or ufo 06:35
dalek ecza: eaf583b | sorear++ | CodeGen.pm:
Allow inline type information in the style of pir::weirdop__PSS
 06:40
ecza: 0c64692 | sorear++ | CompilerDriver.pm:
Fix small bug where Foo.dll would be linked against Foo.dll
ecza: 38b3ab2 | sorear++ | (3 files):
Add multithreading support
06:46 kst joined
ingy greetings 06:49
sorear sadly, my lazy lists are not threadsafe, so hyperops are probably a no go 06:50
also STD.pm6 doesn't seem to honor is export on class definitions (?)
whiteknight what's an optional param, $foo?
sorear yes
(hinthinthint: niecza is now probably the first Perl6 implementation with multithreading capability) 06:51
whiteknight does new() not support optional parameters? 06:52
sorear new() is a completely ordinary class method 06:53
(in particular, it doesn't magically create an object; if you override new, you need to call my $obj = self.CREATE; return $obj; yourslef 06:54
)
06:57 jhuni joined
sorear don't expect much love on Threads.pm6 from me, though 06:57
with my 1 CPU I don't plan to dogfood it often
(yes, it predates HyperThreading) 06:58
dalek ecza: 5b8140e | sorear++ | Threads.pm6:
Fudge Thread & Monitor to be correctly exported
 06:59
ingy star: my $a = ['x', 'y', 'z']; say join("\\n", $a)
p6eval star 2010.07: OUTPUT«x y z␤»
ingy star: my $a = ['x', 'y', 'z']; say $a.join("\\n")
p6eval star 2010.07: OUTPUT«x␤y␤z␤»
ingy star: my $a = ['x', 'y', 'z']; say ($a, '').join("\\n") 07:00
p6eval star 2010.07: OUTPUT«x y z␤␤»
ingy star: my $a = ['x', 'y', 'z']; say (|$a, '').join("\\n")
p6eval star 2010.07: OUTPUT«Capture()<0x4baf9f0>␤␤»
ingy star: my $a = ['x', 'y', 'z']; say (@$a, '').join("\\n")
p6eval star 2010.07: OUTPUT«===SORRY!===␤Non-declarative sigil is missing its name at line 22, near "@$a, '').j"␤» 07:01
ingy how do I do the p6 equiv of join("\\n", @$a, '')
whiteknight when I try to pass an optional parameter to my new() method, I get a Null PMC in invoke() error 07:04
ingy star: say "abc".length 07:15
p6eval star 2010.07: OUTPUT«Method 'length' not found for invocant of class 'Str'␤ in main program body at line 22:/tmp/O0w65cRM3N␤»
ingy star: say length("abc")
p6eval star 2010.07: OUTPUT«Could not find sub &length␤ in main program body at line 22:/tmp/MtXJaJc_bD␤»
ingy star: say len("abc")
p6eval star 2010.07: OUTPUT«Could not find sub &len␤ in main program body at line 22:/tmp/P8BHVY9md9␤»
Tene whiteknight: nopaste?
ingy star: say "abc".len
p6eval star 2010.07: OUTPUT«Method 'len' not found for invocant of class 'Str'␤ in main program body at line 22:/tmp/I2vhAFhJay␤»
whiteknight Tene: I did a git --reset and the problem went away
Tene Heh. 07:16
tylercurtis has been reading the Traits papers too much. 07:17
pragma_ I use subversion and I avoid these problems.
ingy star: say "abc".chars
p6eval star 2010.07: OUTPUT«3␤»
tylercurtis When sorear said "... return $obj; yourslef" I thought "wait... Perl 6 has cascades and yourself?" for a moment. 07:18
That's probably a sign I should sleep. G'night/morning #perl6! 07:19
ingy star: say "abc".chars.Bool
p6eval star 2010.07: OUTPUT«1␤»
ingy star: say "abc".chars.Bool.Str
p6eval star 2010.07: OUTPUT«Bool::True␤»
07:19 isBEKaml joined
ingy star: say "abc".chars.Bool.Int 07:20
p6eval star 2010.07: OUTPUT«1␤»
ingy star: say Bool::False.Bool 07:21
p6eval star 2010.07: OUTPUT«0␤»
ingy star: say Bool::False.Bool.Str
p6eval star 2010.07: OUTPUT«Bool::False␤»
ingy star: say 1 ^ 1 07:31
p6eval star 2010.07: OUTPUT«one(1, 1)␤»
ingy star: say 1 ^^ 1
p6eval star 2010.07: OUTPUT«␤»
ingy star: say 1 ^^ 0
p6eval star 2010.07: OUTPUT«1␤»
ingy star: say 0 ^^ 0
p6eval star 2010.07: OUTPUT«0␤»
07:31 Axius joined
ingy star: say 0 ^^ 1 07:31
p6eval star 2010.07: OUTPUT«1␤»
ingy star: say 1 ^^ 1
p6eval star 2010.07: OUTPUT«␤»
ingy I think that's a bug 07:32
star: say (1 ^^ 1) == (0 ^^0)
p6eval star 2010.07: OUTPUT«Method 'HOW' not found for invocant of class 'Undef'␤ in main program body at line 1␤»
ingy star: say (1 ^^ 1) == (0 ^^ 0)
p6eval star 2010.07: OUTPUT«Method 'HOW' not found for invocant of class 'Undef'␤ in main program body at line 1␤»
ingy star: say (1 ^^ 1).Bool
p6eval star 2010.07: OUTPUT«Method 'Bool' not found for invocant of class 'Undef'␤ in main program body at line 22:/tmp/Ajd_iKitH8␤»
ingy star: say (0 ^^ 0).Bool 07:33
sorear lookig for something?
p6eval star 2010.07: OUTPUT«0␤»
ingy sorear: me?
sorear yes
ingy just noting rakudo bugs
sorear p6eval is not a rt bot 07:34
ingy I didn't say it was
sorear you're embedding bug reports in a long dialog
if you're expecting someone to report them for you... 07:35
ingy am I doing something wrong?
I was just playing with p6eval
I can stop if it's bugging you 07:36
07:36 Axius_ joined
ingy star: say "ok" unless (1 ^^ 1) 07:37
p6eval star 2010.07: OUTPUT«ok␤»
ingy star: say (1 ^^ 1) ?? "foo" !! "bar" 07:38
p6eval star 2010.07: OUTPUT«bar␤»
07:40 kst joined 07:41 Guest23195 joined
szabgab rakudo: sub f($a?) { $a //= 42 }; f(2); 07:43
p6eval rakudo 6be429: OUTPUT«Cannot modify readonly value␤ in '&infix:<=>' at line 1␤ in 'f' at line 22:/tmp/GpjYOMVgtf␤ in main program body at line 22:/tmp/GpjYOMVgtf␤»
szabgab I am just wondering if this should not stay alive as there is no eral assignment needed?
s/eral/real/
ingy star: say index("Cat in hat", "hat") 07:45
p6eval star 2010.07: OUTPUT«7␤»
ingy star: say "Cat in hat".index("hat") 07:46
p6eval star 2010.07: OUTPUT«7␤»
ingy prays for a green bar... 07:49
All tests successful. 07:50
Files=9, Tests=47, 118 wallclock secs ( 0.06 usr 0.04 sys + 112.60 cusr 5.13 csys = 117.83 CPU)
thank you, god... 07:51
wow, less than 2 minutes to run a small test suite
frickin ossum!!!
sorear rakudo, or worse? 07:52
ingy rakudo star 07:53
07:54 Axius joined
ingy swears off of perl 6 development for a few days... 07:54
ingy actually wrote a kickass Perl 5 module today. A Moose substitute if you will. 07:57
search.cpan.org/~ingy/Gloom-0.11/li...oom/Doc.pm 07:58
08:01 Axius joined 08:05 wamba joined
Tene ingy: That might be useful for me. I've been tempted to write something approximately like that at work for a while now. 08:10
08:14 Axius joined 08:21 fod joined 08:27 Axius joined 08:38 kst joined 08:39 finanalyst joined
finanalyst pmichaud: ping 08:53
moritz_: ping
moritz_ finanalyst: pong 08:54
finanalyst moritz_: good morning
moritz_: i am looking at pls, as you suggested. 08:55
08:55 molaf joined
finanalyst moritz_: it seems rakudo automatically includes ~/.perl/lib in @*INC but not ./lib 08:56
moritz_ that's correct
finanalyst moritz_: is there any reason for this?
moritz_ rakudo: say @*INC.perl
p6eval rakudo 6be429: OUTPUT«["lib", "/home/p6eval/.perl6/lib", "/home/p6eval//p2/lib/parrot/2.6.0-devel/languages/perl6/lib", "."]␤»
ingy Tene: try it. it's dead simple. 08:57
moritz_ huh, this one does include lib :-)
finanalyst: the desire that a program should run the same everywhere
the '.' at the end is only a stop-gap solution, and will be removed at some point
needed for Test.pm 08:58
ingy Tene: it's just clever packaging of code that I've been using in most of my modules for many years.
finanalyst moritz_: in the proto README it indicates that "good" modules will have a ./lib directory
sorear finanalyst: perl5 doesn't include lib in @INC either
finanalyst: perl5 people are expected to use -Ilib
and the -l option to prove, which passes -Ilib 08:59
finanalyst so what can be assumed about an environment?
moritz_ finanalyst: yes, and a good module will be installed before use
finanalyst if pls is downloaded, it uses modules in ./lib
moritz_ that's a very broad question; care to phrase it less general?
finanalyst but running it gets an error
moritz_ if I were masak, I'd manipulate @*INC inside the program 09:00
but he doesn't like doing that, it seems :-)
I guess the idea is to install pls too 09:01
finanalyst would it be a "bad" idea for rakudo to have './lib' in the rakudo default?
moritz_ for example bundled with the next R* release
I don't know if it's bad, but it's certainly not the solution we are looking for
finanalyst btw there are several environment variables do not appear to be working (i git pulled rakudo a couple of days ago), like $*CWD and %*ENV 09:03
any special reason for this?
moritz_ $*CWD is simply not yet implemented 09:04
(and I think it's the wrong API anyway)
%*ENV suffers from some parrot limitations
isBEKaml finanalyst: can you tell me what error you get when running pls? 09:05
09:05 masak joined
masak oh hai, #perl6! 09:05
moritz_ having the current working directory in a dynamic variable seems oh-so-wrong
isBEKaml AFAICS, pls generates a makefile that utilises the lib folder of the module to be installed.
finanalyst isBEKaml: cant find App:pls
moritz_ because the current working directory is really a process wide property 09:06
finanalyst masak: good localtime
moritz_ whereas $*CWD is tied to a scope
so things like
masak finanalyst: are you attempting to run proof-of-concept?
isBEKaml hi masak! :)
masak finanalyst: then you need to set PERL6LIB=`pwd`/lib
finanalyst masak: just started working on pls . yes proof-of-concept
moritz_ my @a := gather { for 1..3 { my $*CWD = '/tmp/'; take $_ } }; # access items of @a here 09:07
masak finanalyst: that goes for all non-installed projects with a lib/ dir, by the way.
moritz_ would have to jump back and forth, tied to control flow
finanalyst masak: i though putting BEGIN { @*INC.push './lib'} ;# at start of pls
masak finanalyst: please don't. 09:08
finanalyst masak: why?
masak finanalyst: eventually, we want people to install pls instead.
finanalyst: what you're proposing is a crutch in the code for the developers. 09:09
finanalyst: I'd much prefer that the developers suffer a bit, and find ways to make their development environment efficient.
finanalyst \\a nod
masak adds an alias for export PERL6LIB=`pwd`/lib on his box
finanalyst masak: to make it easier for first timers, maybe an install-pls script? 09:10
isBEKaml finanalyst: fwiw, I haven't had that problem when I ran poc last night. and my p6lib contains only ~/.perl6/lib and the parrot lib path.
moritz_ is it intentional that ufo's Makefile installs the .pir files in blib/lib instead of just blib/ ?
masak moritz_: I was surprised the first time I saw that too. 09:11
moritz_: I don't know. gotta ask ingy.
moritz_ masak: I think that 'make' should also copy the .pm file to blib/, otherwise you need both blib/ and lib/ iin PERL6LIB
ingy: see above
isBEKaml masak: yes, I was surprised about that one too. I saw that last night when I tried building yapsi. 09:12
masak moritz_: aye, probably. fwiw, I haven't been able to get PERL6LIB to work with a blib setup yet.
isBEKaml masak: now that yapsi needs rakudo, we need both the .pm amd .pir to be installed into PERL6LIB. 09:13
masak moritz_: maybe copying the .pm files is what's missing.
moritz_ masak: probably; export PERL6LIB=blib/lib:lib works
"works"
masak yeah, that's what I've ended up doing.
moritz_ masak: btw is it intentional that proof-of-concept doesn't install dependencies first?
or is that just NYI?
masak would someone be so kind as to volunteer to make a patch that copies the .pm files into blib? 09:14
moritz_: it doesn't?
moritz_ masak: it doesn't.
isBEKaml moritz_: huh?
masak moritz_: do you have an example?
isBEKaml moritz_: can you give some module that fails in its dependencies?
masak I mean... installing dependencies is in the *core* of pls.
it's not supposed to fail. ever.
moritz_ I jst tried to install Math-Model
masak tries that 09:15
isBEKaml too
moritz_ nopaste.snit.ch/22798
hm, I have some of the deps in ~/.perl6/lib/ already
masak try clearing out cache/ and poc-projects.state 09:16
moritz_ it's LTA that the log file does not exist
masak yeah, that should Never Happen either. but I can see why that one does.
moritz_ with cleared cache and .state, I still get the same output 09:17
masak it's very encouraging that there's a sudden interest around pls. I'm happy about that. but I will spend today with Buf and pack/unpack, if you don't mind. :)
moritz_ tries again with cleared .perl6/lib/
moritz_ would like to see masak's GSOC project finished :-) 09:18
isBEKaml masak: good luck. When's the hard-pencils-down?
masak tomorrow.
(and thanks)
it's looking good, but I need much of today to have something I can call finished. 09:19
isBEKaml :)
masak moritz_: fwiw, pls doesn't seem to install dependencies here either. :/
will investigate... some other time.
sorear masak: niecza has protoregexes now.
masak sorear: awesome. 09:20
sorear is on an optimizing binge now
masak moritz_: hah! a funny thought struck me. I don't remember writing any deps.proto logic for pls. it's probably the case that the tests do dependencies, but proof-of-concept doesn't. :)
isBEKaml sorear: Festival shopping? :)
sorear mm? 09:22
09:24 jfried joined
isBEKaml sorear: poor joke. never mind. :) 09:29
rakudo: role Foo { method foo { say "role"; }; }; class FooClass does Foo { method build { say "class"; }; }; FooClass.new(); 09:30
p6eval rakudo 6be429: ( no output )
isBEKaml rakudo: role Foo { method foo { say "role"; }; }; class FooClass does Foo { method build { say "class"; }; }; FooClass.build();
p6eval rakudo 6be429: OUTPUT«class␤»
isBEKaml rakudo: role Foo { method foo { say "role"; }; }; class FooClass does Foo { method build { say "class"; }; }; FooClass.new.build(); 09:31
p6eval rakudo 6be429: OUTPUT«class␤»
masak isBEKaml: you never call .foo
isBEKaml is there any chaining order for methods in the role and class thats share the same name?
blah!
masak isBEKaml: class methods override role methods.
isBEKaml masak: you're right!
rakudo: role Foo { method foo { say "role"; }; }; class FooClass does Foo { method foo { say "class"; }; }; FooClass.new.build();
p6eval rakudo 6be429: OUTPUT«Method 'build' not found for invocant of class 'FooClass'␤ in main program body at line 22:/tmp/7sBuUsiaAJ␤» 09:32
isBEKaml rakudo: role Foo { method foo { say "role"; }; }; class FooClass does Foo { method foo { say "class"; }; }; FooClass.new.foo();
p6eval rakudo 6be429: OUTPUT«class␤»
masak isBEKaml: class methods override role methods.
isBEKaml masak: yes, right.
Tene rakudo: role Foo { method foo { say "role"; }; }; class FooClass does Foo { method foo { say "class"; }; }; FooClass.new.*foo(); 09:33
p6eval rakudo 6be429: OUTPUT«class␤»
Tene Just curious.
masak they don't stack on top of them, IIUC. they actually push them out.
rakudo: role Foo { method foo { say "role"; }; }; class FooClass does Foo { method foo { say "class"; self.Foo::foo }; }; FooClass.new.foo 09:34
p6eval rakudo 6be429: OUTPUT«class␤role␤»
isBEKaml callsame?
rakudo: role Foo { method foo { say "role"; }; }; class FooClass does Foo { method foo { callsame; say "class"; }; }; FooClass.new.foo();
p6eval rakudo 6be429: OUTPUT«class␤» 09:35
masak isBEKaml: well, no. since they're replaced, not hidden.
isBEKaml: what's important to realize, I think, is that roles aren't part of the class hierarchy.
isBEKaml: there's no MRO with roles.
in fact, roles are the antithesis of all that.
09:35 kst joined
isBEKaml masak: I just find it a bit odd that classes can mixin roles without any MROs for methods that share the same name. 09:36
masak isBEKaml: that's what roles do.
isBEKaml masak: plainly pushing out role's methods in broad "code"light. :) 09:37
masak (slight terminology nit: classes don't mix in roles, objects do. classes just "do" roles.)
isBEKaml masak: noted.
masak anyway, the underlying reason is this: consider a number of roles being done by a single class. let's say there are no methods in the class hiding the role methods. 09:38
now, when there's a collision between the methods of two roles, you get a compile-time error. 09:39
that's what makes roles awesome.
isBEKaml Roles are the fringe guys! Can I call them waterboys? :)
masak to silence that error, you're supposed to put a method in the class that decides what to do. it might call one method, call the other, or do something more fancy. 09:40
so putting a method in the class actually indicates that you know what you're doing (overriding the conflicting role methods)
isBEKaml rakudo: role Foo { method foo { say "FooRole"; }; }; role Bar { method foo { say "BarRole"; }; }; class FooClass does Foo does Bar { method foo { say "clazz"; }; }; FooClass.new.foo; # trying them out 09:42
p6eval rakudo 6be429: OUTPUT«clazz␤»
isBEKaml rakudo: role Foo { method foo { say "FooRole"; }; }; role Bar { method foo { say "BarRole"; }; }; class FooClass does Foo does Bar { method foo1 { say "clazz"; }; }; FooClass.new.foo1; # trying them out 09:43
p6eval rakudo 6be429: OUTPUT«===SORRY!===␤Method 'foo' collides and a resolution must be provided by the class␤»
masak \\o/
isBEKaml masak++
masak jnthn++
isBEKaml this is what is MRO? or just method chaining the hierarchy? :P 09:44
(I upvoted you, you did jnthn++) :P
masak MRO has nothing to do with roles, but is the mechanism for choosing methods in a class hierarchy. 09:45
isBEKaml (I wasn't speaking technical when I said MRO)
masak class A { method foo() }; class B is A { method foo() }; class C is A { method foo() }; class D is B is C {}; C.new.foo; # which method gets called?
isBEKaml: ah, a joke. I see. 09:46
isBEKaml: no, that was more like method delegation... :)
which is different, since it involves several method calls.
isBEKaml masak: for your poser, I'd guess A.
masak isBEKaml: I believe the answer (under C3 MRO) is B. 09:47
isBEKaml masak: no, C is A , C and then A does it.
masak rakudo: class A { method foo { say "A" } }; class B is A { method foo { say "B" } }; class C is A { method foo { say "C" } }; class D is B is C {}; C.new.foo
p6eval rakudo 6be429: OUTPUT«C␤»
masak isBEKaml: seems you were more right. :) 09:48
isBEKaml: but only one method is called, unless I do .*foo or .+foo
isBEKaml rakudo: class A { method foo { say "A" } }; class B is A { method foo { say "B" } }; class C is A { method foo {say "C" } }; class D is B is C {}; C.new.*foo 09:49
p6eval rakudo 6be429: OUTPUT«C␤A␤»
isBEKaml \\o/
masak: I plainly took a guess from whatever I knew of java. class hierarchies play by calling through the method chain in case of overridden methods. 09:50
rakudo: class A { method foo { say "A" } }; class B is A { method foo { say "B" } }; class C is A { method foo {say "C" } }; class D is B is C {}; C.new.+foo
p6eval rakudo 6be429: OUTPUT«C␤A␤»
masak .*foo and .+foo work the same, but the latter expects at least one method, or it'll fail. 09:51
s/fail/die/
isBEKaml rakudo: class A { method foo { say "A" } }; class B is A { method foo { say "B" } }; class C is A { method foo {say "C" } }; class D is B is C {}; D.new.+foo
p6eval rakudo 6be429: OUTPUT«B␤C␤A␤»
isBEKaml masak: akin to * and + in regexes? 09:52
masak: btw, shouldn't you be working on your GSoC now? (I sometimes find IRC to be too much of a distraction :D) 09:53
masak isBEKaml: yes, akin to * and + in regexes. 09:54
isBEKaml: what makes you think I'm not? :) but you have a point.
isBEKaml masak: Go ahead. :)
TiMBuS std: (state => 'a').perl.say 09:58
p6eval std 31912: OUTPUT«ok 00:01 113m␤»
TiMBuS rakudo: (state => 'a').perl.say
p6eval rakudo 6be429: OUTPUT«===SORRY!===␤Malformed state at line 22, near "=> 'a').pe"␤»
Tene also .?foo, fwiw
masak TiMBuS: just quote 'state' and you'll be fine for now. 09:59
TiMBuS: it's a known rakudobug.
Tene: yes. .?foo is to .foo what .*foo is to .+foo
Tene masak: the regex comparison extends to .?foo too 10:00
masak aye.
TiMBuS ok then masak. wasnt sure if it was a rakudobug or a perl6 thing
masak do we have those (?*+) in some other place of the language, besides regexes and method call modifiers?
&pack has * :) 10:01
Tene we have $*foo and $?foo, but not $+foo, and they don't have regex-like meanings. 10:02
masak we used to have $+foo. and no, I don't count those :)
Tene sigs use $foo? and *@foo, which are kind of similar
I don't remember $+foo
masak I think $*foo used to be globals and $+foo was contextuals. 10:03
Tene I've seen people want a *@foo that always matched at least one item, so there's an argument to be made for adding +@foo in sigs
masak they merged into $*foo being contextuals.
Tene: that's a nice idea.
Tene: on the other hand, empty list already trumps *@foo, so you kind of get that behaviour for free.
Tene masak: you mean, add a multi sig with an empty list there? 10:04
masak aye.
Tene 'k
masak Tene: that was decuded at YAPC::EU 2009. 10:05
Tene I like that.
masak s/cud/dic/ 10:06
ergbhfl.
s/dic/cid/ :)
10:06 finanalyst left 10:08 tadzik joined
tadzik o/ 10:08
masak \\o 10:10
news of Rakudo Star in Castellano. nice. www.programacion.com/noticia/lanzam...erl_6_1850 10:12
maybe we'll see more Castillian programmers here in a while. 10:13
10:18 proclus joined
moritz_ that would be nice 10:19
masak ¡guay! 10:22
10:24 risou joined 10:27 kst joined 10:31 Trashlord joined 10:33 barney joined
masak yay! .encoding now does latin-1 10:38
(locally)
where do you guys stand on allowing more than just 'iso-8859-1' as the $encoding parameter? 10:39
things like 'ISO-8859-1' and 'latin-1' won't ever be used for something else. is this a case where synonyms might be a good thing?
same question about 'utf-8', 'UTF-8' and 'utf8'. 10:40
arnsholt There's an RFC somewhere which lists synonyms for encodings. I think at least allowing some of the synonyms there for utf-8 and latin-1 might be nice 10:41
masak doing 'latin-1' => 'iso-8859-1' for now. and treating everything case-insensitively. 10:42
moritz_ +1 10:44
is there a case where removing dashes could cause name clashes?
or at least the first dash?
masak dunno. 10:45
but I don't want to put too much magic in there from the start.
better to begin conservatively and see where people trip up.
good docs are probably worth much more in this case. 10:46
masak ♥ Perl 6 10:47
die "Unknown encoding $encoding" unless $encoding.lc eq any @KNOWN_ENCODINGS;
moritz_ pushes ufo patch 10:53
hope I resolved the merge conflict correctly :/
tadzik rakudo: (1..100).pick for(1..100) 10:54
10:54 avuserow joined
p6eval rakudo 6be429: ( no output ) 10:54
tadzik segmentation fault here
rakudo: (1..100).pick for(1..100); say 'alive'; 10:55
p6eval rakudo 6be429: OUTPUT«alive␤»
tadzik oh you
oh, not this rakudo
tadzik--
szabgab mb 10:56
wrong window and wrong command 10:57
masak szabgab: but we're happy to have you here :)
hugme: hug szabgab
hugme hugs szabgab
szabgab anyway, just finished 4 hours of screencast creating, result : 10 min
masak \\o/
szabgab++
tadzik masak: you stole my reaction 10:58
szabgab++\\
\\o/
masak watches the screencast: szabgab.com/blog/2010/08/perl6-subr...ators.html
tadzik sounds tempting 10:59
but it's not on the list
awesome intro :) 11:01
masak indeed. 11:03
are the screencasts linked from perl6.org yet?
szabgab masak: yes 11:04
masak good.
11:05 avuserow joined
masak TimToady: loose thought: if Nil gets split into one defined and one undefined form, and thus @a and %h can suddenly be set to be undefined, maybe it's time to make unpassed array and hash parameters undefined? :) 11:08
11:11 avuserow joined
masak agrees with Whiteknight in wknight8111.blogspot.com/2010/01/pr...h-pir.html 11:13
11:14 patcat88 joined
jnthn afternoon, 6folk 11:17
masak jnthn! \\o/
jnthn masak! 11:18
Trashlord hey dudes
masak Trashlord! \\o/
Trashlord how's it going, masak?
hey, you're from sweden, right?
masak aye.
Trashlord cool
masak :)
Trashlord I'm going to move to Finland within the next 6 months 11:19
got family there
masak Trashlord: nice! we could visit each other.
Trashlord so we'll be a lot closer to each other ;p
yeah
masak \\o/
ok, going down for lunch and software updates. see you on the other side. :)
jnthn: Str.encode now does different encodings :)
11:20 avuserow joined
jnthn mantovani: yay! 11:22
masak++
Trashlord: Ooh, where in Findland?
er, Finland :-)
jnthn goes to make coffee so he can type properly :-)
Trashlord haha 11:23
well
I have family in Vantaa
and Lovissa
and friends in Helsinki
jnthn I've only ever been in Helsinki. But I liked it. :-)
Trashlord last time I was there was 8 years ago, when I was 13
I remember going to Linanmakki, which is a big theme park 11:24
I was also in the northern area of Finland, totally awesome
fishing at 2am, the sun still hasn't totally set
so amazing
11:25 redicaps joined
jnthn Yeah, I don't get midnight sun here, sadly, but in the middle of summer it only got dark for a couple of hours, if that. 11:25
Trashlord I think Finland will provide awesome opportunities for personal growth, for me 11:26
jnthn (I'm in the very south of Sweden, so quite a way from the arctic circle).
Trashlord ah
szabgab all this talk about artic circle
Trashlord My aunt says I can study in a university, scholarship from the government 11:27
szabgab I have 38C in my room
Trashlord if that's true, I will complete my high school education, and then move on hopefully to computer science
yes, I'm 21, and I'm a high school drop out, at this point, living in a little village in the mountains of northern Israel
jnthn szabgab: Ouch!!! 11:29
Trashlord: Sounds like nice plans. :-) 11:30
Trashlord yeah ;p 11:32
I can't wait to camp out in a forest, middle of nowhere
or go to a trance party in some forest
11:38 avuserow joined 11:39 redicaps left 11:50 pmurias joined
pmurias ruoso: hi 11:50
11:58 masak joined
masak \\o/ 11:58
hm, next week is release week, innit? 12:05
tadzik today's the the of the Parrot release, isntit?
isBEKaml 6Compiler, 2Star, innit? :)
masak tadzik: no, Parrot releases are on Tuesdays.
tadzik oh 12:06
right
masak second of the month, to be exact.
tadzik ehm, counting from 0?
isBEKaml always the third tuesday.
masak meybe it's the third. 12:07
jnthn: we should go visit Trashlord in Finland and have a hackathon there :)
jnthn masak: \\o/ 12:08
masak: I wonder how many projects we could finnish on that trip...
isBEKaml great! _Finnish_ as much as possible! :) 12:09
isBEKaml tries reading parrot docs 12:10
masak jnthn: you're punbearable. 12:12
jnthn stops before he gets punished 12:13
masak fsvo 'stops' :) 12:14
jnthn distracts himself with implemented named args 12:17
*implementing
12:17 azert0x joined 12:19 whiteknight joined
dalek kudo: 10041cb | (Solomon Foster)++ | src/core/ (2 files):
Implement infix:�<=>�(Num, Num), infix:�cmp�(Num, Num), and

On the plus side, this knocks about 10% off the execution time of Range iteration on Nums. (It's still mysteriously about 50% slower than Range iteration on Ints.) On the minus side, all of these functions are violating the spec by not using the Order enum.		 12:20
masak bah. spec, schmeck.
colomon++
12:24 kst joined 12:27 isBEKaml_ joined
masak szabgab: at szabgab.com/perl6.html, "Next Perl 6 training" is out-of-date. 12:37
szabgab: by the way, in the end I didn't stop by that training session. I hope it went well.
12:42 molaf joined
szabgab masak: I noticed you were not there :) 12:43
masak :)
masak was in Firenze
szabgab TimToady and pmichaud were the trouble makers this time :)
masak oh, good.
moritz_ how many "real" attendends were there?
szabgab 1.5 :) 12:44
masak how many of these were half attendees, and why? :)
szabgab szbalint was the 0.5 as he came in only after lunch 12:45
masak ah. szbalint+
moritz_ lol
12:46 drbean joined
szabgab the viewer stats of the screencasts are very strange 12:46
12:47 Mowah joined
szabgab if I take all the time since my first perl 6 cast , then 15% are women 12:47
if I take the last 10 days then 0.2% are women
12:48 dual joined
szabgab between 27/7 - 2/8 that was 22% 12:49
12:54 rindolf joined 12:57 amoe joined
masak $Config{intsize} # what's the Perl 6 equivalent of this? 12:57
12:58 vpm left 12:59 amoe left 13:09 orafu joined
masak rakudo: sub foo($a, $b) { say $a; say $b }; foo |(42 xx 2) 13:10
p6eval rakudo 10041c: OUTPUT«42␤42␤»
masak sometimes wishes for a listop form of prefix:<|>
C<flatten>, perhaps. 13:11
13:11 lithos joined
jnthn Not sure how that'd be implemented... 13:13
13:15 kst joined 13:28 tewk joined 13:30 sftp joined 13:31 barney joined
masak rakudo: sub foo(*@a) { say shift(@a); say @a.perl }; foo(1,2,3) 13:33
p6eval rakudo 10041c: OUTPUT«1␤[2, 3]␤»
masak shouldn't @a be readonly there?
moritz_ it is 13:39
pmichaud good morning, #perl6
moritz_ it's just not immutable
\\o pmichaud
pmichaud masak: see my discussion last night with TimToady
masak pmichaud: aye, that's surely related. 13:40
pmichaud: what did you conclude? :)
pmichaud basically, iiuc TimToady++ said that mutating @a is erroneous but compilers aren't required to catch it
masak oh right.
because it might be intractable.
pmichaud well, more that it might be inefficient 13:41
masak slowtractable.
moritz_: I find it... unsatisfying... to allow readonly things to be mutable. 13:42
pmichaud well, I'm thinking of going ahead and doing the checking in Rakudo
masak: did you see the discussion re: Nil? 13:44
masak yes.
I'm hoping the fallout of this might be that @param? isn't defined when no corresponding argument was passed in. 13:45
pugssvn r32001 | pmurias++ | [mildew] update the -Cgtk backend to Forest::Tree::Viewer::Gtk2
pmichaud oh, I'm pretty certain Rakudo has that wrong now. 13:46
same for my @param; # @param should be undefined
pmurias szabgab: how many viewers did the screencast have?
masak pmichaud: up until now, they've been defined by spec. :( 13:47
pmichaud they have?
masak pmichaud: my longest RT ticket is about that.
hold on, I'll get it for you.
pmichaud please do
masak pmichaud: rt.perl.org/rt3/Ticket/Display.html?id=64928 13:48
my conclusion from that discussion was that I'll never be happy about the state of things. now there's a new chance to make things right. :)
13:49 macroron joined
pmichaud that discussion is.... weird 13:50
masak it's a bit old, if that's what you mean.
pmichaud I've been thinking that "my %h" initializes %h to be the Hash type object
(thinking recently, at least)
with a WHENCE property that vivifies %h into a true Hash when it's used
szabgab pmurias: on the first 2 days they get about 200 and then 3-4 / day (each screencast) way less than I was hoping for :) 13:51
masak ooh
pmichaud same for %h? as a parameter
masak \\o/
pmichaud would that solve the problems identified in the ticket?
masak oh yes, oh yes 13:52
pmichaud (I'm having trouble following the issues there)
masak dances
masak adds this new development to ticket
pmichaud but I see that as being somewhat orthogonal to the Nil issue :-) 13:53
s/somewhat/mostly/
masak aye.
pmichaud: did you see my summary of the two problems I have with Nil definedness right now?
pmichaud masak: sure, I understand that part -- in fact, I somewhat argued the same things in 2009 (unsuccessfully) 13:54
masak oh, ok.
pmichaud back then, Rakudo was treating Nil as definitely different from an empty Parcel
masak I can see why you hesitate to switch back to a model like that. :)
pmichaud (frankly, there was a lot of confusion in my head as to what Nil represented... finally after cornering TimToady++ at YAPC::NA for a while we concluded that Nil was an empty Parcel. Looks like we were wrong.) 13:55
(but I didn't like the definedness of Nil then either :-)
masak :)
pmichaud yes, I'm mainly hoping that switching Nil back doesn't end up with huge ramifications for a lot of the other parts of Rakudo. In the past it always has. 13:56
(But I think we'll be okay this time.)
masak at some point we had Nil( warn "oops, this loop had no elements!" ) too
pmichaud yeah, that's part of what made me make it type-ish 13:57
masak but that turned into sink ...
pmichaud I speculated last night that Nil might be an alias for the Parcel type object
rakudo: say Parcel.list.perl; 13:58
p6eval rakudo 10041c: OUTPUT«()␤»
pmichaud hmmmmmmmmmmm
rakudo: say (Nil).list.perl;
p6eval rakudo 10041c: OUTPUT«()␤»
pmichaud rakudo: say (Nil, Nil).list.perl;
p6eval rakudo 10041c: OUTPUT«((), ())␤»
pmichaud rakudo: say (Nil, Nil).flat.perl; 13:59
p6eval rakudo 10041c: OUTPUT«()␤»
pmichaud rakudo: say (Parcel, Parcel).flat.perl;
p6eval rakudo 10041c: OUTPUT«(Parcel, Parcel)␤»
pmichaud well, it can't be a true alias
13:59 squeeky left
TimToady Nil still has to print out as Nil :) 14:00
pmichaud print or .perl?
TimToady I still suspect Nil is a singleton value more than a type 14:01
pmichaud I'm wondering if it's just a special instance of Parcel
TimToady could be, but Nil.perl needs to say Nil, not Parcel
pmichaud sure
maybe Nil is just an empty Parcel with a .defined mixin ? 14:02
TimToady huh?
those are both supposed to be undefined now
pmichaud I meant an empty Parcel instance, not the type object
TimToady ah 14:03
szabgab I just read the signatures slides of jnthn, very nice but I got lost starting from Zavolaj :)
pmichaud () but role { method defined() { Bool::False } }
jnthn You could easily sneak a method Str { 'Nil' } in there too :-) 14:05
pmichaud right
although I suspect Nil may be more along the lines of Failure then
so perhaps () but Failure
14:05 kst joined
jnthn szabgab: I imagine it's easier to follow from there with me explaining it. ;-) 14:05
TimToady I wonder if the new Nil can be used to skip an argument and trigger the default: foo(1,Nil,3) 14:07
pmichaud seems plausible 14:08
TimToady Nil basically being a value that every container refuses to store
szabgab jnthn: I guess so 14:09
moritz_ what would say ~(my @a = 1, Nil, 2) do?
pmichaud "1 2\\n"
TimToady Nil goes to () in list context
moritz_ if you can't store Nil in a Parcel, how can you make it trigger a default value in a signature? 14:10
pmichaud you can put Nil in a Parcel
TimToady well, a parcel is "raw" so can handle them
pmurias TimToady: what are the kinds of control flow (things that can cause jumps in code) besides conditionals, loops, exceptions and control exceptions
?
TimToady: i'm trying to figure out where i need to put the edges in my basic block graph 14:11
TimToady pmurias: lemme think about that in the shower
masak pmurias: exit 14:13
pmurias TimToady: ok
masak just implemented the C S L n N v V directives in &pack 14:32
github.com/rakudo/rakudo/commit/a98...8339d4adbd
time to attack &unpack, methinks.
14:33 JimmyZ joined
masak 卓明亮! \\o/ 14:33
JimmyZ 麦高，下午好 14:35
masak 下午好
你好吗?
JimmyZ 还好，最近比较忙，你呢 14:36
masak 好,好
JimmyZ O(∩_∩)O~
masak 我还真忙... 14:37
JimmyZ 呵呵，我准备去睡觉了，晚安，明天还要上班
masak 晚安! 14:38
14:46 kst joined
pugssvn r32002 | moritz++ | [bench] .gitignore 14:47
14:47 dudulz joined 14:48 jt_ joined
jnthn church & 14:49
14:51 seungwon joined 14:52 dual joined
TimToady pmurias: I don't think you'll be able to do basic block analysis until you know exactly which language you're parsing, and how it uses lambdas, and have inlined those lambdas that can be inlined. 14:55
also, I think you'll have to ignore the issue of exceptions, since they're supposed to be exceptional 14:56
and almost anything could throw an exception
14:56 ruoso joined
TimToady also, exceptions themselves aren't control flow, only the stack unwind they might trigger 14:56
also church & 14:57
masak ignoring 'next', 'redo' and 'last' feels kinda debilitating...
pmichaud what should be the result of 9/3 cmp 4 ? 14:58
moritz_ rakudo: say 9/3 cmp 4
pmichaud or, perhaps more interestingly: 90/3 cmp 4
p6eval rakudo 10041c: OUTPUT«-1␤»
moritz_ rakudo: say 90/3 cmp 4
p6eval rakudo 10041c: OUTPUT«1␤»
moritz_ is this a trick question?
pmichaud sorta.
is &infix:<cmp> defined for differing types?
moritz_ I think so 14:59
masak if those types both fall on the real number line, I certainly hope so.
pmichaud okay, how about this one:
masak that feels like a small thing to ask.
TimToady I meant fatal exceptions, not control exceptions, which by definition control control flow :)
moritz_ it would be kinda un-nice of sort() to die on a list of numbers
pmichaud rakudo: say 9/3 cmp 3
p6eval rakudo 10041c: OUTPUT«0␤»
pmichaud rakudo: say 9/3 eqv 3
p6eval rakudo 10041c: OUTPUT«0␤»
TimToady &
moritz_ both seem right to me 15:00
pmichaud ...that tells me that &infix:<cmp> cannot be defined in terms of eqv semantics, as S03 suggests
masak aye.
moritz_ scratch S03 then
masak down with S03!
pmichaud I'm fine with that.
moritz_ eqv can only give True or Fals anyway
pmichaud right
I'm thinking that cmp on Numeric and cmp on String (and cmp on Any) has little to do with eqv
moritz_ aye 15:01
pmichaud okay, here's another question. S03 says that $a <=> $b is defined as +$a cmp +$b
but colomon++ seems to have this completely the other-way-round -- he's defined cmp in terms of <=> 15:02
I think the S03 definition should be the correct one here.
moritz_ it kinda makes sense to me to define something more complicated in terms of something simpler
ie cmp in terms of <=> and leg
but maybe that's just the weird view of misguided scientist 15:03
pmichaud I have trouble thinking of <=> and leg as being the 'simpler' ones
since they're the ones that do explicit casting.
whereas cmp doesn't cast.
(or, at least, shouldn't)
I somewhat see you point, though -- in some sense 'cmp' can be the higher-order comparison when the types differ. 15:04
*your
moritz_ I found <=> and leg alway much easier to explain, so to me they are "simpler" 15:05
then cmp
15:05 seungwon left
pmichaud easier to understand, perhaps, but that doesn't make them more fundamental :-) 15:05
electrons, protons, and neutrons are easier to understand, but not necessarily fundamental :) 15:06
moritz_ never grokked those electrons 15:07
pmichaud anyway, the fact that cmp is defined in terms of <=> is part of what is making Rakudo so slow at the moment.
it's de-optimizing comparisons and type checking.
colomon pmichaud: errr...
not sure how an extra function call there makes much difference one way or the other? 15:08
most of the super common cases are already optimized away anyway.
pmichaud actually, they're not.
in fact, the reason why the previous RangeIter was so slow is because Integers aren't well optimized for comparison. 15:09
colomon that was a problem a month ago.
pmichaud yes, and it's still there. The special-cased InfiniteRangeIter was simply avoiding the slow integer comparisons 15:10
I went back to a normal-ish RangeIter and fixed up integer comparisons and it ended up being faster than even InfiniteRangeIter
moritz_ we need profiling tools. 15:11
pmichaud currently, comparison of two Ints using infix:<after> ends up doing after -> cmp -> <=> -> Bridge -> Num on each integer 15:12
that's..... ugly.
colomon it's also trivially fixable
every other case but before and after has already been fixed 15:13
so far as I know
pmichaud no
even comparing two ints using == ends up doing == -> <=> -> Bridge -> Num
colomon no it does
pmichaud oh, you're right
colomon doesn't
multi sub infix:«<=>»(Int $a, Int $b) {
pir::cmp__III($a, $b);
}
pmichaud anyway, I'd argue that cmp(Int $a, Int $b) should also use the pir::cmp
colomon whoops, wrong one. 15:14
which it does
multi sub infix:«cmp»(Int $a, Int $b) {
pir::cmp__III($a, $b);
}
pmichaud oh
colomon you added that one
pmichaud silly «
colomon I added all the other Int Int and Num Num cmps and <=>s this morning
pmichaud oh, perhaps I didn't git pull
masak colomon++
15:15 justatheory joined
pmichaud anyway, my question remains -- should cmp be defined in terms of <=> or vice-versa? 15:15
the spec says it should be the other way around... and I'm inclined to agree.
colomon what are the practical differences?
masak cmp does feel like the general one.
pmichaud <=> always coerces to numeric
and then does a numeric cmp 15:16
dalek kudo: 21f3104 | moritz++ | build/PARROT_REVISION:
bump PARROT_REVISION to get some testing after the merge of the parrot
kudo: 869b1fc | moritz++ | src/core/ (2 files):
Merge branch 'master' of github.com:rakudo/rakudo
pmichaud it's not always possible (or desirable) to take the arguments to cmp and make them somehow Numeric
in particular, if I have a type that defines .Numeric, then <=> should work for that type without having to overload <=> for it 15:17
moritz_ I don't understand the difference, as long as cmp only re-dispatches to <=> when both args are Numeric
which I think is currently the case; otherwise 'a' cmp 'b' wouldn't work at all 15:18
pmichaud it's more a question of what <=> does when its arguments aren't numeric, I think.
moritz_ rakudo: for <a b c> { say $_ cmp 'b' }
p6eval rakudo 10041c: OUTPUT«-1␤0␤1␤»
moritz_ it coerces
either way
pmichaud moritz_: it coerces and then uses <=> on itself?
colomon I think pmichaud is right. 15:19
pmichaud I mean, $a <=> $b ends up being +$a <=> +$b ?
(when $a and/or $b aren't numeric)
moritz_ yes
pmichaud same for leg, then?
colomon I was thinking of <=> as more fundamental, but actually cmp(Numeric, Numeric) is the fundamental, in some sense.
pmichaud i.e., $a leg $b is ~$a leg ~$b ? 15:20
moritz_ yes
colomon well, hmmm.
moritz_ in my mind, cmp(Numeric, Numeric) and <=>(Numeric, Numeric) behave exactly the same
why does it matter which one is more fundamental? 15:21
colomon moritz_: overloading
pmichaud because of what you may or may not need to define for custom types....
15:21 ruoso joined
moritz_ hm 15:21
I'd buy that argument if we had lift(), and used it
colomon at the same time, it sure seems like a world of suck if <=> is less efficient than cmp is for numbers.
pmichaud actually, I suspect that <cmp> gets used far more often than <=> at the moment 15:22
all of the comparison functions are defined in terms of &infix:<cmp>
colomon right
pmichaud so right now we pay the penalty on things like sort and series
because they're defined in terms of cmp and before and after 15:23
and I don't see a problem with having an optimized <=> for numbers
but the default <=> should probably be defined in terms of cmp 15:24
colomon it certainly seems to me we need to optimize cmp, before, and after for numbers.
pmichaud well, optimizing cmp is probably good enough
but that gets back to "this is what the spec wanted in the first place" 15:25
colomon it's just hard to see how to balance that with not pessimizing <=> for the Numeric, Numeric case AND making it easy to overload the operators for new types.
15:25 kst joined
pmichaud ? 15:25
I don't see the problem.
moritz_ as long as we don't optimize (Numeric, Numeric) but specific types (Int, Int) and (Rat Rat) we should be fine 15:26
colomon by spec, if you define a new Numeric type and a new cmp on it, that should change the operation of <=> for that type as well, right
>?
pmichaud by spec, <=> would be defined in terms of the types .Numeric, and have nothing to do with cmp 15:27
colomon I thought you said <=> was defined as +$a cmp +$b? or is that only if <=> is operating on non-Numeric?
arnsholt Does nqp-rx handle multiple return values?
colomon at any rate, agreed that cmp should be better, and will fix that today after lunch if I get the chance. 15:29
pmichaud <=> is defined as +$a cmp +$b. I suspect that <=> on numerics can be optimized.
colomon right now, I've got to get a small guy (he insists, "baby" no longer) ready for lunch. :)
btw, what dowe need for lift? 15:31
*do we
moritz_ I plan to give lift a shot once we have compile-time lexpads
oh wait, maybe we don't even need that... 15:33
15:33 ashleydev joined
pmichaud yes, I'm not sure how 'lift' enters into this picture. 15:33
oh, because you want <=> to have lifted cmp semantics?
that wouldn't feel right. 15:34
moritz_ huh, my docs/test_summary.times isn't valid JSON 15:37
there are commas missing 15:38
dalek kudo: f100e9d | pmichaud++ | src/core/ (3 files):
Using double-angles around 'cmp' makes them harder to find (e.g., with ack).
 15:39
pmichaud decides to pound on book building for a while. 15:40
15:43 justatheory joined
pmichaud when building the book, I keep getting errors about UsingPerl6.A4.d not being found 15:46
(or UsingPerl6.Letter.d) 15:47
what is it looking for and where?
colomon pmichaud: that's what made moritz_ think of lift this morning, but in general, it seems to be a significant issue for people overloading operators. 15:48
pmichaud colomon: I don't understand.
colomon if you define new versions of the standard operators right now, they don't work with sort (for comparison operators), or any of the meta-ops. and that's just the tip of the iceberg 15:49
pmichaud example?
oh, I understand what you mean by meta-ops, yes. 15:50
colomon someone had one a day or two ago. if you define your own cmp, sort doesn't recognize it unless you explicitly pass it in
pmichaud right
that part I understand. That seems unrelated to the factoring, to me.
colomon well, if <=> is defined in terms of cmp, then it may be an issue if you overload cmp. 15:51
pmichaud *no*
colomon but that's really very much a side issue to the factoring, agreed.
pmichaud the default <=> always uses infix:<cmp>(Numeric, Numeric)
that's the point.
unless prefix:<+> somehow returns something that isn't Numeric. 15:52
overloading cmp for a custom type wouldn't have anything to do with <=>
because <=> always ends up calling cmp(Numeric, Numeric) 15:53
colomon the default <=> (in theory) should always use +$a cmp +$b
pmichaud right
colomon but that's not necessarily cmp(Numeric, Numeric)
pmichaud and +$a (in theory) should always return a Numeric
colomon yes, but cmp may be overloaded for something else.
like Int, Int
pmichaud but that doesn't matter! 15:54
you're saying someone would overload cmp(Int, Int) ?
colomon no, I'm saying someone probably will overload cmp(BigInt, BigInt). 15:55
or whatever.
gotta run, alas. more later.
pmichaud ....why would we have a BigInt?
Int is already BigInt
pmichaud is now *very* confused.
masak colomon: speaks, confuses, and leaves. :) 16:00
huf confuse a rakudo dev in collaboration with confuse a cat ltd 16:01
pmichaud is there anyone around knowledgable about the build system for the book that can help me get past my current "can't build the book" woes?
moritz_ pmichaud: pronik is in #perl6book 16:02
pmichaud: he wrote most of the build system
pmichaud moving to #perl6book 16:03
16:07 kst joined 16:13 15SAAORV0 joined
moritz_ nine: ping 16:14
16:19 Clem joined
masak nom & 16:29
frettled nom nom
frettled is (slowly) finishing off around 400-500 g tenderloin of pork. 16:30
gfldex std: class A { method &.($c) {}; }; 16:33
p6eval std 31912: OUTPUT«Potential difficulties:␤ $c is declared but not used at /tmp/NZDBEfrEXT line 1:␤------> [32mclass A { method &.([33m⏏[31m$c) {}; };[0m␤ok 00:01 114m␤»
gfldex std: class A { method &.($c) { say $c.WHAT }; };
p6eval std 31912: OUTPUT«ok 00:01 116m␤»
gfldex rakudo: class A { method &.($c) { say $c.WHAT }; };
p6eval rakudo 10041c: OUTPUT«===SORRY!===␤Malformed method at line 22, near "&.($c) { s"␤»
16:35 ajs joined 16:38 azert0x joined 16:39 ashleydev joined 16:42 proller joined
gfldex std: use v6; my Any $o; my Code $m; class Foo { method bar(){ say "bar called"; }; }; my $f = Foo.new(); $o = $f; $m = &Foo::bar; $o.$m(); 16:47
p6eval std 31912: OUTPUT«ok 00:01 121m␤»
gfldex rakudo: use v6; my Any $o; my Code $m; class Foo { method bar(){ say "bar called"; }; }; my $f = Foo.new(); $o = $f; $m = &Foo::bar; $o.$m();
p6eval rakudo 10041c: ( no output )
16:48 kst joined
pmichaud gfldex: method bar doesn't appear in the namespace unless you specify 'our' 16:51
gfldex i did asked p6eval because i got a null pmc, what should be fixed with updating rakudo 16:52
16:53 meppl joined 16:56 barika joined, ruoso joined
jnthn back 17:03
17:07 Axius joined 17:09 masonkramer joined
rindolf jnthn: hi. 17:14
jnthn o/ rindolf 17:15
17:32 kst joined 17:46 pnate joined 17:48 whiteknight joined 17:51 Su-Shee joined
Su-Shee yo folks. 17:51
17:51 molaf joined 17:52 mberends joined
rindolf Hi Su-Shee 17:52
Su-Shee: how was your weekend? 17:53
Su-Shee: thanks for following me on Twitter.
Su-Shee rindolf: exhausting, I partied into my birthday. ;)
rindolf Su-Shee: ah.
gfldex happy birthday Su-Shee! 17:54
rindolf Su-Shee: Mazal Tov!
colomon pmichaud: last I heard, someone out there was trying to write a BigInt class to use until we get Int going in the core. But it was just a generic example.
17:55 molaf joined
colomon happy birthday! 17:57
avar wonders if anyone here saw blogs.perl.org/users/aevar_arnfjor_...erl-6.html 17:58
gfldex std: use v6; my Any $o; my Code $m; class Foo { has Int $.counter = 0; our method bar(){ $.counter++; say "bar $.counter times called"; }; }; my $f = Foo.new(); $o = $f; $m = &Foo::bar; $o.$m(); 17:59
p6eval std 31912: OUTPUT«ok 00:01 121m␤»
gfldex rakudo: use v6; my Any $o; my Code $m; class Foo { has Int $.counter = 0; our method bar(){ $.counter++; say "bar $.counter times called"; }; }; my $f = Foo.new(); $o = $f; $m = &Foo::bar; $o.$m();
p6eval rakudo f100e9: OUTPUT«Cannot modify readonly value␤ in '&infix:<=>' at line 1␤ in 'Foo::bar' at line 6559:CORE.setting␤ in main program body at line 22:/tmp/CbX0kSXDtX␤»
gfldex what am i doing wrong? gist.github.com/525747 18:00
pmichaud Su-Shee: happy birthday? what day? 18:03
s/\\?/!/
colomon: even presuming someone creates BigInt (let's call it MyInt for now), I don't see why that poses a difficulty. 18:04
well, I see how the lack of 'lift' might be a difficulty then. 18:05
Su-Shee pmichaud: today. :)
pmichaud Su-Shee: happy birthday. my birthday was 2 days ago. There are a lot of August birthdays it seems :-)
Su-Shee pmichaud: oh, congrats. :) yeah, we also have 4 in the office (out of 9 people ;) 18:06
colomon \\o/
Su-Shee what you too? ;)
mberends happy birthday to y'all :) 18:07
rindolf pmichaud: happy birthday. 18:08
18:12 Eevee joined
colomon pmichaud: BigInt is actually a pretty good choice for an example, because it points up the potential problem nicely. If the BigInt developer writes infix:<cmp>(BigInt, BigInt), should she expect infix:«<=>» to use that infix:<cmp>? It's a significant thing, because generic infix:<cmp>(Real, Real) is only as precise as Num... 18:13
pmichaud colomon: would the BigInt developer also have .Numeric defined?
colomon pmichaud: yes
pmichaud colomon: then <=> would end up using cmp 18:14
because <=> would call .Numeric, the results of which would then be passed to cmp
18:14 kst joined
pmichaud presumably .Numeric would return self in that case 18:14
colomon but without lifting, it won't call cmp(BigInt, BigInt) 18:15
because that doesn't exist in the setting
pmichaud agreed, we'd need lift there to make that ultimately work
colomon right, that's why we were talking about lift
pmichaud 18:05 <pmichaud> well, I see how the lack of 'lift' might be a difficulty then.
colomon gotcha
pmichaud anyway, the person implementing BigInt is free to overload <=> as well. 18:16
having cmp be based on <=> is the exact same problem in reverse :-)
colomon absolutely
pmichaud jnthn: ping 18:19
phenny: tell jnthn what was the trick you used to output pbc files from PIR? I tried it on the plane home from .it and couldn't get it work except in very limited cases 18:22
phenny pmichaud: I'll pass that on when jnthn is around.
18:24 Felipe__ joined 18:25 Felipe__ joined, leprevost joined 18:34 Eevee joined 18:37 fod joined
jnthn pmichaud: pong 18:37
phenny jnthn: 18:22Z <pmichaud> tell jnthn what was the trick you used to output pbc files from PIR? I tried it on the plane home from .it and couldn't get it work except in very limited cases
jnthn pmichaud: You get the eval PMC and write it to a file iirc. 18:38
Oh, stringify it first I think
$S0 = the_eval_pmc
Then write that.
ingy jnthn, pmichaud: gist.github.com/525796 18:40
ingy is writing a little blog post :) 18:41
jnthn (BigInt) I suspect that this may be do-able by the Int class mixing a BigInt role into itself. What we don't want is to always do bigint operations or to have to do a check on every operation to see whether we've got an upgraded Int or one just using the native int type. In essence, it'd somewhat be a type-based decision but we can hide it and claim to be an Int to the world either way.
colomon woo-hoo, only about 100x slower!
ingy colomon: _under_ 2 mins! \\o/ 18:42
colomon ingy: was it worse before?
ingy colomon: I don't know
jnthn 100x slower is bad, but I'm optimistic we can significantly improve on it. :-) 18:43
ingy jnthn: of course you can
it's just a matter of when
colomon jnthn: huh. how would that interact with operators?
ingy I want to make the case that maybe soon is a good time to shift focus towards that for a while
colomon ingy: you sure don't have to convince me 18:44
ingy well you have to pick some goals over others
jnthn ingy: That's why I seem to have disappeared from day to day Rakudo development at the moment.
ingy jnthn: to speed it up, or two program in a fast lang? 18:45
;)
jnthn ingy: I'm actually madly hacking away at a bunch of new meta-model prototyping and working out other changes to get us there. :-)
colomon: The whole way we do operator bodies now is So Wrong And Silly.
colomon: For numeric operations.
colomon jnthn: oh?
jnthn It's going to change drastically.
Yes
Here's what sucks 18:46
18:46 tadzik joined
jnthn We do a multi-dispatch by type 18:46
We often then use Parrot opcodes
ingy jnthn: to be fair, testml does a runtime parse of the testml docs. I suspect that's the main issue.
jnthn Which go and do a multi-dispatch too.
ingy jnthn: that's why I want to get pegex to p6
tadzik o/ 18:47
colomon pegex?
ingy but that's waiting for an external regex engine
colomon: pegex.org
jnthn Oh, actually I think we switched to using the add_nnn variants...maybe
ingy colomon: short answer: p6 rules for all langs
colomon: currently implemented in Perl 5, and Python very soon 18:48
jnthn OK, then it's not quite so bad. But we still have twice as many PMCs (or three times as many perhaps) as we really need when we have an Int object. :)
ingy jnthn: is there any tools I can use to see where the main problems are?
jnthn ingy: Parrot has some profiler that may be able to help. 18:49
colomon I've been wondering what would be required to implement a Rakudo profiler.
jnthn ingy: But not sure how well it works on Rakudo code.
colomon: Well, take the Parrot one + make it aware of HLL annotations if it isn't = not bad :-)
ingy jnthn: I'll need some tuit days for that. Maybe we can pair up at osdcfr
jnthn ingy: Yes, I did before manage to get the profiler to work. :-) 18:50
18:51 dalek joined 18:54 kst joined
sorear good * #perl6 18:56
ingy sorear: whatever
tadzik sorear: o/
ingy ;)
jnthn o/ sorear 18:57
18:57 masak joined
masak ahoj! 18:57
ingy sorear: gist.github.com/525796 I'd like to see niecza's comparison :) 18:58
tadzik lolitsmasak!
ingy hi masak
colomon \\o
masak waves :)
jnthn lawltismasak!
masak Su-Shee: happy belated birthday!
oh, it's today, even. I'm on time :) 18:59
Su-Shee I'm still suffering, hasn't ended yet. ;)
jnthn Happy birthday, Su-Shee :-)
Suffering? Sounds like your doin it rite. :-)
masak .oO( birthday rites... ) 19:00
Su-Shee and I don't even drink ;)
pmichaud jnthn: (write pbc) hmmm
masak what kind of suffering are we talking about here, then?
jnthn Su-Shee: Oh! 19:01
pmichaud: How were you doing it?
Su-Shee masak: the 12 hour partying until 7 suffering, mostly.
pmichaud jnthn: gist.github.com/525830
masak Su-Shee: wow. sounds like you *are* doing it right :)
sorear ingy: it'll be nearly the same as rakudo 19:02
ingy: for test suite-like things, niecza and rakudo runtime are both dominated by the parser 19:03
jnthn pmichaud: youch!
sorear niecza's parser is ~50% slower, but the backend compiler is currently faster
Su-Shee does anybody know how the dancer port is coming along?
jnthn pmichaud: I was generally generating the PIR and then translating it to PBC and writing that in the same proccess.
sorear so maybe 2m 20s
jnthn pmichaud: Not sure why that would make a difference though. 19:04
tadzik Su-Shee: mberends is caring about it
pmichaud jnthn: in this case I'm taking some pretty common and uninteresting PIR, though
mberends Su-Shee: the Dancer port is at the stage of: "OMG there is sooo much to do..." 19:05
pmichaud I can get it to work fine from a .pir file -- just not from NQP, or anything that has libraries loaded
anyway, I have to run errands with @family, so I'll bbl. I just didn't remember if there was anything fancy you did to make it work.
jnthn k
Su-Shee mberends: oh, that stage...
19:05 cono joined
pmichaud anyway, at this point --target=pbc doesn't look at all likely. 19:05
the following *does* work, though 19:06
masak mberends++ # reaching that stage
pmichaud gist.github.com/525841
okay, gotta run
bbl
jnthn pmichaud: I don't remember there being. 19:16
We probably need to get to the bottom of what's wrong there, otherwise we're going to be a bit screwed further down the line. :-/ 19:17
19:18 Minimiscience joined, KenGuru joined 19:22 tylercurtis joined
KenGuru 2(sorear2): i'm tryin to compile niecza on windows. I'Ve installed all dependencies. Mono 2.6.7 and svn are allready installed. I'Ve installed Strawberry 5.12.1 and CPANed all dependencies. getting this error, any ideas: pastebin.com/D1g8F45F 19:22
19:30 Italian_Plumber joined 19:35 kst joined
dalek kudo: c4b6df5 | (Solomon Foster)++ | src/core/ (2 files):
Move the infix:�<=>�(Numeric, Numeric) logic into infix:<cmp>(Numeric,

infix:<cmp>(Numeric, Numeric). Also implement both infix:<cmp>(Real, Real) and infix:�<=>�(Real, Real) in terms of infix:<cmp>.		 19:38
sorear KenGuru: what version of make do you have? 19:45
KenGuru erm, the one from the mono bin
3.81 19:46
sorear gmake? nmake? bsdmake?
KenGuru gnu make 3.81
i386-pc-mingw32 build
running on win7 x64
:D
i should use a linux distro to build ... 19:47
masak is really enjoying using the less frequently used operators +> and +< today 19:48
tadzik what are they?
masak at one point, I had »%» in the code :)
tadzik: binary shifts.
tadzik oh
masak tadzik: I'm writing pack and unpack.
tadzik: in C, they look like this: >> and <<
tadzik yeah, I see the progress on GH. masak++
I know them :) 19:49
masak tadzik: there are a few more commits to come before I pack it in tonight :)
sorear KenGuru: just to confirm, what is line 20 of STD_checkout/Makefile
masak this error in the ng flavour of Rakudo really means "syntax error": "Unable to parse blockoid, couldn't find final '}'" 19:51
19:51 x3nU joined
masak it would be slightly less awesome of the error said what it found instead. :) 19:51
to its credit, the error message does provide the right line and file :) 19:52
KenGuru PERL6LIB=./lib:. 19:53
jnthn masak: slightly mess or slightly more? :-) 19:54
er, less :-)
KenGuru sorear:PERL6LIB=./lib:.
masak jnthn: hm. I meant "slightly less LTA" :)
sorear KenGuru: somehow, PERL6LIB is getting set to .\\lib;. 19:56
which is wrong
:/
jnthn SMA (Slight More Awesome) :-)
masak adopts that at once 19:58
it could also mean Somewhat More Awesome, Subtly More Awesome, or So Much Awesomer :) 19:59
mberends masak: don't pack it in yet! (scnr)
masak mberends: it's going extremely well tonight.
mberends: I'm basically tying up loose ends. 20:00
jnthn oh no
SMA also carries a slight connotation in the first two letters that it may be a bit painful to get the improvement. :-)
mberends masak++ # very valuable work
KenGuru sorear:which path do I have to specify?
masak jnthn: thought that was official Perl 6 policy. :) 20:01
tadzik mberends: you meen HTTP::Body and MIME::Types as Dancer prerequisites, right?
jnthn masak: Will packtest.t become a spectest before you merge? 20:02
masak jnthn: yes. I also have an unpacktest.t here now.
both will become t/spec files.
mberends tadzik: the former maybe, but it could be an internal class rather than an external dependency, The latter probably should be an external prereq. 20:04
jnthn masak: Excellent.
sorear KenGuru: you specified .\\lib;. yourself? 20:05
I thought that was just make being stupid and trying to rewrite paths 20:06
KenGuru 2(sorear2): the default from your makefile was .\\lib:i and i changed it to .\\lib;.
masak jnthn: after that, I thought I'd do prefix:<~^> infix:<~&> infix:«~<» infix:«~>» infix:<~^>, and then call it a night :)
jnthn :-) 20:07
sorear KenGuru: it should just be ./lib:., no i involved
masak pugs: say "foo" ~& "bar"
p6eval pugs: OUTPUT«bab␤»
sorear CursorBase uses split ':' to break up $PERL6LIB
masak dang, Pugs already has those :) 20:08
sorear unfortunately with this big maze of Makefiles my situation is fairly no-win 20:09
KenGuru i'll try setting up a nix ... 20:10
20:11 gbacon joined
sorear I can't just use $Config{path_sep} because, well, the Makefile doesn't have access to it 20:11
maybe I should just switch STD to Module::Build
:-)
KenGuru it's a path separator problem
sorear yes
: is the only path separator STD supports for %PERL6LIB%
KenGuru its an OS issue 20:12
changed the ./lib:. to .\\lib:. and now i've got it running, but now faced with another error ...
sorear ./lib didn't work? 20:13
masak rakudo: say 48 + (0 +< 8); say (48 + 0) +< 8; say 48 + 0 +< 8; # should equal former (since +< has multiplicative tightness), but equals latter
p6eval rakudo f100e9: OUTPUT«48␤12288␤12288␤»
KenGuru no
masak submits rakudobug
KenGuru curiuos
sorear what's the new error? 20:14
20:14 kst joined
masak rakudo: say 48 + 0 +> 8 20:15
p6eval rakudo f100e9: OUTPUT«0␤»
masak also wrong.
jnthn lolkassoperator 20:17
masak it's a good sign that I didn't immediately jump to the conclusion that Rakudo was wrong... as I would have one or two years ago :) 20:18
Rakudo++ 20:19
KenGuru sorear:pastebin.com/t0AVUi7y
mberends sigh. The p5 OpenGL examples seem to be so near to p6, yet so far. On Ubuntu they Just Work with p5. 20:21
20:21 Guest23195 left
sorear KenGuru: oh right, that. 20:26
20:30 drake1 joined
drake1 hello, can you pack for a specific alignment in perl6? 20:30
masak all unpack tests pass! yay 20:31
drake1: funny you should ask that :)
drake1 crazy
mberends masak++ \\o/
masak drake1: presently, no.
sorear KenGuru: have you changed anything else in STD_checkout?
masak drake1: but if it works in Perl 5, there's a big chance we'll get it in the next month or so.
(in Rakudo) 20:32
sorear the fact that it's not working doesn't suprise me, but I can't figure out how you got *to* that error...
masak drake1: if you give me a use case, I can probably fix it for you in shorter time that that.
drake1 masak: I don't think perl5 has it
masak drake1: I know perlpacktut talks about alignments.
drake1: grep for 'Align' in perldoc perlpacktut 20:33
drake1 ok
KenGuru no
masak ok, time to merge with master, before I get too greedy and drift away from it. :)
KenGuru i started xbuild once
svn did a checkout of the std 20:34
drake1 masak: there's a whole chapter about packing C-structures that I didn't read. thanks
KenGuru i missed the cpan dependencies first
grab them via cpan install module 20:35
masak drake1: when you've read it, and come back wanting the same for Perl 6, let me know. :)
KenGuru and did a xbuild again resulting in the error
masak latest changes are up at github.com/rakudo/rakudo/tree/buf 20:36
commencing merge/build/spectest.
KenGuru i wisely not used the program files folder because of windows7 madnes ...
2(sorear2): what os (distrubution) do you use to build niecza? 20:38
20:39 snearch joined
sorear debian (though, this system was installed quite some time ago) 20:39
drake1 masak: in perl5 you have to specify the alignment through padding templates
sorear snarkyboojum has gotten it working on OS X 10.5
not sure what moritz_ is using
masak drake1: aye.
sorear it troubles me that it doesn't work on Windows thoguh
masak drake1: it's crude but it works.
drake1: if you come up with a less crude way, I'd like to hear about it. 20:40
drake1 masak: it's fine enough
sorear I guess the sane thing to do would be to make STD use the filesystem less
drake1 masak: some kind of "align" command
masak drake1: 'some kind of' qualifies as hand-waving :) 20:41
I was thinking more of a grandiouse vision :P
go wild!
drake1 a perl embedded kind, then
masak: like a generic variable for the alignment 20:42
pugssvn r32003 | masak++ | [t/spec/S32-str] added pack.t/unpack.t for the new Rakudo &pack/&unpack 20:43
r32003 |
r32003 | Actually, there was an old pack.t there, but it was all wrong,
r32003 | so bulldozed it.
masak drake1: sorry, I'm not following. it's a bit too abstract.
drake1: I have my brain on a very short leash. it likes things to be concrete.
as in code examples.
KenGuru 2(sorear2): i'll testing it for you, if you're going to do any changes .. 20:44
drake1 $#=32 for 32 bit alignment or whatever. to match the local C compiler it might make sense to use template specifiers with C struct standard padding 20:45
sorear testing to make sure I haven't broken *nix takes 20 minutes and all my memory
so I don't like to change it much
:/ 20:46
masak drake1: a variable such as '$#' sounds like a p5ism :) I'm guessing Perl 6 would want something more like a specialized DSL for packing aligned things. probably with fewer symbols and more words.
drake1: but beyond that, I'm not sure how it would look. 20:47
drake1 masak: something like a struct declaration
masak ooh
maybe even *exactly* like a C struct? 20:48
drake1 pack("int;char;struct test *;" etc
sorear constant $?ALIGN then do caller.hints('$?ALIGN')
masak and it'd just get translated into the right pack/unpack template :)
sorear niecza: sub foo() { say caller.hints('$?ALIGN'); }; { constant $?ALIGN = 8; foo(); }
p6eval niecza 5b8140e: OUTPUT«(timeout)» 20:49
sorear I don't have CALLER::<$?ALIGN> working yet though
niecza: sub foo() { say caller.hints('$?ALIGN'); }; { constant $?ALIGN = 8; foo(); }
p6eval niecza 5b8140e: OUTPUT«8␤»
sorear ok, it only works on a hot cache
interesting
masak hot cache? 20:52
KenGuru 2(sorear2): thx anyway. i'm goin to use *nix for compilation. is there a binary release for niecza?because i'll have to make a fresh install
20:59 kst joined
sorear masak: the first time I ran niecza on p6eval, it timed out, presumably due to virtual memory 21:00
masak sorear: ah, so you didn't do anything in between? I see. 21:01
"Continuations in Natural Language". whoa. www.cs.bham.ac.uk/~hxt/cw04/barker.pdf
drake1 (union int;char)char for union padding. maybe it's best to let it be
masak drake1: depends on one's willingness to implement that. :) 21:02
drake1 yes
masak drake1: would you agree that it'd be interesting to follow the C declaration syntax quite closely? 21:03
drake1 masak: then it's probably faster to implement the binary interface
masak I'm not familiar with that.
drake1 perl XS or something 21:04
masak oh.
ingy blogs.perl.org/users/ingy_dot_net/2...lowly.html
masak sounds like Zavolaj, then.
ingy: thanks for phrasing that as positively as you did :) 21:05
masak would have it as just "Rakudo is slöööööööööw" :)
ingy hähä 21:07
drake1 actually an "open(BINOUT,"|ascii2binout")" with a scanf might be a lot more flexible so why even bother
with the output buffer, it's typically faster than a call chain in C 21:09
sorear ingy: Would you take a look at Niecza's Test.pm6 and tell me if I'm on the right track? 21:10
ingy I was chatting with nothingmuch earlier today. He basically thinks that Perl 6 had no future. But I think really think you guys are doing great things. Who cares if no company ever takes over the world with Perl 6...
-Ofun forever
sorear: sure
21:10 Ken-Guru joined
sorear ingy: I spoke with nothingmuch a couple months ago 21:10
ingy sorear: I should try to run the TestML suite on niecza 21:11
I just think nothingmuch is completely burnt out on programming
sorear ingy: his arguments boiled down to "this is a trainwreck of adding features with no regard for efficient implementation and it will never be usable without an overhaul of process"
ingy seems accurate 21:12
sorear multi dispatch was his favorite target
masak seems to me TimToady is thinking quite a lot about hotspots for optimization.
21:12 p6eval joined
ingy wonders why TimToady would even care about such things 21:13
that's what @jnthn is for
masak ingy: if that kind of division really existed, I would probably agree with nothingmuch.
ingy I guess @jnthn ~~ $TimToady
masak I respect nothingmuch's opinion, but to me it sounds like an outsider peering in, and not quite seeing what's happening. 21:14
perhaps he's been reasing p6l too much :)
ingy masak: totally agree
I was telling nothingmuch about how much energy was in this channel
masak I like -Ofun too, but if I didn't believe what we have will eventually be fast, usable, and in some sense disruptive, I'd probably wander off. 21:15
21:15 azert0x joined
japhb mberends, You mentioned P5 OpenGL examples a while back -- which ones are you referring to? 21:15
ingy I personally have no stake in Perl 6, but the energy here is fantastic for Acmeism
masak I hope to someday be as experienced a programmer as nothingmuch. and I hope that by then, Perl 6 will be the tool that I wield for everyday production tasks :) 21:16
Juerd What's acmeism?
ingy masak: -Ofun means finding out how to get speed in some insanely cool way that nobody has thought of yet
sorear masak: nothingmuch is, or was, the Audrey of #moose. Being jaded is quite out of character for him
masak Juerd: acmeism represents proglang anti-tribalism, IIUC.
ingy sorear: like I said, he's just going through a little programming burnout 21:17
Juerd I know what represents means :) \\o/
All the other words... no idea.
ingy usually @sex is the cure...
Juerd @sex, not %sex?
ingy :D
masak Juerd: tribalism. Python people hating on Perl people hating on PHP people.
Juerd &sex :)
ingy Juerd++ 21:18
masak Juerd: I should have written 'programming language', but I was lazy.
Juerd Oh!
ingy Juerd: acmeism.org
Juerd See... that's why I don't like abbreviations :|
masak Juerd: how's that commit bit coming along?
oh, and anyone else want to vote on the whole .fmt/.format thing?
Tene masak: what's the actual question? 21:19
masak grep yesternight's backlog for it, if you want the backstory.
Tene :P
drake1 I made a fast request object, compatible with bash, python and perl, so they're only used for changes
masak Tene: short story: Juerd wants .fmt to be named .format
Tene I'd see that as a positive change. 21:20
as long as nobody suggests that we support both options.
jnthn t cld b clrr.
masak I count one vote from Tene for .format
Juerd Tene: Or, alternatively, .f 21:22
pmurias no
;)
Juerd But that'd be for historic reasons: printf, sprintf
print $foo.f("%s");
printf("%s", $foo);
masak please no
Tene I'd rather dislike that.
ingy bbl 21:23
jnthn .f would be quite the huffmanization fail.
masak we just got rid of one-char methods for Str. let's not put new ones on Any.
Juerd But my preference would be .format
masak or Cool, as it were.
jnthn .format is fine imo, but .fmt doesn't bother me too much either.
masak same here. 21:24
sorear I like fmt
pmurias likes format
Juerd I think abbreviations should be avoided if the original isn't very long.
masak who introduced .fmt? TimToady?
Juerd Except for standard library/system calls
Tene I don't mind fmt 21:25
Juerd Tene: It's ugly and a scary precedent
Next is msg/mess/mesg
masak it's a bit cute.
sorear That's not a scary precedent
Juerd masak: Cuteness requires vowels.
sorear Please explain to me how % is in any way mnemonic
jnthn Juerd: I feel very uncute. 21:26
Juerd sorear: It's not. % is ugly.
masak hugme: hug jnthn
hugme hugs jnthn
Juerd sorear: % is as bad as fmt, imo.
21:26 becu joined
sorear Juerd: I'm talking about %hash now 21:26
Juerd sorear: The mnemonic is that the character used in the template is also %.
I see.
Tene Juerd: There are many different types of formatting you could do to something. "fmt" isn't an english word already, so it's clear that it has a specific meaning.
Juerd Thought you referred to python's sprintf, which is %
21:26 simcop2387 joined
masak sorear: % in %hash is a visual mnemonic. the two thingies around the oblique line represent key and value, respectively. :) 21:27
Juerd Tene: If that were a good argument, we should avoid all preexisting words.
Tene I often have at least some degree of skepticism for using generic english words for specific language meanings.
sorear If I want vowels with my programming, I know where Java is
Juerd sorear: rkd prl 6
sorear Changing fmt to format sets a terrible preceent 21:28
Tene That way lies "It's all english words that I already know, so *ANYONE* can read this code and know exactly what it means and reason about it without knowing the langauge!!"
Juerd prnt $f, $b;
Tene: At least I know how to *spell* the word.
masak this is the interesting discussion I expected yesterday :)
sorear Juerd: prt is better
Tene Juerd: "fmt" is *a different word*.
sorear or say 21:29
Juerd With abbreviations, there's no way other than just remembering it, to know if it's frmt or fmt or even formt
masak maybe we should go with .form :)
Juerd While they'd all be prononuced as "format" when read out loud.
jnthn $baby.form
Juerd masak: I could certainly live with that.
masak Juerd: it's prefix of .format, *and* a different word.
Tene I pronounce .fmt as "dot eff emm tee" 21:30
sorear I wonder what ever happened to the whole "infinitely customizable language" selling point
use Juerd::Format;
masak sorear: that selling point is still there.
Juerd Tene: You should take into consideration all obvious pronunciations.
Tene: Not just the one you'd use.
masak sorear: but we want Standard Perl 6 to be good enough for many people.
Juerd sorear: I refuse to discuss dialects; the standard language should be neat and consistent regardless of all ways to extend it. 21:31
Tene agrees.
masak that's what I meant.
Juerd fmt looks out of place
masak .oO( No-one tell Juerd about Rat.nude... ) 21:32
Juerd nude rats?
masak oops. :)
Tene Rat.numerator_and_denominator
masak rakudo: say (3/4).nude.perl
Juerd At least it's not Rat.nd
p6eval rakudo c4b6df: OUTPUT«(3, 4)␤»
Juerd Although that'd be consistent with .kv
keva
masak keva! \\o/ 21:33
Juerd In fact, .nd would probably be better.
masak aye.
...contradiction all your arguments so far. :)
Juerd Oh well, I'll just rant about inconsistent identifiers when the language is "done"
And write something like tnx.nl/php about it :)
masak s/ion/ing/
Juerd masak: No. This is a significant abbreviation. 21:34
masak: Which makes it very useful.
afk
masak "Perl 6 is like a training car without brakes." 21:35
Tene .fmt => .put-pieces-where-the-percent-signs-are()
21:35 theclaw joined
Juerd Maybe I care too much about the bikeshed's appearance. 21:36
But I do care.
21:36 theclaw left
masak Juerd++ 21:36
Tene I also agree that .format would be an improvement
Juerd afk
masak so far, .format seems to come out slightly in favour. 21:37
hope Juerd finds his commit bit.
Tene .fmt => .favour
masak .fmt => .the-method-formerly-known-as-fmt
Tene .u formerly
phenny Tene: Sorry, no results for 'formerly'.
Juerd Where's the info about how to check out?
masak Juerd: you can check out anytime, but you can never leave. :) 21:38
Tene check out rakudo?
Juerd Tene: Specs
Was that the pugs svk thing?
Tene They're in the pugs svn repo
/docs/Perl6/Spec/ 21:39
masak Juerd: I'd use git svn nowadays. much preferable.
Tene svk is dead, iirc
Juerd Is this documented somewhere?
masak Juerd: I can set up a small workflow for you, if it'd help.
Juerd Yes please!
Tene svn co svn.pugscode.org/pugs/docs/Perl6/Spec/
masak does it
Juerd I haven't used anything like this in ages
masak Juerd: don't listen to Tene; you'll just get the svn :)
Tene trac.parrot.org/parrot/wiki/git-svn-tutorial 21:40
masak oh, good.
someone's already done all the work for me. Tene++
Tene some of that applies, some not.
21:47 tadzik joined
tadzik sorry to interrupt, but I really prefer .fmt :) 21:47
masak one more vote for .fmt
TimToady: sub foo { lol: say "OH HAI" }; goto lol # can I goto into a sub? 21:48
sorear masak: you need to add a ()
masak fair enough. 21:49
jnthn sub foo and sub foo() became equivalent a while back, no?
(I know that sub foo had used to imply a default *@_ as a sop to Perl 5, but I thought we'd killed that.)
masak sorear: where does control flow go after printing "OH HAI"?
sorear jnthn: according to STD.pm6, sub foo means sub foo($_ is parcel = $OUTER::_) { }
masak: S04 doesn't say.
masak sorear: that's why I'm asking. 21:50
jnthn sorear: Heh.
sorear: That feels like an implementation detail.
sorear S04:1188 says that you can't goto into a sub with parameters
jnthn I may be wrong, but I find it very surprising.
masak I wouldn't be sad if one couldn't goto into a sub.
the sub, after all, isn't part of the "normal flow" of the program. 21:51
sorear anyone want to comment on STD.pm6+5110?
jnthn masak: otoh, it fits the argumentless block pattern. :-) 21:52
masak: Guess it could be argued both ways.
masak jnthn: well, as opposed to all the other cases, a sub is only exited through the call stack, it never falls out into its outer blocks like all those other blocks. 21:53
jnthn troo
dalek psi: 7674823 | masak++ | doc/LOLHALP:
[doc/LOLHALP] added 'goto'
pmichaud back again, briefly 21:56
masak pmichaud: .fmt or .format? :)
jnthn or .form
;-)
masak or .f
jnthn No, not .f
:-)
masak or .bikeshed
pmichaud masak: I actually have no preference on that one. 21:57
sorear I propose .floormat
pmichaud if pushed, I'd give a slight preference to .fmt simply because it's shorter, and things that have .fmt on them tend to eat up lots of characters on my 80-char terminal anyway
but I wouldn't consider it a "full vote in favor of .fmt"
masak pmichaud: fwiw, you just put into words what I feel is good about .fmt 21:58
but I also don't want to put in a full vote in favour of .fmt, because I see where Juerd is coming from.
pmichaud masak: right. Before there was .fmt we had "sprintf", and .format isn't really any longer than that :-) 21:59
so.... I can argue that both ways. :)
masak :)
pmichaud besides, given that sprint-formatting strings tend to be fairly cryptic, perhaps it's good that they end up with a cryptic name :-) 22:00
maybe it should be called .wtf
("write this formatted".... :-)
jnthn "way to format"
:D
oh, even better
\\o/
sorear How much is known about perl6 threads?
pmichaud although I suppose that .WTF should be defined at some point in the spirit of .HOW, .WHY, etc. 22:01
masak .wtf -- 'write template format'
lue ohai o/
pmichaud sorear: very little. We declared at yapc::eu that one of rakudo's big pushes for the next few months would be to come up with a true parallel implementation of hypers
to sort of push the envelope on threading and parallelism in general
(and perhaps push Parrot a bit on the same) 22:02
sorear I have threads, but I'm not sure how hypers should work
My implementation of lazy lists are not locked for multithreaded shifting, and I doubt tey should be
pmichaud anyway, I'm not aware of any strong notions regarding p6 threads.... in the past it's always been "we hope someone will explore this space a bit for us" 22:03
sorear more generally, I feel there has been a bit of semantic drift for hyperops 22:04
now the common use is stuff like lines() >>=>>> 1
which wouldn't benefit from parallelism at all
masak sorear: oh, there are still other common uses.
jnthn Just because that use doesn't, doesn't mean no uses would. 22:05
masak sorear: I say @paths>>.=substr(...) the other day, which I really liked.
sorear I feel like there should be a lightweight vector op
masak er, I mean substr
subst :)
sorear: and >> isn't it?
sorear and it probably deserves more Huffman weight than the "I have 10,000 elements and processing each one will take 5ms) 22:06
masak: we don't have enough AI to tell whether dropping the parrellel-hammer is a good idea, we need programmer input
jnthn pragmii could work for that 22:07
*pragmas
sorear tries to refresh his memory on the Haskell implementation of parallel vector operators, misgoogles and finds an AK47 replacement parts page
masak :D 22:08
"we don't know what happened to this Perl 6 implementor -- he just went postal all of a sudden" 22:09
22:09 tom_tsuruhara joined
pmichaud is a little surprised to see "went postal" in an international context :-P 22:09
guess that meme spread pretty far :)
jnthn ..."went postal"?
<confused look> 22:10
pmichaud en.wikipedia.org/wiki/Going_postal
jnthn Must be an American thing. :)
pmichaud see the wikipedia entry :)
jnthn lol
masak proof that Sweden is more American than jnthn :)
jnthn OK, I'd not have ever guessed that's what it meant. :-) 22:11
lue "I went postal when the electricity was out"
masak spectests almost done. looking good.
pmichaud jnthn: but do you agree there's a chance that one of the implementors could "go postal"? ;-) I blame TimToady++
at next year's yapc::eu.... LOOK OUT!!!!! MUWAHAHAHAHAHAHA 22:12
jnthn pmichaud: Next time in meatcleaverspace...
pmichaud "WHO SAYS RAKUDO IS TOO SLOW?!??!?!? *pew* *pew* *pew*"
masak a Perl 6 implementor googles for threads, and comes up with AK47 replacement parts instead... I'd say there's a significant chance. :)
pmichaud I'd blind you all with my laser pointer but it got confiscated by airport security in Pisa :-(
jnthn In Latvia, you'd just have got inte-riga-ted about it. :-) 22:13
masak pmichaud: airport security is a bully whom nobody dares to put down.
pmichaud masak: including me, as it turns out. I let them have the pointer, even though it was actually a gift with a fair amount of personal value to me. Wasn't much else to do at 22h00 though.
I'll get a replacement somehow. 22:14
masak "I'm sorry, you're not allowed to bring that hand lotion aboard your plane. Now take off your shoes. Stand there. Turn around. Let me feel you up."
lue
.oO(airport security has *all* the guns. Don't mess with them)
jnthn Flying sucks.
It's just unavoidable for some journeys.
pmichaud lue: airport security has all of the *declared* guns. They might not have *all* of them :)
masak I googled a bit for "security theater" a while ago. it helped.
pmichaud yes, I got the full pat-down treatment at LHR also 22:15
jnthn Sorry you had to experience Heathrow. :/
pmichaud but obviously their computers indicated that I'm working on Perl 6, and therefore I'm a likely threat to "go postal" :-P
sorear
.oO(I wonder what cheap mass teleportation will do to customs)
jnthn That would be awesome!
pmichaud locally: 22:16
jnthn Airport security seem very unbothered with me most of the time. 22:17
pugssvn r32004 | masak++ | [t/spec/S32-str/encode.t] switch back to slurpy .new
pmichaud pmichaud@plum:~/rakudo$ ./perl6
> say 48 + 0 +< 8
48
jnthn Dunno why.
I'm not complaining though. :-)
masak pmichaud: you fixed it? great!
jnthn \\o/
pmichaud++
masak then I can remove some parentheses.
pmichaud trivial fix, actually.
just spectesting it now.
masak pmichaud: you'll need to pull before you push :)
pmichaud I just did a pull. 22:18
masak you'll need to pull again. :)
pmichaud indeed, so I do.
restarting spectest
after pull
okay, time for me to go some important Lego building work
*go do
jnthn
.oO( could we build Perl 6 in lego? )
 22:19
pmichaud jnthn: yes, we can
moritz_ would love to
jnthn \\o/
pmichaud I was brainstorming earlier today how I can cross Perl 6 and Lego
masak not sure if this is praise or sharp criticism: www.formspring.me/halcy/q/965794626
dalek kudo: 83b2cdf | masak++ | (3 files):
[Buf] worked on &pack, .unpack and .encode

  * The directives they know about: A H C S L n N v V x
  * .encode now recognizes its $encoding parameter
  * It can handle utf-8 (default), latin-1, and ascii
lue
.oO("Please sir, we need to see that lego brick. Is it... a 2x4 ?")
masak \\o/
moritz_ \\o/ indeed 22:20
pmichaud masak: I'd count it as praise.
afk
jnthn masak++ # yay
masak: The pencil. You must soon lay it down. 22:21
;-)
masak pmichaud: I've heard worse versions of it. a friend of mine relayed "well, they've taken all the modern programming concepts, an OO system, and all manner of nice stuff, and glued it... onto *Perl*"
jnthn: well, I plan to do the operators and then lay it down. I won't even need tomorrow, despite the deadline being 19:00 UTC. 22:22
jnthn I woudln't say we glued it on. It's more like whisked it in.
lue .oO[ this year's perseid meteor shower makes me want to find a comet whizzing by in December, and build a snowman to go with it :) ]
masak jnthn: which is good, because I have $WORK to do too :)
jnthn masak: Excellent.
masak: Yes, I have $dayjob to attend to tomorrow too, I suspect. :-) 22:23
masak lue++ # that's an AWESOME idea
lue: take a photo!
jnthn That was still my favorite ever RT ticket. :-)
masak jnthn: it'll surely go into my 1024 talk :) 22:24
lue
.oO(find projected comet path, find 'P 6' constellation, make snowman, see comet,tilt the camera's left down, take picture.)
 22:25
masak "Snowman-comet: a Match made in heaven" 22:26
22:27 hercynium joined
jnthn I still want to give a talk on Perl 6 grammars just so I can use that pun in a talk title. :-) 22:27
22:27 hercynium joined, kst joined
masak jnthn: please do! 22:27
tadzik Juerd: SYN 22:30
jnthn Maybe it should be the one that I do after "I KnowHOW to build a metamodel!"
lue rakudo: Doctor.WHO 22:31
p6eval rakudo c4b6df: OUTPUT«Could not find sub &Doctor␤ in main program body at line 22:/tmp/_uhYgWmQkD␤»
Tene hehe 22:34
Juerd tadzik: nack
tadzik Juerd: noticed my feather account request maybe? 22:35
Juerd tadzik: Yes
afk
lue prefix:<?> checks for defined-ness, right?
tadzik oh, ok
Juerd: fine, just up to this moment I still had a feeling that this ##### email address is a prank :) 22:36
lue rakudo: my $a = 3; say ?$a; say ?$b; 22:37
p6eval rakudo c4b6df: OUTPUT«===SORRY!===␤Symbol '$b' not predeclared in <anonymous> (/tmp/9N2f74jP8P:22)␤»
lue rakudo: my $a = 3; my $b; say ?$a; say ?$b;
p6eval rakudo c4b6df: OUTPUT«1␤0␤»
masak lue: no, it's just that "undefined" is one of the false values. 22:40
lue: 0 and '' would yield False as well.
lue here's another pun: m/smart/ # or at least the beginnings of one 22:41
masak rakudo: say .perl given "foo".comb>>.ord >>+^<< "bar".comb>>.ord 22:45
don't ask :)
p6eval rakudo c4b6df: OUTPUT«[4, 14, 29]␤»
22:50 awwaiid joined
lue rakudo: my @a = 1,2,3; my @b = 4,5,6; say (@a «[+]» @b); # what happens when you combine hyperops and reduction ops ? 22:54
p6eval rakudo 83b2cd: OUTPUT«579␤»
masak lue: the [+] parses as just + 22:55
lue: generally, in infix position, [op] parses as just op
lue darn.
masak std: »[+] 1, 2, 3 22:56
p6eval std 31912: OUTPUT«[31m===[0mSORRY![31m===[0m␤Bogus statement at /tmp/plSLueiyJS line 1:␤------> [32m<BOL>[33m⏏[31m»[+] 1, 2, 3[0m␤Parse failed␤FAILED 00:01 113m␤»
Tene I was a little bit disappointed earlier that I couldn't transform a lists of lists with .map([*] *) as short for .map({[*] @^a})
lue I guess the other way around ( [«+»] ) wouldn't do anything either? 22:57
lue ponders how to do infinite cardinals in Perl 6 22:58
masak std: my @a; my @b; @a [«+»] @b
p6eval std 31912: OUTPUT«ok 00:01 115m␤» 22:59
lue rakudo:my @a = 1,2,3; my @b = 4,5,6; say [«+»] @a, @b 23:01
p6eval rakudo 83b2cd: OUTPUT«21␤»
jnthn ponders how to do 1 pope in Perl 6 23:02
lue :) .oO(what just happened?)
rakudo: my @a = 1,2,3; my @b = 9,8,7; say [«+»] @a, @b
p6eval rakudo 83b2cd: OUTPUT«30␤»
lue rakudo: my @a = 1,2,3; my @b = 4,5,6; say [«+»] @a, @b; say [+] @a, @b; 23:03
p6eval rakudo 83b2cd: OUTPUT«21␤21␤»
lue rakudo: say uc 0x2a 23:05
p6eval rakudo 83b2cd: OUTPUT«42␤»
masak testing -- ingy 23:07
testing -- thingy
ingy meh
masak: I irssi always hilights thingy and stringy
for me 23:08
mberends japhb: the OpenGL examples in cpansearch.perl.org/src/CHM/OpenGL-.../examples/ seemed like good candidates for trying out with Zavolaj, The large number of functions and named constants is daunting, however.
sorear ingy: what do you think of my p6take on Test::Builder? 23:09
ingy sorear: sorry, not there yet... 23:10
sorear: ok, I got niecza_eval working 23:12
Tene The common mishilight in this channel for me is "flatTENEd" 23:13
ingy sorear: rebuilding niecza 23:16
Tene: do you know the cure?
Tene ingy: Yes.
ingy: I just don't care. I find it a little entertaining when I get mishilights. 23:17
sorear ingy: I was actually expecting you to just skim the tile. 23:18
file
ingy sorear: yeah, but if I'm going in there... 23:20
sorear also, several features are missing because I last touched Test.pm6 before I had arrays and hashes working 23:21
23:21 s1n joined
sorear is reading the DPH papers trying to figure out if this can be translated to hyperops 23:21
23:22 mahdi joined
mahdi hi all 23:23
sorear hello
mahdi This is my first visit to this channel
sorear Welcome.
tadzik hello mahdi
mahdi thanks
hi :)
masak stop by anytime! 23:25
if only to say hi. 23:26
sorear contemplates hacking his copy of Linux to simulate a multiprocessor by not allowing any thread to take more than 25% of timeslaces
ingy masak: did you get rid of the Makefile in the ufo repo?
masak: I seem to remember make install-ing ufo 23:27
masak ingy: aye. ufo now self-hosts.
ingy: you can run ufo on ufo to get a Makefile in your ufo.
ingy you might want to document that 23:28
masak: and a one line Makefile would do that :P
well maybe 2 lines
masak ingy: I'll document it, when tuits present themselves. 23:29
sorear make sucks. no way to get at the platform path separator. 23:30
ingy sorear: take it back, or I won't give you my review 23:31
make is ingy's best friend for life
> time ufo 23:33
real\t0m13.363s
sorear make is great... I need to stop selectively remembering negatives 23:34
masak ingy: recent discussions about pure-p6 make alternatives have made me think about how to obviate ufo. 23:37
don't tell the aliens.
ingy sorear: Can't locate CompilerDriver.pm in @INC 23:38
I see...
sorear the probability of me complaining about a piece of software is directly proportional to how often I have to use it 23:39
ingy: trying to use it from outside the build dir? 23:40
I haven't gotten around to handling that 23:41
there are several different places where path search need to work
ingy sorear, masak: I tried this: gist.github.com/526092
I could hack on ufo to better support other perl6 impls
sorear STD.pm6 wants to find syml/Test.pm6.syml, niecza wants to find Test_ast.store, gmcs wants to find Test.dll, mono wants to find Test.dll and Test.dll.so 23:42
ingy ufo is somewhat rakudo specific
sorear I'm not totally sure how I want to support cross-directory importing
ingy the --target=pir --output=blib/lib/TestML/Parser.pir part anyway 23:43
masak ingy: for now, ufo, is *intentionally* Rakudo-specific :)
ingy bah
why?
sorear if you're ok with using the builtin NULL setting and are willing to install Kernel.dll into the mono system libraries folder, niecza can probably be made to work outside builddir easily
ingy I need to remember my promise to myself to not look at perl 6 for at least a week! 23:44
I got temporarily sidetracked to play with p6 in mid-June! 23:45
YOU BASTARDS!
masak ingy: because I know of no other implementation that needs a build system right now.
ingy I'm not sure what a build system has to do with running make test and make install 23:46
I was pretty excited to use ufo to run a niecza make test on my modules
masak ingy: :)
ingy but I'll work on it next week
masak ingy: ok, depending on how non-invasive such a patch would be, I'd accept it.
ingy Python here I come! 23:47
23:47 pochi joined
ingy masak: have you seen me ever do anything invasive?! 23:47
it's against my religion
masak ingy: Spiffy :) 23:48
ingy Spiffy never hurt anyone
nor invaded anyone
anyway Gloom is the new Spiffy 23:49
masak :)
ingy it's as tame as a pussy cat
jnthn Gloom? 23:50
Comes with a free Nirvana MP3 to help you be more gloomy!
ingy jnthn++ 23:51
jnthn: Gloom is the Great Little OO Module for p5
lue
.oO(I don't want your freevil MP3s !)
ingy it was born yesterday
after a 10 year pregancy 23:52
more or less
I'm about to do an IO::All release to CPAN. Looks like I've done one in 2004, 2006, 2008 and 2010 23:53
IO::All is even 23:54
even if some people think it's odd
tadzik :) 23:55
ingy sorear: Test.pm6 looks like a great start 23:56
sorry for getting distracted...
maybe we can try it in Rakudo
pugssvn r32005 | masak++ | [t/spec/S03-operators/buf.t] added 23:59
r32005 |
r32005 | For Rakudo, which now handles these operators.

Powered by Cro and the Raku® Programming Language.

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

Thank you!