»ö« | perl6.org/ | nopaste: paste.lisp.org/new/perl6 | evalbot usage: 'perl6: say 3;' or rakudo:, alpha:, pugs:, std:, or /msg p6eval perl6: ... | irclog: irc.pugscode.org/ | UTF-8 is our friend!
Set by lichtkind on 5 March 2010.
TimToady everyone agrees that parrot is slow :) 00:00
lue and that any speed issues are amplified on my piece of 20th century history :)
TimToady you think you've got problems, I've got an Amiga 1000 00:01
still runs, too
lue you're not using it right now, are you? 00:02
TimToady nope
lue
.oO(poor Larry :) )
what would the regex be for filtering out all but dna bases? (ACTG) 00:04
00:05 felipe joined
TimToady .comb(/<[ACTG]>/.join or maybe .trans('ACTG' => 'ACTG', :d), but that's probably not implemented 00:05
rakudo: say "I WANT A CAREER IN ACTING!!!".comb(/<[ACTG]>/).join 00:06
p6eval rakudo 1eef08: OUTPUT«ATACAACTG␤»
TimToady rakudo: say "I WANT A CAREER IN ACTING!!!".trans('ACTG' => 'ACTG', :d)
p6eval rakudo 1eef08: OUTPUT«too few positional arguments: 2 passed, 3 (or more) expected␤current instr.: '_block14' pc 29 (EVAL_1:0)␤» 00:07
lue TimToady++
TimToady hmm
lue (S05 isn't exactly clear on the basic regex stuff)
(from what I skimmed)
TimToady don't think it knows :d
rakudo: say "I WANT A CAREER IN ACTING!!!".trans('ACTG' => 'ACTG', '\0..\x1ffff' => '') 00:09
p6eval rakudo 1eef08: OUTPUT«Attributes of type 'INTVAL *' cannot be subclassed from a high-level PMC.␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
TimToady cool
00:12 SmokeMachine joined
BrowserUk TimToady: Can I read more about "p6's model is that everything outside the lexical scope of the async {} is shared, and everything inside isn't" somewhere? 00:12
TimToady not really, but it's the design direction we've been moving for years 00:15
why we don't like localized globals anymore, for instance
and why most of the dispatching we do is now lexical rather than via packages 00:16
00:16 muixirt left
lisppaste3 lue pasted "DNA to Hex (Halp!)" at paste.lisp.org/display/98853 00:16
TimToady so it's kinda threaded (sorry) throughout the various synopses
basically, there are are very few globals left in Perl 6 00:17
lue It just won't work (problem explained in the nopaste)
TimToady it's also implicit in the notions of feed operators behaving a lot like Unix pipes 00:18
which is more of an Erlangish notion of threading
as for synchronizing stuff outside of async, there's rudimentary docs on STM somewhere out there, hmm 00:19
lue I think it could be a stupid overflow error :( (paste.lisp.org/display/98853)
TimToady we'd like to go lockless as much possible
some of these ideas are sketched in S17 00:20
BrowserUk Okay. Sounds interesting. Time I re-read the synopses again. Is that STM doc P6 related? (I looking for keyowrds to search on) 00:21
TimToady look for Contend/Maybe/Defer in S17
[Coke] tries not to take it personall.
*y
lue rakudo: say :16<7fffffff> 00:22
p6eval rakudo 1eef08: OUTPUT«2147483647␤»
lue rakudo: say :16<7fffffff>.fmt('%b')
p6eval rakudo 1eef08: OUTPUT«1111111111111111111111111111111␤»
BrowserUk Okay, thanks.
lue Even with a 64-bit computer (which I ssh to), it seems I won't be getting to "decode" large strings of DNA _that_ easily :/ 00:23
TimToady tries to envision what taking it impersonally would mean... 00:26
ruoso .oO( while STM is mentioned in docs... most of the threading and data-sharing model is undefined... ) 00:29
TimToady
.oO(I used the word "sketch" on purpose)
sjohnson
.oO(I wonder if this does anything on some remote site?)
 00:30
ruoso one Idea I once considered was wiring Perl 6 around Postgresql MVCC to implement STM 00:31
it kinda sucks when running just one core... but will rock when used in 60 cores 00:32
lue anything would rock in 60 cores :) 00:33
TimToady if only that were so
00:33 lest_away is now known as lestrrat
TimToady but maybe the definition of "rock" varies 00:34
sjohnson megathreated apps
apparently new budget-conscious 6 core cpus from AMD are "ya get what you pay for" CPUs
ruoso the news are even more interesting... 00:35
I've recently saw that we're about to see servers with arrays of arm processors
lue
.oO(I bet they only cost a leg :) )
 00:36
TimToady yes, but the interest rate goes way up in 5 years
ruoso the hard part on STM is getting the semantics straight... 00:38
lue why does @out.push(:4($temp).fmt('%x')); fail? ($temp being a 4-digit string, i.e. "0132")
All i get is []
ruoso when to begin and commit a transaction...
TimToady ruoso: the hard part is defining what you mean by "software", "transaction", and "memory".
ruoso heh
lue rakudo: my @out=(); my $temp="1103"; @out.push(:4($temp).fmt('%x')); say @out; 00:40
p6eval rakudo 1eef08: OUTPUT«53␤»
lue *sigh* failure on my end :/
TimToady shell quoting?
lue no, it's in a .p6 file. 00:41
should I nopaste it?
TimToady quail shooting?
yes, as long as you don't ask for permission first
00:42 jonrafkind left
lisppaste3 lue annotated #98853 "New version" at paste.lisp.org/display/98853#1 00:42
lue there you are. What could be going wrong? 00:43
(the last line in the loop is just a debugging command) 00:44
... I probably used substr() incorrectly :/
TimToady your >= should be <= 00:45
lue oh. I defined when to stop. :P
TimToady why not just use the list returned by comb 4 values at a time?
use of loop() is usually a smell 00:46
lue comb returns a list? :)
TimToady why did you put the .join :)
lue I changed it, and got a redeclaration of $i err.
I copied off of masak, and that part was not explained. :) 00:47
TimToady try putting spaces around your infix operators too. $i<= is going to misparse
lue ah. 00:48
BrowserUk My concerns about STM are: 1) The process of chasing out everything called indirectly from within a contend block to ensure it is undoable; is impossibly costly for an interpreter. 2) With fat internal structures, even reading a scalar can change it's state: $x = '1 ffff'; if( $x == 0 ){...} Now $x has changed. How do you undo that? 00:49
lue tries to find a clever way to pull 4 values of a list at a time w/o loops
00:51 rgrau_ left
TimToady BrowserUk: wrt 1, yes, STM implies global state change tracking. It also is something of a toy, so far, so we're adopting a wait-and-see attitude towards it 00:51
mostly we're trying to get rid of most of the obvious need for locks in the first place through adequate scoping primitives
lichtkind TimToady: can you give a metaop the op by an variable? 00:52
TimToady re 2, that's another thing that is a bit different in p6
a value is allowed to cache a coercion like that, but it is not supposed to change its fundamental type or value
lichtkind: no, use the corresponding functional interface for that sort of thing 00:53
lue wonders...
TimToady reduce, crosswith, etc
lichtkind ah thanks
ruoso BrowserUk, as I see it, it's a all-or-nothing... the STM will only work as far as the memory is managed by it... it might intentionally repeat side-effects of things outside STM boundaries 00:54
lue rakudo: my @a = (1,10,34); say ([.fmt('%x')] @a)
p6eval rakudo 1eef08: OUTPUT«Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 11␤current instr.: 'perl6;Regex;Cursor;FAILGOAL' pc 1696 (ext/nqp-rx/src/stage0/Regex-s0.pir:932)␤»
lue rakudo: my @a = (1,10,34); say [.fmt('%x')] @a
p6eval rakudo 1eef08: OUTPUT«Confused at line 11, near "say [.fmt("␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
TimToady all we can say right now is that, if we adopt STM, we think we know what main syntax is. :)
but it's far from being mandated for 6.0
the currently specced syntax was, however, prototyped in pugs, and seems to work right 00:55
ruoso but we might have very small boundaries for more low-level operations in order to provide lock-free access to data 00:56
sorear discovers Perl6::Compiler.compile("foo", :parsetrace)
ruoso STM-begin-commit-boundaries, I mean
sorear There will be TIMTOWDI. STM is not a panacea
ruoso sure sure sure....
but the question of "how do you share data" is still in the table 00:57
TimToady lue: metaops may only have literal operator names, not random code
and whenever you get "Confused" from rakudo, STD is likely to give you a more specific error
std: my @a = (1,10,34); say [.fmt('%x')] @a
p6eval std 30559: OUTPUT«===SORRY!===␤Method call found where infix expected (omit whitespace?) at /tmp/TO8LydnsBr line 1:␤------> my @a = (1,10,34); say [⏏.fmt('%x')] @a␤ expecting any of:␤ bracketed infix␤ infix or meta-infix␤Parse failed␤FAILED 00:01 111m␤»
TimToady which may or may not be a correct error :) 00:58
sorear also, ./perl6 --trace -e '2+2'
TimToady std: my @a = (1,10,34); say [ .fmt('%x') ] @a 00:59
sorear I wonder how many other undocumented options there are
p6eval std 30559: OUTPUT«===SORRY!===␤Two terms in a row (preceding is not a valid reduce operator) at /tmp/gpWW7FSsZ8 line 1:␤------> my @a = (1,10,34); say [ .fmt('%x') ] ⏏@a␤ expecting any of:␤ bracketed infix␤ infix or meta-infix␤ statement modifier
..loop␤Parse fa…
TimToady that's the error I expected to get :)
lue Yeah. I wanted the power of metaops with the function I wanted :/
maybe 6.1, if nothing else ?
TimToady as I *just* got through telling lichtkind++, use the functional form for that
BrowserUk TimToady: I'm wary of what pugs could do, because of the deep analysis power of GHC. Can't quite see fitting that into an interpreter.
ruoso rakudo: my @a = (1,10,34; say @a.reduce({ .fmt('%x') })
p6eval rakudo 1eef08: OUTPUT«Confused at line 11, near "my @a = (1"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
ruoso rakudo: my @a = (1,10,34); say @a.reduce({ .fmt('%x') }) 01:00
p6eval rakudo 1eef08: ( no output )
TimToady BrowserUk: indeed, which is why we're holding back on anything resembling a mandate there
ruoso rakudo: my @a = (1,10,34); say @a.reduce({ $^a.fmt('%x') })
p6eval rakudo 1eef08: ( no output )
lichtkind hej i got ++ :) 01:01
TimToady: i see no table for these methods can you just tell me them once so I make such table
BrowserUk TimToady: Okay. I'll keep reading and keep an open mind :) 01:02
I kinda wanna say: "There's a better way", but believing it and proving it are quite different things. 01:03
01:03 bbkr_ joined
TimToady [op] is reduce(), Xop is crosswith, Zop is zipwith. We haven't defined functional forms for all the others, but I imagine we will 01:04
01:05 SmokeMachine left
TimToady BrowserUk: if it's your hot button, you're welcome to contribute to the design, since obviously what we have so far is somewhat ramshackle on that front. 01:06
lichtkind TimToady: have we one for [\ ] ?
TimToady you'd have to ask colomon that one, I think
vague recollection that he called it triangle()
lichtkind colomon: really triangle? 01:08
lue yeah, i've heard [\ ] being called triangle. 01:09
lichtkind have we a hyper method?
BrowserUk ruoso: I know continuations are supposed to be able to allow fast return to earlier states--though I wonder about the memory costs for storage. But STM that allows (undoable) externals side-effects are dangerous. And I'm really non-plussed by how you implement Global State Tracking.
TimToady negation is a bit strange in that $a !op $b wants to turn into not($a op $b), so !op is more like * !op * anti-currying 01:10
function composition, really
and we have all the hyper variants too. don't know what those are called inside rakudo currently
BrowserUk Do you try to catch potentially side-effectful transactions at compile time? Or bug out and die at runtime?
TimToady or if they even have function names
the hypers might still be hardwired and non-extensible 01:11
BrowserUk: we'd like to have some clue about what has side effects at compile time, though certainly not in a Haskellian sense
lichtkind TimToady: what you said so far is now in www.perlfoundation.org/perl6/index....aoperators
sorear BrowserUk: do you understand what it means for concurrency to not be on this month's agenda? 01:12
TimToady after all, we can't determine "Useless use of" messages without knowing what is pure
sorear: it's okay, we need someone to champion this area, so please try not to scare my BrowserUk away. :)
unless you wanna do it? :) 01:13
BrowserUk sorear: If that means shut up. Sure thing. Bye
01:13 BrowserUk left
sorear ...wow 01:13
TimToady yes, well, most of the threads on Perl Monks that are deeper than 30 will have him as one of its sides :) 01:15
but what you said did come across as rather snippy
snarkyboojum thinks BrowserUk needs a hug
sorear I don't like it when people proactively whine 01:16
TimToady I like it when people care
lue hugme: hug BrowserUk
hugme hugs BrowserUk
sorear what's the point of <terminator>? 01:17
lue rakudo: say chr(64)
p6eval rakudo 1eef08: OUTPUT«@␤»
lue rakudo: say chr( :4<10> )
p6eval rakudo 1eef08: OUTPUT«␤»
TimToady the point is to be an extensible set in subgrammars
lichtkind slice() is alias to @@ ? 01:18
TimToady though the interplay of <terminator>, <stopper> and $*GOAL is a bit more complicated than strictly necessary
sorear I'm playing around with one of masak's bugs
he's defined a macro named ]
snarkyboojum just stumbled across www.ozonehouse.com/mark/periodic/ again - not completely up to date but *beautiful* :)
sorear term:<]>
now, 'say ]' 01:19
TimToady terminators take precedence over infixes
or terms
sorear the parser matches ] as <terminator>, closes the arglist, then dies
TimToady that is, they ignore LTM considerations
sorear What is the justification for this?
Also, is anyone else going after BrowserUk
?
TimToady too lazy to calculate all the follow sets :) 01:20
and a sneaking suspicion that anyone who defines operators starting with terminators is really Asking For It already
sorear: no point in chasing BrowserUk, I think 01:21
lue methinks chasing someone virtually is a bit difficult. 01:22
TimToady he likes to argue, and really believes his signature of "Silence betokens consent", though occasionally it seems like it reduces to "I always get last word". Hence the "deep" discussions on PM. 01:23
but generally he's quite cogent, if a curmudgeon.
lue snipperwhapper!
TimToady Get onto my lawn!!! 01:24
lue Bless those meddling kids!
TimToady sorear: despite my comments earlier, I'm not all that eager to sweep all of the p5 community into p6 as soon as possible. :) 01:25
in fact, Camelia is specifically designed to keep some of them out... 01:26
sorear Hah.
lue how do I write to a (text) file? the whole IO mess in the spec isn't that helpful...
sorear lue: spew
TimToady: Why have terminators at all? Why isn't $?GOAL enough?
TimToady $*HANDLE.print should work
$*GOAL is only a literal string currently 01:27
<stopper> is only the goal for the innermost slang that is actually derived, and is sometimes not a literal string
<terminator> is the larger set of things we should probably stop on 01:28
not because the current language is specifically looking for it, but because the presence of one of them almost certainly indicates a syntax error
lue spew isn't in the spec :/
01:29 isBEKaml left
TimToady it was a joke 01:29
use print
as I said above
lue it certainly would make sense though :) 01:30
01:30 bbkr_ left
TimToady for instance, $$ is always a valid terminator even when you're looking for ] 01:30
01:30 wknight8111 left
snarkyboojum lue: I've seen spew() implemented in one of masak's projects somewhere I think :) 01:30
TimToady valid in the sense of stopping any further progress, at least 01:31
you generally don't want to parse past EOF :)
well, you might if you're writing malware, I suppose...
lue bleh. How doth I use $*HANDLE? (google search of spec isn't being very nice :/)
TimToady please be more specific than "use" 01:32
lue well, how do I create a file to write to? That's what I want to do.
(I'd prefer binary file IO, but it is NYI) 01:33
TimToady my $handle = open($filename, :w); $handle.print: @stuff
sorear I was under the impression that spew was the opposite of slurp
TimToady well, in real life it is...
but print is already a spewer
also, spew is perhaps a bit too colorful for people with sensitive stomachs... 01:35
slurp is only culturally disgusting, not viscerally 01:36
lue or we can define its palindrome: pruls()
TimToady N.B. yes, there are varieties of spew that are not, strictly speaking, vomit 01:37
we might also end up with @stuff ==> qp/file/ doing the right thing, in an ingyesque sense 01:38
sorear rakudo: my regex foo { <[a-l]> } 01:39
p6eval rakudo 1eef08: OUTPUT«Obsolete use of hyphen in enumerated character class;in Perl 6 please use .. instead at line 11, near "l]> }"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
sorear rakudo: my regex foo { <[a..l]> }
p6eval rakudo 1eef08: ( no output )
lue is now off to "decode" some DNA! :) 01:40
01:43 agentzh joined
dalek kudo: 6e98d29 | (Solomon Foster)++ | src/core/operators.pm:
Refactor infix:<...> a bit and make it handle some edge cases in the non-numerical world.
 01:46
01:49 rv2733 left, snarkyboojum left
diakopter is it safe to peer out yet? 01:59
01:59 kst left, Psyche^ joined 02:00 kst joined
lue what do you mean by that? 02:00
diakopter there seemed to be a scuffle earlier
lue oh, that.
I think the butterfly said stop. 02:02
02:03 Patterner left, Psyche^ is now known as Patterner
s1n 0../-985 02:06
oops, sorry 02:07
02:08 Heame joined 02:11 plobsing joined
lue woah. genes DQ831955 and DQ831956 (which are closely related to one another) both have this bit in their sequence: SAMÆ 02:12
reads like "SAME". That's weird, because (bar a couple of characters at the end) the unicode "translation" of those two genes _are the exact same_ 02:13
02:21 justatheory joined 02:26 am0c joined
lue afk 02:29
02:35 patspam left 02:40 dduncan joined 02:41 cdarroch left
dduncan question - is it possible in Perl 6 to create and use an object ostensibly of some class Foo without declaring the Foo class? 02:42
for example, could I say something like $x = Foo.new( a => 5, b => 7) and have it just work?
assuming that the behaviour is similar to as if $x then held a hash, but it had metadata indicating Foo? 02:43
basically, how far one could go using Perl 6 for structured data without declaring any user-defined classes 02:44
diakopter C# calls those anonymous types; I don't know about Perl 6. 02:46
dduncan yes, I had also been wondering what the terminology was for these things
such as whether this is the same as typeless objects or not 02:47
diakopter I mean, it's not much different from a hash literal, but differs by how the members are accessed... whether the object responds to ^members or to .keys I guess 02:48
dduncan what I was hoping to be able to have is something that has semantics like a hash literal, in that you could just have named elements that are not coerced in anyway, and you could also say .WHAT or something on it and get the class name you specified with your new(), but that the class doesn't have to be defined anywhere, almost like the class name was a trait 02:49
and then therefore the role of classes is somewhat demoted to existing more for the purpose of applying constraints rather than being an enabler of functionality 02:50
diakopter if 'Foo' is in the source code anyway, why not predeclare it
oh, you're wanting it inferred/autodeclared 02:51
dduncan one idea is to be able to have arbitrarily complex values and move them between different environments that may not have all of the same classes
such as moving data in a communications channel between different programs 02:52
diakopter these environments are on the same process? or will use some kind of serialization?
dduncan they could be different processes ...
my main point is it would be nice to have arbitrarily complex data that can be understood fully without any context provided by class definitions ... 02:53
diakopter yaml would be the thing for that, since it supports cyclical references
dduncan and I'm thinking of it being more abstract than a simple serialization format
diakopter but Perl 6 doesn't specify how to serialize object initializers
afaik.
my point when mentioning yaml is that you've got to have something that handles cyclical/nested references 02:54
and you'll have to have something *like* yaml to do that
dduncan one use case is say one could define code for using a type, and it just works, and then later one could declare the class and the code continues to have the same behaviour, and all the class addition added essentially was constraints
lichtkind good night 02:55
dduncan so in this case we're all just within a single program
02:55 lichtkind left
dduncan essentially, make objects not depend on classes, but having classes can enhance objects that otherwise don't depend on them 02:55
I'm also working under the assumption that all relevant routines are not methods 02:56
diakopter wants a more precise use case
dduncan anyway, this is more of a mental exercise, I don't really need it answered
diakopter dduncan: how's muldisd 02:57
02:57 Tolkien joined
dduncan I'm almost at a major milestone 02:57
diakopter reads the Changes file and gasps
dduncan true, the Changes is verbose, but I'll be truncating it when the language is more stable 02:58
I find Changes useful to go back and find out when and how I changed things 02:59
diakopter, as you might guess, my question related partly on how I describe how Muldis D works, and it makes sense to phrase the question in terms of Perl 6 because I try to make them alike in many ways 03:00
03:01 Heame left, JimmyZ joined 03:10 Tolkien left 03:14 kst left 03:16 kst joined 03:20 JimmyZ_ joined 03:22 patspam joined, JimmyZ left, JimmyZ_ is now known as JimmyZ
TimToady dduncan: anonymous classes are perfectly fine in Perl 6, though rakudo might not be quite there yet 03:23
dduncan thank you 03:24
colomon Something like my $a = class { ... }
?
TimToady std: my $a = class { ... };
p6eval std 30559: OUTPUT«ok 00:01 110m␤»
TimToady rakudo: my $a = class { method foo { say "howdy"; } }; say $a.new.foo 03:25
colomon \o/
p6eval rakudo 1eef08: OUTPUT«howdy␤1␤»
colomon rakudo++
TimToady looks pretty anonymous to me
pugssvn r30560 | colomon++ | [t/spec] Additional series tests. 03:26
r30561 | colomon++ | [t/spec] Unfudge log tests in real-bridge.t. 03:27
dduncan FYI, here's a hello-world type example, completely to spec v0.125.0 in 2 hours: darrenduncan.net/Muldis-DB/scraps-n...amples.txt 03:28
anything that looks like Perl 6 generally has the same meaning
dalek kudo: 9709f85 | (Solomon Foster)++ | src/core/ (4 files):
Rename the default $base versions of .log to .log-e and then create a simple .log with $base method that simply calls .log-e twice.
 03:29
diakopter dduncan: I thought you were looking for a way to create objects *without* declaring a "class", anonymous or otherwise. 03:30
(and then add behavior to the objects "later")
TimToady can always mix in roles too
dduncan by behaviour I mainly meant constraints
TimToady but that's really making an anonymous class underneath
dduncan eg, you can write initially less strict code and then add strictures
eg, declaring the class is like adding 'use strict' or 'use warnings', very loosely speaking ... 03:31
or a better example ...
diakopter None Dare Call It Uncompilable
dduncan is write this initially in code: my $x = Foo.new(...)
and then later change it to: my Foo $x = Foo.new(...) for example 03:32
er ...
TimToady sounds like roles to me 03:33
03:33 _jaldhar left
dduncan actually, one context I'm thinking of is runtime changes 03:33
03:33 _jaldhar joined
dduncan eg, when you have a program that runs without stopping for months, and you want to change its code while it is running 03:34
TimToady sounds more Erlangish
dduncan that is the main context, analogous to a DBMS
which could be up for years, but users keep changing the code at runtime
which is what data-definition, changing tables or stored procedures etc at runtime is 03:35
Muldis D copies a lot from Perl 6, but it is designed to run in such an environment
part of the idea is that you could drop and add data types in terms of which your data is defined, without having to first delete and re-add your data
the data can continue to exist in a defined state while the data types it was defined with don't exist or are mutating 03:36
given all the runtime magic Perl 6 is capable of, I assumed I'd be able to implement Muldis D over it
TimToady you sort of have to never forget type info though
dduncan at the low level in Muldis D, the type name of data is actually in the data 03:37
eg, pretend each actual value is like an array ref whose first element is a type name and whose second is a structure having its attribute values
that detail can be optimized away, but conceptually it is that way
and so the values still have their type names and attribute names physically attached to them 03:38
and types serve more the role of constraints
if the database field must hold a value of type Foo, then iff Foo is defined then it will ensure that the constraint is met, and if Foo isn't defined it will let anything in that is tagged as Foo 03:39
TimToady progressive nominal typing, eh?
dduncan I was wondering what the best terminology was to describe what I was doing? 03:40
is "progressive nominal typing" a known term or did you make it up now as a description?
03:40 alester joined
dduncan can Google ... 03:40
TimToady made it up
sounds like you're moving from structural to nominal typing as types get GC'd
dduncan perhaps ... have to look up exactly what those terms mean 03:41
I have to finish up something quickly, back in maybe 30 min
TimToady cioa
*ciao
diakopter fights my last week self 03:42
03:43 am0c left
dduncan well, I just added this sentence to the spec ... can be changed later ... 03:51
One might say that Muldis D is using I<progressive nominal typing>, where values in databases move between structural and nominal typing as user-defined types whose names match their declared types come in and out of existence or mutate.
sorear ponders the ecological niche of Perl 6 .trans 03:53
I always rationalized Perl 5 tr/// as a restricted form of regexes that could be compiled down to a 256-element lookup table 03:54
03:55 chitragupt left
spinclad with unicode one begins to want that table sparse 03:56
03:57 chitragupt joined
TimToady there are also aspects of LTM matching to it, potentially 03:57
it's more like an alternation in some ways
sorear Does ~~ have special syntactic cases or is it always just $rhs.ACCEPTS($lhs) ? 03:58
03:58 patspam left
spinclad if all the alts are one char long, which of them is longest? more specific i'll buy though 03:59
TimToady there is the potential of lhs's to be longer than one char
in the list of pairs form 04:00
also, what is a "character" these days?
spinclad well, you know, thingy
longer than one: ah, yeah, that rings a distant bell in my forgettery 04:01
'fully composed whatzit' 04:02
diakopter phenny: tell pmurias I got parametric type declarations working. That is, they're compiled to fields/locals in the CIL. You can't use .new on them yet, tho. 04:10
phenny diakopter: I'll pass that on when pmurias is around.
diakopter perlesque: my List[int] $b; say(33254) 04:11
p6eval perlesque: OUTPUT«33254␤»
spinclad perlesque: my List[int] $l; $l = (2, 4, 7); say($l); 04:12
p6eval perlesque: OUTPUT«␤Unhandled Exception: System.ArgumentException: An element with the same key already exists in the dictionary.␤ at System.Collections.Generic.Dictionary`2[TKey,TValue].Add (.TKey key, .TValue value) [0x00000] in <filename unknown>:0 ␤ at
..Sprixel.PrototypeChain`2[TKey,TValue].AddH…
diakopter heh, no 04:13
dalek meta: r238 | diakopter++ | trunk/Sprixel/ (4 files):
[perleque] parametric types. yes.
spinclad perlesque: my List[int] $l; $l = List[int].new(2, 4, 7); say($l);
diakopter perlesque won't have list literals
p6eval perlesque: OUTPUT«␤Unhandled Exception: System.ArgumentException: An element with the same key already exists in the dictionary.␤ at System.Collections.Generic.Dictionary`2[TKey,TValue].Add (.TKey key, .TValue value) [0x00000] in <filename unknown>:0 ␤ at
..Sprixel.PrototypeChain`2[TKey,TValue].AddH…
diakopter as I said to pmurias, .new doesn't work yet
spinclad perlesque: my List[int] $l; $l.push(2); say($l); 04:14
p6eval perlesque: OUTPUT«␤Unhandled Exception: System.MissingMethodException: Cannot find method push in Type␤ at TriAxis.RunSharp.TypeInfo.FindMethod (System.Type t, System.String name, TriAxis.RunSharp.Operand[] args, Boolean static) [0x00000] in <filename unknown>:0 ␤ at
..TriAxis.RunSharp.Operand.Invok…
diakopter and it's not initialized.
and it's Add in the CLR
spinclad perlesque: my List[int] $l; $l.Empty; $l.Add(2); say($l);
diakopter my List[int] $l; $l.Add(2); say($l);
p6eval perlesque: OUTPUT«␤Unhandled Exception: System.MissingMemberException: Cannot find property Empty in Type␤ at TriAxis.RunSharp.TypeInfo.FindProperty (System.Type t, System.String name, TriAxis.RunSharp.Operand[] indexes, Boolean static) [0x00000] in <filename unknown>:0 ␤ at
..TriAxis.RunSharp.Opera…
diakopter what is this "Empty"
04:15 chitragupt left
diakopter for the billionth time, perlesque is not a Perl 6 implementation. the types will not correspond _AT_ALL_ 04:15
spinclad make it Empty, so .isEmpty is True?
diakopter just the syntax will correspond.
esque
04:15 chitragupt joined
diakopter $l is a clr null reference, there. 04:15
spinclad well, .Empty wouldn't be good Perl 6 either. just guessing around here. 04:16
perlesque: my List[int] $l; $l.Add(2); say($l);
p6eval perlesque: OUTPUT«␤Unhandled Exception: System.MissingMethodException: Cannot find method Add in Type␤ at TriAxis.RunSharp.TypeInfo.FindMethod (System.Type t, System.String name, TriAxis.RunSharp.Operand[] args, Boolean static) [0x00000] in <filename unknown>:0 ␤ at
..TriAxis.RunSharp.Operand.Invoke…
diakopter ah, oh, oops.
that's a screwup
04:16 kst left
diakopter by me 04:16
pmurias won't like that 04:17
:)
so I'd better fix it
04:17 kst joined
spinclad perlesque: my List[List[int]] $l; say(3); 04:18
p6eval perlesque: OUTPUT«3␤» 04:19
spinclad :)
diakopter perlesque: my Dictionary[string, P6Object] $l; say(3);
p6eval perlesque: OUTPUT«␤Unhandled Exception: Can't resolve type P6Object`1.␤»
diakopter hrm
oh
perlesque: my Dictionary[string, P6object] $l; say(3);
p6eval perlesque: OUTPUT«␤Unhandled Exception: Can't resolve type P6object`1.␤»
diakopter wheh 04:20
perlesque: my Dictionary[string, Sprixel.Runtime.P6object] $l; say(3);
p6eval perlesque: OUTPUT«␤Unhandled Exception: Can't resolve type Sprixel`1.␤»
diakopter hrm.
spinclad perlesque: my int $i; my List[int] $li; my List[List[int]] $lli; $li.Add($i); $lli.Add($li)
p6eval perlesque: OUTPUT«␤Unhandled Exception: System.MissingMethodException: Cannot find method Add in Type␤ at TriAxis.RunSharp.TypeInfo.FindMethod (System.Type t, System.String name, TriAxis.RunSharp.Operand[] args, Boolean static) [0x00000] in <filename unknown>:0 ␤ at
..TriAxis.RunSharp.Operand.Invoke…
diakopter it won't find Add ever, yet 04:21
:(
so, just the syntax is there :)
bedtime
spinclad sleep well, thanks for the toy
.oO{ the toymaker is greatly offended that something he's poured countless lives into is called a mere 'toy' 04:23
}
.oO{ the lad, cruelly unaware, just likes playing with it. } 04:24
perlesque: my int $i; my List[int] $li; my List[List[int]] $lli; $lli.Add($i) 04:25
p6eval perlesque: OUTPUT«␤Unhandled Exception: System.MissingMethodException: Cannot find method Add in Type␤ at TriAxis.RunSharp.TypeInfo.FindMethod (System.Type t, System.String name, TriAxis.RunSharp.Operand[] args, Boolean static) [0x00000] in <filename unknown>:0 ␤ at
..TriAxis.RunSharp.Operand.Invoke…
spinclad i presume if Add were found then this would be a typefail, while the previous two would pass. 04:26
04:26 molaf joined 04:28 sunnavy left 04:29 meppl joined 04:30 sunnavy joined
sorear wonders if it makes sense to have spectests for reasonable REPL behavior 04:33
s1n is the REPL behavior in the spec? or is that implementation specific? 04:34
04:35 patspam joined 04:39 _jaldhar left
spinclad thinks it would be premature to mandate a REPL; an implementation might be a cross-compiler, say, with no interactive capabilities. 04:39
04:39 _jaldhar joined, cotto left
sorear spinclad: standard options are always a possibility 04:39
s1n i agree with spinclad, i think any repl behavior should be up to an implementation 04:40
spinclad sure; one can spec, 'if you implement a REPL, it should work like this', and then have tests that are gated on that feature-test
TimToady any implementation that can run a BEGIN can run a REPL 04:41
spinclad if you spec, then spectests are yes
04:41 cotto joined
TimToady because the BEGIN runs on the compiler, not the cross-compiled runtime 04:41
sorear The command line invocation syntax for the 'perl6' executable is specced
a repl seems less constraining than that
TimToady a repl just needs an eval-me-now hook in the grammar after each complete statement
sorear and the ability of the parser to work on-line ? 04:42
TimToady STD currently allows for a .moreinput method, and it's closely related to that 04:43
basically, you just assume that the match of $ on a line transition isn't necessarily the final answer, if .moreinput can add to the buffer 04:46
this is untested, though, and I'm sure there are spots where the rest of the program would have to be tweaked to make it work 04:47
also, one should have some way of backing out of a statement with a syntax error 04:48
04:48 molaf left
spinclad after some mistakes it's easier to start fresh.. 04:49
TimToady shure
you mean like Perl 5? :P
spinclad having MONKEY_PATCHED the wrong thing badly 04:50
TimToady yes, well, kill 9,$*PID is also a bit problematic
spinclad relive and relearn 04:51
TimToady the repl should save a log so you don't have to copy/paste the screen 04:52
spinclad 'but this is a quality-of-implementation issue, and not in the language spec' 04:53
TimToady well, the particular method call that calls an AST might be specced, since we need it for BEGIN and constant folding and such
spinclad sure, that sounds core. REPL features like DropMeInto$EDITOR are more CPAN-module level, and dribble files are useful enough to be in pretty much everyone's module, i expect. 04:57
04:57 patspam left
dduncan okay, so I uploaded 0.125.0 to PAUSE and it said it was indexed at 21:58, so hopefully the next search.cpan refresh in a half-hour will show it all HTMLized 05:05
sorear interestingly, rakudo already has a place for after-statement-compiled hooks 05:06
a TimToadian REPL could be implemented without *too* much pain
05:06 steinberg left
sorear it would, however, involve cutting ourselves somewhat away from PCT 05:06
05:09 kaare joined
sorear pines for protolexpads 05:10
05:10 kaare is now known as Guest42448 05:11 TimToady sets mode: +vvv buubot dalek hugme, TimToady sets mode: +vv ilogger2 IRSeekBot, TimToady sets mode: +vvv lisppaste3 p6eval phenny, TimToady sets mode: +v pugssvn, steinberg joined
plobsing protolexpads? 05:12
sorear aye
they're like regular lexpads but... prototypical 05:13
compile time constants live in them
as do the initial values of regular lexicals in some confusing corner cases
05:14 TimToady sets mode: +oooo sorear [particle] [Coke] arnsholt, TimToady sets mode: +oooo cognominal colomon frettled japhb, TimToady sets mode: +oooo Juerd spinclad szabgab Tene, TimToady sets mode: +o xinming_
sorear is currently implementing reducer macros for masak 05:15
I almost think reducer macros should have a different keyword
TimToady you mean tree rewriters? 05:16
sorear yes
category macro: macro term:<foo> is parsed(rx:s{ <identifier> ** "," }) { # no arguments, but $/ and $<identifier> are valid here
TimToady or possibly some system resembling syntactic categories 05:17
sorear reducer macro: macro infix:<~~> ($lhs, $rhs) { # $/ is not valid here, and we aren't involved in the parsing of the symbol
05:17 chitragupt left 05:18 masak joined
masak good morning, #perl6. 05:18
phenny masak: 05 May 19:04Z <sorear> tell masak - postfix: macros not working is not a bug, it turns out that operator-like macros need a completely different implementation and I haven't implemented it yet. The ][ thing is troubling, though.
TimToady in some sense they're already selected by taking ASTs as args
but perhaps infix: is getting a bit stretched
sorear TimToady: so... if there's a signature, tree rewriter, else parse extender?
masak: Update: TimToady says macro names shouldn't start with anything in <terminator> (], }, )) 05:19
TimToady masak: arbitrary restriction because I'm mean :)
05:19 chitragupt joined
masak define "shouldn't" :) 05:20
"it just won't work" or "you'll get an error"?
sorear TimToady: expand "stretched"
TimToady not sure merely having a sig is the distinguishing feature, but it might be enough, especially taken with types
have to think about it some
sorear masak: I could make it a fatal error easily enough. Generally speaking, macros need a lot more awesome error power 05:21
TimToady well, currently we really only have infix:sym<op> and infix:<op>
maybe there are others
sorear What else could we have?
Also, aren't :sym<op> and :<op> synonymous?
TimToady dunno
no, they aren't 05:22
infix:sym<+> is the rule that parses the + symbol
infix:<+> is the name of the operator
masak phenny: tell szabgab that I don't know if we've touched on this before, but both .HOW and .WALK are 'internal' in some sense (thus their uppercaseness). use 'nextsame' in the normal case: gist.github.com/391806
phenny masak: I'll pass that on when szabgab is around.
TimToady so one is a regex, and the other is just a function
masak phenny: tell szabgab and the fact that this wasn't obvious from the spec shows the need for a tutorial just like yours. :) szabgab++ 05:23
phenny masak: I'll pass that on when szabgab is around.
TimToady infix:tree<+> is possible, maybe
or some such
infix:ast<+>
it's probabably a feature to keep "infix" in there somehow 05:24
spinclad infix:<+>:ast (and infix:<+>:sym) ? keeps the 'infix:<+>' part together 05:26
TimToady I've been trying to avoid double adverbial names so far...
spinclad (but that perturbs things for R*, so not now?)
05:27 Guest42448 is now known as kaare_
TimToady though they are possible in theory 05:27
well, and Foo:auth<>:ver<> is certainly double adverbial
in some ways its easier for the <infix> multi dispatcher to just look for infix:sym though 05:28
and the <+> is the least specific part of it
s/least/most/
sorear I'm afraid you've lost me.
How does multi dispatch enter into this at all?
TimToady when you match <infix> it calls all the infix:sym* rules in some pecking order 05:29
primarily defined by LTM
as per recent clarification to S05
it's a form of multi dispatch
05:30 justatheory left
TimToady so the <op> part "varies most rapidly" from that viewpoint 05:30
sorear what about :ast and tree and so on? 05:31
spinclad a double adverb (in either order) makes a bit more sense to me: a :sym or :ast adverb with a value of <+> doesn't say the same as :name<+> (with 'name' blank) which could better be invariant...
TimToady well, it seems like the operator itself is the most commonly used one, so the that has the null key on the pair.
having a key of "sym" or "ast" is just a differently keyed pair 05:32
sorear what does infix:ast<+> mean and what namespace does it live in?
TimToady it's just a name that cannot be confused with either the operator or the parse rule 05:33
what namespace it lives in depends on its declarators
05:33 chitragupt left
spinclad (yes, null key is what i meant by "with 'name' blank") 05:33
TimToady it doesn't install anything special in the current grammar like infix:<op> does
sorear ah, so you're thinking that, instead of macro infix:<+>, I should write macro infix:sym<+> or macro infix:ast<+> ? 05:34
TimToady on that level, it's just a name that the reducer knows to look for
05:34 chitragupt joined
TimToady maybe 05:34
earlier I was suggesting that macro is really more of a scope declarator 05:35
so you might even write macro regex infix:sym<+>
and it would put it into the current grammar
that is, it would go in the 'my' slot
the word 'macro' would go where 'my' currently goes, that is 05:36
since its scope is really the current grammar, which is lexically scope at run time but really dynamically scoped at compile time
but have to think through the ramifications of such a change 05:37
esp in terms of exportable macros
05:37 gfx joined
TimToady anyway, the idea is to put it somewhere with a unique name/location that the normal reducer can find it 05:39
without having to paw through a bunch of unrelated named, hopefully
masak makes a second attempt at the Russian consulate
05:39 masak left
TimToady if the name to call can be generated directly from the node in question, then we can just call it without reflection 05:40
so the reducecheck for infix:sym<op> could always just be infix:ast<op>, instead of setting it explicitly in the :sym
dunno how well that generalizes outside the operator precedence parser though 05:41
sorear outside the operator precedence parser, macro actions run at reduce time
TimToady current reducecheck mechanism could conceivable take anonymous routines
but forcing a name works against that
sorear there is no conflict between reducer macros and parser macros for statement:, because they run at the same time with the same arguments 05:42
05:42 iblechbot joined
TimToady well, but reducecheck is really only needed for opp, since other reductions are done right at the time the rule reduces 05:42
sorear you seem to be talking about two distinct solutions at the same time
TimToady I always talk about many solutions at the same time :)
sorear one involving :sym and :ast, the other involving macro sub and macro regex
TimToady those are really independent choices on some level 05:43
sorear three, if you count the signature version
independant, how so?
what does macro regex infix:ast<+> { } mean?
TimToady both the :sym and the :ast rule would live in the same grammar class, I suspect 05:44
so 'macro regex infix:sym<op>' is the parser, but could also be 'macro method infix:sym<op>'
since regexes are really just methods in disguise 05:45
the :ast would very likely just be 'macro method infix:ast<+>'
if you just want to run it
in this view, 'macro' would just be a way to augment the current slang somehow 05:46
and method vs regex, and various names that get called at various times, are indpendent
'macro regex infix:ast<+>' would be nonsense 05:47
unless you really want to write your tree transformer as a regex
'macro method' is much more likely
sorear viewing macro as a scope declarator has some nicities, but it seems to miss the fact that a macro is a concerted augmentation of the grammar, the action engine, and the lexical scope
TimToady yes, but the current macro syntax is really klunky when you try to put all those things in 05:48
it's not clear where the sweet spot is for sugar
sorear I concede that having to wedge half of the macro into "is parsed" is not very sweet 05:49
TimToady and which notations require you to say things that you might not need this time
yes, that's a smell
though you have to admit that in encourages people to use existing rules :) 05:50
05:50 PZt left
TimToady a system of names that interrealate and can generate defaults for missing ones seems like the most flexible approach, and :sym vs :ast might be that 05:50
05:50 alester left
TimToady or, if there's some block we always have to have, make optional phaser-like annotations for extras 05:51
don't really have any phasers in regexen yet tho
:my $stub = BEGIN {...}; is closest we have currently 05:52
and maybe the regex isn't the 'master' block in any case 05:53
up till now, it's whatever returns the AST to the parser that is the master
maybe that could have a PARSE phaser or some such
just brainstorming now 05:54
more like braindrizzling, given my current mental state :)
anyhoo, the design of p6 has to to where it is primarily because we haven't just assumed that the first thing we think of is the best thing 05:55
or the second, or the third...
s/to to/got to/
gah, brane fried 05:56
sorear putting the regex as the master is the most natural in an order of operations sens
but it has the disadvantage of making for very topheavy code
TimToady but many of these operators have a default way of parsing
sorear our regexes are pretty space-efficient
yes
05:56 jonasbn joined
TimToady so it feels wrongish to me 05:56
but I agree with you about 'is parsed' also feeling wrongish 05:57
some of the previous thinking included 'augment slang { your regexen here }' 05:58
which is a good way to basically mix in a bunch of new methods
sorear related to this is the question of how "use COBOL" is supposed to work
TimToady that would be defining things from the grammar viewpoint primarily
05:59 steinberg left
sorear there isn't anything in the macro stuff about how a "use" can radically change an existing slang 05:59
TimToady yes, that sort of thing is essentially an exported 'supersede slang'
just kinda clobber the current %*LANG braid of languages and replace it with COBOL equivalents 06:00
it's possible that augment/supersede are the wrong declarator, but the right idea
since they aren't really about monkey typing something in place
sorear it's not clear how to do that clobbering using a symbol, though
TimToady more like 'constant %?LANG = stuff that gets back into %*LANG' 06:01
sorear oooh
constant works for ?-twigil stuff?
brilliant
TimToady well, somehow it has to change the type of the current $¢ in the parser, anyway 06:02
currently it flows the other way
if you have a $*foo variable in the compiler
and you say $?foo and there isn't one predefined
it assumes you mean a snapshot of the current $*foo value
sorear well, we have "gmagic" compiler-contextuals ($?LINE)
"smagic" ones like %?LANG make equal sense 06:03
TimToady std: say $?CURPKG
p6eval std 30561: OUTPUT«ok 00:01 109m␤»
TimToady that's actually just snapshotting $*CURPKG
spinclad std: say %?LANG
p6eval std 30561: OUTPUT«ok 00:01 108m␤»
TimToady logically we have a $*LINE, but that's not actually how we do lines 06:04
$?LINE is really $¢.pos.lineof
sorear right, hence "gmagic"
$*LINE is synthesized on the spot
TimToady anyhoo, grammar mods really just need to clobber the type of $¢ 06:05
what :lang() does in STD, and also .cursor_fresh if you give a new language
or more subtley, .tweak with a mixin
in rakudo there is currently a .LANG I believe 06:06
but it's probably a bit more brittle
06:06 gfx left
sorear back to pure macro land - how about "macro" <deflongname> <trait>* [ "parsed" '{' <regex> '}' | "slang" '{' <statementlist> #`(as if in a grammar) '} | "reduce" <signature> <blockoid> ]* 06:06
well, NQP-rx doesn't have roles
which makes proper slanging impossible
spinclad could that be, the next $¢ produced is of the new type? 06:07
sorear so my macros would look vaguely like Pascal subprograms
moritz_ good morning
TimToady yes, Cursors are immutable, so any method that rebless the next Cursor makes a new language
only in proper English
moritz_: o/ 06:08
sorear macro event
parsed { <identifier> }
spinclad \o: moritz_
sorear reduce ($/:) { ... }
;
no, I'm confusing myself 06:09
TimToady it would be nice to use some syntax similar to things we already have :)
sorear yes.
06:10 uniejo joined
sorear but we don't really have any syntax for "stuff with N semi-optional parts" 06:10
TimToady I will let it stew in my backbrain.
sorear or do we?
TimToady well, phasers, switches, related function/method names
spinclad (N optional nameds)
TimToady but not try {} finally {] style, by design 06:11
those became phasers
sorear a PARSE phaser could work 06:12
one other thing I don't really like
TimToady but not clear whether nested lexical scopes buy us anything there
sorear two, actually 06:13
TimToady well, there's 'don't really like' vs 'really don't like' :)
sorear one is that, in my model, 'macro foo' defines a reducermacro which has no choice but to reuse the same syntax &foo would use
06:13 chitragupt left
sorear the "declarator" macro pattern has to use term:<foo> 06:14
which is slightly smelly, all those term:s
moritz_ should macro foo { } be synonymous to macro term:foo { }?
TimToady might or might not be a problem, depending on how big the sweet spot is, and whether it makes hard things impossible
sorear two, separating the regex from the ast-builder means that captures have to be bundled into $/ and unpacked as a hash, it would be nicer if they could be lexically propagated somehow
TimToady moritz_: maybe, but that's a sugar 06:15
and maybe it should be reserved for listops
moritz_ TimToady: he, Perl 6 is full of it :-)
sorear moritz_: no, because then I'd need a new syntax for listop macros
moritz_ ok
06:15 Ross joined
moritz_ makes sense 06:15
moritz_ retracts his proposal
06:15 Su-Shee joined
TimToady on two, have to think about that when I'm awaker 06:16
spinclad sorear: two, like grammar and actions? 06:17
TimToady but sounds like a good spot for phasers seeing their outer scope
06:17 chitragupt joined
spinclad (i mean, i guess, the same applies to grammar and actions) 06:17
TimToady by default actions trigger when their associated rule terminates, but it gets more comlplicated with the opp 06:18
sorear rakudo doesn't currently have a syntactic category for list prefix ops
TimToady which is why Actions.pm (in STD's directory) has funny actions for PREFIX, INFIX, etc
those are actually called from EXPR
(the opp) 06:19
sorear I wonder if it should.
TimToady STD currently considers them prefix ops that mutate themselves magically into terms
but they could be as easily considered terms that happen to embed their own arglists 06:20
06:20 chitragupt left
TimToady would remove a special case from EXPR 06:20
not gonna decide that when my brane is much though
sorear in rakudo, all list prefix ops are parsed by term:sym<name>
TimToady *mush, even 06:21
that's perhaps saner, and what I was considering
nice to know that's what rakudo does
06:21 TiMBuS joined
TimToady it's a bit like the distinction between module declaratoins that take a block, and moudle declarations that take semicolon and chew up the rest of the file... 06:22
but now I'm just kinda free associating 06:23
moritz_ std: { class Foo; }
p6eval std 30561: OUTPUT«===SORRY!===␤Semicolon form of class definition not allowed in subscope;␤ please use block form at /tmp/qQCVtHqq48 line 1:␤------> { class Foo⏏; }␤ expecting any of:␤ name␤ trait␤Parse failed␤FAILED 00:01 107m␤»
moritz_ std++
06:23 viklund joined
TimToady it's not quite right yet though 06:23
std: ( class Foo; )
p6eval std 30561: OUTPUT«ok 00:01 107m␤»
JimmyZ std: augment my class foo { ... } 06:24
p6eval std 30561: OUTPUT«===SORRY!===␤Malformed augment at /tmp/ZPEvpVCG10 line 1:␤------> augment ⏏my class foo { ... }␤ expecting any of:␤ scoped declarator␤ typename␤Parse failed␤FAILED 00:01 107m␤»
TimToady you can still fool it
JimmyZ std: my augment class foo { ... }
p6eval std 30561: OUTPUT«===SORRY!===␤Malformed my at /tmp/YoJqBs5KpM line 1:␤------> my ⏏augment class foo { ... }␤ expecting any of:␤ scoped declarator␤ typename␤Parse failed␤FAILED 00:02 107m␤»
TimToady my and augment are both scope declarators and can't be neste
d
06:24 [mark] joined
JimmyZ how to augment non-global classes? 06:25
TimToady augment must name an *existing* class
sorear how to augment anonymous classes?
TimToady so must supersede
JimmyZ std: class foo { ... }; my augment class foo { ... }
p6eval std 30561: OUTPUT«===SORRY!===␤Malformed my at /tmp/UG9ZsAKzHj line 1:␤------> class foo { ... }; my ⏏augment class foo { ... }␤ expecting any of:␤ scoped declarator␤ typename␤Parse failed␤FAILED 00:01 107m␤»
JimmyZ std: my class foo { ... }; augment class foo { ... }
p6eval std 30561: OUTPUT«===SORRY!===␤Can't augment class foo without MONKEY_TYPING at /tmp/oPHn4AQHaH line 1:␤------> my class foo { ... }; augment class foo ⏏{ ... }␤Check failed␤FAILED 00:01 107m␤»
sorear ok, here's what I'll do 06:26
spinclad -> bed # .oO{ dream of adding a display argument to (some variant of) pir::newclosure --> dream of newclosure mess magically working, and pink unicorns, and stuff }
sorear 1. The bodies of all macros run at reduce time, with the matched portion of the subexpression in $/, and any outside portions (for prefix, postfix, infix, postcircumfix, and list prefix) in positionals / named 06:27
06:27 chitragupt joined
JimmyZ TimToady: S12 says, 'In order to discourage casual misuse of these declarators, they are not allowed on global classes unless you put a special declaration at the top: use MONKEY_TYPING;', but how to use it with non-global clasess? 06:27
TimToady well, STD hasn't caught up with the 'global' part yet 06:28
so it's a stdbug
sorear 2. Any macro which is attached to a syntactic category may give an "is parsed" flag, which is used as the parse rule prefixed with <sym> 06:29
3. If you wish to do things at category match time, put a closure in the parser. This is deemed a rare edge case.
TimToady well, 3 is perhaps 'augment slang {...}' time 06:30
or whatever syntax we do that with
sorear TBD.
TimToady I presume 2 means the is parsed rule is only for the infix itself, and relies on the opp to supply the args
sorear yes 06:31
TimToady seems sane-ish in my current branestait
06:31 chitragupt left
TimToady note that this is currently thought to be 'is reparsed', if it includes <sym> 06:32
but maybe that won't hold over the long haul
sorear what makrudo currently does is to set the regex to / <sym> <!PARSED__++> / and inject the is parsed regex as !PARSED__++ 06:33
TimToady if 'is parsed' is defaulting to just <sym> most of the time anyway
sorear which is a tad hackish but works
06:33 chitragupt joined
sorear oh, hmm 06:33
IMHO removing implicit <sym> from is parsed would make it more intuitive
Let's suppose I want to write a macro which expands to a method definition, but the method's name and signature are variable. Can I do this with quasiquotes? 06:34
06:34 Ross left
TimToady yes, I remember you asking that earlier 06:34
perhaps if we could tag unquotes with the proper category somehow 06:35
sort of type casting thing maybe
it's important to keep the parser knowing what to expect next
06:35 chitragupt left, sh0ck_ joined
sorear Despite what it says in the Pugs notes, I have no qualms about exposing Match objects to macros 06:36
06:36 sh0ck_ is now known as sh0cked
sorear Standing here and looking forward, I am convinced that anything that can parse full Perl 6 is going to be functionally equivalent to a Perl 6 regex engine 06:36
so it can support $/ fine
TimToady well, have to keep our matches and asts straight, but types our our friend, maybe
in a sense, we want unquote to have a return sig 06:37
or some way to insert the appropriate Match/Ast object at that point, given something non-match/ast-ish 06:38
but I may be talking gibberish
I agree that the current unquote may not be quite right
sorear giving unquote a return sig isn't quite enough
TimToady in the sense that the compiler needs to know the returned type at compile time, and the innards may not supply that
06:39 chitragupt joined, dduncan left
sorear having to put <starter><starter><starter> into every protoregex is probably going to wreak havok with LTM 06:39
especially since it runs contrary to TTIAR
and the abilitiy to distinguish terms from ops
TimToady for missing function name, for instance, you want to reassure the compiler that this unquote returns a <longname>
it's kinda like you want to put regex into the unquotes and hoist those outward into the is parsed as appropriate 06:41
TimToady now no longer has a mush brane, but a puddle brane
"Hey, I think it's been raining cats and dogs." "How d'ya know?" "I just stepped in a poodle." 06:43
06:43 chitragupt left
TimToady zzz & # thunk 06:43
06:43 Ross joined, Ross left, Ross joined 06:44 chitragupt joined
sorear phenny: tell TimToady - oh dear, I think I just found a fatal flaw 06:46
phenny sorear: I'll pass that on when TimToady is around.
sorear phenny: tell TimToady - my-macro-which-works-just-like-sub foo(&infix:<¢>) { 2 ¢ 3 } # During parse, the compiler doesn't know the lexical environment of the <blockoid> and can't parse the operator correctly. 06:47
phenny sorear: I'll pass that on when TimToady is around.
06:47 kst left 06:48 kst joined
TimToady reawakens briefly and point sorear at 'lift', which was invented for this, and has nothing to do with Haskell 06:51
phenny TimToady: 06:46Z <sorear> tell TimToady - oh dear, I think I just found a fatal flaw
TimToady: 06:47Z <sorear> tell TimToady - my-macro-which-works-just-like-sub foo(&infix:<¢>) { 2 ¢ 3 } # During parse, the compiler doesn't know the lexical environment of the <blockoid> and can't parse the operator correctly.
TimToady or maybe not 06:52
rethunk&
sorear hmm. I think parsing modules has something to do with dynamic_wind - from a certain perspective, we need to not see the contextualized modifications to the parsing context 06:55
06:58 steinberg joined
moritz_ TimToady: should MyGrammar.parse($str) set $/? 06:58
07:03 Ross left 07:09 xabbu42 joined, xabbu42 left, xabbu42 joined 07:15 chitragupt left 07:16 chitragupt joined 07:24 chitragupt left 07:25 chitragupt joined 07:26 envi^home joined 07:31 tylerni7 left 07:33 tylerni7 joined 07:47 flip214 joined, kst left 07:48 kst joined 07:54 bbkr_ joined, k23z__ joined 07:57 prfc0 joined, chitragupt left 07:58 bbkr_ left, chitragupt joined
sorear What are %*COMPILING, %COMPILINGPACKAGES, and %STUBCOMPILINGPACKAGES used for? 08:02
moritz_ I guess the STUB ones are those that don't warn/barf about redefinition 08:04
08:06 chitragupt left 08:08 chitragupt joined 08:22 plobsing left 08:23 kst left, kst joined
flip214 A short question: is there some reflection for a grammar? 08:25
I'm asking because I'm thinking about a commandline program, that uses the grammar to parse the commandline parameters and the input (STDIN, load file) - similar to the various configuration interfaces. 08:26
moritz_ grammars are just like classes
flip214 And if there was a way to find out which token didn't match/would match next that could be used for readline completion, too - looking what would be expected next, and print it's representation
moritz_ so you can use the same introspection for grammars that are available for classes 08:27
flip214 Only for user-specified names there'd have to be callback functions, to look which names would make sense here.
Has someone already done a readline completion for grammars?
moritz_ but I don't think the level of detail you need is specced or implemented
STD.pm's operator precedence parser has something like that 08:28
it tells you what kind of things were expected
std: 1 + **
p6eval std 30561: OUTPUT«ok 00:01 110m␤»
moritz_ std: 1 + ^^
p6eval std 30561: OUTPUT«===SORRY!===␤Preceding context expects a term, but found infix ^^ instead at /tmp/J2yu0n9ZGg line 1:␤------> 1 + ^^⏏<EOL>␤Parse failed␤FAILED 00:01 108m␤»
08:29 masak joined
masak five minutes into the train journey, I realize I've forgotten the passport. :( so no visa today. the entire journey to Russia and my participation in DevConf now hang by a thread. 08:32
moritz_ :/
masak not the best morning so far. 08:33
moritz_ can't you backtrack to obtain your passport?
08:35 zzz_ joined
masak I have backtracked, if only to get back to $work. but that puts me outside of their opening hours. 08:35
moritz_ :( 08:36
08:36 yinyin joined, zzz_ left 08:39 PZt joined
viklund it takes patience to get a visa at the russian embassy 08:42
masak patience mixes badly with lack of time. :/
viklund you really need to be there when they open, if there's still a queue when they close they just send the queue home...
masak oh well, it seems I'll get a last chance tomorrow...
viklund good luck 08:43
moritz_ /me wonders if Russia doesn't like visitors
masak viklund: I know about the queue :)
moritz_: at the very least, they seem to want to make sure that when you want to go there, you really mean it. 08:44
moritz_ aye
viklund maybe they're afraid of drop-outs from Europe
moritz_ it seems that setting $/ in the closure that you pass to .subst isn't easy 08:45
$/ is lexical
so you'd have to set a lexical on a sub-to-be-called
and it seems that parrot doesn't support that notion 08:46
I'll ask on parrot-dev, but glancing over the docs wasn't very encouraging 08:47
08:48 chitragupt left 08:49 flip214 left, chitragupt joined
mathw Morning 08:50
08:51 chitragupt left 08:55 chitragupt joined 09:07 lestrrat is now known as lest_away, chitragupt left 09:09 chitragupt joined
masak mathw: \o 09:09
mathw: have you seen that named enums exist a little again?
moritz_ rakudo: enum A <a b c>; multi x(a $x) { say $x }; x a 09:10
p6eval rakudo 1eef08: OUTPUT«0␤»
masak rakudo: multi foo(0 $x) { say $x }; foo 0
p6eval rakudo 1eef08: OUTPUT«0␤»
masak rakudo: multi foo(0 $x $y) { say $y }; foo 0 09:11
p6eval rakudo 1eef08: OUTPUT«Malformed multi at line 11, near "foo(0 $x $"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
09:14 ktne joined
masak is sad to see in the logs that BrowserUk gave up so easily :/ 09:18
moritz_ huh? when? what did I miss? 09:19
masak grabs link 09:20
irclog.perlgeek.de/perl6/2010-05-06#i_2298685
moritz_ thanks 09:22
09:24 prfc0 left
moritz_ you can have both good discussions and good fights with BrowserUk on perlmonks, IME :-) 09:26
luckily good discussions more often 09:27
masak: in your opinion, should MyGrammar.parse($str) set $/? 09:28
09:28 chitragupt left
masak moritz_: good question. 09:29
moritz_: I could ack my modules and see how often I make use of that.
I think I use it in a few places at least. 09:30
moritz_ the tests also seem to assume that in a few places
masak (1) if it didn't, how would one do to actually set $/ ? (2) what's the worst thing about .parse setting $/ ?
09:31 chitragupt joined
moritz_ (1) $/ := G.parse(...) probably 09:31
or maybe assignment even
(2) clobbering an existing $/
masak yes of course. 09:32
but so does m//
they do feel a bit similar.
09:33 mikehh left, yinyin left
masak so far I've found the 'sets $/' assumption in code in November and Text::CSV. 09:34
moritz_ well, I'll try to implement it
09:41 chitragupt left, chitragupt joined, PZt left 09:45 lest_away is now known as lestrrat 09:47 chitragupt left 09:51 arthur-_ left, chitragupt joined
moritz_ > grammar G { token TOP { bar } }; G.parse('bar') && say "go to $/" 09:53
go to bar
09:55 dakkar joined
pugssvn r30562 | moritz++ | [t/spec] unfudge a test for $/ after .parse 09:58
mathw masak: awesome 10:00
dalek kudo: 7563a50 | moritz++ | src/core/ (2 files):
Grammar.parse sets $/ in the caller's scope
10:08 chitragupt left 10:10 chitragupt joined
masak brief mention of Perl 6 in a Spolsky talk: use.perl.org/~jozef/journal/40345 10:16
he seems to be running under the assumption that Perl 6 implies no Perl 5. 10:17
moritz_ that's a common assumption 10:19
JimmyZ can't visit youtube because GFW :(
moritz_ we'll need a good slashdot article to refute that believe
masak YouTube is petty borgeoise and must be resisted!
JimmyZ: my suggesting it might not be apropos, but have you ever considered circumventing GFW? I hear there are ways. 10:26
10:26 ruoso left
moritz_ which are illegal, as I've heard 10:27
JimmyZ Yes, and slow.
moritz_ so it's no good talking about that in a publicly logged channel :-)
JimmyZ No, there is no law on it.
masak I was only talking about considering it :P
JimmyZ It's not illegal
moritz_ JimmyZ: then I've read a mis-informed article the other day 10:28
which is no surprise
the quality of reports about "far east" isn't very good here
masak it's much better than it was 300 years ago :)
JimmyZ the law is not on paper, just on someone's mouth :) 10:29
hejki the quality of media about "anything" isn't very good anywhere
moritz_ hejki: depends on the medium
hejki with little extrapolation it doesn't
:)
moritz_ I regularly read good Perl 6 coverage on planetsix :-)
masak well, naturally, if you feel the need to circumvent the belovelent state's protection mechanisms, then you're already a bit suspect.
moritz_: I tend to find planetsix very biased. don't believe all you read. :P 10:30
JimmyZ nope, there is no law on paper. just on mouth of some departments
Su-Shee the chief scalability techie guy of flickr also mentioned Perl 6 in a talk (rant why django sucks at some django conference :) 10:31
masak Su-Shee: URL?
moritz_ masak: much of the bias on planetsix is clearly marked as personal opinion
Su-Shee masak: *headscratch* lemme think about it..
masak moritz_: I was mostly joking. 10:32
moritz_: but a non-Perl6er could very well look at Perl 6 blog posts and think 'biased'.
moritz_ masak: I know; still I have a habit of giving serious answers to joke quetions and statements
masak now who would ever do such a thing... :P
Su-Shee masak: ah, here: www.youtube.com/watch?v=i6Fr65PFqfk 10:33
masak Su-Shee: thank you.
aesop o/
Su-Shee well the problem is that "the environment" just doens't mention p5/p6. I read plenty web stuff for example which just doesn't mention any perl.
but the smalltalk traits folks mention p6 plenty! 10:34
masak aesop: \o 10:35
Su-Shee: they do? cool!
Su-Shee masak: scg.unibe.ch/research/traits 10:36
also I like to point out that this is within the context of academic research which might be a realm worth pushing into with bleeding edge features. 10:37
masak nice.
10:38 chitragupt left
JimmyZ good video :) 10:39
masak sorear: thank you thank you thank you for getting implementation-based discussion of macros going. I've been waiting for this for years! 10:41
10:41 chitragupt joined
moritz_ non-bikeshed-discussions++ 10:41
Juerd moritz_: But... 10:43
moritz_: We need to discuss the color of the non-bikeshed.
Green!
Su-Shee JimmyZ: it's pretty funny.
masak is a non-bikeshed a shed for non-bikes or a non-shed for bikes? 10:44
10:44 snarkyboojum joined
Juerd masak: I think it is. 10:44
masak :)
aesop lol
masak snarkyboojum: \o
Juerd \o without digits feels weird.
snarkyboojum masak: o/ 10:45
Juerd So incomplete, like \x
masak is that a lawn chair waving?
aesop lol
masak senses a possible miscommunication
Juerd: digits/ \o # better? 10:46
10:46 JimmyZ left
Juerd masak: Er... :) 10:47
masak Juerd: oh, I see now!
Juerd: you were referring to the lack of fingers on the backslash-arm. :)
Juerd Not really :)
masak maybe there's something up in Unicode to help you out. :P
Juerd \o127 10:48
masak or you could use quotes: "\o
Juerd: (yes, I know) :P
just exploring the misunderstanding space. 10:49
Su-Shee btw roles and traits.. I have to do a presentation and I want to show a "Member" class (the usal "Person" stuff - we actually have those though :) and I was thinking to illustrate roles with something like Member does OutputJSON, OutputXML, DumperObject and so on - would one actually use Roles that way?
masak Su-Shee: I'm no expert, but sure. 10:50
Su-Shee to compare it against "Member inherits from Object the methods output_as_json.." and so on.
masak: noone is. ;)
masak Su-Shee: it does sound a bit like what I discussed in my 'Ovid is right' post.
10:50 kst left
Su-Shee I'm trying to illustrate the "behavoir" point clearly marked with "does" and why this is a practical thing. 10:51
masak: url?
masak use.perl.org/~masak/journal/40107
10:52 rv2733 joined
Su-Shee the interesting stuff is that in some smalltalk traits papers OO with traits/roles essentially goes into the direction of using classes for their inheritance capabilities and their attributes, but take much more methods out of classes and in traits/roles instead. not sure what to make of that. 10:52
masak neither am I. 10:53
I touch upon the possibility of that at the end of the post.
still on the fence, really. 10:54
Su-Shee essentially you would only include methods into classes which really really really make sense in any child derived and where inheritance is obvious. all others -> trait/role
10:55 kst joined
Su-Shee that is probably the key aeh sentences: Classes are primarily for instance management, not code reuse. Consider using roles when you simply want to factor out common code. 10:55
10:57 chitragupt left
masak *nod* 10:57
I understand it, but I don't yet grok it.
Su-Shee actually, I going via the semantics. I wrote down our Member stuff and tried to actually distinct between "Member has foobar" and "Member does foobar" which ended up with much more "does" then "has" 10:58
10:59 chitragupt joined
Su-Shee (afterwards I was sure that I plainly didn't get OO - yet again. ;) 10:59
11:02 chitragupt left 11:04 chitragupt joined
masak Su-Shee: if roles are anything like the basic parts of OO design, it's more a question about seing past one's 'naturalistic' preconceptions of the various concepts ("Of course a board game has players, and not the other way around! That's how it is in Real Life!") and instead focus on what combination of features gives the most functional/maintainable/extensible architecture. 11:13
Su-Shee yeah, but usally perl has its language stuff for a reason :) 11:15
11:15 mikehh joined
masak even so, even with the defaults poking you in a sensible directions, a lot of (sometimes crucial) OO design choices are left up to the programmer. 11:16
good OO design is one of those "I know it when I see it" phenomena.
Su-Shee well I know bad design when I designed myself into a corner. ;) 11:18
masak nods in recognition
Su-Shee feeling like a genius having an epiphany usally is also a bad sign. ;) 11:19
masak phenny: tell szabgab oops -- was talking about this: szabgab.com/talks/perl6/perl6-class...tance.html 11:20
phenny masak: I'll pass that on when szabgab is around.
11:22 kaare__ joined
masak phenny: tell szabgab , as for the TODO question in szabgab.com/talks/perl6/perl6-classx.html -- the ultimate reason is that the normal accessors return an lvalue (that you can assign to). 'course, the accessors could be extended to take an optional argument, but it makes sense to encourage the assignment syntax, which is clearer anyway. 11:23
phenny masak: I'll pass that on when szabgab is around.
11:24 chitragupt left, chitragupt joined 11:25 kaare_ left 11:27 snarkyboojum left, snarkyboojum joined 11:32 agentzh left
masak moritz_: Hitomi in Web.pm, and Yapsi also use $/ immediately after parse. so that's four cases where I assume it gets set, and 0 cases so far where I don't. :) 11:35
11:35 chitragupt left
Su-Shee masak: the guy who implemented the javascript traits module, defines it like so: "A trait is essentially a group of pure methods that serves as a building block for classes and is a primitive unit of code reuse. In this model, classes are composed from a set of traits by specifying glue code that connects the traits together and accesses the necessary state." 11:35
(which reads pretty understandable..) 11:36
masak aye.
11:36 chitragupt joined
masak but it doesn't address the question of how this meshes with inheritance. is inheritance even needed under such a model? when faced with a choice between making something a role or a class, all other things being equal, what should one choose? 11:37
moritz_: hm, actually, that's two cases in Hitomi. so 5-0. 11:38
11:39 chitragupt left
moritz_ masak: implementation is already pushed 11:42
masak yes, I saw that. moritz_``
moritz_++
11:42 chitragupt joined 11:43 clintongormley joined
Su-Shee masak: I think, I still would use inheritance to get from the common to the special. in our Member stuff, my base Member class would just know an id and a usename (because that's the only stuff really needed mostly) and go to more specialized Members like Admins from there. 11:46
11:46 snarkyboojum left
moritz_ Admin sounds like a fine role 11:49
11:49 arthur-_ joined
Su-Shee ok, then intern and extern and guest member. which are only distringuished by attributes they have. 11:53
takadonet morning all
Su-Shee though we have like 4 different admins ;)
moritz_ enum Admin <bart home marge lisa> 11:54
or so :-)
11:55 iblechbot left
masak takadonet: \o 11:56
takadonet masak: Is jnthn still on holidays?
masak takadonet: yes. 11:58
moritz_ I'm pretty sure he is
11:58 chitragupt left 12:04 jnthn left, chitragupt joined, mathw left, bloonix left, broquaint left 12:05 bloonix joined, broquaint joined 12:06 chitragupt left, Gothmog_ left 12:07 chitragupt joined 12:08 kst left, kst joined 12:09 bloonix left 12:11 mathw joined 12:17 jnthn joined, chitragupt left, bluescreen joined, bluescreen is now known as Guest14064, bloonix joined, chitragupt joined
moritz_ rakudo: say 12:17
p6eval rakudo 1eef08: OUTPUT«␤»
12:17 Guest14064 left
moritz_ std: my %h = { a => b }␤say %h 12:17
p6eval std 30562: OUTPUT«===SORRY!===␤Undeclared routine:␤ 'b' used at line 1␤Check failed␤FAILED 00:01 110m␤» 12:18
moritz_ std: my %h = { a => 'b' }␤say %h
p6eval std 30562: OUTPUT«ok 00:01 109m␤»
moritz_ rakudo: my %h = { a => 'b' }␤say %h
p6eval rakudo 1eef08: OUTPUT«a b␤␤»
moritz_ huh, are there two newlines in the output?
12:22 mathw left 12:25 chitragupt left 12:27 chitragupt joined
masak phenny: tell szabgab re szabgab.com/talks/perl6/perl6-scalars.html -- 'undef' doesn't exist anymore. instead, it's all different values of undefined types. an untyped variable will start its existence holding the value Any(), the Any type object. 12:30
phenny masak: I'll pass that on when szabgab is around. I'll have to use a pastebin, though, so your message may get lost.
12:32 mathw joined
masak phenny: yeah, it's growing to be quite a lot, I know. :) 12:32
12:32 ruoso joined, TiMBuS left, Gothmog_ joined, chitragupt left 12:33 chitragupt joined 12:34 chitragupt left 12:35 chitragupt joined
masak phenny: tell szabgab re szabgab.com/talks/perl6/perl6-boole...sions.html -- both those zeros are the correct output. however, they should both be typed as Bool, and (0 xor 0) gives an Int right now in Rakudo. that's wrong. 12:36
phenny masak: I'll pass that on when szabgab is around. I'll have to use a pastebin, though, so your message may get lost.
masak I believe there's a ticket for it.
here: rt.perl.org/rt3/Ticket/Display.html?id=73280 12:37
moritz_ yes; I just don't know if it's golfed that much
masak it's talking about ^^
but I'd bet the same goes for infix:<xor>
rakudo: say (90 xor 4).WHAT
p6eval rakudo 1eef08: OUTPUT«Method 'WHAT' not found for invocant of class 'Undef'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
masak pugs: say (90 xor 4).WHAT
moritz_ we're not talking about the same thing, I guess
p6eval pugs: OUTPUT«Bool␤»
masak rakudo: say (0 xor 0).WHAT 12:38
p6eval rakudo 1eef08: OUTPUT«Int()␤»
masak updates ticket subject
Rakudo dies when something goes wrong with &open, right? is that the long-term Perl 6 behavior? I don't see anything about the expected behavior in S32/IO. 12:45
arnsholt I'd argue that it should throw an exception of some kind on error 12:47
Magical return values are evil
masak you must be new here :)
unthrown exceptions are all over the place in Perl 6.
that's what &fail does, in essence.
arnsholt True. But you get an exception when you try to evaluate it, no? 12:48
So it's essentially a lazy exception (which I think is a neat idea)
It's the C-style magical return values I don't care for
masak ah.
indeed.
I got to thinking about it because szabgab's code does if (open ...) { ... } else { ... } 12:49
12:49 chitragupt left
arnsholt &fail is quite cool because it will only raise an exception when the resource is actually needed 12:49
masak ...which I guess will work if the exception is returned unthrown.
arnsholt Although fail-early semantics would probably be nice in some settings as well
12:50 chitragupt joined
moritz_ that'll be available with 'use fatal "fail"' or so, iirc 12:51
masak and, for more fine-grained control, 'or die' will still work fine.
12:56 chitragupt left
arnsholt moritz_: Good plan 12:56
12:56 chitragupt joined
moritz_ masak: 'or die' is not awesome, because it requires you to write your own awesome error messages 12:56
12:57 webdexter joined
webdexter hello 12:57
masak moritz_: hm, true. $! will not contain the error.
moritz_ hi
masak webdexter: hi!
12:59 chitragupt left
webdexter masak : iam totally new in perl, juste would like to modify a source from an open source project, I would like to replace a line that said if this field begin with 192.168 "if ($fields[0]=~/^192\.168\..*/)" by a line that said if field end with -color . I mean now the format of my field is "192.168.1.1-red" and I would like to say : if end is red then ... if end is yellow then .. 13:00
masak mberends: hey, three weeks ago you changed "Its only purpose" to "It's only purpose" in the proto README. don't do that. :-P
webdexter: sounds like you want #perl, not #perl6.
well, you may want us too, but you don't seem to know it... :)
webdexter masak : yes problably wrong channel sorry :) but I have #perl :Cannot join channel (+r) - you need to be identified with services (Iam quiet new on irc) 13:01
masak webdexter: I can help you on privmsg if you want, but general Perl 5 help is OT for this channel.
13:02 chitragupt joined
hejki webdexter: masak is right, but you could do sth like if ($fields[0] =~ /^192\.168\.\d+\.\d+-(\S+)/) { if ($1 eq "red") { .. } elsif ($1 eq "yellow") { .. } } :)= 13:02
pumpin' sum iron & 13:04
masak hejki: and if this'd been the #perl channel, mst would have told you to "PLEASE USE ENGLISH. sth IS A CONCEPT FROM THE DBI MODULE"
:)
hejki use English; is for nubcakes :)
masak writes down 'nub... cakes...' 13:05
hejki also i have unfinished jobs, laters :)
13:07 chitragupt left, SmokeMachine joined, plobsing joined 13:08 chitragupt joined
masak phenny: tell szabgab re szabgab.com/talks/perl6/perl6-undef-defined.html -- 'defined' still exists (both in sub and method form), but 'undef' is gone. 13:11
phenny masak: I'll pass that on when szabgab is around. I'll have to use a pastebin, though, so your message may get lost.
masak phenny: tell szabgab same thing with szabgab.com/talks/perl6/perl6-list-...nment.html -- $z will be Any(). also, the 9 will cause an overflow warning in real Perl 6. 13:14
phenny masak: I'll pass that on when szabgab is around. I'll have to use a pastebin, though, so your message may get lost.
13:15 xinming_ is now known as xinming 13:18 kaare__ left 13:20 snarkyboojum joined
moritz_ poor szabgab 13:20
masak: you know, you can reach him by privmsg too
masak I didn't plan on passing him that many messages. it... just happened... 13:21
13:23 JimmyZ joined
JimmyZ \o, I sent a patch rt, but nothing is there, seems that rt blocked me again. 13:24
bbkr rakudo: my %h = { "a" => %h }; say %h.values.perl; #this one loops badly
p6eval rakudo 1eef08: OUTPUT«maximum recursion depth exceeded␤current instr.: 'parrot;P6protoobject;VTABLE_name' pc 1298 (runtime/parrot/library/P6object.pir:764)␤»
moritz_ JimmyZ: it sometimes takes a few minutes to hours, even for non-blocked users 13:25
JimmyZ 4 hours...
bbkr submits 13:26
moritz_ bbkr: that's already known
masak JimmyZ: I see your patch: rt.perl.org/rt3/Ticket/Display.html?id=74942
moritz_ bbkr: and there's a rather old ticket about it, iirc
colomon is it really a bug?
moritz_ masak: didn't you apply that patch?
bbkr thanks for info moritz_ :) 13:27
masak moritz_: I think it might have been a very similar one.
moritz_ I'm talking about 376196dffcd4fa8ea337d4be7c62f4593494c952
13:27 viklund left
masak moritz_: yes. but that one contained only one hunk. this one contains three. 13:28
JimmyZ masak: I can't see it in p6c 13:29
13:30 chitragupt left
masak JimmyZ: what moritz_ said about delays. 13:30
JimmyZ I wan't see it on RT, but RT lists the oldest one in the first 13:31
er, want
moritz_ JimmyZ: if you click on the "created" column heading, you can revert the sort order 13:32
JimmyZ I don't know how to do it. I couldn't find the button 13:33
moritz_ you need to be logged in to do that :( 13:34
bbkr how to delete hash key? S09 mentions only declarations.
moritz_ in Perl 6 with %hash<key>:delete
in rakudo with %hash.delete('key')
JimmyZ :( 13:35
bbkr works, thanks.
moritz_ in general it's bad if rakudo deviates from the spec, but I really hate the :delete thing
the verb going at the end is like reversed yodaspeak 13:36
JimmyZ I want del %hash<key>
moritz_ excuse me, reversed yodaspeak the verb going at the end is
JimmyZ ;)
moritz_ JimmyZ: you won't get it 13:37
bbkr indeed, :delete is ugly and hard to remember
13:37 PZt joined
JimmyZ I can decalre it by myself 13:37
moritz_ have fun trying
it would need to some kind of a macro
aesop i like the %hash.delete('key') better
moritz_ because %hash<key> is evaluated before del() is called 13:38
so the association with the hash is lost
(which is why the spec doesn't use it)
bbkr rakudo: my %h = {"a" => "b"}; %h{"a"} = %h.delete("a"); say %h.perl; # is it bug? right side should be evaluated before assignment and then %h{"a"} initialized on left side.
p6eval rakudo 1eef08: OUTPUT«{}␤»
moritz_ in perl 5 it's a syntactic special form
JimmyZ Yeah, %hash.delete('key') looks better
aesop slightly more readable 13:39
bbkr votes for %hash.delete('key1', ...)
JimmyZ More OOP
blames it to TimToady
my poor english again :( 13:40
bbkr std: my %h = {"a" => "b"}; %h{"a"} = %h.delete("a"); say %h.perl; 13:42
p6eval std 30562: OUTPUT«ok 00:01 111m␤»
bbkr std: my %h = {"a" => "b"}; %h{"a"} = %h.delete("a"); say %h.keys;
p6eval std 30562: OUTPUT«ok 00:01 109m␤»
moritz_ STD.pm6 just checks syntax 13:43
and it's valid method call syntax after all :-)
masak moritz_: I always thought Perl 6 (spec) still has both .delete and :delete, because TimToady wanted the latter form only but hasn't decided to scrap the former one yet.
ditto :exists.
and yes, I don't like those adverbial forms either. 13:44
mathw I don't understand them
13:44 quester_ joined
mathw They seem to serve no purpose 13:44
mathw nearly wrote 'serve no porpoise' which is probably true
moritz_ masak: grepping for 'delete' should only one mention that's not an adverb form
and there it explicitly says 13:45
The interpretation of @indices is subject to change. The normal way to delete
is by applying a ":delete" adverb to any subscripting operation.
masak nod. 13:47
moritz_ I'm kinda inclined to guerilla-add it in 13:48
masak I still seem to recall TimToady not being entierly decided on the issue.
maybe that's wishful thinking or projection, though :P
bbkr this ":delete" form looks weird if you want to perform operations on deleted value. for example how this will look in :delete form: '%h.delete("a", "b", "c").[2].say' ? 13:50
%h{"a", "b", "c"}:delete.[2].say ?
colomon seems to me I remember someone explaining to me why the adverb form was better back when I first joined #perl6. but I've got not the faintest memory of what the explanation was... 13:51
masak bbkr: probably. might be that parens are needed, I'm not sure.
colomon: let us know if you rediscover it! :)
moritz_ colomon: might be because it uses the same subscripting syntax and semantics as "normal" postcircumfix:<{ }>
colomon masak: I also seem to remember TimToady being down on adverbs in general...
masak seems to have a problem with idioms involving 'down' this week 13:52
colomon: 'being down on' is positive, right?
colomon no, negative.
bbkr masak: and parens make code looks messy. OO syntax .delete(@keys) does not cause confusion. 13:53
masak oh.
13:54 iblechbot joined 13:55 pmurias joined
colomon if you say %h<a b>:exists, is it either <a> or <b> exists, or both? 13:56
13:57 rgrau_ joined
moritz_ "yes" 13:57
hejki :D
moritz_ I'd assume "both", but I'm not sure
colomon masak: "being down with" is is an older "hip" way of saying positive. Just a weird English idiom thing. 13:58
diakopter pmurias: sorry I broke perlesque
pmurias: fixing
14:01 chitragupt joined
pmurias diakopter: hi 14:05
phenny pmurias: 04:10Z <diakopter> tell pmurias I got parametric type declarations working. That is, they're compiled to fields/locals in the CIL. You can't use .new on them yet, tho.
pmurias diakopter: why do i talk to to get shell access to the p6eval box? 14:06
bbkr colomon: most OO programmers would think this way %h<a b> returns an Parcel, but Parcel does not have "exists" method, so what is going on?
if it's %h.exists("a", "b") then it's very clear what it does 14:07
diakopter pmurias: me, since it's mine
moritz_ %h.exists('a' & 'b') would be clear 14:08
pmurias could i have a shell account? ;)
moritz_ or %h.exists('a' | 'b')
bbkr besides - OO form can be used with junction: %h.exists(1|2|3) and it's also very understandable. :exists form
is messy
colomon bbkr: I agree wholeheartedly. Like I said, I can't even remember the alleged reasons %h<a b>:exists was supposed to be better.
pmurias diakopter: so i can add mildew-perlesque to the evalbot 14:09
bbkr right! moritz_ is a mindreader :P
diakopter pmurias: privmsg me
dalek meta: r239 | diakopter++ | trunk/Sprixel/src/ (2 files):
[perlesque] attempt to fix sub declarations return parameterized types
14:09 webdexter left
pugssvn r30563 | pmurias++ | [mildew] added App::Persistent support to the new mildew executable 14:10
14:10 rgrau_ left
diakopter boggles to think about trying to contain an oil leak 1.6km underwater 14:12
14:13 xabbu42 left 14:14 kst left, kst joined
moritz_ get as much $heavy_stuff on it as possible? 14:15
thousands of tons of concrete? 14:16
diakopter www.cnn.com/2010/US/05/06/gulf.oil....tml?hpt=T1
moritz_ really depends on pressure with which the oil comes up 14:17
aesop russia is recommending we use underwater nukes
moritz_ that sounds a bit... unpredictable
and probably not very healthy for the sea life, if any is left 14:18
PerlJam the oil isn't very healthy either. 14:19
14:20 alester joined
PerlJam (not to say that the damage is comparable, but it might be given the long-term effects of hydrocarbons on the reproductive processes) 14:20
moritz_ but what's the chance of actually closing the leak with a nuke? 14:21
JimmyZ %h.exists 'a' | 'b'
diakopter wow; cpanminus is speedy-seeming 14:22
masak JimmyZ: for method calls, you have to use a colon: %h.exists: 'a' | 'b' 14:23
JimmyZ rakudo: say: 'hello'
p6eval rakudo 1eef08: OUTPUT«Confused at line 11, near "say: 'hell"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤» 14:24
PerlJam moritz_: I think the russians may have a bit of experience with tactical explosions of this nature, so I would guess the chances are close to 80% or 90%
moritz_: with a 100% chance of killing lots of stuff in the area
14:24 plobsing left
diakopter yeah, but who's to say the rubble would land at all on top of the oil vent; maybe it would just create a very deep crater, and the oil vent would still be exposed to the sea 14:25
moritz_ that's my fear as well
diakopter same reason my nuke-the-volcano idea wasn't sound
14:26 plobsing joined
PerlJam diakopter: I would imagine it's not about rubble, but more about creating high enough temperatures that the "dirt" fuses into a solid mass and caps the leak 14:26
diakopter oh 14:27
moritz_ well, if they do try it, I'm happy it's not close to Europe 14:28
14:28 kaare joined, uniejo left, kaare is now known as Guest13256
PerlJam is only a few miles from the Guld of Mexico and just a few hundred miles from where the leak is. 14:30
arnsholt moritz_: Water is actually very good at absorbing radiation, from what I hear 14:31
masak std: say: "hello"
p6eval std 30563: OUTPUT«ok 00:01 109m␤»
masak JimmyZ: but it parses as a label.
moritz_ still, it sounds like casting out devils by Beelzebub
arnsholt True, true 14:32
masak people are still attempting that? hah, noobs! 14:35
"Some people, when confronted with a devil, think 'I know, I'll use Beelzebub.' Now they have two devils." 14:37
moritz_ that hurt. 14:38
masak sorry :/ 14:39
it was a bit of a *SCNR* thing...
masak stops being a nuisance and does something useful 14:42
moritz_ masak: speaking of useful work... have you thought about where you do your gsoc project? 14:43
masak: iirc the recommend you do it in branch
so that it's easy to tell which changes were part of your gsoc work 14:44
and then sync to trunk now and then, or so?
14:44 lrnperl6 joined
lrnperl6 morning all 14:44
moritz_ or just work in master, and in the end cherry-pick all your branches, pack them in a zip file, and upload it as your "deliverables"?
masak moritz_: haven't given it much thought. doesn't seem like a big issue to collect commits in the end, though. 14:45
lrnperl6: \o
moritz_: I'll probably end up doing atomic commits in master, and bigger things in branches.
14:47 stepnem left, plobsing left 14:52 BrowserUk joined 14:58 quester_ left 14:59 sh0cked left 15:00 stepnem joined
bbkr rakudo: my %h = {my => "x"}; # should this parse? 15:06
p6eval rakudo 1eef08: OUTPUT«Malformed my at line 11, near "=> \"x\"}; #"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
masak rakudo: class A { my submethod foo() { say "OH HAI" }; method bar() { foo(self) } }; A.new.bar # just realized that this works!
p6eval rakudo 1eef08: OUTPUT«OH HAI␤»
szabgab phenny, so how can I listen to my messages?
phenny szabgab: 05:22Z <masak> tell szabgab that I don't know if we've touched on this before, but both .HOW and .WALK are 'internal' in some sense (thus their uppercaseness). use 'nextsame' in the normal case: gist.github.com/391806
szabgab: 05:23Z <masak> tell szabgab and the fact that this wasn't obvious from the spec shows the need for a tutorial just like yours. :) szabgab++
szabgab: 11:20Z <masak> tell szabgab oops -- was talking about this: szabgab.com/talks/perl6/perl6-class...tance.html
szabgab: 11:23Z <masak> tell szabgab , as for the TODO question in szabgab.com/talks/perl6/perl6-classx.html -- the ultimate reason is that the normal accessors return an lvalue (that you can assign to). 'course, the accessors could be extended to take an optional argument, but it makes sense to encourage the assignment syntax, which is clearer anyway.
[Sorry, some messages were elided and lost...]
szabgab phenny, thanks 15:07
masak phenny: you said you'd nopaste them!
15:07 nihiliad joined
bbkr std: my %h = {my => "x"}; 15:07
p6eval std 30563: OUTPUT«ok 00:01 113m␤»
szabgab masak, thank you too :-)
masak you're welcome. 15:08
hoping to see an updated course some day! it has really good material; just needs updating. 15:09
rakudo: class A { my method foo() { say "OH HAI" }; method bar() { foo(self) } }; A.new.bar 15:10
p6eval rakudo 1eef08: OUTPUT«OH HAI␤»
masak TimToady: S12: 'A subroutine call considers only visible subroutines (including submethods) of that name.' how does that square with the above?
15:11 lrnperl6 left, kst left
TimToady these days, should say, "including methods and submethods declared with a lexical name 15:11
masak ah, ok. 15:12
moritz_ btw TimToady++ # multi method/proto unification
I just wonder if it should be s/empty body/stub/
so that proto foo($x) { ... } does the magic re-dispatch, not foo($x) { } 15:13
TimToady I considered that, but it just didn't feel quite right for some reason
moritz_ s:2nd/<before foo>/proto /
TimToady has to do with people thinking the ... means "nextsame", and then trying to use "..." to mean that 15:14
... is not a placeholder in that sense
moritz_ ... is "explicitly empty" in my mind
TimToady it's not a strong reason 15:15
moritz_ and it's not a strong sentiment on my part eitehr
15:15 JimmyZ left
moritz_ I just read it and though "huh, ... would fit better" 15:15
TimToady either way could be made to work, or maybe even both 15:16
moritz_ are proto tokens ever executed?
TimToady being a proto kinda turns the whole body into a sort of macro anyway
15:16 kst joined
moritz_ proto token { } # empty regex disallowed 15:17
proto token { <...> } # feels much better
PerlJam indeed
15:17 pmurias left
TimToady I now think <...> is kinda redundant with {...} 15:17
proto token { {...} }
moritz_ <...> feels more declarative 15:18
{...} is procedural
15:18 jonrafkind joined
TimToady one can view the nextsame stub both ways too, I guess 15:19
either as a "do this" or as an "insert here"
and how do we get all the nextsame/callsame variants into regexese
<nextsame> and <callsame> I guess 15:20
moritz_ somehow proto tokens feel quite different than multi subs
they execute all candidates in parallel
TimToady not really 15:21
masak regex matching executes quite differently than ordinary code.
TimToady they run a DFA as part of trying to decide which one to *really* run
moritz_ well, notionally at least
bbkr Pair constructor stringifies key. so @t=>"x" is the same as ~@t=>"x" - key is Str type. Should it work on Junctions as well? (1|2|3)=>"x" crashes in Rakudo (beside fact, that it's just stupid code)
moritz_ bbkr: I'd expect it to autothread, and return a junction of Pair objects 15:22
TimToady bbkr: pair constructors aren't stringy
or aren't *supposed* to be
moritz_ rakudo: say (1 => 3).key.WHAT
p6eval rakudo 1eef08: OUTPUT«Int()␤»
moritz_ and they aren't
just hashes stringify
the keys 15:23
bbkr rakudo: my %h = {(1|2|3) => "x"};
p6eval rakudo 1eef08: OUTPUT«Odd number of elements found where hash expected␤current instr.: '&die' pc 17868 (src/builtins/Junction.pir:393)␤»
15:23 masak left
bbkr i meant Pair constructor in hashes, like above 15:23
moritz_ the error comes from the hash assingment, not from the pair composition
rakudo: say ((1|2|3) => "x").perl
p6eval rakudo 1eef08: OUTPUT«any(1 => "x", 2 => "x", 3 => "x")␤»
moritz_ that looks right so far
bbkr I get it now. thanks! 15:24
moritz_ and I really don't know what should happen if you assign a junction to a hash
bbkr me neither :) i'm just trying weird but simple code snippets. 15:26
in this case it's not a bug. my %h = {(1|2|3) => "x"}; is the same as my %h = {any(1 => "x", 2 => "x", 3 => "x")}; and it's indeed odd number of elements
tedv how do you remove the any() call? Ie. convert a junction to the union of all potential values 15:28
moritz_ unless juctions are supposed to flatten out in list context, and a list of pairs is fine to assingn to a hash
tedv: you don't. If you want to, it tells you that you should really be using Sets (which are NYI, of course...) 15:29
15:29 PacoLinux left
PerlJam bbkr: my %h{Junction} = { (1|2|3) => "x" }; # should succeed (eventually :) 15:30
15:30 PacoLinux joined
tedv That's a rather cavalier attitude 15:30
The operation is mathematically well defined
moritz_ maybe I misunderstood the question
tedv who is to say that I won't receive a junction from some library function and want to convert it to a set?
moritz_ do you want to flatten a nested junction?
tedv yeah
moritz_ or do you want to extract the eigenstates?
PerlJam tedv: you would do so explicitly though, not implicitly. 15:31
tedv I mean how would you convert any(a,b,c) to the set of (a,b,c)
[particle] any(a,b,c).Set ??
moritz_ there used to be a .eigenstates method
tedv yeah, I'm just curious what the explicit operator or function is called
moritz_ that was was removed
for the reason I stated above
and because of much bikeshedding on p6l :( 15:32
tedv because perl knows better than me why I'm writing perl code, got it
TimToady any() is not a set, it's a set of sets
15:32 Ross joined
PerlJam eigenstates is still in S32-setting-library/Containers.pod 15:32
moritz_ PerlJam: as a private method, no?
TimToady only all() is a reasonable approximation of a set
PerlJam moritz_: oh, aye.
moritz_ anyway, you can write a utilty
TimToady and use of junctions as sets turned out to be an "attractive nuisance" 15:33
15:33 rgrau_ joined
tedv Not disagreeing with that 15:33
It's just that perl traditionally gives you enough rope to hang yourself, so why not that piece of rope? :)
moritz_ sub eigenstates(Mu $x) { my @e; -> Any $y {@e.push: $y}.($x); @e)
TimToady because we were already getting tired of throwing ropes to people
moritz_ (might be a race condition one junctions evaluate in parallel) 15:34
tedv mortiz_: The order of @e won't be guaranteed I think
er yeah roughly what you said
moritz_ tedv: sure; junctions don't have an intrinsic order
tedv which brings us back to sets instead of arrays 15:35
15:35 szabgab left
[particle] on bags, for repeated values 15:35
*or
tedv yeah
TimToady those are all specced, but not very implemented
moritz_ should sets autothread like junctions? 15:36
tedv So what are some practical uses for junctions? When I heard them described, they seemed like a genuinely useful feature but I had trouble thinking of how I'd use it in everyday coding.
moritz_ tedv: mostly as matchers
tedv and just assumed it was because my mind didn't have practice thinking in that manner 15:37
TimToady most common: if $a == 1 | 2 | 3
tedv you mean more general form regular expression?
moritz_ tedv: I mean things like "grep"
tedv ie. operates on any data type, not just strings
moritz_ rakudo: say (1..10).grep(3|6)
p6eval rakudo 1eef08: OUTPUT«36␤»
moritz_ rakudo: say (1..10).grep(3|6).join(', ')
p6eval rakudo 1eef08: OUTPUT«3, 6␤»
tedv rakudo: say (1..10).grep(3|6|25) 15:38
p6eval rakudo 1eef08: OUTPUT«36␤»
moritz_ ok, it's a fairly contrived example
tedv the more interesting case
so that's useful for set intersection
[particle] rakudo: say (1..10).grep(6|3|6..8).join(', ')
p6eval rakudo 1eef08: OUTPUT«3, 4, 5, 6, 7, 8␤»
moritz_ rakudo: say (1..10).grep({$_ < 8}|{$_ % 2}).join(', ')
p6eval rakudo 1eef08: OUTPUT«1, 2, 3, 4, 5, 6, 7, 9␤»
[particle] *cough*
moritz_ rakudo: say (1..10).grep({$_ < 8}&{$_ % 2}).join(', ')
p6eval rakudo 1eef08: OUTPUT«1, 3, 5, 7␤»
[particle] i ran afoul of precedence
moritz_ look ma, junctions of code blocks
[particle] rakudo: say (1..10).grep(6|3|(6..8)).join(', ') 15:39
p6eval rakudo 1eef08: OUTPUT«3, 6, 7, 8␤»
TimToady rakudo: say ~(1..10).grep: 2 | /0/
p6eval rakudo 1eef08: OUTPUT«2 10␤»
15:40 molaf joined
TimToady rakudo: say $*IN ~~ :r & :f 15:40
p6eval rakudo 1eef08: OUTPUT«Method 'r' not found for invocant of class 'IO'␤current instr.: 'perl6;Enum;ACCEPTS' pc 508921 (src/gen/core.pir:78424)␤»
TimToady aww
rakudo: say $*IN ~~ :e & :f
p6eval rakudo 1eef08: OUTPUT«Method 'e' not found for invocant of class 'IO'␤current instr.: 'perl6;Enum;ACCEPTS' pc 508921 (src/gen/core.pir:78424)␤»
TimToady rakudo: say "/etc/passwd" ~~ :e & :f 15:41
p6eval rakudo 1eef08: OUTPUT«1␤»
[particle] rakudo: IO.^methods.say
p6eval rakudo 1eef08:
..OUTPUT«eofinslinessayprintgetgetcprintftcloseslurpfmtbytesacosecsinhNumericasinhcosechtrimasectrim-leadingchracotanhfloorsechasechrandtruncateroundsqrtasinsplitmatchexpcoshacoshwordssigncotanatan2lcfirstseccislogsamecasesprintfcapitalizesinflipp5choplcsubstacosechabsacotanchompindexceilin…
TimToady which isn't quite right
moritz_ yes, it's still auld style in rakudo
TimToady but illustrates the junctional matching of file tests
as for autothreading, if a set in list context returns its elements, then any of the other autothreaders can be used with sets 15:43
any(%set) == 42
dalek meta: r240 | diakopter++ | trunk/Sprixel/src/Perlesque/Declarations.cs:
[perlesque] fix ConstantValue
 15:44
TimToady or, with the conjectural each(), Set(each(%set) + 1)
15:44 molaf left
TimToady but making sets autothread natively might or might not make sense 15:45
at one time there was speculation that ranges would autothread 15:46
diakopter spinclad: hi
TimToady maybe autothreading is a mixin of some sort, someday 15:47
except a mixin should be knocking the type of the object around
and autothreading means putting the type outside of Any
diakopter perlesque: my List[int] $l; $l.Add(2); say($l); # yay, it does the right thing now (NRE) 15:49
p6eval perlesque: OUTPUT«␤Unhandled Exception: System.Reflection.TargetInvocationException: Exception has been thrown by the target of an invocation. ---> System.NullReferenceException: Object reference not set to an instance of an object␤ at TopLevelFrame.Exec () [0x00000] in <filename unknown>:0 ␤ at
..F…
dalek meta: r241 | diakopter++ | trunk/Sprixel/ (3 files):
[perlesque] more fixes
meta: r242 | diakopter++ | trunk/Sprixel/src/Perlesque/Declarations.cs:
[perlesque] remove debug
15:58 justatheory joined 16:02 PerlJam left, PerlJam joined 16:05 kst left 16:11 kst joined 16:12 pmurias joined
pmurias diakopter: hi 16:13
16:15 cdarroch joined, cdarroch left, cdarroch joined 16:18 [mark] left 16:19 pmurias left 16:20 jonasbn left 16:21 pmurias joined
pmurias ruoso: hi 16:24
diakopter hiu
pmurias diakopter: did you get -Bperlesque to work?
fixing mildew... 16:26
pugssvn r30564 | pmurias++ | [mildew] add dependencies to dist.init 16:28
r30565 | pmurias++ | [mildew] update to work with current perlesque
16:31 mikehh left
ruoso pmurias, hi... 16:31
pmurias if Mildew::Backend::Perlesque requires a path to Perlesque to work how can i feed it early enough so that it can run tests? 16:33
16:34 stephenlb joined
pmurias ruoso: and one problem with loading VAST nodes using Module::Pluggable is if a VAST node is removed and the user install a new Mildew version on top of an old one 16:34
16:35 mikehh joined 16:38 envi^home left 16:39 lichtkind joined
lichtkind colomon: .triangle is alias to [/ ] ? 16:40
moritz_ that doesn't look right 16:41
triangle is [\op]
and rakudo current implements that as reduce(..., :triangle) or so 16:43
diakopter pmurias: I guess just assume it's in the PATH 16:45
pmurias: at least assume that mono is in the PATH 16:46
so maybe perlesque.exe's path is in PATH_PERLESQUE
I dunno.
16:47 isBEKaml joined
lichtkind moritz_: yes i mistyped it, i meant [\ op] 16:50
moritz_ and no space allowed :-)
16:50 molaf joined
lichtkind grtsch 16:51
pmurias diakopter: is it possible to install mono generated executables in the path?
diakopter actually yes in linux you can register .exe as binary executable 16:52
but it's a lot slower 16:53
than a shell script wrapper
but of course you lose most the shell args
16:53 BrowserUk left
pmurias diakopter: when do i loose the shell args? 16:53
diakopter I don't know.
pmurias in a shell script wrapper i get to keep them 16:54
diakopter I guess I don't know how to pass-thru "all of my args" to another invocation
huf via shell? "$@"
diakopter ok.
oh. :) /me ignorant in these ways
huf dont worry, the shell's a dick anyway 16:55
diakopter in cmd.exe it's %*
pmurias so maybe assuming a perlesque in the shell is sensible? 16:56
s/shell/PATH/
diakopter sure, if there's a perlesque that's a #!/bin/sh wrapper 16:57
for mono /path/to/perlesque.exe
mono /path/to/perlesque.exe "$@"
17:02 meppl left, BrowserUk joined 17:03 jjore left
ruoso pmurias, hmmm... each VAST node could refuse to load if Mildew version is not X 17:03
17:04 patspam joined 17:05 sorear left, sorear joined 17:07 pmurias left 17:11 jjore joined, buubot left 17:12 buubot joined, kst left 17:13 kst joined 17:14 ktne left 17:15 rgrau_ left 17:17 jjore left
sorear phenny: tell masak that "may get lost" means "will get lost" 17:21
phenny sorear: I'll pass that on when masak is around.
17:29 dakkar left
colomon lichtkind: moritz_++ has it right. 17:30
isBEKaml moritz_: There's a small thing in proto's index.tmpl that appears to be a typo. causes justified paragraph to be displayed in hanging text. Was that intended?
lichtkind colomon: thanks
17:31 patspam left
isBEKaml moritz_: #about p... :) 17:31
17:32 patspam joined 17:33 patspam left
jnthn o/ folks 17:38
jnthn is back from the Icelandic countryside
isBEKaml jnthn: yo!
jnthn No internets out there. :-) 17:39
isBEKaml must have been quite a relief... ;) 17:40
isBEKaml hides...
jnthn I enjoyed it. :-) 17:41
For a few days.
:-)
Just hiked, camped, relaxed. :-) 17:42
colomon :) 17:43
isBEKaml :)
TimToady :)
diakopter : )
TimToady why the long face?
isBEKaml maybe he has a really long face.. ;)
he smiles well.. :D
diakopter the better to smell you with 17:51
17:52 sundar_afk joined, rv2733 left 17:53 sundar_afk is now known as sundar 17:57 jonasbn joined 17:58 jonasbn left
tedv newbie question: if I run "make spectest" from a rakudo checkout, will that automatically load the latest version of the pugs unit tests, or do I need to do that myself? 17:58
jnthn It automaticky does an svn up (or an svn co the first time), so you get the latest. 18:01
*automatically
ga
jnthn needs to learn to type again
pugssvn r30566 | lwall++ | [S03] small clarifications to usage of brackets around infixes
TimToady blame it on the ash in the keyboard
jnthn ;-) 18:02
I actually have seen The Volcano.
I still can't pronounce its name, however.
TimToady tedv: were you the one who was asking about whitespace in metaoperators the other day? I'd write those as >>[op]>> instead. 18:03
rakudo: say ~( (1,2,3) >>[=>]>> 42 ) 18:04
p6eval rakudo 1eef08: OUTPUT«Confused at line 11, near "say ~( (1,"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
18:04 cognominal left
TimToady well, I would if rakudo recognized it :) 18:04
std: say ~( (1,2,3) >>[=>]>> 42 )
p6eval std 30565: OUTPUT«ok 00:01 113m␤»
colomon rakudo: say ~( (1,2,3) »=>» 42 ) 18:05
p6eval rakudo 1eef08: OUTPUT«1 42 2 42 3 42␤»
colomon rakudo: say ( (1,2,3) »=>» 42 ).perl
p6eval rakudo 1eef08: OUTPUT«[1 => 42, 2 => 42, 3 => 42]␤»
TimToady btw, shortest way to get a bunch of true hash values is likely %hash{@keys}»++
tedv TimToady: Yes, I was, primarily because of how >>=>>> loops
looks*
diakopter rakudo: say ( (1,2,3) >>=>>> 42 ).perl
p6eval rakudo 1eef08: OUTPUT«[1 => 42, 2 => 42, 3 => 42]␤»
tedv but that's more because I'm a dirty American without chevrons on my keyboard. 18:06
colomon TimToady: That's "true" but not "True", right?
tedv I suppose >>[=>]>> is clearer
diakopter std: say ( (1,2,3) >>=>>> 42 ).perl
p6eval std 30565: OUTPUT«ok 00:01 111m␤»
diakopter o_O
colomon rakudo: say ( (1,2,3) >>=>>> 42 ).perl
p6eval rakudo 1eef08: OUTPUT«[1 => 42, 2 => 42, 3 => 42]␤»
TimToady std: say ~( (1,2,3) >>>>> 42 )
p6eval std 30565: OUTPUT«===SORRY!===␤Missing << or >> at /tmp/zudoOoseH6 line 1:␤------> say ~( (1,2,3) >>>>>⏏ 42 )␤ expecting infix_circumfix_meta_operator__S_267LtLt_GtGt␤Parse failed␤FAILED 00:01 111m␤»
TimToady hmm
std: say ~( (1,2,3) >>[>]>> 42 )
p6eval std 30565: OUTPUT«ok 00:01 111m␤»
TimToady rakudo: say ~( (1,2,3) >>[>]>> 42 ) 18:07
p6eval rakudo 1eef08: OUTPUT«Confused at line 11, near "say ~( (1,"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
18:07 jjore joined
jnthn Didn't implement that form just yet. 18:07
TimToady oh, yeah
jnthn It's easy enough, I suspect.
18:07 patspam joined
diakopter std: say ~( (1,2,3) >>[=<]>> 42 ) 18:07
TimToady just recursion in the infixish rule
tedv is >>>>> supposed to work? I wouldn't be surprised if that just happened to be indistiguishable to the parser for some reason
p6eval std 30565: OUTPUT«===SORRY!===␤Unable to parse bracketed infix at /tmp/V6iCrxUsDj line 1:␤------> say ~( (1,2,3) >>[⏏=<]>> 42 )␤Couldn't find final ']'; gave up at /tmp/V6iCrxUsDj line 1:␤------> say ~( (1,2,3) >>[=⏏<]>> 42 )␤Parse
..failed␤FAILE…
colomon rakudo: say ( (1,2,3) >><>> 42 ).perl
p6eval rakudo 1eef08: OUTPUT«[Bool::True, Bool::True, Bool::True]␤»
tedv can you unspace it too? 18:08
TimToady not currently, I don't think
tedv rakudo: say ( (1,2,3) >>\ >\ >> 42 ).perl
p6eval rakudo 1eef08: OUTPUT«Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 11␤current instr.: 'perl6;Regex;Cursor;FAILGOAL' pc 1696 (ext/nqp-rx/src/stage0/Regex-s0.pir:932)␤»
TimToady not sure it really helps readibility, over [op]
tedv yeah
clearly not in that case
TimToady [op] makes a nice "pill" 18:09
tedv I got the impression that unspace was most useful when you have a long list of foo.bar foo.baz, and want everything to line up nicely
jnthn TimToady: Aye, I or @other will no doubt get to it in due course. :-)
tedv er foo.bar and longfoobar.bar
jnthn is feeling more healthy and relaxed than he has for ages, and will be back and hacking in a couple of days. :-) 18:10
18:10 ShaneC1 joined, ShaneC1 left
TimToady since unspace has to be checked for specifically in the grammar, I tend not to put it in any old where 18:10
but only where there's demand
and no better way to do it
tedv I think it's the type of feature that will become more expected with language age, but won't see much use in the beginning. 18:11
TimToady its primary use is to put whitespace before a postfix, really
tedv what I like about perl 6 is the sheer amount of operators that encode algorithms that are simple to describe in natural language
like "add the numbers in this list"
[+] @x
Except if you're programming in C your whole life, you probably thought something more like... 18:12
int total = 0; for (int i = ......
and it takes time for people to think using new operators
unspace kind of seems the same to me
Juerd tedv: I think most people think sum(@x)
diakopter TimToady: for the strongly-typed (including user classes) subset of Perl 6, I need some way of representing up-casting and down-casting 18:13
tedv okay well I'm sure I can find a more complicated example that doesn't have a natural foo() wrapping function
any kind of cartesian list joining for example 18:14
diakopter TimToady: .. just casting, I mean. obviously the type-checker/compiler knows whether it's going to down-cast or up-cast
Juerd tedv: [<=]
colomon rakudo: say ([=>] 1..10).perl 18:15
p6eval rakudo 1eef08: OUTPUT«1 => 2 => 3 => 4 => 5 => 6 => 7 => 8 => 9 => 10␤»
TimToady diakopter: what's the problem with MyType($thing)?
diakopter oh
forgot about that
colomon What does MyType($thing) do? 18:16
diakopter that's certainly better than (MyType)$thing
TimToady rakudo: say Str(42).WHAT
p6eval rakudo 1eef08: OUTPUT«Could not find sub &Str␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
TimToady hmm
rakudo: say 42.Str.WHAT
p6eval rakudo 1eef08: OUTPUT«Str()␤»
TimToady same as that, in theory
diakopter std: say str(42).WHAT 18:17
p6eval std 30565: OUTPUT«===SORRY!===␤Undeclared routine:␤ 'str' used at line 1␤Check failed␤FAILED 00:01 110m␤»
diakopter oops
std: say buf8(42).WHAT
p6eval std 30565: OUTPUT«ok 00:01 110m␤»
diakopter std: say int128(42).WHAT
p6eval std 30565: OUTPUT«===SORRY!===␤Undeclared routine:␤ 'int128' used at line 1␤Check failed␤FAILED 00:01 110m␤»
diakopter microsoft has 128-bit windows running on something from Intel, supposedly 18:18
TimToady std: say int64(42).WHAT
p6eval std 30565: OUTPUT«ok 00:01 110m␤»
18:18 rv2733 joined 18:22 kst left, meppl joined
pugssvn r30567 | lwall++ | [S09] speculate 128-bit types and see if we get away with it 18:22
18:23 kst joined
pugssvn r30568 | lwall++ | [STD] use same rules for interpolating @foo[] into regex 18:25
r30568 | [CORE] add some 128-bit types
diakopter hah 18:27
you don't think Unicode will go 128-bit? :P 18:28
mathw Depends... how close are they to using up the 64 bit space? :) 18:29
18:29 molaf left
diakopter
.oO( I think everyone on the planet needs each's own glyph )Oo.
 18:29
(kidding) 18:30
PerlJam diakopter: you kid, but one day it will be so. 18:31
18:32 k23z__ left, jjore left
Juerd diakopter: Every grain of sand can has its own with 128 bits I guess :) 18:34
18:35 jjore joined
TimToady me tries to picture an application that maps between glyphs and IPv6 addrs 18:37
Su-Shee well you'll just compare ::::::::: and everything else is easily filled by masks. ;) 18:38
slavik1 Microsoft can write code for ARM ... they wrote something for 128bit?
diakopter nt was built for arm & ppc, yeah 18:39
18:43 patrickas joined 18:46 patrickas left
diakopter hrm. us stock markets down 11%. 18:49
colomon ouch 18:50
diakopter er, sorry, down only 8.5%. but it's now recovered to down 6% 18:51
18:52 patrickas joined
colomon that's still plenty bad. 18:54
moritz_ something's wrong with stock markets in general 18:55
if within one day the economy loses 8% of its value
diakopter google finance crashed
colomon CNN thinks it is Greece.
moritz_ or the perceived value
18:56 cognominal joined
diakopter price := perceived value 18:59
moritz_ even then 19:01
TimToady part of the valuation is betting on whether a fourth catastrophe would compound the current three 19:06
moritz_ doesn't know which counting scheme produces three current catastrophies
TimToady I'm counting Greece, Iceland, and the Gulf of Mexico 19:07
[particle] you forgot PHP
moritz_ 1) situation in Iraq 2) Situation in Afghanistan 3) general financial 4) Gulf of Mexico 5) Greece
Su-Shee welcome to the world of 2500 years of educated crystal ball polishing :) 19:08
TimToady 1 & 2 are long-term situations already fed into the equation
19:09 tri1 joined
moritz_ in terms of human casualties 1 & 2 are much more severe than Greece by far 19:10
TimToady I'm referring more to things that were unexpected 3 months ago
moritz_ ok :-)
TimToady well, obviously Greece has been festering for longer than that too 19:11
moritz_ phenny: tell masak I'm inclined to reject maspalio++'s patch that changes the proto web script to Web::Scraper. Reason: the documentation of Web::Scraper is so incomplete that I don't think I could maintain that script. Thoughts?
phenny moritz_: I'll pass that on when masak is around.
Su-Shee moritz_: in terms of politics, they're the usal run-of-the-mill middle east stuff. remember that there already were like 20 years of sowjets in afghanistan.
[particle] greece has been festering, and ireland was on the brink but is recovering. portugal and spain aren't so lucky 19:13
patrickas moritz_ too bad about Web::Scraper, parsing HTML through using a regex is unfortunately going to stay there for longer than I would have hoped!
TimToady anyway, to bring this back on topic, we need to raise the perceived value of Perl 6 :) 19:18
19:18 Ross left
[particle] a well-marketed and community-supported release of a perl 6 compiler will certainly raise the perceived value 19:22
Su-Shee just had the word "brand" on her keyboard. :) 19:23
pugssvn r30569 | lwall++ | [sigspace.t] don't use [abc] to mean 'abc' since it looks like old-school character class
r30570 | lwall++ | [repeat.t] don't use [abc] to mean 'abc' since it looks like old-school character class 19:26
r30571 | lwall++ | [basic.t] same thing, change [ed] to 'ed' 19:31
19:32 cognominal left
moritz_ patrickas: I know, but I can at least maintain them 19:35
patrickas: if others mostly agree that moving to Web::Scrape is a good thing, then I'm fine with it 19:36
19:37 kst left
pugssvn r30572 | lwall++ | [rx.t] more appearance-of-cclass warning suppression 19:37
patrickas moritz_: Unfortunately I have no idea about Web::Scrape
19:38 kst joined 19:42 Ross^ joined
pugssvn r30573 | lwall++ | [99problems-61-to-70.t] move sub count outside of its block so it's visible to later blocks 19:43
19:43 ash_ joined 19:45 meppl left
pugssvn r30574 | lwall++ | [99problems-21-30.t] move sub combination outside of its block so it's visible to later blocks 19:46
19:46 ash_ left 19:53 Ross^ left 19:54 Ross^ joined 19:55 ash_ joined
pugssvn r30575 | moritz++ | [t/spec] fudge mixin.t for rakudo 19:55
r30576 | moritz++ | [t/spec] fudge BUILD.t for rakudo 20:00
dalek kudo: 5713af2 | moritz++ | t/spectest.data:
run two more test files
colomon i\o/ 20:01
what's the passing count now?
ash_ rakudo: use MONKEY_TYPING; augment grammar Perl6::Grammar { };
moritz_ >before :-)
p6eval rakudo 1eef08: OUTPUT«Method 'compose' not found for invocant of class ''␤current instr.: 'perl6;Perl6;Grammar;!class_init_11' pc 352 (EVAL_1:27504175)␤»
moritz_ ash_: Perl6::Grammar is an nqp-level class, and not built with the "right" meta model 20:02
ash_ I figured as much
still shouldn't die like that... i don't think
moritz_ right 20:03
ash_ rakudo: slang Foo { }; # also acts funny 20:04
p6eval rakudo 1eef08: OUTPUT«Could not find sub &Foo␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
20:04 Ross^ is now known as Ross
moritz_ it just parses both slang and Foo as sub calls 20:05
and dies at the one that's executed earlier
ash_ ah, i didn't realize it was parsing slang as a sub call
i figured Foo was, but not slang
moritz_ any bareword that's not recognized somehow is parsed as a listop/subcall
ash_ I was trying to see if any of docs/metamodel.pod's examples worked 20:08
I also just noticed that in one of the examples in the file someone wrote a '.' for string concat, i think
github.com/rakudo/rakudo/raw/master...amodel.pod line 296 20:09
moritz_ aye 20:10
20:10 cognominal joined 20:11 lisppaste3 left, lisppaste3 joined
TimToady std: Str => $name . ⏏"()" 20:12
p6eval std 30575: OUTPUT«===SORRY!===␤Variable $name is not predeclared at /tmp/eDEklNOq3f line 1:␤------> Str => $name⏏ . ⏏"()"␤Unsupported use of . to concatenate strings; in Perl 6 please use ~ at /tmp/eDEklNOq3f line 1:␤------> Str => $name .
..⏏⏏"()"[…
TimToady std: Str => $name . "()"
p6eval std 30575: OUTPUT«===SORRY!===␤Variable $name is not predeclared at /tmp/iqEIeV0zTz line 1:␤------> Str => $name⏏ . "()"␤Unsupported use of . to concatenate strings; in Perl 6 please use ~ at /tmp/iqEIeV0zTz line 1:␤------> Str => $name .
..⏏"()"…
dalek kudo: badc61d | moritz++ | docs/metamodel.pod:
[docs] perl5o found by ash_++
ash_ i know you can use $foo.?bar(); to try to call bar if it exists, is there a way to do that with a sub too? 20:13
for instance, module Baz { our sub bar { ... } }; Baz::?bar(); or something? 20:14
moritz_ try { bar() } :-)
ash_ what if i don't know which package its in? try { my $a = 'PackageName'; $a::bar() } ? 20:15
20:15 kst left
moritz_ $::($a)::bar() iirc 20:15
TimToady subs are not called from packages 20:16
they are in lexical scopes
moritz_ unless they are 'our', no?
TimToady then it's the lexical alias that is called
the dispatcher never looks in the package these days
unless you specifically tell it to
20:17 clintongormley left
ash_ rakudo: module Foo { our sub bar { say 'bar' } }; my $a = 'Foo'; $::($a)::bar(); # so that is not correct? 20:17
p6eval rakudo 1eef08: OUTPUT«Indirect name lookups not yet implemented at line 11, near "::bar(); #"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
20:17 tri1 left
TimToady that is explicitly telling it to look in a package 20:17
so that's okay
ash_ okay 20:18
TimToady I was complaining about try { bar() } which always knows at compile time
ash_ oh, got ya
I am just trying to see what else might be missing if I wanted to make a custom metaclass to support my own phasers 20:19
TimToady we'd like our candidate lists to be static by default
as soon as you put a mutable package there, the lists are no longer static
20:20 snearch joined
TimToady there are various ways to force the lookup to wait until run time, and those can run as slow as they like. :) 20:21
so I don't care if, say, ::($name)() looks in the current package, if it can't find it lexically 20:22
we'd just like the default to be highly optimizable 20:23
ash_ that makes sense
make the common case fast
moritz_ colomon: current test case: 31871
*count 20:24
20:24 kst joined
colomon moritz_: damn, do I wish I didn't have pressing $work stuff, so I could go after those next 129.... 20:24
patrickas how come rakudo.de/progress.png seems to be really not up to date! 20:25
20:28 Su-Shee left
Tene patrickas: it's updated by hand, and the person who does that has been otherwise busy. 20:28
20:33 SmokeMachine left
colomon moritz_: interestingly, got 31876 passes here. 20:33
20:34 chitragupt left
ash_ which any glaring issues with gist.github.com/392673 as a possible functionality of COMPOSED and INHERITED ? 20:37
20:40 [Coke] left 20:41 patrickas left 20:44 cdarroch left, snearch left 20:46 [Coke] joined 20:52 SmokeMachine joined 21:01 orafu left, orafu joined 21:15 Trashlord left 21:17 wknight8111 joined, kst left
sjohnson 2hi dudes 21:17
lichtkind hai 21:19
sjohnson ... perl. is there anything it can't do?
21:20 Sumeragi joined 21:23 Ross left 21:25 kst joined 21:28 astinus left, chitragupt joined 21:29 astinus joined, KyleHa joined
KyleHa rakudo: if Nil.defined { say "yes Nil.defined" } else { say 'no Nil.defined' }; if NIl.defined ~~ True { say 'yes Nil.defined ~~ True' } else { say 'no Nil.defined ~~ True' } 21:31
p6eval rakudo 1eef08: OUTPUT«no Nil.defined␤Could not find sub &NIl␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
KyleHa f Nil.defined { say "yes Nil.defined" } else { say 'no Nil.defined' }; if Nil.defined ~~ True { say 'yes Nil.defined ~~ True' } else { say 'no Nil.defined ~~ True' }
rakudo: if Nil.defined { say "yes Nil.defined" } else { say 'no Nil.defined' }; if Nil.defined ~~ True { say 'yes Nil.defined ~~ True' } else { say 'no Nil.defined ~~ True' }
p6eval rakudo 1eef08: OUTPUT«no Nil.defined␤yes Nil.defined ~~ True␤»
KyleHa What's going on here? 21:32
I'd expect those two to agree.
21:33 sundar left 21:34 cdarroch joined, cdarroch left, cdarroch joined, Sumeragi left
sjohnson i think there are two types of perl people. ones who match with /string/, and ones who use m/string/. i am the latter 21:36
you can tell a lot about a man (or woman) by which camp they belong to in this case 21:37
21:37 Guest13256 left
Tene what about those that match with m{} ? 21:38
sjohnson i was JUST going to say 21:39
that
using m{} with paths is the WAY to perl happiness
Juerd I prefer m[] over m{} 21:42
And I use s[][], but s[]{}e 21:43
sjohnson hmm, maybe i'll start using the square brackets 21:48
Juerd vim doesn't highlight it well but I've gotten used to that. 21:49
Delimiting strings or regexes with {} feels like Tcl to me :) 21:50
I don't like tcl
21:51 KyleHa left
arnsholt Juerd: What's the problem with vim's highlighting of brackets? 21:53
Juerd arnsholt: It highlights the RHS of s[][] as code instead of string contents 21:55
That is, sometimes. Most of the time it just works
When it doesn't, scrolling forth and back usually fixes it 21:56
arnsholt I see
21:56 ruoso left
arnsholt Confuses it with array deref I suppose 21:56
21:56 isBEKaml left
Juerd Oh, and when you use non-equal delimiter pairs it doesn't highlight the second one as a string 21:57
But don't fix that. I've come to rely on it, for s[]{}e :)
21:57 iblechbot left
diakopter std: sub s() { }; s{4}{4} 21:57
p6eval std 30576: OUTPUT«===SORRY!===␤Unexpected block in infix position (two terms in a row, or previous statement missing semicolon?) at /tmp/iVUe8lrlIY line 1:␤------> sub s() { }; s{4}⏏{4}␤ expecting any of:␤ bracketed infix␤ infix or meta-infix␤
..quantifier␤Parse…
arnsholt Juerd: Nice one. Good exploitation of a bug and turning it into a feature =) 21:58
Juerd juerd.nl/i/1b2f35cb05486cf1beaab6ce2e648708.png # example 22:00
Oh, and it doesn't understand qq[]
I almost forgot that one
22:02 SmokeMachine left
diakopter std: s{4}{4} 22:04
p6eval std 30576: OUTPUT«===SORRY!===␤Unexpected block in infix position (two terms in a row, or previous statement missing semicolon?) at /tmp/3IdmycEol5 line 1:␤------> s{4}⏏{4}␤ expecting any of:␤ bracketed infix␤ infix or meta-infix␤ quantifier␤Parse
..failed␤FAILE…
diakopter std: s[4]{4}
p6eval std 30576: OUTPUT«===SORRY!===␤Unexpected block in infix position (two terms in a row, or previous statement missing semicolon?) at /tmp/Ph9bfExBaW line 1:␤------> s[4]⏏{4}␤ expecting any of:␤ bracketed infix␤ infix or meta-infix␤ quantifier␤Parse
..failed␤FAILE…
diakopter std: s[4]{4}e
p6eval std 30576: OUTPUT«===SORRY!===␤Unexpected block in infix position (two terms in a row, or previous statement missing semicolon?) at /tmp/ArZSlpVD7E line 1:␤------> s[4]⏏{4}e␤ expecting any of:␤ bracketed infix␤ infix or meta-infix␤
..quantifier␤Parse failed␤FAIL…
22:05 [particle] left 22:11 hercynium joined
sjohnson Juerd: you probably noticed some hilight problems which is to be expected... with v5.10.0 stuff 22:13
Juerd 5.10.0 of what? 22:15
22:17 alester left 22:27 kst left, kst joined
sjohnson Juerd: given / when || say call in perl 5.10.0 22:35
Juerd I don't use those yet 22:36
aesop use what?
ash_ perl 5.12 is out, I rather like some of its changes, small things, like use v5.12; also does use strict; 22:39
snarkyboojum rakudo: say 'Perl 6'.pred ~ '.10+ features' 22:40
p6eval rakudo 1eef08: OUTPUT«Perl 5.10+ features␤»
ash_ they added the ... 'yada yada' op in 5.12 from perl6 22:42
Juerd ash_: I'm not too happy about the optional bugfix.
A bugfix that can be enabled lexically 22:43
Which means that if you do enable it, it still doesn't work correctly in modules.
ash_ I am not sure what you mean, which optional bugfix? 22:44
Juerd ash_: use feathure 'unicode_strings' 22:45
s/feathure/feature/
22:45 ethel left 22:46 snarkyboojum left
ash_ ah, the unicode updates 22:46
Juerd If I understand correctly, the regex charclasses are changed regardless of the pragma 22:47
But I haven't tried 5.12 yet
22:56 nihiliad left 23:04 jhuni joined 23:13 chitragupt left 23:14 chitragupt joined 23:20 chitragupt left 23:21 chitragupt joined 23:23 kst left 23:36 kst joined, chitragupt left 23:37 chitragupt joined 23:47 TiMBuS joined, [particle] joined
lichtkind i currently do :) 23:47
sorear I'd like to see an editor with STD-highlighting
lichtkind you speak to me? 23:48
sorear (Does anyone know what's wrong with STD_highlight? It crashes when fed STD.pm6, yet STD.pm6 can parse itself)
23:49 kst left 23:50 kst joined, ruoso joined
diakopter sorear: I think STD_highlight got bitrotten 23:51
23:58 lestrrat is now known as lest_away
sjohnson rakudo: say " cow ".trim-trailing 23:59
p6eval rakudo 1eef08: OUTPUT« cow␤»
sjohnson rakudo: say " cow ".trim-leading
p6eval rakudo 1eef08: OUTPUT«cow ␤»
sjohnson rakudo: say (" cow ", " pig ", " chicken").trim-leading
p6eval rakudo 1eef08: OUTPUT«cow pig chicken␤»

Powered by Cro and the Raku® Programming Language.

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

Thank you!