»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg p6eval perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend!
Set by sorear on 4 February 2011.
00:00 tokuhiro_ joined
japhb_ more web benchmarks, or more benchmarks in general? 00:00
00:00 cognominal left
japhb_ Yeah, more work per test is definitely important. 00:00
supernovus japhb: web benchmarks. I want to test the adapter-based services (SCGI and FastCGI) with at least two HTTP servers, and compare all three of the Perl 6 libraries against their Perl 5 counterparts. 00:02
japhb_ nodnod
Perl 5's Benchmark.pm had a useful feature that (in addition to specifying a fixed number of iterations per test), you could specify a minimum run time per test, and it would iterate a couple times to determine an iteration count that would meet that minimum run time. 00:03
supernovus japhb_: That's pretty cool! I just recently re-discovered the 'ab' tool (thanks to bbkr++ for that) and figured it would be a neat idea to see how my web app libraries compare to one another :-) I'll add in more Perl 6 libraries that I didn't write as I built up the benchmarks, the first iteration is limited to the three libraries I have the most experience with (being the author helps...) 00:06
japhb_ heh 00:07
supernovus Anyway, time to go for this evening, have a good one! :-) 00:08
00:08 supernovus left 00:10 cognominal joined
dalek rl6-bench: f264e1b | (Geoffrey Broadwell)++ | TODO:
More feature TODOs
 00:14
00:19 cognominal left 00:27 cognominal joined
cognominal r: my &ords = -> $_ { .split('')>>.ord }; my ($a,$b) = "ABCDE\nABDCE".split: "\n"; say +((ords($a) >>+^<< ords($b)).grep: * != 0) 00:28
p6eval rakudo 7ddece: OUTPUT«2␤»
cognominal this is a top ugly way to calculate a hamming distance (for that problem : rosalind.info/problems/hamm/) is there anything cleaner? 00:29
00:34 cognominal left 00:39 fridim_ left
TimToady my ($a,$b) = "ABCDE\nABDCE".split: "\n"; say [+] $a.comb Zne $b.comb; 00:39
phenny TimToady: 15 Oct 22:25Z <pmichaud> tell TimToady BTW, one of the girls on our LEGO robotics team really likes Camelia as the mascot. :-)
TimToady nr: my ($a,$b) = "ABCDE\nABDCE".split: "\n"; say [+] $a.comb Zne $b.comb; 00:40
p6eval rakudo 7ddece, niecza v22-13-g0d2a99e: OUTPUT«2␤»
japhb_ phenny, tell cognominal Str.ords already exists in Rakudo
phenny japhb_: I'll pass that on when cognominal is around.
00:41 cognominal joined
TimToady cognominal: see backlog 00:41
cognominal thx
phenny cognominal: 00:40Z <japhb_> tell cognominal Str.ords already exists in Rakudo
TimToady well, that's assuming they're the same length, of course 00:42
japhb_ TimToady, that's specified in the problem cognominal++ referenced
TimToady there was some sentiment to remove .ords
japhb_ I really like .ords, myself. 00:43
(conceptually, I mean -- I haven't looked at Rakudo's implementation thereof)
TimToady it may have been chrs that people were upset with, but they go together
00:44 cognominal left
TimToady ords might be faster than comb 00:47
and ords could conceivably be lazy enough not to create an array of the whole string 00:48
well, so could comb
in either case should just be a lightweight iterator pointing into the (presumably immutable) string 00:49
00:52 mjreed joined 00:54 cognominal joined, MikeFair left 00:56 anuby joined, cognominal left 01:01 MikeFair joined 01:06 cognominal joined 01:09 MikeFair left, cognominal left 01:13 fgomez left 01:14 hypolin joined 01:16 MikeFair joined 01:20 cognominal joined 01:22 cognominal left
skids For NCI with APIs that allow you to provide an allocator hook, is there a parrot/nqp function that is worth handing off, e.g. one that will free/GC the given allocation? 01:22
( I assume there is, but which one is preferred?) 01:23
01:26 fgomez joined 01:32 cognominal joined
colomon rn: my %a = b => c; say %a<b>:defined 01:41
p6eval rakudo 7ddece: OUTPUT«===SORRY!===␤Confused␤at /tmp/fc9Qc2Hr4w:1␤»
..niecza v22-13-g0d2a99e: OUTPUT«===SORRY!===␤␤Undeclared routine:␤ 'c' used at line 1␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza/boot/lib/CORE.setting line 1437 (die @ 5) ␤ at /home/p6eval/niecza/src/STD.pm6 line 1147 (P6.comp_unit @ 37) ␤ at /home/p…
colomon rn: my %a = "b" => "c"; say %a<b>:defined 01:42
p6eval rakudo 7ddece: OUTPUT«===SORRY!===␤Confused␤at /tmp/vUl_IsB5d3:1␤»
..niecza v22-13-g0d2a99e: OUTPUT«Unhandled exception: Excess arguments to KERNEL Hash.postcircumfix:<{ }>, unused named defined␤ at <unknown> line 0 (KERNEL Hash.postcircumfix:<{ }> @ 1) ␤ at /home/p6eval/niecza/lib/CORE.setting line 183 (postcircumfix:<{ }> @ 5) ␤ at /tmp/TbPHRX3VDP l…
colomon rn: my %a = "b" => "c"; say %a<b>:exists
p6eval rakudo 7ddece: OUTPUT«===SORRY!===␤Confused␤at /tmp/TOSQgJz7NG:1␤»
..niecza v22-13-g0d2a99e: OUTPUT«True␤»
colomon rn: my %a = "b" => "c"; say %a<b> ~~ defined
p6eval rakudo 7ddece: OUTPUT«===SORRY!===␤CHECK FAILED:␤Calling 'defined' will never work with no arguments (line 1)␤ Expected any of:␤ :(Mu \x)␤»
..niecza v22-13-g0d2a99e: OUTPUT«===SORRY!===␤␤Unsupported use of bare 'defined'; in Perl 6 please use .defined if you meant $_, or use an explicit invocant or argument at /tmp/Ac1SODgJma line 1 (EOF):␤------> my %a = "b" => "c"; say %a<b> ~~ defined⏏<EOL>…
colomon rn: my %a = "b" => "c"; say %a<b>.defined 01:43
p6eval rakudo 7ddece, niecza v22-13-g0d2a99e: OUTPUT«True␤»
01:43 popl left 01:50 tokuhiro_ left
skids rn: my %a = "b" => "c"; say %a.exists("b") # knew I had seen at least one form of "exists" working 01:50
p6eval rakudo 7ddece, niecza v22-13-g0d2a99e: OUTPUT«True␤»
01:50 tokuhiro_ joined 01:52 FROGGS_ joined 01:54 tokuhiro_ left
colomon the sad thing is, I probably implemented the niecza version of that. stupid memory. 01:55
01:55 flightrecorder left 01:56 FROGGS left, flightrecorder joined 02:01 cognominal left
sorear colomon: actually I added .exists(), and btw it's not spec 02:13
the spec form is :exists 02:14
colomon sorear: I knew that it wasn't spec.
02:14 daniel-s left
colomon sorear++ # adding it for rakudo-compatibility 02:14
02:22 japhb_ left
skids Interestingly, the method form IS specced for Array 02:22
02:22 daniel-s joined
kurahaupo ii, 02:26
sorear o/ kurahaupo 02:27
kurahaupo sorry, random window refocus 02:28
02:30 daniel-s left 02:35 mjreed_ joined 02:36 mjreed_ left 02:38 mjreed left
colomon Ooo, just published my 99th post on justrakudoit. 02:56
03:03 gootle joined, cibs left, leprevost joined 03:13 xinming_ joined 03:14 am0c joined 03:16 xinming left
TimToady whois Xmental 03:31
just checking to see if a known userid here...
sorear I think you wanted /whois 03:35
TimToady yes, that's what I did next, but you can't see that bit... 03:42
nothing, btw
it's after that that I googled for it
(we're discussing an RC vandal)
am0c hello #perl6, there is no "NativeCall" thingy in perlcabal.org/syn/ although there is an article about it ( perl6advent.wordpress.com/2010/12/1...om-perl-6/ ) 03:50
is implementation or spec about it removed temporarily? 03:51
TimToady implementations are free to prototype things like this; they may be specced when they prove to be universal across implementions 03:52
sorear am0c: NativeCall has never been part of the generla spec
am0c oh, I see.
so, it means that more implementations make us happy yey! 03:55
03:56 benabik left, benabik joined 03:59 thou joined 04:01 kurahaupo left 04:07 leprevost left 04:11 telex left 04:17 kurahaupo joined 04:19 telex joined 04:28 telex left 04:32 hash_table left 04:35 telex joined 05:00 aharoni joined 05:03 kaleem joined 05:06 thou left 05:09 kevouze joined 05:22 sftp left, sftp joined 05:32 birdwindupbird joined
moritz colomon: re justrakudoit.wordpress.com/2012/10/...-actions/, what about make $/.hash.values[0].ast ? 05:39
pmichaud yay, I've finally found a program to reproduce the parrot readline bug in pure parrot 05:41
05:42 bryntum_support joined
moritz \o/ pmichaud++ 05:42
pmichaud gist.github.com/3897389 05:43
I'll create a ticket for it
05:44 bryntum_support left
dalek kudo/nom: 637aad0 | moritz++ | docs/ChangeLog:
update ChangeLog
 05:44
05:44 SamuraiJack joined
sorear pmichaud: o.O 05:45
masak pmichaud++
pmichaud: on my setup, it doesn't even show the prompt until I hit ^D (and then it runs the whole program without waiting for input) 05:48
pmichaud ick, does that mean the flush issue is still present?
masak I... probably have a very old Parrot. :/
updating.
pmichaud github.com/parrot/parrot/issues/861 05:50
masak I'm now on bleeding Rakudo, which uses Parrot RELEASE_4_8_0. still no prompt until ^D. 05:55
pmichaud wtf? why is bleeding rakudo using RELEASE_4_8_0? 05:56
sorear no, the causality is backward.
Rakudo is bleeding because masak stabbed it with RELEASE_4_8_0 05:57
pmichaud rakudo nom HEAD should be on RELEASE_4_4_0
nqp has RELEASE_4_4_0 in tools/build/PARROT_RELEASE, unless my git repos are way out of sync
moritz well, 4.8.0 > 4.4.0 05:58
masak my Rakudo just built on 2012.09.1-39-g637aad0.
moritz and we have no mechanism in the build system to enforce the downward bump
sorear is the nqp build system smart enough to auto-downgrade?
moritz no
sorear oh
pmichaud we don't do maximum version numbers, no.
masak should I bump my Parrot down to 4.4.0?
maybe just nuke the parrot/ subdir? 05:59
pmichaud masak: you definitely don't want to be on 4.8.0, no matter what :-)
masak heh :)
masak nukes
sorear pmichaud: just bc of io-rewrite?
pmichaud sorear: I'm pretty sure io-rewrite is the culprit, yes. 06:00
I know I ended up spending several days on the 2012.09 star release because of the bugs in the 4.8.0 release :-/
sorear pmichaud: I mean, are there other showstopper bugs besides readline? 06:01
pmichaud sorear: I don't know... I don't think many people have been using 4.8.0 with later rakudoes
i.e., it's a largely untested configuration.
06:01 ServerSage left
sorear does it pass spectest at least? 06:02
pmichaud yes, it passes spectest
but neither spectest nor parrot tests were able to catch the autoflush or readline bugs
masak I/O is always going to be undertested. it's a square peg to unit testing's round hole. 06:03
pmichaud I'm thinking the 2012.10 release needs to remain on 4.4.0-4.7.0 06:04
sorear I am tempted to do the following experiment:
remove autoflush support from Perl 5. run "make test". see what kinds of tests catch it, if any
pmichaud immediately after the release (and assuming parrot gets the readline bug fixed), we can try bumping PARROT_REVISION for a month to see if any other I/O bugs pop up 06:05
06:05 ServerSage joined
pmichaud anyway, bedtime here...bbl 06:08
sorear gnight
masak 'night, pm 06:09
PerlJam++ is releasing 2012.10. any volunteers for 2012.11? 06:10
FROGGS_ masak: any special knowledge required, other than just to be able to run the test suite? 06:12
masak that's about it. 06:14
06:14 flightrecorder left
masak see docs/release_guide.pod for the process in detail. 06:14
it's actually very straightforward.
FROGGS_ I know that document, since I needed it to recompile for testing 06:15
what blog-accounts would I need?
masak I'm back on Parrot 4.4.0. z.pir now prompts directly.
FROGGS_: no blog accounts needed. 06:16
one just emails p6c.
on 4.4.0, the readline bug doesn't manifest. I guess because it's pre-io-rewrite?
FROGGS_ masak: thanks 06:17
so I might volunteer, if no other wants to do it 06:18
masak FROGGS_: please add yourself to github.com/rakudo/rakudo/blob/nom/..._guide.pod under "Planned 2012 releases" 06:20
FROGGS_ k, I will in a few minutes
brb
06:20 FROGGS_ left
sorear CIA postmortem, have not finished reading: esr.ibiblio.org/?p=4540 06:28
bonsaikitten I've put an irker instance online to cover for cia's disappearance 06:30
it's an interesting fail because there were no backups and the hoster did wipe that VM after getting an explicit NAK from the customer
doy i can't be the only one around here who gets twitchy just by instinct when people write a blog post talking about overengineering and "what happened to simplicity" and whatnot 06:39
esr: still pretty irritating 06:40
sorear the current version of dalek is a tail -f bot
there is a file on feather3, anything you cat >> it will show up here 06:41
bonsaikitten doy: amusing: esr used xml for writing the manpages
doy: ... if that's not overengineering ...
06:42 zk_ joined
bonsaikitten apart from that irker is pretty much the minimum needed for structured notifications 06:42
doy well, that's the thing
06:42 kaleem left, zk_ left
doy "overengineering" in about 90% of the cases just means "i don't understand all of your use cases" 06:42
06:43 birdwind1pbird joined
bonsaikitten hmm, yes and no 06:43
if you look at the cia codebase it's multiple different languages for no reason
there's strange use of XML (and it smells even like pseudo-XML) that feels a bit unneeded 06:44
it does cover a lot more ground than irker (webUI, statistics etc)
still ... why write one part in python, then one part in erlang and nail it together with something else as translation layer? 06:45
doy "that feels a bit unneeded" is my point 06:46
"feelings" are generally not very productive in this sense
granted, this is more relevant on projects that haven't failed 06:47
but still
bonsaikitten usually that "code smell" feeling has a base, but it takes lots of time to formalize it
intuition is not precise, but has a good enough hit rate to be used as a starting point for investigations 06:48
doy sure
but there's this tendency to use that intuition as a reason to just be dismissive and feel superior about your design skills 06:49
again, in this particular case it may be justified (i don't know, i haven't seen the actual code), but that kind of language just tends to set off other kinds of alarm bells in my head
06:50 cibs joined 06:51 sizz left
bonsaikitten I'm not a good designer, just better than average ;) 06:51
06:52 flightrecorder joined, sizz joined 06:58 birdwind1pbird left 06:59 birdwind1pbird joined, mucker joined 07:00 sizz_ joined, sizz left
masak the topic of software design interests me. 07:01
though the words "architecture" and "architect" are even less strictly defined than the term "object orientation" :)
sorear overengineering is what happens when you change your requirements 10 times and wind up with software that does 80% things you don't need because inadvertant cross-dependencies prevented you from fully adapting the software to the removal of requirements 07:02
bonsaikitten there are funny patterns like "second system effect" 07:03
basically "let's rewrite it to fix $issues - oh, why not add $features - oh, let's add EVERYTHING"
masak *cough* 07:04
sorear bonsaikitten: have you heard the perl 6 motto?
bonsaikitten sorear: ... yes? 07:05
07:06 brrt joined
masak skimming the post, it seems to me that esr does have a point about the complexity and overengineering. 07:08
ooh, and "I'd take that patch" (from the comments). nice, haven't seen that meme before. 07:09
"To paint this purely as a case of some idiot developing a recklessly complex architecture would be misleading, and I think much less interesting than the truth: This is a great case study of what happens when a service is well-liked but nobody steps up to maintain it. It grows organically to the extent that it can, then collapses." 07:11
nice summary.
and that's where a deliberate architecture, and an architect, would have been helpful. 07:12
07:14 mucker left 07:17 hoelzro|away is now known as hoelzro
bonsaikitten basically someone driving the project forward 07:21
07:23 FROGGS joined
brrt hmm, reading it i can't get the feeling but 'this shouldn't have been hard' 07:23
masak brrt: which can be said about a lot of software, I guess. 07:24
bonsaikitten usually it is not hard 07:27
just takes time, and motivation
masak some people would classify things that take time and motivation as "hard". 07:29
brrt setting up an architecture like that would always have taken time 07:32
and anything that takes time pretty much takes motivation, too
07:35 flightrecorder left, FROGGS left, flightre1 joined, FROGGS joined 07:38 kevouze left 07:40 awwaiid left 07:41 bbkr left 07:42 awwaiid joined
brrt btw, testing the string representation of an object for some value 07:43
thats not always a good idea, right
07:46 birdwind1pbird left
masak no, but it gets you a long way. 07:46
I have a question about the separation of lexing and parsing.
brrt ask :-) 07:47
as if there is a better channel
masak is it the case that such a separation grows less realistic/desirable with the amount of slangs/language switching going on?
is that why neither Perl 5 nor Perl 6 makes an effort to separate lexing and parsing? 07:48
moritz yes
masak can some sort of maxim be formed here, such as "if you think you're going to do lots of DSLs, don't bother with the separation of the lexing and parsing steps"?
moritz fwiw Perl 5 does try to separate lexing and parsing
and the result is madness :-)
masak 'cus it seems to me that such a separation is a good idea, when possible. but it also seems to me that language switching makes it cumbersome or impossible. 07:49
moritz unless the different slangs can be lexed by the same rules 07:50
07:50 sirrobert left
moritz (which is usually not the case) 07:50
masak well, if you try to do the separation when there is lots of language switching going on, what you end up with is probably madness. or half-hearted compromises, like "scan ahead for (what looks like) the closing delimiter of this regex".
right. 07:51
07:51 erkan left
moritz which kinda kills you if you want to allow regex modifiers to go at the end of the regex :-) 07:51
07:52 erkan joined, erkan left, erkan joined
sorear masak: separation works best in a language like C where the lexer and parser can work asynchronously, with (ideally) zero feedback from the parser to the lexer 07:52
masak ooh, feedback.
07:52 kurahaupo left
masak hadn't thought of it in terms of feedback, but yeah, that's what's going on. 07:52
sorear (in practice most C parsers use feedback, but that's more a matter of working around yacc limitations than something fundamental to C)
masak right.
even a C parser has to build symbol tables and stuff. 07:53
sorear (specifically, C compilers like the lexer to spit out different tokens for things like TYPEDEF so that int32_t foo, bar can parse sanely)
07:54 sirrobert joined
arnsholt Requiring symbols to be declared makes your grammar context sensitive, so hacking around it is easier =) 07:54
sorear in modern C there's also a preprocessor wedged between the parser and the lexer 07:55
pre-standard C ran the preprocessor before the lexer, wth hilarious consequences
masak I suppose there's a connection between "context sensitive" and "slangs", too. 07:56
sorear #define assert(x) if (!x) carp("x");
bonsaikitten ah, stories from the dark ages 07:57
sorear watches masak slowly turn green
masak sorear: oh wow.
that is... so wrong. 07:58
tadzik :)
sorear morn tadzik
arnsholt masak: I think it was you who introduced me to the presentation style where you have a few big words on each slide, but I can't remember what it was called. Help? =) 07:59
brrt sorear: can't really see whats wrong
arnsholt brrt: It'll probably expand the x in the string, which I don't think happens in modern C
sorear brrt: just pointing out that in pre-standard C the preprocessor worked at the character level, so the occurence of x in the string is expanded 08:00
brrt ah, i see
modern c has an option for that iirc
##x or so
sorear in modern C the preprocessor runs after the lexer and mostly only touches identifier tokens
it never modifies strings except through the #x formation 08:01
masak arnsholt: Takahashi?
arnsholt Takahashi, yeah. Think that's it
masak en.wikipedia.org/wiki/Takahashi_method
I think I was directly inspired by au++.
arnsholt I'm attempting to use it for the presentation I'll give in Copenhagen of my research 08:02
sorear there's also #define FOO(x) __foo__/**/x
arnsholt It's a bit scary to not be able to rely on lots and lots of bullet points though
sorear never used this, the incantation may be a bit more involved than I remember
08:04 Caldrin left
jnthn mornin' 08:05
arnsholt o/
sorear o/
brrt \o
tadzik sorear: g'morning 08:06
jnthn Perl 6 *does* separate lexing and parsing, in my view. It's just that it doesn't make the grammar writer do it. It figures it out by looking at declarative prefixes and constructing NFAs where it can. 08:07
08:09 fhelmberger joined
sorear does not think that longest declarative prefix semantics and lexers are at all the same thing 08:11
jnthn sorear: Squint harder.
;-)
I'd say that if you hit something that is entirely declarative, though, it counts as a token. 08:12
(In the lexing sense)
masak jnthn: I disagree, for (I believe) the same reasons as sorear. 08:16
interleaved "lexing" through LTM and declarative prefixes does not count as "separate lexing" in my book.
sorear I look at it in more of an LL(\infty) way - the parser is allowed lookahead when deciding what option to take 08:18
an interesting failure mode that I've seen once or twice in niecza is that if you have too long of declarative prefixes, the \infty can be realised and now your parse is quadratic because each "rule" looks-ahead all the way to the end of the file 08:19
masak hm, never had that. 08:20
sorear maybe my p6 demotivation of late is a laws and sausages thing 08:22
masak if so, just remember that some things are inherent to the Perl 6 design, and others to a particular implementation. 08:23
seems to me the branch you went down that didn't lead anywhere in particular was instrumental in sucking some motivation out of you. 08:24
as for me, it reminds me of the phenomenal velocity with which Niecza has grown. 08:35
imagine if we'd had Niecza in 2002 :)
(and yes, I know it has a lot of dependencies we didn't have back then, such as STD.pm6) 08:36
bonsaikitten hehe
masak: that's a rather funny thought experiment 08:37
sorear masak: OTOH, then we'd be stuck with prefix:<=> and &want 08:38
masak ugh 08:40
08:40 sjohnson left
jnthn
.oO( module IdeasWeRejected; sub prefix:<=>($a) is export { $a.get } )
 08:41
sorear to be fair a lot of my gripes with perl6 boil down to "it's not designed to support niecza" :D
(a unit-at-a-time weakly optimizing compiler for a target with unusable coroutines) 08:42
brrt what about a better target
08:55 am0c left 09:00 dakkar joined 09:01 flightre1 left 09:02 FROGGS left 09:23 kresike joined
kresike hello all you happy perl6 people 09:23
09:23 hoelzro is now known as hoelzro|away
jnthn o/ kresike 09:25
moritz \o *
brrt hi kresike
kresike jnthn, moritz, brrt o/ 09:26
masak kresike! \o/ 09:29
kresike masak, o/ 09:30
sorear what about less than happy perl6 people? 09:33
brrt they can explain their cause of unhappiness 09:34
and we might all learn a bit
masak we encourage open and healthy debate, as long as people's opinions are similar enough to what we believe we believe :P 09:39
09:51 hoelzro|away is now known as hoelzro, SamuraiJack_ joined 09:54 SamuraiJack left 10:00 hypolin left 10:02 daxim joined, brrt left 10:05 anuby left 10:12 domidumont left
lumi__ r: gist.github.com/3898515 10:23
p6eval rakudo 637aad: OUTPUT«(signal SEGV)»
lumi__ That's my attempt at an overload trait, and it segfaults
jnthn == 10:25
?
You wanted eq?
It shoudln't segfault
But still, == ain't gonna work
lumi__ I probably wanted eq, yup :P
But I brought it up because it segfaults
jnthn Yeah 10:27
file it in RT, I'll see if I can reproduce
It otherwise looks like fine code
10:53 erkan left 10:58 pmurias joined 11:02 replore joined 11:03 kaleem joined 11:05 Su-Shee_ joined 11:07 Su-Shee left, Su-Shee_ is now known as Su-Shee 11:09 kaleem left, flightrecorder joined 11:10 kaleem joined 11:12 brrt joined 11:17 replore left
brrt \o #perl6 11:17
11:18 replore_ joined
pmurias hi 11:19
jnthn: ping 11:20
nqp: my $a; my $b; $a := 1; $b := 1; say($a =:= $b);
p6eval nqp: OUTPUT«0␤»
pmurias rakudo: my $a; my $b; $a := 1; $b := 1; say($a =:= $b);
p6eval rakudo 637aad: OUTPUT«True␤»
pmurias jnthn: I have resumed porting nqp to js, and I found the following difference between rakudo and nqp in the test suit 11:21
jnthn Doing =:= in NQP with something that's value-type-ish is likely a bad idea.
In fact, the thing we use =:= for in NQP is not really what Perl 6 wants it to mean... 11:22
pmurias what do you use =:= in NQP for? 11:23
jnthn "are the objects on each side the same object"
As in, same memory address.
pmurias so two different identical literals in =:= are different objects 11:24
?
jnthn Yes because NQP doesn't have a constants table for such things
It does however have native types these days
pmurias should I duplicate that in nqp-js? 11:25
jnthn The native types, or lack of constant table? :)
Looking at S03 though, NQP using =:= for the meaning it currently has is never gonna be consistent with Perl 6 semantics.
And what Rakudo did above is rather suspect also.
git st 11:26
oops
pmurias jnthn: should I implement the lack of constant table?
which is different from not implementing a constant table as js already has one ;) 11:27
jnthn pmurias: No
pmurias: Let's for now say that 1 =:= 1 in NQP is undefined behavior
I should hope we don't rely on it anywhere...
I think all the =:=s really mean "same object" when it's a reference type. 11:28
Maybe we should change to === but then people will also expect 1 === 1 to be true.
pmurias it's named eqaddr
jnthn Yeah
I don't think Perl 6 has an operator that means what NQP is doing with =:= at the moment. 11:29
Probably because defining operators with undefined semantics in many cases is a bad idea ;-)
pmurias the test suit needs to be changed
jnthn Yeah, but need to decide what to change it to first :)
pmurias looking at ack output, rakudo doesn't seem to use =:= for comparing values 11:37
=== seems more suitable
but having it undefined on values seems confusing 11:38
jnthn I'm not implementing === in real NQP
er, real === in NQP 11:39
pmurias change it to something we don't have in perl6? 11:40
jnthn Maybe we should just make them all nqp::eqaddr($a, $b) :-)
And if you do it on something that it's not sensible to do it on, you get a non-sensible answer :)
pmurias nqp::eqaddr seems much better 11:41
colomon o/ 11:42
pmurias \o 11:43
moritz "As we say in French, when life gives you stateful functions, make le monade." -- rgs 11:47
masak :) saw that one earlier today.
one of the finest puns in quite a while.
it has a certain... je ne sais quoi. acidity? :)
Su-Shee moritz: I laughed at that this morning too.. :) 11:48
moritz labs.ripe.net/Members/john_bond/wi...ng-ripe-65 # if you hold a RIPE conference, you'll surely have enough competent people to debug the wifi :-) 11:49
Su-Shee different level of "network" :) 11:50
layer :)
11:51 timbunce joined
timbunce My "Perl at SkyCon'12" slides got featured on the slideshare homepage www.slideshare.net/Tim.Bunce/perl-at-skycon12 - it includes an overview of the state of perl5 and perl6. 11:52
11:52 tokuhiro_ joined
timbunce There's also a screencast at blip.tv/timbunce/perl-at-skycon-12-6391951 11:52
moritz timbunce++ 11:53
timbunce moritz++ for the clean JSON::Tiny module code I featured.
Makes a lovely 3 slide set. 11:54
Su-Shee *hihi* and of course it starts with xkcd.. :)
timbunce The lack of procedural code is particularly impressive.
pmurias jnthn: should I change nqp? 11:55
timbunce Su-Shee: Randall Munroe was the closing keynote speaker at the conference :)
Su-Shee timbunce: haha. so everyone added an xkcd, because there is of course an xkcd for every subject and purpose by now.. ;)
masak timbunce++
jnthn pmurias: Feel free to change the =:= usages to nqp::eqaddr 11:56
masak Su-Shee: Mr Munroe should make an xkcd for the state of there being an xkcd for every subject and purpose... ;)
11:57 flightrecorder left
jnthn lumi__: gist.github.com/3898844 plus a local fix I'm spectesting works now 11:58
Su-Shee masak: I see a department of redundancy department tasting xkcd coming up. :)
monty xkcd! 11:59
masak timbunce: those Perl Data Anatomy diagrams are gorgeous.
dalek kudo/nom: 9c15141 | jnthn++ | src/binder/multidispatch.c:
Various bits of SEGV-protection in multi-dispatch.
 12:00
timbunce masak: Agreed. I'm not sure what tool was used for those (or if the current ill guts maintainer knows how to update them) 12:01
masak timbunce: very, very nice slides. thanks for sharing those. 12:03
jnthn timbunce++ # nice
masak they're nice to the point of me wondering whether, were I ever in the position of presenting Perl 5 and 6 in that way, I might not simply ask for permission to use those slides straight off.
pmichaud good morning, #perl6
masak pmichaud! \o/
timbunce I'm happy for anyone to use any of my slides. That deck was partly distilled from my earlier perl myths talks. 12:04
masak I recognize some of those slides, yes.
jnthn hi, pmichaud 12:05
timbunce The most fun in the talk was during the Q&A when I was demonstrating the scope of CPAN by asking the (mostly non perl) audience for "any topic" and typing it into the metacpan.org search box to see the matching suggestions. Proved to be impressive.
masak I can imagine :) 12:06
that's a neat way to convey the impressiveness of CPAN.
timbunce Someone asked for "world peace" - I suggested they work on that one themselves.
Su-Shee that's just one half of the impressiveness..
I have xx years old platforms where I only do cpan Foo::Bar for whatever I need and _nothing_ else and it all works and all installs and all runs.. 12:07
timbunce I liked the tag line "Get ahead of the revolution: perl6.org" 12:08
pmichaud excellent slides timbunce++
masak r: say "I have $x$x years old platforms..." for 1..5
p6eval rakudo 637aad: OUTPUT«===SORRY!===␤Variable $x is not declared␤at /tmp/uIdII7N0ff:1␤»
timbunce pmichaud: thanks
masak r: say "I have $_$_ years old platforms..." for 1..5
p6eval rakudo 637aad: OUTPUT«I have 11 years old platforms...␤I have 22 years old platforms...␤I have 33 years old platforms...␤I have 44 years old platforms...␤I have 55 years old platforms...␤»
pmichaud I recognize some code examples in those slides :) 12:09
timbunce :) 12:10
12:10 pmurias left
masak cards++ 12:10
12:10 ggoebel left
colomon moritz++ # omg, make $/.hash.values[0].ast // []; works in both rakudo and niecza. big win! 12:10
moritz colomon: even $/.values[0].ast might work 12:11
12:13 domidumont joined 12:17 bbkr joined
bbkr rakudo: printf "%d"; # known bug? 12:17
p6eval rakudo 9c1514: OUTPUT«Null PMC access in get_integer()␤ in sub sprintf at src/gen/CORE.setting:2166␤ in sub printf at src/gen/CORE.setting:2171␤ in block at /tmp/lpFWKhquug:1␤␤»
tadzik not surprising bug, at least 12:21
pmichaud bbkr: rt #106594
rt.perl.org/rt3/Ticket/Display.html?id=106594
tadzik considers toying around with our own printf()
bbkr pmichaud: thanks 12:22
12:22 birdwindupbird left 12:24 birdwindupbird joined
brrt popquiz, why are 6 bit operations stupid? 12:28
(context, a virtual machine) 12:29
32-bit instruction word, 6 bytes for the operation
6 bits 12:30
masak brrt: insufficient context. 6 bits seems plenty given that brainfuck only has 8 operations in total. only 3 bits required to encode those. 12:31
brrt 6 bits are plenty
the point is efficiency 12:32
getting the first 6 bits takes at least an AND operation
as in, instruction & 0b111111
making it 8 bits, its a full byte 12:33
colomon moritz++: you are quite right.
jnthn tadzik: Please do :)
tadzik: One that can handle bigint too :)
tadzik aye
jnthn: do you think a grammar is a proper way to tackle this? 12:34
brrt context is this nice link: rsms.me/2012/10/14/sol-a-sunny-litt...chine.html
tadzik it feels Perlsixy
brrt: yeah, 6 bits seems a bit underthought
pmichaud tadzik: I wouldn't use a grammar, no. 12:35
tadzik for efficiency reasons?
jnthn Is it actually that complex? 12:36
tadzik not really
pmichaud that, and it just strikes me as being easier to scan for %'s
tadzik well, if it were mere %'s
but then you have %5d's
and %-5d's, stuff
pmichaud you still scan for the %'s
tadzik sure 12:37
pmichaud then you look at what comes next to figure out what to do
arnsholt You could comb $format-regex I guess?
Although that might count as grammars
tadzik arnsholt: but then you get only %'s, not the rest
well, I'll experiment and see for myself
pmichaud tadzik: oh, you could comb for /%[dxo]/ patterns
(I still don't recommend it, but it would work)
tadzik I could, yes
but then it's a matter of substituting them with new stuff 12:38
so at least 2 passes, no?
arnsholt Feels like it should be doable in a single pass
tadzik yeah, I just don't see how I'd do that with .comb
arnsholt True 12:39
jnthn At present, regex is probably the fastest way to scan through a string
tadzik without making it unneceserilly complex
right
yay, fun project on the horizon :) 12:40
jnthn yay
pmichaud you could do .split( /pat/, :all)
that would still be one pass
12:40 benabik left
tadzik hm, I never knew about :all 12:40
12:40 benabik joined
tadzik indeed, split, map, join then 12:40
12:41 leont joined
tadzik will try different approaches and see how they work out 12:41
12:43 jaldhar_ is now known as jaldhar 12:48 mucker joined 13:00 hash_table joined 13:04 Psyche^ joined 13:06 telex left 13:07 Patterner left, Psyche^ is now known as Patterner 13:08 telex joined 13:09 rindolf joined, hoelzro left 13:10 rob___ joined, rob___ is now known as hoelzro 13:13 skids left 13:14 flightrecorder joined, FROGGS joined 13:15 sirrobert left 13:22 GlitchMr joined 13:25 PacoAir joined 13:26 grondilu joined
grondilu rn: my @words = <foo bar>; say "foo" ~~ / ^ < @words > $ /; 13:26
p6eval niecza v22-13-g0d2a99e: OUTPUT«Potential difficulties:␤ @words is declared but not used at /tmp/AFcbwDjjUz line 1:␤------> my ⏏@words = <foo bar>; say "foo" ~~ / ^ < @␤␤Match()␤»
..rakudo 9c1514: OUTPUT«#<failed match>␤»
grondilu r: my @words = <foo bar>; say "foo" ~~ / ^ < {@words} > $ /; 13:27
p6eval rakudo 9c1514: OUTPUT«#<failed match>␤»
grondilu n: my @words = <foo bar>; say "foo" ~~ / ^ < {@words} > $ /;
p6eval niecza v22-13-g0d2a99e: OUTPUT«Potential difficulties:␤ @words is declared but not used at /tmp/W40e7m7NU1 line 1:␤------> my ⏏@words = <foo bar>; say "foo" ~~ / ^ < {␤␤Match()␤»
grondilu Can't I interpolate inside a < > in regex?
moritz D/w 20 13:29
13:29 leont left
masak grondilu: see S05. 13:29
(not an "RTFM" so much as "I don't know/don't remember offhand")
jnthn No, just as you can't in normal Perl 6. 13:30
masak oh, it's [|@words] or something, isn't it?
jnthn Something like that, yeah
grondilu masak: ok, sorry 13:31
13:32 benabik left
grondilu rn: my @words = <foo bar>; say "foo" ~~ / @words /; 13:33
masak no apology needed.
p6eval niecza v22-13-g0d2a99e: OUTPUT«#<match from(0) to(3) text(foo) pos([].list) named({}.hash)>␤»
..rakudo 9c1514: OUTPUT«#<failed match>␤»
grondilu according to line 1014 of S05 the above should match, shouldn'it? 13:34
lumi__ Is there some trick to exporting trait_mod:<is> ?
grondilu oh ok I've just realized that niecza found a match. 13:35
masak lumi__: don't think so.
grondilu: yeah. probably just unimplemented in Rakudo.
or implemented but not to spec.
moritz lumi__: it might need a new rakudo 13:37
13:37 pmurias joined
moritz r: module A { multi trait_mod:<is>() is export { } }; import A; 13:37
p6eval rakudo 9c1514: ( no output )
moritz seems to work 13:38
star: module A { multi trait_mod:<is>() is export { } }; import A; 13:39
p6eval star 2012.07: ( no output )
moritz star: module A { multi trait_mod:<is>(say 42 ) is export { } }; import A; trait_mod:<is>() 13:40
p6eval star 2012.07: OUTPUT«===SORRY!===␤Invalid typename in parameter declaration at line 2, near " 42 ) is e"␤»
moritz star: module A { multi trait_mod:<is>() is export { say 42 } }; import A; trait_mod:<is>()
p6eval star 2012.07: OUTPUT«42␤»
moritz r: module A { multi trait_mod:<is>() is export { say 42 } }; import A; trait_mod:<is>()
p6eval rakudo 9c1514: OUTPUT«42␤»
moritz seems no trick is necessary
13:41 simo_ joined
lumi__ moritz: I'm having trouble with importing it from another file, not sure what exactly is going on, but my multi doesn't appear on the candidate list 13:42
13:42 simo_ left
moritz lumi__: how young/old is your rakudo? 13:44
13:44 tokuhiro_ left, hash_table left
moritz lumi__: I suspect it might need patch Perl-22-g031a497 13:44
13:44 tokuhiro_ joined 13:45 grondilu left
lumi__ Head from after jnthn++'s recent fix to my segfault 13:47
13:49 tokuhiro_ left
felher Hm, i know i can interpolate a variable into a regex with / $variable /. Is there also an easy way to interpolate an attribute of an object into the regex? / $object.strAttribute / doesn't work, because it interpolates $object.Str, and takes '.strAttribute' as part of the normal regex. 13:53
FROGGS so you are searching for Perl 5's \Q\E? 13:55
moritz FROGGS: try $obj.strAttribute()
lumi__: I've just put a simple trait_mod:<is> into a file, exported it and used it just fine
(also on HEAD)
GlitchMr {$object.strAttribute} or $object.strAttribute() 13:56
It should work
Oh, wait, {} is assertion
moritz {} in a regex is just action 13:57
<{ }> might do it, not sure
felher GlitchMr, moritz: yep. {} is an action and () in a regex is treated das a (disallowed) null-pattern 13:58
even after $object.strAttribute
lumi__ I'm getting this weirdness: Method 'SET_FILE_LINE' not found for invocant of class 'X::AdHoc'
I guess maybe my rakudo is sick
FROGGS moritz: can you try to add "use NativeCall" after you import the trat_mod:<is> from that file and see if it still sowkrs?
13:58 replore_ left
FROGGS s/sowrks/works/ 13:58
felher tries <{ }> 13:59
FROGGS didnt update my rakudo for a while now, so I'm not sure that the import-bug is still there
GlitchMr This is confusing. If {} interpolates in strings, why it shouldn't it regexpes?
felher moritz++ # awesome, thnx :)
hoelzro GlitchMr: that may be a bug
13:59 stopbit joined
moritz lumi__: that indicates you might throw an exception inside the trait 14:00
lumi__: try calling it at run time, and see if you get a better error message
GlitchMr It's a feature as far I know, but confusing
How often you use (?{}) in Perl 5 regexpes anyways?
moritz doesn't need interpolation into regexes very often 14:01
FROGGS r: use NativeCall 14:02
p6eval rakudo 9c1514: OUTPUT«===SORRY!===␤Could not find NativeCall in any of: /home/p6eval/.perl6/lib, /home/p6eval/nom-inst/lib/parrot/4.8.0-devel/languages/perl6/lib, /home/p6eval/nom-inst/lib/parrot/4.8.0-devel/languages/perl6/custom/lib␤»
FROGGS sad
jnthn FROGGS: r is just the compiler 14:03
star: use NativeCall;
p6eval star 2012.07: ( no output )
FROGGS ohh
jnthn That's how you get at the distro
FROGGS but 2012.07 is way to old for that...
jnthn oh, hm
FROGGS jnthn: did you fix that trait_mod import problem? 14:04
pmurias GlitchMr: putting in prints for debugging purposes is one use of (?{ }) 14:05
jnthn FROGGS: Hmm...I fixed something about trait mod in 031a49747 but I think you mean something else. 14:06
14:06 kaare_ joined
FROGGS hmmm, k, I'll recompile and check 14:10
moritz FROGGS: I still can't import trait_mod:<is> from two different modules, no 14:16
FROGGS moritz: k, thanks
jnthn Is that a general multi issue?
GlitchMr pmurias: isn't debugger for that? 14:17
FROGGS jnthn: I think so
jnthn: the problem is here: github.com/rakudo/rakudo/blob/nom/...ld.pm#L268 14:19
it should throw an X::Import::Redeclaration, but instead should merge the symbols and add dispatchees 14:20
s/should/shouldn't)
meh
jnthn Yeah
Only if things are all onlystars I guess though
pmurias GlitchMr: I haven't tried Regexp::Debugger
FROGGS I tried to hunt that down, but after a week and a half I gave up
jnthn Otherwise I guess the different protos could conflict.
pmurias GlitchMr: and I prefer prints to the perl5 code debuggers 14:21
jnthn FROGGS: I think it's just an NYI
FROGGS right, thats why there is a # TODO ;o)
I tried to find the point where I would be able to handle the dispatchees, but had no luck 14:22
all I can say is if you uncomment that exception, you will get the symbols/dispatchees of the module you import at last 14:23
jnthn Right
14:24 rindolf left 14:25 bbkr__ joined 14:27 bbkr left
moritz what I currently wonder is 14:29
star: use NativeCall; class A is export { }
p6eval star 2012.07: ( no output )
moritz how does that work?
I'd thought that NativeCall exports a trait_mod:<is> 14:30
how does it work that the 'is export' from the setting is still visible?
colomon moritz: it's a multi of trait_mod:<is>, isn't it? 14:31
moritz colomon: yes, that much is clear
14:31 sirrobert joined
FROGGS moritz: you can have the trait_mod:<is> from core and another module 14:32
moritz does it mean that importing multis merges the multis into the existing multis?
FROGGS but not core + module + module
moritz FROGGS: correct, and I don't understand why
FROGGS me neither
colomon seems like it should always work no matter how many modules are involved? (are we talking rakudo or spec here?) 14:33
moritz colomon: that much is also clear
the question is why core + module + module doesn't work in rakudo, even though core + moduel works
since core + module works, there must be a merging operation somewhere (though maybe implicit through scoping or so)
FROGGS maybe there is a separate dispatcher list for core and modules? 14:34
jnthn moritz: I think the answer is horrible.
moritz: It's because the CORE multi candidates are included in the dispatcher that is exported.
moritz eeks
jnthn moritz: Because that's how lexical multis work. 14:35
(they incorporate outer candidates into their candidate list)
moritz so on merging, we have to take care not to create dispatch conflicts between the same candidate exported through two different routes
FROGGS btw, I made debugging output so I could see what functiona are called when loading and importing modules, its just crazy (for me), every core module is calling stuff from modules that calls stuff from the calling modules and so on
just like crazy
jnthn What? :) 14:37
moritz: Something like that.
moritz: Or we make sure that exports don't end up including things that they "inherited"
sirrobert If there's a file I eval a fair bit (for my test suite), is there a way to compile it and read the compiled version in? 14:38
(eval seems slower than I'd expected)
moritz sirrobert: make it a module, precompile it
sirrobert moritz: ok. precompiling modules is currently broken for me. 14:39
moritz :(
sirrobert well, actually, I haven't checked in like a month... maybe it's working =) 14:40
(checking)
jnthn You have to be doing some fairly specific things to hit pre-comp bugs. 14:41
14:41 skids joined
sirrobert it's a natural skill =) 14:41
14:41 ggoebel joined
colomon seems to recall that he has hit pre-comp bugs as well... 14:41
sirrobert running my test suite on the precompiled bit now. 14:42
moritz colomon: ABC did, last I looked
colomon moritz: right
jnthn User-defined operators are the "easy way" to hit the issue
colomon moritz: I don't even try to pre-comp any of my modules these days... I always do something like prove -e "perl6 -Ilib" t/
14:43 grondilu joined
sirrobert !! perfect tests on Rakudo 2012-09 =) I turned off precomp in the 2012-07 when I was hitting the issues. 14:43
everyone++ 14:44
and the tests ran in 1/98457639875349739 the time =)
14:44 hash_table joined
jnthn precomp ain't perfect yet for sure, but the cases where it works are much more common than the cases where it doesn't 14:44
sirrobert jnthn: I had hit something that broke all of my tests 14:45
(I wrote a testing framework to do spec-testing for a REST API)
jnthn ooh, nice
sirrobert probably something in there went weird
jnthn (the framework, not the bug)
sirrobert heh nod
it's super slick
it allows introspection to produce api spec docs out of the test suite =)
(*I* like it, anyway ;) 14:46
moritz sounds nice :-)
the precomp bugs are usually all-or-nothing 14:47
sirrobert when we get some solid footing (business-wise) I'll see about releasing it and some other stuff we built
moritz either everything works perfectly, or you get a compile-time error
sirrobert heh nod
that was my experience
moritz (at least that's my experience so far)
jnthn Yes, and I'd rather have it all-or-nothing that really weird behavior. 14:48
14:49 benabik joined 14:52 kresike left 15:01 bbkr_ left 15:02 bbkr_ joined
rjbs- 2/lm 15:04
15:05 lorn left 15:06 lorn joined 15:08 replore joined 15:09 colomon_ joined
skids reposes irclog.perlgeek.de/perl6/2012-10-16#i_6065721 question now that folks are awake 15:09
15:10 colomon left, colomon_ is now known as colomon
jnthn skids: Can I have a concrete example of such an API? 15:10
skids jnthn: things like: void *mhash_end_m(MHASH thread, void* (*hash_malloc)(size_t)); 15:11
The malloc is used to create the memory for the digest, which is pointed to by the return value. 15:12
15:12 mtk left 15:13 att joined
jnthn skids: Maybe arnsholt++ has some idea but I'm not aware of any such function we have that'd do anything useful there. 15:14
moritz what is the problem, the callback or the malloc? 15:17
skids No problem, most APIs that offer these forms also offer forms that just internally call malloc.
The question is whether there's a benefit from using the other form, for example, getting a malloc that can DESTROY itself (eventually, even if NYI). 15:18
15:18 mtk joined
jnthn moritz: Did you mean to add changelog entries under New in 2012.09.1? 15:19
moritz: Or did you want a 2012.10 heading? :)
(I mean for latest 3 entries)
15:19 fgomez left 15:22 benabik_ joined
dalek kudo/nom: 7372a10 | jnthn++ | src/core/ (8 files):
Eliminate various pir:: usages.
 15:23
kudo/nom: c2005e6 | jnthn++ | src/Perl6/Grammar.pm:
Fix most of the Pod tests on Win32.
15:24 grondilu left 15:25 benabik left, benabik_ is now known as benabik 15:27 hoelzro is now known as hoelzro|away 15:28 benabik_ joined 15:30 sjohnson joined, sjohnson left, sjohnson joined, benabik left, benabik_ is now known as benabik 15:32 rindolf joined, fgomez joined, brrt left 15:33 MayDaniel joined
jnthn r: $*ARGFILES.lines() 15:34
p6eval rakudo 9c1514: ( no output )
jnthn r: say $*ARGFILES.lines()[0]
p6eval rakudo 9c1514: OUTPUT«Land der Berge, Land am Strome,␤Land der Äcker, Land der Dome,␤Land der Hämmer, zukunftsreich!␤Heimat bist du großer Söhne,␤Volk, begnadet für das Schöne,␤vielgerühmtes Österreich,␤vielgerühmtes Österreich!␤␤Heiß umfehdet, wild umstritten␤liegst dem Erdteil du inmi…
jnthn r: say $*IN.lines()[0]
p6eval rakudo 9c1514: OUTPUT«Land der Berge, Land am Strome,␤Land der Äcker, Land der Dome,␤Land der Hämmer, zukunftsreich!␤Heimat bist du großer Söhne,␤Volk, begnadet für das Schöne,␤vielgerühmtes Österreich,␤vielgerühmtes Österreich!␤␤Heiß umfehdet, wild umstritten␤liegst dem Erdteil du inmi…
jnthn r: say $*IN.lines().perl
p6eval rakudo 9c1514: OUTPUT«("Land der Berge, Land am Strome,\nLand der Äcker, Land der Dome,\nLand der Hämmer, zukunftsreich!\nHeimat bist du großer Söhne,\nVolk, begnadet für das Schöne,\nvielgerühmtes Österreich,\nvielgerühmtes Österreich!\n\nHeiß umfehdet, wild umstritten\nliegst dem Erdt…
jnthn r: say open('README', :r).lines[0] 15:35
p6eval rakudo 9c1514: OUTPUT«open is disallowed in restricted setting␤ in sub restricted at src/RESTRICTED.setting:2␤ in sub open at src/RESTRICTED.setting:5␤ in block at /tmp/HbelFTiQq_:1␤␤»
jnthn r: say CORE::open('README', :r).lines[0]
p6eval rakudo 9c1514: OUTPUT«No such method 'Any' for invocant of type 'Str'␤ in at src/gen/BOOTSTRAP.pm:814␤ in any at src/gen/BOOTSTRAP.pm:811␤ in block at /tmp/OUvIHGkHN2:1␤␤»
jnthn Hm, oddness. lines works for me... :S 15:36
moritz jnthn: I wanted 2012.10 15:44
jnthn: but it seems I've messed it up
jnthn moritz: ok :)
std: 'abc' ~~ /:iabc/ 15:46
p6eval std 77327a4: OUTPUT«===SORRY!===␤Unrecognized regex modifier :iabc at /tmp/bZn_zdAHo2 line 1:␤------> 'abc' ~~ /:iabc⏏/␤Check failed␤FAILED 00:00 44m␤»
15:56 thou joined 15:57 alester joined 15:58 ComWT left
sirrobert is there a way from "prove" to run just a specific test? 15:58
moritz prove -e perl6 t/yourtest.t 15:59
sirrobert like, 32 tests planned, skip until you get to test 14 and run that one.
err assertions, I mean
moritz erm, no
sirrobert ok =)
moritz often individual tests depend on previous tests
and prove just executes some binaries; doesn't need to be perl 16:00
anything that prints plan 1;\nok 1 - your dad\n is fine
sirrobert nod 16:04
just wondering if it was possible for those times that don't =)
easy enough to comment out the ones I want to skip 16:05
mistype
moritz easy enough to mistype :-) 16:06
16:06 silug left, silug joined
sirrobert heh 16:06
masak sirrobert: if your tests are parseable enough, you could autogenerate test files containing subsets of your tests, and run those. 16:07
sirrobert ahh, good idea. Yeah, I tend to put sets of subtests in anonymous blocks
thanks
moritz has an evil idea 16:08
sirrobert ....? 16:09
do tell =)
moritz sub test($name, &block) { block() }
so you write
test 'first label', { # first block of tests here }
test 'second label', { # second block of tests here }
sirrobert hmmmm good =) 16:10
moritz and then you can make sub test() more sophisticated, to only call tests whose blocks are in @*ARGS
and call them all if @*ARGS is empty
sirrobert nice
I'll slip that into my to-do =)
my current test framework lets me do API tests this way: 16:11
moritz sub test($name, &block) { block() if !@*ARGS || $name eq any @*ARGS }
[Coke] you could just steal "fudge".
sirrobert $suite.test: 'description', GET, $url, %query-params, %body-params;
lumi__ You also need a count for skipping, I think
16:11 sjohnson left
sirrobert and a separate file handles the expected results so my test file is clean 16:11
moritz lumi__: unless you use planless testing
jnthn r: :(::T $x) 16:12
16:12 sjohnson joined
p6eval rakudo c2005e: ( no output ) 16:12
moritz r: :(::T $x); say $x.WHAT
p6eval rakudo c2005e: OUTPUT«===SORRY!===␤Variable $x is not declared␤at /tmp/1oIy7PsXFp:1␤»
moritz r: my :(::T $x); say $x.WHAT
p6eval rakudo c2005e: OUTPUT«===SORRY!===␤Malformed my␤at /tmp/IsFMGbqDMR:1␤»
sirrobert What is ::T? 16:13
16:13 leont joined
skids Type capture en passant. 16:13
lumi__ r: my ::T $x;
p6eval rakudo c2005e: ( no output )
jnthn Type capture
It used to explode when you wrote the above. Now it doesn't.
jnthn is doing some RTs :)
sirrobert I don't understand the response =)
moritz r: my ::T $x = 24; say $x
p6eval rakudo c2005e: OUTPUT«Cannot assign to a readonly variable or a value␤ in block at /tmp/h4FK6PMgL1:1␤␤»
sirrobert ohhh
lumi__ r: my ::T $x; say $x.WHAT
p6eval rakudo c2005e: OUTPUT«Cannot call a method on type variable T␤ in any find_method at src/gen/Metamodel.pm:295␤ in sub say at src/gen/CORE.setting:7268␤ in block at /tmp/NoSNyivfRM:1␤␤»
jnthn I don't think it's very useful in a my :)
r: say \(1, 1) ~~ :(::T, T) 16:14
p6eval rakudo c2005e: OUTPUT«True␤»
jnthn r: say \(1, 'lol') ~~ :(::T, T)
p6eval rakudo c2005e: OUTPUT«False␤»
jnthn Even works :)
16:14 leont left
dalek p: 3cfcb96 | jnthn++ | src/QRegex/P6Regex/Grammar.nqp:
Detect invalid internal regex mods.

Includes the case where the invalid mod starts with the name of a recognized one.		 16:16
kudo/nom: aeb4515 | jnthn++ | tools/build/NQP_REVISION:
Bump NQP_REVISION to get regex modifier fix.
 16:18
kudo/nom: 7731d28 | jnthn++ | src/core/MapIter.pm:
Avoid an unbox-as-Int failure in MapIter.

Note that this may not be quite the right semantics, but it's better than exploding over the eventual unbox_i of Inf we can otherwise get.
16:18 pmurias left 16:19 pochi left
lumi__ I've uploaded github.com/lumimies/override-perl6 - it does that whole "No such method 'SET_FILE_LINE'" when the exception is thrown 16:19
Oh, which is a bad thing
16:21 Hor|zon joined
jnthn moritz: ^^ may be one you've an answer for :) 16:22
sjn \o 16:25
sjn is at Hackeriet (the Oslo hackerspace) talking about Perl6 o/
got a question here... what kind of parsing style does rakudo use? (recursive decent?) 16:26
sirrobert =)
jnthn recursive indecent
sjn :)
jnthn It's a kind of PEG 16:27
16:27 rogerrogerstein joined
TimToady nr: sub foo { [+] gather '' ~~ / ^ { take 1..4; return 0 } / }; say foo 16:27
p6eval niecza v22-13-g0d2a99e: OUTPUT«10␤»
..rakudo c2005e: OUTPUT«0␤»
TimToady I think this is a nieczabug
sjn jnthn: kind of? 16:28
jnthn It's recursive descent but alternations are decided by transitive LTM, meaning that unlike typical recdescent it won't have to try anything like as many branches.
And it also has a bottom-up expression parser too
16:33 spider-mario joined, kaleem left
Hor|zon Question: am I the only one having really confusing conversations about perl6 roles because of conflation of interface and mixin? 16:35
it seems to me that you need some experience of both concepts separately to make _sensible_ use of roles 16:36
sirrobert Hor|zon: my conversations about roles have worked well when I say: "composable class elements" 16:37
then just emphasize "composable" for a while.
TimToady added a validating version to rosettacode.org/wiki/Roman_numerals...de#Perl_6, but it runs into the nieczabug above
16:38 dakkar left
Hor|zon sirrobert: lol, I was trying to explain things to a new developer, and I realised that there was no way he wouldn't create a mess, because roles themselves come without handle bars 16:40
TimToady Hor|zon: that's why we picked the word "role", because it's a concept that doesn't commit to one or the other. If you use a role from a class, it's interface; if you use it to wrap an object in a new type, it's a mixin.
Hor|zon you usually want interface type roles, or mixin type roles, and then sometimes a *Tiny* bit of the other, but the grey area is huge
jnthn Most of my roles aren't interface-type or mixin-type these days. 16:41
doy mine either
Hor|zon jonathan-type roles? 16:42
TimToady well, they're also generics, generally speaking
as such they can suggest implementation, which interfaces don't
you can also think of them as closed classes :) 16:43
It would seem that roles have many roles. :D
jnthn Hor|zon: To me, roles are primarily about factoring out common concerns that can be composed into many classes. A class should have a responsibility and completely fulfil it. A role is free of the need to have a responsibility, and can focus on being a reusable thingy. 16:44
TimToady Hamlet is a role, but it has to be instantiated by an actor, the class. 16:45
jnthn Example from Rakudo implementation: grammars, roles and classes can all have methods, so the functionality related to having methods is factored out into a role.
TimToady We all know that Hamlet is irresponsible. :)
16:45 Pleiades` left
jnthn A "method haver" is useless on its own, but the concept and its implementation is reusable. 16:45
sirrobert jnthn: Whereas a superclass is a hierarchical extraction of duplicated functionality, a role is a non-hierarchical extraction.
masak TimToady: agree on the Nieczabug. I'll submit it. 16:46
sirrobert jnthn: (in support of what you were saying =)
Hor|zon yeah, it's definitely a better concept, but I'm starting to think having some type of mixin and interface syntax/module (implemented on top of roles), which makes it obvious would be nice to have, it could just be another module
that way when you see a role explicitly, you know it's grey area code, but when it's only a mixin or only an interface it's also obvious 16:47
masak TimToady: github.com/sorear/niecza/issues/153 16:48
TimToady masak++
doy that just always feels like an artificial limitation to me - what if you have an interface, and then decide it would be useful to provide a default implmentation of one of the methods (because it ends up almost always doing the same thing)? 16:49
Hor|zon I'm just finding that I end up writing in comments "this is meant as an interface" with my roles
if I could just alias "role" to "interface", I could get rid of the stupid comment
TimToady well, if all your methods are stubs, it's an interface
16:51 sjohnson left
Hor|zon it all depends on your intentions, present and future, I'm just saying I could make it very clear from the outset using a different keyword 16:51
16:51 sjohnson joined, sjohnson left
TimToady compared to p5, we do give more names to things in p6, but naming a degenerate case seems to me to not all that terribly useful 16:51
*not be
16:52 Pleiades` joined
TimToady if it's useful in your situation, we'll probably have macros to give aliases to things like that 16:52
Hor|zon I agree, not terribly useful, so maybe not core, but as an optional extra
ah yes, macros
cool
TimToady aliases are a good (or at least un-bad) spot to use single-word textual macros, I suspect 16:53
jnthn had pondered making installing any name in EXPORTHOW lead to the creation of a package_declarator that parses that name and uses the appropraite meta-object. 16:54
16:54 sjohnson joined, sjohnson left, sjohnson joined
jnthn Then you could also do the meta-programming to make sure that there was no way to supply anything in an interface besides a stub. 16:54
well, you can do that bit today
but the extra names bit you can't
it may be too magical though ;) 16:55
TimToady and may fall out of exportable macros anyway
Hor|zon too magical? ;)
TimToady or more generally, exportable grammar
jnthn TimToady: You still need to get the right entires in %*HOW 16:56
(that is, we could make you need to explicitly do the syntax tweak but it'd be additional)
TimToady sure, that's more than an alias will do, but grammars are a bit more powerful than that
jnthn TimToady: I guess in this I'm assuming that the EXPORTHOW package will stay around as the way to do these things... :) 16:57
16:57 replore left
TimToady "use Python;" will have to do even more work than that :) 16:57
jnthn Oh, I had a question about STD...
std: { sub infix:<lolz>($, $) { }; a lolz b; }; a lolz b
p6eval std 77327a4: OUTPUT«===SORRY!===␤Undeclared routines:␤ 'a' used at line 1,1␤ 'b' used at line 1,1␤ 'lolz' used at line 1,1␤Check failed␤FAILED 00:00 44m␤»
jnthn er 16:58
std: { sub infix:<lolz>($, $) { }; 1 lolz 2; }; 1 lolz 2
p6eval std 77327a4: OUTPUT«===SORRY!===␤Two terms in a row at /tmp/1LajtzUe_o line 1:␤------> infix:<lolz>($, $) { }; 1 lolz 2; }; 1 ⏏lolz 2␤ expecting any of:␤ infix or meta-infix␤ infixed function␤ statement modifier loop␤Parse failed␤FAILED 00:00
..46m…
jnthn That's what I wanted
So, the grammar tweaks are lexically scoped. In Rakudo that currently isn't true
r: { sub infix:<lolz>($, $) { }; 1 lolz 2; }; 1 lolz 2
p6eval rakudo c2005e: OUTPUT«===SORRY!===␤CHECK FAILED:␤Undefined routine '&infix:<lolz>' called (line 1)␤»
16:58 birdwindupbird left
jnthn I mean, it gets it but not as a syntax problem like it should. 16:58
TimToady so what's the question?
jnthn I was looking at how STD makes this work and got...confused.
I saw that it twiddles what's in %*LANG<MAIN> 16:59
Also that we immediately mix into the current cursor
But how does that lead to the cursor we're parsing the statements with in the outer scope work?
TimToady we throw away that cursor at the end of the current lexical scope 17:00
17:00 erkan joined, erkan left, erkan joined, vlixes joined
jnthn But there's more than one cursor object created at this point, at least in the way QRegex has it. 17:00
TimToady that's how STD does all of its scoping and backtracking; the cursor carries everything, so you merely throw away cursors
jnthn OK, but when you hit statementlist, that makes a cursor 17:01
Calling statement creates another one, etc.
When one statement is a declaration that installs a categorical, it can mix into the current cursor, but that's the cursor for, say, routine_def
Now, I can see how everything we parse inside of the routine sees the derived grammar 17:02
TimToady see line 1325
jnthn looking
TimToady each statement starts in the current language in MAIN
so it comes down to dynamic scoping at that point
jnthn Oooooh!
hm, and I can't find where $*LASTSTATE is declared? 17:03
Or what it's about 17:04
The bless makes sense to me and answers my question, though.
jnthn guesses it's the bless that really matters here
TimToady there's a 'local $::LASTSTATE;' in CursorBase 17:05
jnthn Is there a reason it's <!!{ $*LASTSTATE = $¢.pos; $¢ = %*LANG<MAIN>.bless($¢); }> rather than just { $*LASTSTATE = $¢.pos; $¢ = %*LANG<MAIN>.bless($¢); }
LTM related?
TimToady yes
<!...> is transparent to LTM
<?...> isn't
jnthn *nod* 17:06
TimToady nor is {}
jnthn I'm pretty sure we have it that way already
jnthn tries to work out what $*LASTSTATE is actually for
TimToady so I guess it's an oversimplification to say that STD does all its scoping by throwing away cursors, though that's essentially what the dynamic scoping is managing
jnthn yeah, I'd been looking in cursor_fresh and all that lot for some odd magic :) 17:07
well, and mixin
I'm relieved that what's going on is basically sane :)
LASTSTATE looks suspiciously fossily to me
TimToady ./STD.pm5:$SIG{__WARN__} = sub { die @_," statement started at line ", 'Cursor'->lineof($::LASTSTATE), " 17:08
jnthn Oh
OK, I guess it's not important for what I'm doing here
OK, thanks, that clears things up nicely for me :) 17:09
TimToady it's an old debugging aid
though not necessarily obsolete, if I stick a "warn" in somewhere
I should probably take it out of STD
haven't used it in years 17:10
[Coke] jnthn: any ideas on github.com/perl6/nqp/issues/59 ? 17:11
TimToady jnthn: the 'bless' is part of the magic; the important thing to remember is that we never actually return a cursor up the tree (except to cheat); all info that travels up the syntax tree is (at least notionally) copied out of a dying cursor into (a clone of) the outer cursor that initiated the subparse. 17:13
jnthn takes it as license to ignore it for now
[Coke]: looking
TimToady: Yes, the "never return a cursor up the tree" thing was what was bothering me about how on earth STD was doing it :) 17:14
n: class EEK {method Numeric() {return 3}}; my $a:= EEK.new(); say(+$a); 17:15
p6eval niecza v22-13-g0d2a99e: OUTPUT«3␤»
jnthn grr
nqp: class EEK {method Numeric() {return 3}}; my $a:= EEK.new(); say(+$a);
p6eval nqp: OUTPUT«No applicable candidates found to dispatch to for 'Numeric'. Available candidates are:␤␤current instr.: 'Numeric' pc 3370 (src/stage2/gen/NQPCORE.setting.pir:1584) (src/stage2/NQPCORE.setting:171)␤»
jnthn nqp: class EEK {multi method Numeric() {return 3}}; my $a:= EEK.new(); say(+$a);
p6eval nqp: OUTPUT«No applicable candidates found to dispatch to for 'Numeric'. Available candidates are:␤␤current instr.: 'Numeric' pc 3370 (src/stage2/gen/NQPCORE.setting.pir:1584) (src/stage2/NQPCORE.setting:171)␤»
jnthn [Coke]: I have to wonder if it's not something to do with Parrot vtable oddness 17:16
TimToady yes, it's the fact that ; can behave as a language change, such that everything following the ; is logically a subparse of what came before
semicolons are sort of a legal fiction of "all statements are created equal"
jnthn nqp: class EEK {method Numeric() is parrot_vtable('get_integer') {return 3}}; my $a:= EEK.new(); say(+$a);
p6eval nqp: OUTPUT«No applicable candidates found to dispatch to for 'Numeric'. Available candidates are:␤␤current instr.: 'Numeric' pc 3370 (src/stage2/gen/NQPCORE.setting.pir:1584) (src/stage2/NQPCORE.setting:171)␤»
TimToady but declarations are more equal than others
jnthn nqp: class EEK {method Numeric() is parrot_vtable('get_number') {return 3}}; my $a:= EEK.new(); say(+$a);
p6eval nqp: OUTPUT«3␤»
jnthn [Coke]: Workaround is ^^ 17:17
jnthn gets a fix for ignorecase literals in the LTMer 17:18
tadzik .u ｣ 17:20
phenny U+FF63 HALFWIDTH RIGHT CORNER BRACKET (｣)
tadzik Perl 6, exercising your fonts since 2001, or something
jnthn tadzik: I did a Pod arsing patch earlier that fixed most of the test fails of Pod parsing on Windows.
tadzik: Not everything passes yet but it's closer at least. 17:21
17:21 kurahaupo joined
jnthn Feel free to review it to make sure it's not insane. 17:21
oh, *parsing. :)
Hor|zon lol, the jury is still out on which is truer
:P
jnthn :D 17:22
dalek p: 3629cd6 | jnthn++ | src/ (2 files):
Handle case-insensitive literals in LTM.
tadzik :) 17:23
jnthn: I suppose it's alright :)
dalek kudo/nom: 90fad9b | jnthn++ | tools/build/NQP_REVISION:
NQP_REVISION bump for :i/LTM fix.
 17:24
masak jnthn: it totally should have been called "lexing and arsing". :P 17:25
jnthn
.oO( Butt all I did was a typo... )
 17:27
Hor|zon github needs perl6 highlighting
17:27 mucker left
Hor|zon wonder how to go about it... 17:27
masak Perl 6 highlighting is Perl 6-complete.
17:28 sudokode left
skids
.oO(but then instead of "bison" we'd have "arsen")
 17:28
jnthn If it manages a better approximation than a Perl 5 syntax highlighter it's a step forward :)
Hor|zon so, this would be the project to patch github.com/github/linguist 17:30
ugh ruby...
TimToady just write a ruby backend for p6 :) 17:31
masak How Difficult Can It Be?(TM) 17:33
TimToady I see you stuck to ASCII there...
17:33 sudokode joined
skids I wonder whether a completely heuristic-based generic syntax highlighting algo could yield "good enough" results across a wide variety of languages. 17:34
masak TimToady: ™
TimToady skids: not without some kind of "am I getting warmer?" oracle 17:35
masak au++ # App::Uni
Hor|zon on files of a certain length I'm sure it's good enough
masak would like a generic "am I getting warmer?" oracle
skids Well, using existing highlighters and codebases, one could train it to get close to those results. 17:36
TimToady a heuristic highlighter will likely have a terrible time with Perl's term vs operator distinction
skids How critical is getting that particular bit of highlighting correct? 17:37
TimToady it's one thing to hide a namespace behind a sigil, it's another to hide a namespace behind a syntax node
do you want to use a variable named 'x'? 17:38
or 'gt', or 'cmp'?
skids Well, the question is would I be thrown off horribly if x wasn't blue.
TimToady and is / a division or the start of a regex? 17:39
jnthn well, it's promotion too
er, ww
TimToady do you want a regex like / x y z / to highlight the x as if it were an infix?
Hor|zon I don't think there's enough Perl 6 code in existence to train anything well enough either way 17:41
TimToady just saying a purely heuristic algo is unlikly to figure this out; it's okay if a highlighter makes a few mistakes, but I doubt we can have one of those currently without giving it some strong hints about "warmer"
sirrobert ooc: highlighter for what context? 17:43
TimToady imagines someone patient enough to learn Perl 6 merely by being rapped on the knuckles every time there's a syntax error (without seeing the message)
sirrobert: github, I believe
sirrobert TimToady: ah
17:44 erkan left 17:45 qazplm joined
skids In an editor context, "warmer" feedback would be easier to implement. 17:45
TimToady well, more likely you'd get "cooler" feedback, like a "Guess Again" button the user can hit 17:46
17:47 fglock joined
TimToady or a right-click menu item you can put on the offending passage 17:47
skids Of course, introducing the human means the feedback could quite often be entirely wrong. 17:48
TimToady otoh, you might end up with users trying to teach the parser how to parse syntactically incorrect code, which would also be madness
I think we're in agreement there... :)
arnsholt skids: Sounds like you want some kind of Markov model highlighterh (re: heuristic highlighter) =)
sirrobert "Then I applied myself to the understanding of wisdom, and also of madness and folly, but I learned that this, too, is a chasing after the wind."
17:49 kurahaupo left
TimToady but a cord of three strands is not easily broken 17:49
sirrobert huh, segfault. I don't get those often 17:50
Hor|zon seems linguist actually uses pygments for highlighting, which is python, and allows custom lexers, hoorah
17:50 Chillance joined
TimToady "a time to search, and a time to give up for lost" There's parsing in a nutshell... 17:50
Hor|zon ( noruby moresanity )++ 17:52
17:52 hoelzro|away is now known as hoelzro 17:59 mucker joined 18:02 fhelmberger left 18:04 GlitchMr left
TimToady felher: note that <{ EXPR }> will assume that you're trying to compiles a regex, so it won't work with a literal string that has metachars, if you just want string matching 18:12
use $(EXPR) or "{EXPR}" to get string matching, depending on what rakudo implements
at worst, you'll have to use a temp var
nr: my $_ = "foo bar"; say m/ <{ $_ }> / 18:14
p6eval niecza v22-13-g0d2a99e: OUTPUT«Potential difficulties:␤ Useless redeclaration of variable $_ (see line 0) at /tmp/5L1LqSVnZV line 1:␤------> my $_ ⏏= "foo bar"; say m/ <{ $_ }> /␤␤Match()␤»
..rakudo 7731d2: OUTPUT«===SORRY!===␤Redeclaration of symbol $_␤at /tmp/p8_BfrCztk:1␤»
TimToady nr: $_ = "foo bar"; say m/ <{ $_ }> /
p6eval niecza v22-13-g0d2a99e: OUTPUT«Match()␤»
..rakudo 7731d2: OUTPUT«#<failed match>␤»
TimToady nr: $_ = "foo bar"; say m/ "{ $_ }" /
p6eval niecza v22-13-g0d2a99e: OUTPUT«#<match from(0) to(7) text(foo bar) pos([].list) named({}.hash)>␤» 18:15
..rakudo 7731d2: OUTPUT«Null PMC access in find_method('Stringy')␤ in regex at /tmp/8YLkZPevlM:1␤ in method match at src/gen/CORE.setting:4105␤ in block at /tmp/8YLkZPevlM:1␤␤»
TimToady ooh, rakudobug
nr: $_ = "foo bar"; say m/ $( $_ ) /
p6eval niecza v22-13-g0d2a99e: OUTPUT«#<match from(0) to(7) text(foo bar) pos([].list) named({}.hash)>␤»
..rakudo 7731d2: OUTPUT«#<failed match>␤»
TimToady and another :)
felher: looks like your best bet is a temp scalar for now 18:16
nr: my $x = "foo bar"; say m/ "{ $x }" / 18:17
p6eval niecza v22-13-g0d2a99e: OUTPUT«Unhandled exception: Unable to resolve method match in type Any␤ at /tmp/MNJtdv2eHy line 1 (mainline @ 4) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4212 (ANON @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4213 (module-CORE @ 579) ␤ at /home/…
..rakudo 90fad9: OUTPUT«No such method 'match' for invocant of type 'Any'␤ in block at /tmp/JqBUydpkuI:1␤␤»
masak submits rakudobug 18:18
TimToady nr: my $x = "foo bar"; $_ = $x; say m/ "{ $x }" /
p6eval niecza v22-13-g0d2a99e: OUTPUT«#<match from(0) to(7) text(foo bar) pos([].list) named({}.hash)>␤»
..rakudo 90fad9: OUTPUT«｢foo bar｣␤␤»
TimToady looks like m is clobbering $_
felher TimToady: well, "{ $x.y }" seems to work for my purposes here.
TimToady: thanks! I can't believe i didn't try "{ }" :)
TimToady++
TimToady yes, the problem seems to be losing $_
is probably desugaring m// to $_ ~~ // or some such 18:19
except you'd think that'd still work
nr: my $x = "foo bar"; $_ = $x; say m/ $( $x ) /
p6eval niecza v22-13-g0d2a99e: OUTPUT«#<match from(0) to(7) text(foo bar) pos([].list) named({}.hash)>␤» 18:20
..rakudo 90fad9: OUTPUT«#<failed match>␤»
TimToady that one seems like an unrelated bug
probably not parsing $() as a variable there
masak nod 18:21
TimToady nr: my $x = "foo bar"; $_ = $x; say m/ $( 1 2 ) /
masak submits rakudobug
p6eval niecza v22-13-g0d2a99e: OUTPUT«===SORRY!===␤␤Unable to parse argument list at /tmp/3VwM6MRlmY line 1:␤------> my $x = "foo bar"; $_ = $x; say m/ $(⏏ 1 2 ) /␤Couldn't find final ')'; gave up at /tmp/3VwM6MRlmY line 1:␤------> my $x = "foo bar"; $…
..rakudo 90fad9: OUTPUT«#<failed match>␤»
TimToady yes, it thinks it's just $ and () separatelhy
sirrobert All tests successful.
=)
TimToady no more challenges in life, eh? 18:22
sirrobert Challenge Accepted. ;)
18:25 mucker left, mucker joined 18:27 erkan joined, erkan left, erkan joined
sergot hi o/ 18:31
masak sergocie! \o/
tadzik hello :) 18:32
sergot \o/ I'm back. :)
masak welcome back \o/
masak .oO( I again come to here )
TimToady \.\o\O(evolution) 18:33
\.\o\O\.(revolution) 18:35
18:38 vlixes left 18:47 daxim left 18:51 PacoAir left 18:52 PacoAir joined
dalek d: 8a3fdfd | larry++ | / (3 files):
remove LASTSTATE hack
 18:53
d: a426055 | larry++ | boot/ (2 files):
rebootstrap
 18:56
19:16 fuleo2 left 19:24 benabik left 19:25 benabik joined 19:26 PacoAir left, PacoAir joined 19:32 birdwindupbird joined
dalek p: a98083d | jnthn++ | src/QAST/Operations.nqp:
Map nqp::rebless => pir::change_type.
 19:34
p: 5fab963 | jnthn++ | src/ops/nqp.ops:
Ensure reblessing to the same type is cheap.
jnthn star: { use Grammar::Tracer; grammar Foo { token TOP { ^ abc $ } } }; grammar Bar { token TOP { ^ abc $ } }; Foo.parse('abc'); Bar.parse('abc'); 19:38
p6eval star 2012.07: OUTPUT«TOP␤* [37;42mMATCH "abc"␤TOP␤* [37;42mMATCH "abc"␤»
19:38 daxim joined 19:42 SamuraiJack_ left
dalek kudo/nom: 670363b | jnthn++ | tools/build/NQP_REVISION:
Bump NQP_REVISION to get nqp::rebless.
 19:43
kudo/nom: 0d7f83c | jnthn++ | src/Perl6/ (2 files):
Use nqp::rebless in a couple of places.
kudo/nom: dbd8130 | jnthn++ | src/Perl6/Grammar.pm:
Localize %*LANG and %*HOW in statementlist.

This means that custom meta-objects really do only apply in the lexical scope that they are imported into.
rakudo/nom: 919d967 | jnthn++ | src/Perl6/Grammar.pm:
rakudo/nom: Apply any %*LANG<MAIN> tweaks to statements.
19:48 orafu joined 19:49 supernovus joined
supernovus Is there any interest in an indexOf type method being added to Array ? Concept @c.index-of($a); # if $a is found in @c, its index position is returned. Otherwise False is returned. 19:51
tadzik we have is-prime, I don't see why we wouldn't want index-of :>
FROGGS ya, I use that often in PHP and always missed it in Perl 19:52
sirrobert supernovus: I'd use it.
19:52 fglock left
daxim its doc should come with an admonishment to use a hash instead 19:53
19:53 mucker left
sirrobert hashes are interesting because they're less flexible than arrays of pairs 19:54
19:54 fgomez left
sirrobert I recently had a good use case for an array of pairs, but a hash wouldn't work. 19:55
(though I would have liked to use a hash =)
flussence well there's .first, maybe .index-of should be called .first-index
supernovus daxim: there are times when it is very useful to be looking through an Array for a certain object. In the current case I have come across this, the nodes of an XML element are stored in @.nodes. I want to be able to do a $xml.insert-before($existing-node, $new-node); without having to know the position of $existing-node before hand. For this case I've written my own lookup function, but this is a common enough feature, it's in Jav 19:56
ascript 1.5, PHP, etc.
sirrobert flussence: makes sense to me
19:56 fgomez joined
diakopter which type of equals would index-of use? 19:57
sirrobert eqv
?
supernovus What is the difference between eqv and === ? My current code is using === but the basic idea is to find the same object. 19:58
sirrobert r: {a => 1, b=>{c=>2}} === {a => 1, b=>{c=>2}} 19:59
p6eval rakudo 90fad9: ( no output )
sirrobert r: {a => 1, b=>{c=>2}} eqv {a => 1, b=>{c=>2}}
p6eval rakudo 90fad9: ( no output )
sirrobert r: say {a => 1, b=>{c=>2}} === {a => 1, b=>{c=>2}}
p6eval rakudo 90fad9: OUTPUT«False␤»
sirrobert r: say {a => 1, b=>{c=>2}} eqv {a => 1, b=>{c=>2}}
p6eval rakudo 90fad9: OUTPUT«True␤»
diakopter may e index-of could optionally take a comparator WhateverCode 20:00
TimToady maybe it could just take a smartmatch like it oughta
then you can say * === or * eqv if you wanna
sirrobert TimToady: how would that work for this... 20:01
r: say {a => 1, b=>{c=>2}} ~~ {a => 1, b=>{c=>2}}
p6eval rakudo 90fad9: OUTPUT«False␤»
sirrobert syntactically, I mean
dalek d: cda80a9 | jnthn++ | CursorBase.pmc:
Couple of tabs => spaces, to deconfuse the reader.
TimToady r: say {a => 1, b=>{c=>2}} ~~ (* eqv {a => 1, b=>{c=>2}}) 20:02
p6eval rakudo 90fad9: OUTPUT«True␤»
sirrobert I didn't know you could do that
cool
TimToady r: say {a => 1, b=>{c=>2}} ~~ (* === {a => 1, b=>{c=>2}})
p6eval rakudo 90fad9: OUTPUT«False␤»
sirrobert so it's comparing the first hash to an expression of equivalence? 20:03
rather, applying it to
TimToady it's right there at S03:3593 20:04
sirrobert thanks (I consume the specs in pieces sorted by relevance and filtered by awareness =) 20:05
diakopter i eat the S like a cow 20:07
supernovus Okay, that's pretty cool. Using first-index($a); will use default smart matcher semantics, and first-index((* === $a)); will use === semantics (via the smart matcher operator inside first-index()) 20:08
skids wonders what's so horrible about @a.pairs.grep({ $_.value op $thingy })>>.key[0];
TimToady why do all that extra work when .first().key will work 20:09
sirrobert skids: xkcd.com/378/
supernovus skids: well for one, compared to @a.first-index($a); that's damn ugly and more boiler plate than should be needed for a simple operation. 20:10
TimToady @a.pairs.first($a).key is not much longer
skids Well, how ugly is this going to be after allowing someone to express which op to use? 20:11
sirrobert the cognitive load is a fair bit higher =)
20:11 fgomez left
TimToady in fact, the .pairs.first is exactly same length as .first-index, so the only difference is the .key 20:12
supernovus skids: @a.first-index((* === $a));
sirrobert supernovus: are the inside () needed? =)
would @a.first-index(*===$a) work?
TimToady well, and the need to do *.value in there instead of *
supernovus sirrobert: nope, I just tried it without them and it works fine
TimToady might be worth it to save both a .key and a .value 20:13
sirrobert supernovus: nice
diakopter which equals does .first use
supernovus @a.first-index(* === $a) will work fine
TimToady diakopter: if it doesn't use smartmatch, it's a bug
sirrobert TimToady: why use a *.value?
TimToady: what does that mean in array context? 20:14
TimToady because a .pair won't === its value
[Coke] jnthn++ #workarounds
supernovus sirrobert: for the .pair usage you need *.value because * is referring to the Pair object itself.
sirrobert ah, got it... in the case of an array of pairs 20:15
TimToady list of pairs, but yeah
supernovus I'd like to implement first-index() using a plain old loop() and see if there is any speed difference between that technique and the .pairs.first technique.
skids
.oO(we should decide now how to punish the first person to ask for ".second")
TimToady right now loop will tend to beat anything else in speed, because our (non-)loop optimizations suck 20:16
no, no, we punish the second person to ask for .second 20:17
diakopter heh 20:18
supernovus TimToady: Ah, as my library-specific version is using a loop() and just returning on the first successful match, which should be fairly fast.
jnthn How does it know what to loop up to?
sirrobert @a.nth(3, $a); ;)
dalek kudo/nom: 14e575b | jnthn++ | docs/ChangeLog:
Update ChangeLog.
jnthn (If the answer is ".elems" or something, then it won't work out well with laziness) 20:19
TimToady .[0] // last works pretty well, last I checked
supernovus jnthn: on my library specific implementation it looks like: loop (my $i=0; $i < @.nodes.elems; $i++) 20:20
TimToady 'course, if the list contains undefs, that wouldn't be exactly right...
jnthn supernovus: Yeah, we shouldn't put something like that into CORE
20:20 PacoAir left
jnthn supernovus: 'cus the moment it hits a lazy list it explodes 20:20
Also because we can use a native int
TimToady you'd want something more like .gimme(1)
supernovus jnthn: good point. I hadn't thought of lazy lists in this use case. My current implementation is better left to the library specific use. 20:21
20:21 grondilu joined
jnthn supernovus: Well, if your use case won't have infinite lists you're probably just fine :) 20:22
TimToady of course, if you know the list is already reified, that's the fastest way to get through it
or the fastest way through the reified bits, if partially reified 20:23
so maybe we want something like .elems that tells us how much is reified, so we can just batch that much o fit
of it, even 20:24
supernovus Well, I guess @c.pairs.first(*.value === $b).key; is the best way to do it currently as a generic feature. For my library, I'll stick with the specialized version, as the number of elements in @.nodes is known up front (and if an XML document has an infinite number of nodes in it, there's a bigger problem than my library...)
20:24 grondilu left
TimToady you'd like the generic form to optimize itself into the fast form whenever possible 20:24
jnthn TimToady: I think Rakudo's list impl may well be doing something like that already. It certainly has a reifified part and gimme will return right away when it can. 20:25
20:27 birdwindupbird left 20:29 birdwindupbird joined
diakopter r: say $_!X::a 20:32
p6eval rakudo 919d96: OUTPUT«===SORRY!===␤Method 'is_trusted' not found for invocant of class 'Perl6::Metamodel::PackageHOW'␤»
jnthn hm, a bit lta. 20:33
A package can't have methods, so it can't trust, so a private method call on one makes no sense. 20:34
20:37 FROGGS left 20:38 flightrecorder left, kurahaupo joined
masak submits rakudobug 20:43
good night, #perl6 20:45
TimToady o/
sirrobert wave
supernovus good night masak!
jnthn 'night, masak 20:46
[Coke] can someone comment on whether rt.perl.org/rt3/Ticket/Display.html?id=74946 is an actual bug or a misunderstanding?
dalek kudo/nom: 2fca74a | jnthn++ | src/Perl6/Actions.pm:
Don't explode over missing is_trusted; diakopter++.
20:47 _edwin joined
sirrobert what is this doing: 20:47
r: "foo" ~~~~~~~~ /f/
p6eval rakudo 919d96: ( no output )
sirrobert r: say "foo" ~~~~~~~~ /f/
jnthn Waving.
p6eval rakudo 919d96: OUTPUT«False␤»
sirrobert heh
jnthn std: "foo" ~~~~~~~~ /f/
p6eval std cda80a9: OUTPUT«===SORRY!===␤Expecting a term, but found either infix ~~ or redundant prefix ~␤ (to suppress this message, please use space between ~ ~) at /tmp/Q3s3c_LNQe line 1:␤------> "foo" ~~~~⏏~~~~ /f/␤Parse failed␤FAILED 00:00 42m␤»…
sirrobert r: say "foo" ~~~~~ /f/
p6eval rakudo 919d96: OUTPUT«False␤»
sirrobert r: say "foo" ~~~~~~~~~~~~~~~~~~~~~~~~~~ /f/
p6eval rakudo 919d96: OUTPUT«False␤» 20:48
jnthn It's parsed as
_edwin hi perl6ers!
jnthn "foo" ~~ (~ ~ ~ ~ ~ ~ /f/)
I suspect
hi _edwin
sirrobert what is a single tilde?
_edwin rn: class Foo { has @.bytes of int8; }
sirrobert hi _edwin
p6eval niecza v22-13-g0d2a99e: OUTPUT«===SORRY!===␤␤Two terms in a row at /tmp/oh8SjnnfK2 line 1:␤------> class Foo { has @.bytes ⏏of int8; }␤␤Parse failed␤␤»
..rakudo 919d96: OUTPUT«===SORRY!===␤Cannot call 'trait_mod:<of>'; none of these signatures match:␤:(Routine:D $target, Mu:U $type)␤:(Mu:U $target, Mu:U $type)␤␤at /tmp/FGp09pW0sv:1␤»
jnthn sirrobert: stringify
sirrobert jnthn: heh oh yeah
_edwin hm, locally I get Null PMC access from rakudo
jnthn _edwin: We don't provide int8 yet
_edwin rn: my @bytes of int8; 20:49
p6eval rakudo 919d96: ( no output )
..niecza v22-13-g0d2a99e: OUTPUT«===SORRY!===␤␤Two terms in a row at /tmp/LnHED8x7E_ line 1:␤------> my @bytes ⏏of int8;␤␤Parse failed␤␤»
TimToady rn: my @bytes is Array of int8;
p6eval niecza v22-13-g0d2a99e: OUTPUT«===SORRY!===␤␤Trait name not available on variables at /tmp/qvPQvxuVOK line 1:␤------> my @bytes is Array ⏏of int8;␤␤Two terms in a row at /tmp/qvPQvxuVOK line 1:␤------> my @bytes is Array ⏏of int8;␤…
..rakudo 919d96: ( no output )
_edwin jnthn, in attributes, or generally? 20:50
moritz generally
TimToady 'of' only modifies a type on the left
iirc
jnthn [Coke]: I think the ticket is right, off hand.
TimToady std: my $foo of int8; 20:51
p6eval std cda80a9: OUTPUT«ok 00:00 42m␤»
TimToady I guess STD doesn't mind it
_edwin I thought in case of @ the default type is Array and that would be modified by the "of"
jnthn I think Rakudo doesn't mind it syntactically, it just hasn't much idea about variable traits.
_edwin: No, the "of" would be modifying what the array contains. 20:52
TimToady nr: my @array of Int;
p6eval niecza v22-13-g0d2a99e: OUTPUT«===SORRY!===␤␤Trait of not available on variables at /tmp/ryaFM7fZxy line 1:␤------> my @array of Int⏏;␤␤Potential difficulties:␤ @array is declared but not used at /tmp/ryaFM7fZxy line 1:␤------> my ⏏@a…
..rakudo 919d96: ( no output )
TimToady should mean the same as my Int @array, or my @array returns Int 20:53
_edwin rn: my Int @array;
p6eval rakudo 919d96: ( no output )
..niecza v22-13-g0d2a99e: OUTPUT«Potential difficulties:␤ @array is declared but not used at /tmp/BeRH8BSe4e line 1:␤------> my Int ⏏@array;␤␤»
TimToady or my @array[* --> Int]
(maybe)
jnthn (maybe) ;)
Agree with the others though :)
_edwin ok, I can live with my Int @array for the time being, thanks 20:54
but I'd like to reproduce that Null PMC access... 20:55
TimToady eval my %a = ('a',1,'b',2); $a{a} = delete $a{a}; print join ' ', %a
buubot_backup TimToady: a 1 b 21
TimToady [Coke]: perl 5 agrees 20:56
_edwin seems to happen only in the REPL
jnthn _edwin: If you can make it consistently explode in the repl, please file it in RT.
20:57 hash_table left
[Coke] TimToady: why is it "b 21" ? 20:57
TimToady it's b 2
1 is the final return value
eval 42
buubot_backup TimToady: 42
TimToady eval print "42" 20:58
buubot_backup TimToady: 421
[Coke] danke.
TimToady not terribly well distinguished, admittedly
20:58 aindilis left
TimToady eval say 42 20:59
buubot_backup TimToady: ERROR: syntax error at (eval 20) line 1, near "say 42"
TimToady buu, hiss!
sorear eval $[
buubot_backup sorear: ERROR: Can't locate arybase.pm in @INC (@INC contains:) at (eval 20) line 2. BEGIN failed--compilation aborted at (eval 20) line 2.
sorear eval $]
buubot_backup sorear: 5.016000
sorear eval use 5.016; say 42
buubot_backup sorear: 42 1
21:00 benabik left
_edwin reported rt.perl.org/rt3/Public/Bug/Display...?id=115310 21:05
21:07 skids left 21:11 birdwindupbird left
_edwin std: s[0]*] 21:11
p6eval std cda80a9: OUTPUT«ok 00:00 43m␤»
_edwin does this really mean something, or is it a stdbug (as diakopter suggested iirc) 21:12
jnthn It's a cute form of s/0/*/ :)
_edwin I don't see how 21:13
jnthn std: s[0][*]
p6eval std cda80a9: OUTPUT«===SORRY!===␤Unsupported use of brackets around replacement; in Perl 6 please use assignment syntax at /tmp/kvXsCWRdcH line 1:␤------> s[0]⏏[*]␤Parse failed␤FAILED 00:00 42m␤»
jnthn I suspect it should produce that. :)
21:14 bruges left 21:15 alester left, bruges joined
_edwin std: sub s { ... }; s[0] 21:15
p6eval std cda80a9: OUTPUT«===SORRY!===␤Missing assignment operator at /tmp/veNSKfjzmx line 1 (EOF):␤------> sub s { ... }; s[0]⏏<EOL>␤Parse failed␤FAILED 00:00 43m␤»
sorear n: s[0]*]
p6eval niecza v22-13-g0d2a99e: OUTPUT«===SORRY!===␤␤Malformed assignment operator at /tmp/yi7TeX3YbQ line 1:␤------> s[0]*⏏]␤␤Parse failed␤␤»
21:18 kaare_ left
_edwin std: sub s { ... }; s() 21:18
p6eval std cda80a9: OUTPUT«ok 00:00 42m␤»
_edwin std: sub s { ... }; s();
p6eval std cda80a9: OUTPUT«ok 00:00 42m␤» 21:19
diakopter r: say $_."{ return 55 }"()
p6eval rakudo 2fca74: ( no output )
_edwin huh? I get different behavior using private messages
diakopter n: say $_."{ return 55 }"() 21:20
p6eval niecza v22-13-g0d2a99e: OUTPUT«Unhandled exception: Illegal control operator: return␤ at /tmp/UiO_klH9Lq line 1 (mainline @ 5) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4212 (ANON @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4213 (module-CORE @ 579) ␤ at /home/p6eval/nie…
_edwin std: sub s { ... }; s(); 21:21
p6eval std cda80a9: OUTPUT«===SORRY!===␤Confused at /tmp/ajeD8vghxk line 1:␤------> sub s { ...⏏ }; s();␤ expecting any of:␤ POST␤ infix or meta-infix␤ infixed function␤ postcircumfix␤ postfix␤ postfix_prefix_meta_operator␤ postop␤ statement
..modifier loo…
21:22 aharoni left
_edwin no see no difference 21:22
21:22 rindolf left 21:25 rindolf joined
diakopter n: s[a]*=] 21:25
p6eval niecza v22-13-g0d2a99e: OUTPUT«Unhandled exception: Unable to resolve method subst in type Any␤ at /tmp/ADMJ6D7ql6 line 1 (mainline @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4212 (ANON @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4213 (module-CORE @ 579) ␤ at /home/…
diakopter sorear: ^^
supernovus As fun as the name Exemel is for my library, would its purpose be more easily obvious if I renamed it XML?
sorear yes, a bit 21:26
diakopter it requires the equals sign
sorear n: ]
p6eval niecza v22-13-g0d2a99e: OUTPUT«===SORRY!===␤␤Unexpected closing bracket at /tmp/CaZtbHQlIh line 1:␤------> <BOL>⏏]␤␤Parse failed␤␤»
21:28 benabik joined 21:29 benabik left
_edwin oh, I somehow got a no-break space in there. that makes the difference 21:29
evil character
supernovus Okay, as of now, Exemel is reborn as XML (although the github project will forever be 'exemel'). No more word-plays for my project names, I promise ;-) 21:39
21:40 mtk left
sorear supernovus: I do not get the word play there 21:41
_edwin what is the difference between (*%s) and (**%s)? did not find it in the spec
supernovus sorear: the original name was pronounced the same as saying the letters X M L. 21:42
diakopter sorear: any idea why it parses with the equals sign but not without?
sorear diakopter: no 21:44
21:44 spider-mario left
jnthn sleep & 21:45
sorear night
21:49 supernovus left
_edwin nr: sub foo(Pair *@pairs) {}; foo(1 => 2); 21:49
p6eval niecza v22-13-g0d2a99e: OUTPUT«Potential difficulties:␤ @pairs is declared but not used at /tmp/027ac7pp2t line 1:␤------> sub foo(Pair *⏏@pairs) {}; foo(1 => 2);␤␤»
..rakudo 2fca74: OUTPUT«Nominal type check failed for parameter '@pairs'; expected Positional but got Array instead␤ in sub foo at /tmp/ip2Sl_zZ_k:1␤ in block at /tmp/ip2Sl_zZ_k:1␤␤»
_edwin nr: sub foo(Pair *@pairs) { say @pairs }; foo(1 => 2);
p6eval niecza v22-13-g0d2a99e: OUTPUT«1 => 2␤»
..rakudo 2fca74: OUTPUT«Nominal type check failed for parameter '@pairs'; expected Positional but got Array instead␤ in sub foo at /tmp/YTpkftNtWo:1␤ in block at /tmp/YTpkftNtWo:1␤␤»
_edwin rakudobug?
sorear _edwin: yes 21:50
I have no idea what's going on but that error message is ridiculous
21:51 att left
_edwin as Array does Positional, right? 21:51
21:52 kst left
sorear exactly 21:52
21:53 kst joined
_edwin seems to be a general problem with typed slurpies 21:54
not only slurpy @, just typed @ seems enough 21:55
21:58 fgomez joined 22:03 Chillance left
_edwin oh, i duplicated a masaked bug 22:04
22:09 hoelzro is now known as hoelzro|away 22:11 sjohnson left 22:15 fgomez left, hash_table joined
_edwin updated rt.perl.org/rt3/Ticket/Display.html?id=113964 22:16
22:18 _edwin left 22:19 rindolf left 22:24 whiteknight joined 22:33 stopbit left, benabik joined 22:35 MayDaniel left, fgomez joined 22:40 skids joined 22:46 steampunkey joined, steampunkey left 22:47 paxcoder joined
rjbs- How's Perl 6 going? :) 22:47
What's awesome lately? 22:48
paxcoder rjbs-, i reckon better than python 3 ;-P
22:48 rjbs- is now known as rjbs
paxcoder rjbs-, rules, check em out 22:48
22:48 rjbs left, rjbs joined
rjbs Have they gotten much cooler lately? 22:48
paxcoder rjbs, i know regexes did
rjbs I should update my local rakudo build. I'm sure it's ages out of date.
Orrr.... maybe it doesn't even exist anymore. 22:49
diakopter <blink>
sjn rjbs: $ cpanm Rakudo::Star 22:50
works fine. :)
paxcoder shun the rakudo non-haver shuuuuun.
rjbs sjn: nice.
sjn blames mst
rjbs Yeah, I remember seeing that, now that you mention it.
paxcoder jk, i don't have it. heck, i don't know perl. any perl. i just reinvented the whole rules thing (not implemented) so i thought i'd see how you did them. 22:51
rjbs, when you exec some code in ?{} in a regex, can it access the FSM somehow? 22:52
umm, anyone, not just rjbs 22:53
diakopter what FSM
rjbs In p5? What specifically are you trying to do?
paxcoder diakopter, the one that matches the input against the regex
rjbs, p6 (any, really) 22:54
diakopter <facedesk>
paxcoder rjbs, informative. as i said, i don't code perl.
rjbs can't help in p6.
paxcoder diakopter, problem?
diakopter not anymore :P
paxcoder diakopter, you just looked up FSM, didn't you?
diakopter please don't pick a fight.. I was merely pointing out that there might not be a FSM 22:55
22:55 japhb_ joined
paxcoder diakopter, well, can it access state of whatever it is that does and does the matching? 22:56
sorear
.oO( well, how else do you explain the trend in heights?)
paxcoder is this lang for registered users only? 22:57
*channel
diakopter I hope not
22:57 paxcoder is now known as perlOblivious
sorear paxcoder: are you new here? 22:57
perlOblivious that should help.
sorear, as i've said before, i know no perl. so, very much so. 22:58
diakopter the Perl 6 grammar implementations generally are recursive descent, with some NFA for optimization of patterns that are declarative
user code makes it non-declarative, so code can't be run during the NFA 22:59
Perl 5, I have no idea 23:00
lol at "lang for registered users only"; that's cute :D
(but no, the channel isn't either; it doesn't have the +i by its name here)
rjbs Plenty such languages have existed! :)
perlOblivious diakopter, what are "declarative patterns"? 23:01
diakopter patterns that can be reduced to an automata. see the last occurence of 'declarative' in perlcabal.org/syn/S05.html 23:03
perlOblivious i guess rules (which I reckon are like BNF) could be called declarative, using the definition which makes Prolog declarative programming (eg. pattern matching instead of controlling flow ith ifs and cases)
diakopter *automaton
*occurrence 23:04
declarative portions don't cause side effects 23:05
procedural ones do
perlOblivious diakopter, right, but are you really saying rules cannot execute code? cannot a rule contain a regex which executes something in ?{} ? 23:06
diakopter sure; it just can't be optimized to an NFA or participate in longest-token-matching 23:07
(ltm, which you can read much about in that url above) 23:08
perlOblivious let me make a paste to see what my question is. 23:11
sorear perlOblivious: a rule which executes code does not produce a NFA. 23:12
perlOblivious sorear, that's alright. but something has to process it. 23:13
sorear perlOblivious: the rule is compiled directly into code to do the processing. there is no runtime representation of the rule, it is simply bytecode or machine code 23:14
perlOblivious sorear, hmm. i guess that would mean "no"
the code executed cannot influence the matcher 23:15
diakopter well, it can if the matcher changes its behavior based on external state modified by the code it runs 23:16
TimToady this is starting to sound like a conversation where people are using different definitions for words 23:17
perlOblivious diakopter, but how would it? it's predefined.
let me finish writing that code snippet 23:18
diakopter TimToady: words definitions conversation are using people sound different 23:19
TimToady: also, it cannot be determined whether I know what I'm saying, nor whether I am accurate 23:20
TimToady puts diakopter into a Skinner box to see how he behaves. 23:21
TimToady hopes diakopter doesn't notice that it's really a Schrödinger box... 23:22
japhb_ That's one hell of a randomized negative stimulus 23:28
TimToady well, it's a double-blind study, so if I tell diakopter he's in the control group, there's a 3/4 chance he'll believe me. :) 23:33
perlOblivious actually code is better :-/
back to code
23:36 tokuhiro_ joined 23:37 thou left
perlOblivious diakopter, here: paste.debian.net/plainh/9078fce9 23:47
literate programming ;-) 23:48
crap i forgot to decrement count
TimToady perlOblivious: oh, we do that sort of thing by choosing which cursor to return 23:49
perlOblivious paste.debian.net/plainh/fe39ddb7
(just replaced n with count where it decrements)
Timbus, you manipulate the matcher's cursor? 23:50
TimToady or returning no cursors for failure
no, cursors are immutable
perlOblivious is cursor a pointer to the current char/group/etc?
TimToady a cursor holds the current match state; if you advance the position, you just return a different cursor with the new position 23:51
to fail, return ()
in STD at least, you can return a lazy list of cursors to control future backtracking points
perlOblivious Timbus, () is like nil or smth? 23:52
how is the cursor passed into the function? how do you return cursor-1?
TimToady yes, just an empty list
23:52 armitage81 joined
diakopter perlOblivious: TimToady, not Timbus :P 23:53
TimToady same way, in fact the <after FOO> assertion matches backward in just that way
perlOblivious Timbus, sry, i have a crappy complete, and am used to 3 chars + <tab>
Timbus asdfasd
diakopter :)
TimToady there are many people on this channel (hi lurkers!) 23:54
not that Timbus++ is a lurker
Timbus :o
well i mostly am
perlOblivious Timbus, i'm sorry, i'm trying to get it... 23:55
grr. TimToady *
Timbus, do you just return -1 to go one position back? If not, where is the cursor stored? Is it implicitly passed, what's its name? 23:56
perlOblivious facepalms
TimToady, ^
23:57 sjohnson joined, sjohnson left, sjohnson joined
perlOblivious TimToady, can you write me a function that will just advance the cursor to skip the next char (effectively making it a '.') ? 23:58
TimToady if $¢ points to the current cursor, then movement backwards is $¢ = $¢.cursor_fresh($¢.pos - 1)
same thing, only with a +
perlOblivious you must know that i don't know perl. so $􏿽xA2 is the matcher object or smth? 23:59
TimToady there is no *the* matcher object; there are many
perlOblivious or is there some special cursor entity in perl that i'm not aware of?
TimToady we're talking perl 6 here; perl 5 is completely different

Powered by Cro and the Raku® Programming Language.

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

Thank you!