»ö« | perl6-projects.org/ | nopaste: sial.org/pbot/perl6 | evalbot: 'perl6: say 3;' | irclog: irc.pugscode.org/ | UTF-8 is your friend!
Set by Tene on 14 May 2009.
00:10 DemoFreak left 00:15 snarkyboojum left 00:23 kate21de1 left
pmichaud japhb: it would need to be "use OpenGL:from<parrot>;" (no space before :from) 00:25
still it may not quite work yet.
japhb pastebin? 00:27
sial.org dead?
or rather, the pastebot ...
pmichaud: Can you explain why 'use()' gets called twice here? gist.github.com/128344 00:29
pmichaud probably once at compile time, and then again when Parrot compiles the PIR. 00:33
00:34 lichtkind left
pmichaud We don't have the BEGIN semantics completely worked out yet in the implementation (although I know how to do it -- just haven't had a chance to implement it) 00:34
japhb Ah, OK
pmichaud Parrot doesn't give me a way to "compile this batch of PIR but don't load or execute it"
japhb nodnod 00:35
DanielC Speaking of Parrot PIR... 00:37
pmichaud: Did you see that Q:PIR doesn't like sprintf?
pmichaud sprintf is fine -- the Q:PIR gets confused by the % 00:38
It sees the %0 and tries to do an argument substitution.
DanielC Ok. 00:39
pmichaud try "\x2502x instead
or actually
DanielC Well, I already solved the particular problem by doing the conversion in Perl 6, but I thought I'd mention the problem.
pmichaud Q:PIR {{ ... $S0 = sprintf "\x{25}02x", $P0
yes, it's a known issue. 00:40
DanielC ok
pmichaud it's not one I'm likely to do much about, since sprintf is a little uncommon and we have a workaround.
DanielC Not the top priority I'm sure.
Meldrake_ rakudo: say +<1 2 3 4> 01:12
p6eval rakudo 86aeaf: OUTPUT«4␤»
Meldrake_ rakudo: say +<1 2 3 4>.WHAT
p6eval rakudo 86aeaf: OUTPUT«Use of type object as value␤0␤»
Meldrake_ Is this a bug? 01:13
01:13 donaldh left, donaldh joined
DanielC It is parsing it as +(<1 2 3 4>.WHAT) 01:13
Meldrake_ Oh, ok.
New here, just trying stuff out.
Thanks
DanielC rakudo: say (+<1 2 3 4>).WHAT
p6eval rakudo 86aeaf: OUTPUT«Num()␤» 01:14
DanielC trying is great
I only started playing with Perl 6 just over two weeks ago.
Meldrake_ So how come it's a Num(), not an Int()?
DanielC dunno
Meldrake_ OK :-) 01:15
DanielC It does seem odd. Obviously the list length will not be an decimal.
Meldrake_ Yeah, that's what I thought
DanielC rakudo: say <1 2 3>.elems.WHAT
p6eval rakudo 86aeaf: OUTPUT«Int()␤»
DanielC The + simply means "numerical context". 01:16
Whereas .elems is a method, so it gives a more sensible answer.
But still... I'd expect an array in list context to be an Int.
rakudo: my $a = <1 2 3>; say $a.WHAT 01:17
p6eval rakudo 86aeaf: OUTPUT«Array()␤»
Meldrake_ Me too. I guess I need to learn PIR
DanielC rakudo: my $a = +<1 2 3>; say $a.WHAT
p6eval rakudo 86aeaf: OUTPUT«Num()␤»
DanielC rakudo: my Int $a = +<1 2 3>; say $a.WHAT
p6eval rakudo 86aeaf: OUTPUT«Type mismatch in assignment; expected something matching type Int but got something of type Num()␤in Main (/tmp/8cgdQ3938W:2)␤»
DanielC Interesting.
pmichaud currently rakudo treats + as always returning Num 01:25
DanielC What does the spec say? 01:26
01:26 agentzh left
pmichaud well, clearly + $foo numfies 01:26
numifies
but it's not always an Int
01:26 agentzh joined
Meldrake_ rakudo: say +<1 2 3>.perl 01:26
01:26 wayland76 joined
p6eval rakudo 86aeaf: OUTPUT«0␤» 01:26
pmichaud in this particular case, the spec isn't explicit. But it's likely that +<1 2 3 4> will return an Int -- rakudo just doesn't have that in place yet.
for example... 01:27
rakudo: say +<3.4>;
p6eval rakudo 86aeaf: OUTPUT«3.4␤»
pmichaud Rakudo gets that one right. :-)
DanielC <3.4> is not a list of one element? 01:28
pmichaud it is not.
DanielC k
pmichaud <3.4> is the same as ('3.4') which is the same as '3.4'
01:31 wayland joined, silug left, wayland76 left, wayland is now known as wayland76
wayland76 pmichaud: just asking; has that makefile change happened yet? 01:32
Tene japhb: it's still "use OpenGL:lang<parrot>;" in rakudo ATM. I haven't fixed it to be spec-compliant yet. 01:34
01:34 frobnitz left
pmichaud wayland76: it's in a branch, I'm still awaiting testers. 01:34
We know that it currently fails on win32 but I don't have a system to be able to test/update.
japhb Tene: I fixed that before dinner, but I've got a few more bugs to clear out this evening before it's all working.
wayland76 pmichaud: ok, thanks :)
Tene japhb: I'm mostly settled down for now... I'll have a little spare time this evening. 01:37
japhb k
Tene So, lemme know. I'm glad to do whatever needs done. 01:38
japhb YEAH! OpenGL in Perl 6 BAYBEEE! 01:40
OK, now to rip out hacks, and clean up messes ...
01:41 frew|work joined 01:42 clkao left
pmichaud WTF?!? 01:49
sorry, wrong chan.
01:49 wayland76 left 01:51 hcchien joined
sjohnson1 heh 01:53
01:54 donaldh left, donaldh joined
frew|work pmichaud: hey, I'm getting a (I think) apache error when I try to go to dallas.p6m.org 01:59
pmichaud that's gotta be coming from some other system than mine 02:01
02:01 frew|work is now known as frew
frew if I go to p6m.afoolishmanifesto.com it works 02:02
02:04 eternaleye joined
pmichaud your system may be doing hostname-based virtual hosting 02:08
so when it sees a hostname of dallas.p6m.org, it doesn't know where to forward
frew hmm. Ok, let me try something 02:09
indeed. If I type in the ip it doesn't work. 02:11
02:16 clkao joined 02:20 Whiteknight left 02:22 eternaleye left 02:23 eternaleye joined 02:24 DanielC left 02:28 sjohnson1 is now known as sjohnson 02:54 cookys left 03:01 FurnaceBoy_ joined 03:05 pmichaud left 03:13 FurnaceBoy left 03:20 donaldh left, donaldh joined 03:29 Meldrake` joined, Meldrake_ left 03:30 silug joined 03:34 Molaf left, Molaf joined
dalek kudo: a2b8ceb | pmichaud++ | src/parser/actions.pm:
Bare parens should return Nil. TimToady++
 03:38
03:40 xinming joined 03:50 Meldrake` left 03:54 meppuru joined 04:10 meppl left 04:11 buubot left, buubot joined 04:15 meppuru left 04:20 buubot left, mberends left, buubot joined 04:41 FurnaceBoy_ left 05:06 synthEEEE joined 05:14 Molinari joined, Molinari left 05:22 synth left 05:29 Sunbeam is now known as H1N1, H1N1 is now known as Sunbeam 05:31 Sunbeam is now known as Roxas 05:38 justatheory left 05:39 cognominal left 05:40 sri_kraih joined 05:46 sri_kraih_ joined 05:49 frew left 05:52 ihrd joined 05:54 ihrd left 06:04 sri_kraih left 06:09 cognominal joined 06:13 cognominal left 06:33 finanalyst joined 06:39 kate21de joined 06:41 kate21de left 07:03 DemoFreak joined 07:13 dakkar joined
japhb Anyone with Rakudo commit rights around? 07:14
Tene I am. 07:19
07:20 donaldh left
Tene japhb: ping 07:20
japhb here
07:21 donaldh joined
japhb hold on ... 07:21
Tene: can you apply and commit gist.github.com/128494 please?
the old code didn't work at all, but wasn't on a path that got used. 07:22
(yet, that is)
Tene japhb: why did you change 'load-language' to load_language?
07:22 xinming_ joined
japhb Tene: because that's the op you taught me to use: gist.github.com/124453 (and, since it works, apparently the correct one :-) 07:23
Like I said, the existing code plain didn't work ... it just had never been tested. 07:24
Tene japhb: 'load-language' is defined in src/parrot/misc.pir and just runs the load_language op, but from the 'parrot' HLL, to work around a parrot bug.
japhb shrugs 07:25
Tene I'll make the first change. That's according to spec, and I just hadn't gotten around to it.
japhb Well, when my parrot stuff goes in, we'll see if it was needed or not. It could have been a conflation of bugs, I suppose.
Tene But using the load_language op has a side-effect of any dynpmcs ending up define din the HLL that it was run from.
so it needs to be executed from the 'parrot' HLL to work with any language that uses dynpmcs, like tcl 07:26
dalek kudo: d2b24d8 | tene++ | src/builtins/eval.pir:
Fix the keyword for loading a foreign library (japhb++)
 07:27
07:28 masak joined
masak saluton, actioncamels. 07:28
viklund_ o/
japhb masak: o/
Tene Hmm... problems... 07:29
hi masak
masak \o/
07:29 viklund joined
viklund o/ 07:29
masak: as you might have seen, November now accepts utf8 ;) 07:30
masak viklund: I'm looking at it now. 07:31
viklund++
japhb Tene: FYI, I'm in the process of dotting i's and crossing t's for Parrot side of fixes.
masak viklund: I will look at the commits, and then I will be back for questions. :)
viklund it was a bit tricky
without pmichaud_++, may light forever shine on his path, it would have been very hard for me to do 07:32
he has patience
Tene japhb: I can confirm that rakudo's use ...:from<>; does work. 07:33
japhb Tene: fair enough; I'm willing to believe that I made two changes at a time and thought it was only one while debugging. :-) 07:34
Tene And with one more update, classes will work too. 07:35
viklund rakudo: say chr(229), chr(9787) 07:37
rakudo: say chr(229)~ chr(9787)
p6eval rakudo a2b8ce: OUTPUT«å☻␤»
rakudo a2b8ce: OUTPUT«\xE5☻␤»
finanalyst @seen jnthn 07:38
lambdabot jnthn is in #perl6. I don't know when jnthn last spoke.
finanalyst jnthn: ping
Tene japhb: gist.github.com/128499
viklund finanalyst: jnthn is on vacation 07:39
dalek kudo: e61569f | tene++ | src/parser/actions.pm:
Oops... fix the keyword in one more place too.
finanalyst viklund: thanx
just read jnthn's blog on .wrap & .unwrap
japhb Tene: nice!
viklund yes, that's cool, the first thing I tried was wrapping a multi, it didn't work :( 07:40
finanalyst looked at the spec, which gives .restore not .unwrap for rewinding
masak buubot: ack unwrap
buubot Couldn't match input.
masak buubot: spack unwrap
buubot Couldn't match input.
finanalyst buubot: ack restore 07:41
buubot Couldn't match input.
07:41 xinming left, xinming_ is now known as xinming
finanalyst .restort S06:2309 07:42
s/restort/restore/
would there be a way to .wrap all the subs in the current package? Suppose I want to debug by tracing when subs are called and their signatures? 07:44
Tene iterate over $?PACKAGE 07:45
iirc
07:45 rfordinal joined 07:47 jferrero joined
finanalyst what about current scope? 07:48
masak rakudo: <a b d d>.grep({$^a ~~ $^b}).perl.say 07:50
p6eval rakudo e61569: OUTPUT«too few arguments passed (1) - 2 params expected␤in method Any::grep (/tmp/y69Yw73kJe:1)␤called from Main (/tmp/y69Yw73kJe:2)␤» 07:51
masak should the above worK?
Tene Huh.
ENOCLUE
viklund masak: if it worked, what would you expect the return to be? <d d>, <d> or <d>? 07:54
I would say <d d>...
masak aye.
I'd expect it to proceed pairwise, like a for-loop would.
viklund yup 07:55
masak and return all the matching pairs.
viklund where pairs in this case aren't :pairs...
masak not unless you ask for it, no. 07:57
the default is still flat lists.
Matt-W Good morning 07:58
masak good morning, Matt-W. 07:59
08:01 barney joined
Matt-W masak: so what does infix:< 08:03
infix:<~~> do
(~ is too close to enter)
masak Matt-W: um. it smartmatches, I guess...
Matt-W oh 08:04
doh
I know that
Matt-W needs caffeine, clearly 08:05
so you're expecting grep with a two-arg block to proceed pairwise? That would be nice. 08:06
although personally I don't like these $^ placeholder things
I'm a big convert to pointy blocks
mmm... pointy 08:07
08:07 riffraff joined
viklund mm... pointfree 08:17
Matt-W hah 08:18
Matt-W ponders the use of pointfree style in Perl 6
hmm 08:20
you can't
well, not with the same elegance as in something like Haskell
since we have to call .assuming to produce curried forms
Tene You could define a sub trait that would create additional multis for insufficient args that return a curried sub. 08:21
sub foo($a, $b, $c) is pointfree { ... }
you could also potentially try it with exceptions 08:22
Matt-W mmm
that could be interesting
I think it's a style that's particularly suited to Haskell's syntax 08:23
but actually it might be possible to write a useful function composition operator
use introspection to wire up the parameters behind the scenes
08:35 bacek left 08:38 DanielC joined
viklund o/ 08:39
DanielC hi
moin
@karma 08:41
lambdabot You have a karma of 14
masak moin, DanielC. 08:45
DanielC moin masak 08:48
masak hm, I need to pull the ethernet cable for a while, because I just restarted my Firefox session, and it crashes if it tries to load all the tabs. 08:53
08:53 masak left
Matt-W lo DanielC 08:53
DanielC o/ 08:54
08:54 araujo left 08:56 clintongormley joined
japhb pmichaud_: Would you mind bumping PARROT_REVISION to 39521 ? 09:03
japhb really gone now ...
09:04 masak joined
viklund @seen moritz 09:15
lambdabot I saw moritz leaving #perl6 6m 20d 15h 31m 1s ago, and .
viklund @seen moritz_
lambdabot moritz_ is in #perl6. I last heard moritz_ speak 1d 16h 14m 16s ago.
masak moritz_ was on vacation until a few days ago. maybe he's only partly back still. 09:17
viklund I'm looking at his json thingy
DanielC masak: How do you tell git to remove a file from the repository? Can you just delete the file? 09:23
masak DanielC: git rm.
DanielC error: 'checksum.pir' has local modifications 09:24
:-( How do I restore checksum.pir so I can delete it?
viklund git co checksum.pir
DanielC git: 'co' is not a git-command 09:25
DanielC reads git --help
viklund git checkout checksum.pir
masak DanielC: take heart! these initial pains will pass, and then git will become your best friend.
viklund has command aliases, need to remember that when helping others
masak viklund: skurk. giving examples with aliases in them. :)
DanielC viklund: thanks!
viklund blushes
DanielC masak: I already have a best friend. His name is Darcs :-)
masak that's another nice friend, I've come to understand. 09:26
DanielC Darcs and git actually feel very very similar.
Branching is easier in Darcs, and merging is more flexible, but they are generally similar. 09:27
09:28 synthEEEE left
masak I need to play around a bit more with Darcs. 09:29
DanielC It has less features, but is much easier to use. 09:30
That has to do with the fact that the Darcs model is that every copy of the repository is a branch. 09:31
masak phenny: tell pmichaud_ I'm mystified by [perl #66546]. It would be nice if you could take a look at it and determine if it's inherently easy or inherently difficult to fix.
phenny masak: I'll pass that on when pmichaud_ is around.
DanielC That makes many things a lot simpler, but it prevents a few obscure features that I couldn't name right now but I know they exist.
masak DanielC: the git model is not that different. every commit is a potential branch. branches are just names for particular sequences of commits. 09:32
DanielC Git and Darcs do feel generally similar.
They even share the "push" and "pull" notation. 09:33
masak: Could you do a "git pull" on the parrot-module-lib and confirm that t/03-checksum.t is gone and instead we have t/03-module_data.t ? 09:34
masak DanielC: you have both. 09:35
DanielC stupid git
I said "git mv". Or at least I thought I did.
masak seems you didn't. :) 09:36
DanielC ah, no I didn't.
masak git++ :)
Tene danielc++
09:36 ssm left
DanielC I'm used to Darcs, which detects a whole bunch of things for you, so you don't make this sort of mistake. 09:36
Tene in darcs, if you copy a file, it's an error? 09:37
DanielC define copy?
cp foo bar
?
Darcs looks for changes to every file that it knows about in the repository and asks you which changes you want. 09:38
Tene I understood that you implied that you ran 'git cp' instead of 'git mv'
DanielC ah, no, I used plain "mv"
Darcs would have said "this file is gone, do you want to record that?"
and then I say "oops"
Tene yes, git will do that
DanielC It didn't for me. 09:39
Tene git add --patch
you need to ask for it
DanielC Darcs is more friendly :-)
Tene if you just say "add this one file, and then make a commit", that's all it does.
DanielC Darcs shows you the individual changes to the file.
Tene eh, I'd be quite annoyed by being asked about all of my changes every time I try to commit.
Yes, that's what add --patch does
DanielC If you change two lines in a fie, you can tell Darcs to record one but not the other.
09:39 ssm joined
Tene exactly 09:40
DanielC In Darcs you can specify the individual files, if that's what you want, and the -a ("all") flag says "take all changes".
It works really well, trust me.
Tene that's exactly what 'git add' does. :)
DanielC No, it doesnt. 09:41
Tene Yes, it does work well.
DanielC I have used git and darcs, you have only used git.
Can you please trust me that I can't explain Darcs in IRC when I'm trying to do work?
masak hey, calm down! :)
viklund this parrot has ceased to be...
Tene 'kay
masak please defer your 'my SCM is more usable than yours' contests to #git or #darcs, where people are better equipped to argue against you. :P 09:43
DanielC I didn't even say that Darcs is better. I said that it is better at some things.
09:43 Yoi joined
DanielC It is undoubtedly worse at others. It has fewer features. 09:44
I got a bit stressed because I was trying to work too.
Discussing the merits of SCMs requires more depth than I can afford right now.
I got frustrated when someone basically says "I don't know the other scm but my one is just as good". 09:45
So we had a bad convergence of events, resulting in an unhappy Daniel. 09:46
masak DanielC: I'm sure Darcs is a very good SCM, despite being more minimal than git.
Tene DanielC: I see what you meant now.
masak but don't dismiss the way git does thing offhand. I agree with Tene that what git does makes sense too.
s/thing/things/ 09:47
Tene "darcs record" without arguments asks you about each change
DanielC I Git is a fabulous program.
I know it can do things git can't do. I can't remember what they are.
oops 09:48
I know it [git] can do things Darcs can't do. I can't remember what they are.
Matt-W The thing that always confuses me in git is how to take a file that's been changed locally and change it back to the version before those changes
like svn revert
DanielC darcs unrecord
masak Matt-W: git checkout.
Matt-W masak: last time I tried that, it was a no-op 09:49
as in, it didn't change the file at all
masak Matt-W: I know, it confused me in the beginning also.
Tene DanielC: darcs' patch theory stuff is what you're thinking of. darcs is rather better at pulling down only some changes from some remote.
masak Matt-W: it's like, git misbehaves subtly the first half-year you use it. then it suddenly starts to make sense.
DanielC Tene: I know patch theory :-)
09:49 DemoPhreak joined
Tene DanielC: oh, nm, i misread. 09:50
DanielC Tene: It is more than just cherrypicking.
Matt-W masak: but when people say 'git checkout' to do like svn revert, and then it doesn't do anything when you say 'git checkout that-file', what are you supposed to do?
viklund the json grammar/action pair produces strange results...
Matt-W masak: it's tempting to think the person who told you to do it is wrong
Tene DanielC: i read "Darcs can do things that git can't, but I don't remember what they are."
masak Matt-W: no, that's the way you do it.
Matt-W: but you have to know what you're doing. :P
Matt-W masak: well it didn't work
what's to know?
if you say it's 'git checkout that-file'
and it doesn't work
then are you wrong? 09:51
or is git broken?
masak Matt-W: could you nopaste a reproduced case where it doesn't work?
Matt-W possibly
masak would be interesting to see.
09:51 explorer__ joined
Matt-W I ended up giving up and just stashing it 09:51
DanielC Tene: I wrote wrong the first time. I know someone on the Darcs list once gave me an example of something a bit obscure that git could do and was not possible with Darc's simpler model.
Matt-W although now I'm going to have to figure out how to delete the stash
I'd found websites which said 'git checkout' as well
DanielC Tene: But it was not something I understood. It was definitely an "advanced" feature.
09:51 agentzh left
Matt-W but I assumed they'd missed something out when it didn't work 09:51
09:51 explorer__ left
Matt-W because they weren't well-written in the first place 09:51
overly brief 09:52
09:52 agentzh joined
Tene Matt-W: nopaste.snit.ch/16880 09:53
Matt-W: like that?
viklund Matt-W: were you trying to git checkout the file you git mv'd?
Matt-W Tene: that is not what happened
viklund: no
viklund hmm'
viklund trying to figure out why it didn't work
did you have anything in the index? 09:54
Matt-W git pull -> make local edits to file -> git pull later on -> wanted to revert edits to file -> git checkout file -> file unchanged
Tene Matt-W: did you commit the edits?
viklund aha
masak Matt-W: maybe you ended up in the no-man's-land between branches.
09:55 tomaw- joined, Yoi left
Matt-W Tene: no 09:57
Tene: I didn't want them anymore
10:03 DemoFreak left 10:14 lichtkind joined, lichtkind left 10:16 payload left 10:52 payload joined 10:55 Muixirt joined
Muixirt hi 10:55
10:58 kane_ joined
Roxas !UP 11:05
ew amsg sucks in xchat 11:08
sorry
11:08 hanekomu joined
DanielC I should convert my PHP code to LOLCODE. That would be fun. 11:09
Muixirt the code snippet while 1 { }; leaks a lot of memory, is that a problem of rakudo too?
DanielC dunno 11:10
There are two people who know Rakudo really really well. One of them went on vacation yesterday and the other is in Texas and is not around here yet.
11:14 payload left
masak Muixirt: fwiw, I'm running Rakudo now, and it does seem to leak memory on that one-liner. 11:17
350Mb and rising...
Muixirt it was mentioned some days ago 11:18
but i'm really puzzled, i even created a blog to write on this ;-) 11:19
masak a whole blog, just for this? :)
what's it called, 'The while 1 { } blog'?
11:20 donaldh left
viklund_ that would be a good blogname actually 11:20
"while 1 { ... }"
DanielC @seen mberends
lambdabot I saw mberends leaving #perl6 7h 24s ago, and .
masak viklund_: I was thinking the same thing. :) 11:21
11:21 donaldh joined
masak viklund_: or you could name it "Leaking memory in an infinite loop" ;) 11:21
viklund_ ehm 11:23
for @*THOUGHTS { .blog }
masak :)
Muixirt masak, done ... muixirt.wordpress.com/
masak Muixirt: :D
Muixirt: extra kudos if you actually fix the problem. :) 11:24
Muixirt: (fwiw, I'd go with viklund_++'s suggestion, it's really nice) 11:25
maybe even 'for self.thoughts { .blog }'... 11:26
Muixirt you might have a look at nopaste.snit.ch/16881 the generated pir output of that not so infinite loop 11:28
i wonder why one have to do new $P20, "Int" endlessly 11:29
near loop27_test:
masak one doesn't have to do that, I bet.
it's just that this kind of optimisation hasn't been highly prioritized, between implementing new features and fixing bugs. 11:30
patches welcome! :)
11:36 mberends joined
mberends howdy #perl6 11:37
DanielC o/
masak greetings, mberends.
mberends masak: good day! just having a nice break from furniture shifting ;) 11:38
DanielC, your latest refactoring all works here now.
masak my $nice-break = @furniture.shift;
DanielC mberends: Great.
mberends: Is there a reason why we should do everything in PIR? The YAML part would be much easier in Perl. I wonder if here is a way to mix Perl and PIR to do what we want. 11:40
mberends yesterday's Class 'MD5' not found error was caused by debian having libssl0.9.8 installed but not libssl-dev.
DanielC ah
Ubuntu++ 11:41
mberends DanielC, there may some corner cases where is does not apply, but in general we are trying to make something to fit into the chassis of Rakudo. Dependencies on anything Perl might lead to Catch-22, chicken-or-egg, and suchlike. I'd rather not risk it. 11:42
masak DanielC, mberends: if you do a YAML parser/emitter, please (1) make it a separate module, or at least a separate unit of code somehow, so others can re-use it, and (2) consider getting in touch with ingy -- he hangs out on #yaml on irc.perl.org -- because he and I have been discussing writing a YAML parser in p6 before.
oh, ingy 's even here on the channel. :) 11:43
DanielC masak: I was thinking of just writing a parser for a *very* limited subset of YAML. Not something usable in the general case.
masak DanielC: been there, done that. 11:44
DanielC masak: I wouldn't ind writing something usable in the general case, but then it would have to be written in Perl.
masak it works, but it feels risky.
DanielC s/ind/mind/
11:44 finanalyst left
masak DanielC: proto does quite a lot of limited YAML parsing. 11:44
mberends masak, as discussed at the Mausoleum, a full blown YAML toolset is often overkill, and a quick and dirty little bastard yaml parser will often suffice. DanielC: +1
DanielC masak: Yeah. But can you imagine trying to write a general parser for anything interesting in an assembly language. 11:45
masak DanielC: PIR is actually not so bad. ask pmichaud_ :)
DanielC: (he wrote PGE in PIR)
mberends: I don't disagree. I'm just pointing you to the knowledgeable people, should you decide to go general. 11:46
DanielC masak: PGE is a perl-like grammar, like? Or is it the full Perl 6 grammar?
mberends masak, I'm rather enjoying PIR programming :) re: yaml, indeed yes
masak no-one would be happier than I if we could standardize on a quick and dirty little bastard YAML parser.
DanielC: it's a full P5/P6 regex/grammar engine. the Perl 6 grammar is meant to run on top of it. 11:47
DanielC masak: I wonder if PGE can (in principle) be used to make a YAML parser usable from PIR. 11:48
masak DanielC: sure.
DanielC It may be an interesting project. 11:49
On the other hand, a YAML parser that only uses Perl 6 would be usable to other implementations.
masak DanielC: I'm browsing around in months-old code from when I last attempted this. found this comment: # <rjbs> the best advice to someone about to write a yaml parser is "don't"
DanielC If I wanted to write a general YAML parser, I think I' pick Perl 6.
masak: What does that advice mean? 11:50
masak DanielC: that writing a YAML parser is harder than it first appears.
DanielC ok
It looks hard enough as it is.
masak that's why I advise you to befriend the #yaml people.
DanielC Not many people there. 11:51
ingy is there.
mberends for parrot-module-lib a subset of yaml is desirable but not even required. "What's the simplest thing that could possibly work?" Some domain specific data structure. The yaml is for hacking FTW.
masak mberends: ack.
DanielC A restricted YAML is pretty close to what I would pick for a domain-specific markup language. 11:52
masak DanielC: .perl and eval(slurp $file) goes a long way, though. 11:53
DanielC I would make a few non-YAML alterations, but that's all.
masak DanielC: it works out of the box.
11:53 szabgab left
DanielC masak: Yes, but only if you are using Perl :-( (remember.. PIR) 11:53
masak oh. right. 11:54
DanielC But PIR has arrays and hashes.
Maybe we can use that.
masak DanielC: you could copy the Array.perl and Hash.perl source code from Rakudo. :)
it's open source.
DanielC :-) 11:55
Hey, Ingy is one of the YAML designers.
masak yes.
DanielC cool 11:56
mberends hears the "back to work" signal
masak I almost, but not quite, sat down with him after NPW and ported YAML to Perl 6.
DanielC JSON is easier to parse than YAML. Maybe we should use that. 11:57
11:58 Molaf left
DanielC Then again, even a JSON parser would be hard in PIR, so I would just end up restricting it anyways. 11:58
masak DanielC: using PGE grammars and actions for this kind of parsing would help a lot, I think.
DanielC y 11:59
11:59 mberends left
DanielC I don't suppose there is a tutorial for PGE anywhere. 11:59
masak DanielC: www.pmichaud.com/2006/pres/yapc-par...start.html 12:02
might be a bit outdated, but should get you started.
DanielC *click*
masak the Wikipedia article was quite nice, too. 12:05
DanielC Wikipedia article on PGE?
masak it's probably a bit dated as well, because it makes it sound as if Pugs is still in active development...
en.wikipedia.org/wiki/Parser_Grammar_Engine
DanielC *click*
Maybe we can make a parser and use JSON. 12:06
JSON is much simpler than YAML.
Matt-W a lot of the wikipedia stuff on Perl 6-related tech is outdated
DanielC hacks together a sample file in JSON
Matt-W we should make an effort to update it
masak yes, we should.
12:10 elmex joined
DanielC masak: What is "proto" in the context of PGE? 12:16
proto infix:+ is looser('term:') {...}
masak DanielC: S05:1048.
DanielC k
masak 'if you predeclare a proto regex, you can write multiple regexes for the same category, differentiated only by the symbol they match. The symbol is specified as part of the "long name".' 12:17
DanielC I have read that paragraph a few ties, an seen the example, and I'm still confused. 12:18
For starters, what is a category?
masak DanielC: in the context of the example, the token 'sigil'. 12:19
DanielC k...
masak DanielC: heard of multisubs?
DanielC yeah, I'm going to post a blog about mulltisubs today :-)
coincidentally...
masak DanielC: this seems to be the corresponding thing for regexes/tokens/rules.
DanielC token sigil:sym<$> { <sym> } 12:20
The first problem I have with that is that it looks recursive.
masak yeah.
no.
DanielC Second, the :sym<$> throws me off.
12:21 kane_ left, bejuryu joined
masak I'll let the people who have more of a clue explain the details. fwiw, I read that as 'token sigil, where <sym> is "$"' 12:21
DanielC based on longest-token matching <--- 12:22
Ok, that explains the multi part better.
"The advantage of writing it this way is that it's easy to add additional rules to the same category in a derived grammar" <-- ok 12:23
masak I wish more of STD.pm worked like that. 12:25
I looked into adding a new statement postfix keyword the other week, and it doesn't look very easy at present.
people will want to do such things.
DanielC masak: Could you help me with something? pastebin.com/m51c438d <-- What's wrong on line 25? 12:29
"Statement not terminated properly at line 25," 12:30
masak DanielC: try adding a semicolon on the line above.
DanielC I did, no change.
masak DanielC: oh! infix:over should be infix:<over>.
those <> are not optional.
DanielC ah!
oops 12:31
thanks
masak np.
oh, and cool! Rats.
DanielC I assumed it was just if you wanted "funny" characters like + or !
masak DanielC: no, they're part of the sub name.
I'm a bit surprised infix:over parses, actually. 12:32
rakudo: sub infix:over() { say "OH HAI" }
p6eval rakudo e61569: ( no output )
masak rakudo: sub infix:over() { say "OH HAI" }; infix:over();
p6eval rakudo e61569: OUTPUT«Statement not terminated properly at line 2, near ":over();"␤in Main (src/gen_setting.pm:0)␤»
DanielC Now it complains about the gcd function: Null PMC access in find_method() in in sub gcd called from sub infix:over. 12:33
But I tested gcd individually and it worked.
masak DanielC: try and reduce it to a minimal test case. 12:34
might be a bug.
DanielC starts reducing
Matt-W go on! make masak happy!
masak :)
DanielC WTF? The problem goes away when I remove the == functions. 12:36
It is definitely the == functions. If I comment them out, the program stops complaining. 12:37
Is == a multi-function? 12:38
Any one of the infix:<==> functions is enough to make the program barf. 12:39
Matt-W can you get it down to barfing on having just an infix:<==>?
DanielC reduces more
Hmm... I *have* to call the "over" function to reproduce the error. 12:40
DanielC keeps reducing 12:41
masak rakudo: sub gcd() {}; sub infix:<over>($a, $b) { gcd }; 4 over 5
p6eval rakudo e61569: ( no output )
masak rakudo: sub gcd() {}; sub infix:<over>($a, $b) { gcd }; 4 over 5; multi sub infix:<==> (Int $a, Int $b) {} 12:42
p6eval rakudo e61569: ( no output )
masak hm, not as simple as that, it seems...
DanielC Indeed, it isn't.
masak keep reducing; I've gotta go soon.
I hereby bequeath to you the holy power of rakudobug reporting.
Infinoid Don't spend it all in one place. 12:43
DanielC This is bewildering. If I remove the gcd line it stops complaining.
masak not so surprising, given that that line caused the error. 12:44
DanielC gcd uses ==
Maybe that's where the problem is.
yeah
now I can reduce a lot more. 12:45
masak rakudo: role A {}; multi sub infix:<==>(A $a, A $b) {}; 42 == 42
p6eval rakudo e61569: OUTPUT«Null PMC access in find_method()␤in Main (/tmp/5dgviLoPNU:2)␤»
masak there you go. :)
masak submits rakudobug
DanielC ? 12:46
ah
rakudo: role A {}; multi sub infix:<==>(A $a, A $b) {}; "42" eq "42" 12:47
p6eval rakudo e61569: ( no output )
DanielC This is annoying though. How am I going to make my blog post now without an == operator?
masak :) 12:48
you'll find a way.
DanielC :-P
masak gotta go. see y'all.
12:48 masak left
DanielC bye 12:48
12:50 jonathanturner left 12:51 Whiteknight joined
Matt-W you'll have to wait until the bug is fixed I suppose 12:51
time to start nagging people
12:54 szabgab joined 13:07 kane___ joined 13:13 kane___ left 13:14 jferrero left, FurnaceBoy joined 13:17 alanhaggai joined 13:19 mizioumt joined 13:20 skids_ joined 13:21 frew joined, beggars joined 13:23 frew_ left 13:24 alanhaggai_ joined 13:26 alanhaggai left, alanhaggai_ is now known as alanhaggai 13:29 lichtkind joined 13:35 Muixirt left, jantore left, felipe left, dmpk2k left, DQuest left, Gothmog_ left 13:37 kane___ joined, dmpk2k joined 13:38 DQuest joined, mattp joined, ruoso joined 13:43 jantore joined 13:45 decasm joined
ruoso Hello! 13:45
13:48 [particle] left, [particle] joined, Muixirt joined 13:51 ihrd joined, ihrd left 13:54 kane__ left 14:01 araujo joined
Matt-W $self->{'stats'} = @_; 14:12
^ too much Perl 6!
(I'm sure you all spot the bug)
DanielC actually...
? 14:13
viklund scalar context for @_ => returns length
Matt-W yes
DanielC ugh!
Matt-W this was not the intent
later on I had a rather surprising message noting that I can't use '2' as an array reference
DanielC $self->{'stats'} = \@_; <--- is this what you wanted?
Matt-W yes indeed
DanielC Perl 5 is painful compared to Perl 6. 14:14
Matt-W It's amazing how quickly it gets into your head
oh nice
now I have
Can't locate object method "year" via package "Fri Jun 12 15:12:57 2009" 14:15
Matt-W sighs and goes to fix that too
14:17 beggars left
viklund there's an error somewhere in moritz_ json perl6 thingy, it's hard to pindown... 14:18
14:18 pmurias joined
pmurias ruoso: hi 14:18
Matt-W viklund: it's good for your brain :) 14:19
viklund yes, well, but it's strange
'\n\t' becomes '\\\\\"' 14:20
Matt-W umm 14:21
that is
err
Matt-W scratches his head
pmurias ruoso: is the list of roles in ^!does flattened?
viklund indeed 14:23
clintongormley how complete is the regex engine? 14:27
is it supposed to work as spec'ed?
ruoso pmurias, they are just the direct applied roles
pmurias, they don't include parent's compositions 14:28
viklund scratches his beard 14:31
Matt-W clintongormley: no, it doesn't quite work as specced. Some things are missing, and the most noteable omission is that it doesn't do longest-token matching yet 14:32
clintongormley ahh still not?
pmurias ruoso: should every punned class instance have an ACCEPTS method?
or only the one instance which is directly punned?
clintongormley i remember reading about TimToady's frustrations at getting that working, but i thought he'd conquered it 14:33
ruoso pmurias, hmmm.. good question... I guess the punned class... 14:36
14:38 rfordinal left
pmichaud_ good morning, #perl6 14:38
phenny pmichaud_: 09:31Z <masak> tell pmichaud_ I'm mystified by [perl #66546]. It would be nice if you could take a look at it and determine if it's inherently easy or inherently difficult to fix.
pmurias ACCEPTS method is ovewritten by a lot of things, that would mean we would need something different from Signature.ACCEPTS to check if something is a signature
ruoso pmurias, I guess it could simply delegate it to the meta 14:39
14:41 payload joined
ruoso pmurias, I guess ACCEPTS is a multi method 14:42
viklund good morning pmichaud
ruoso rakudo: my $a = :($a, $b); say $a ~~ Signature 14:43
p6eval rakudo e61569: OUTPUT«too few arguments passed (0) - 2 params expected␤in Main (/tmp/SfvHXBxtNs:0)␤»
viklund thanks for the help yesterday with that utf8 <-> latin1 thing
ruoso hah!
pmurias ruoso: ? 14:44
pmichaud_ viklund: you're welcome. We troubleshooted that bug a bit more in Parrot and found a few more issues to deal with. :-|
ruoso rakudo didn't solve that question as well
pmichaud_ rakudo: say Signature;
p6eval rakudo e61569: OUTPUT«Signature()␤»
viklund yes, I followed that with interest...
ruoso that's actually a good question... 14:45
pmichaud_ rakudo: sub foo() { }; say &foo.signature ~~ Signature;
p6eval rakudo e61569: OUTPUT«1␤»
ruoso I can't really think of anything smart to solve it
rakudo: sub foo($,$,$) { }; say &foo.signature ~~ Signature
p6eval rakudo e61569: OUTPUT«1␤»
14:45 moritz_ sets mode: +o pmichaud_
viklund anyway, we have utf8 working in November now... 14:45
pmurias ruoso: do we want to assume type checking and smart matching is the same thing? 14:46
if so an ACCEPT method on an undefined Signature could only check if the argument does the Signature role
* ACCEPTS
ruoso that's kinda of an general assumption in Perl 6
I mean...
rakudo: my $a = :($a, $b, $c); say $a.defined 14:48
14:48 justatheory joined
p6eval rakudo e61569: OUTPUT«too few arguments passed (0) - 3 params expected␤in Main (/tmp/PMfD6mf7T5:0)␤» 14:48
pmichaud_ I don't think rakudo handles the :(...) syntax yet.
ruoso ah...
that explains it
pmichaud_ rakudo: sub foo($a, $b, $c) { }; say &foo.signature.defined; 14:49
p6eval rakudo e61569: OUTPUT«1␤»
ruoso rakudo: sub foo($a, $b, $c) {}; say &foo.signature ~~ (1,2,3)
p6eval rakudo e61569: OUTPUT«0␤»
ruoso ok... I guess it should have been a 1 there
pmichaud_ actually it would need to go the other way 14:50
14:50 hexmode joined
pmichaud_ (1,2,3) ~~ &foo.signature 14:50
but Rakudo doesn't really do captures yet
ruoso oh righht...
but it is supposed to be done with ACCEPTS, rigth? 14:51
pmichaud_ it still thinks that (1,2,3) is a List (and doesn't grok the new relationship between List and capture)
I'm not sure if rakudo is using ACCEPTS for signature binding yet
I'm sure it _can_ 14:52
ruoso not binding, but checking
pmichaud_ oops, you're correct -- &foo.signature ~~ (1,2,3) also works (I needed to look a little farther down the smart match table) 14:53
clintongormley damn, the JS on the synopses makes the pages all but unusable, at least in firefox
ruoso binding is SMOP goes with .BIND
pmichaud_ clintongormley: try svn.pugscode.org/pugs/docs/Perl6/Spec (no js)
clintongormley ta
Matt-W I've never had any trouble with them on firefo 14:54
clintongormley pmichaud_: Muuuuch better
Matt-W: my firefox hangs for 30 seconds at a time 14:55
Matt-W hmm
not seen that
how odd
viklund noscript ftw
14:55 exodist joined
clintongormley are heredocs not implemented yet? 14:56
moritz_ not in rakudo 14:57
pmurias ruoso: there is a Any Type type membership $_.does(X)
moritz_ STD.pm parses them
clintongormley ta
moritz_: welcome back! long time
pmurias ruoso: row in the smartmatch table
ruoso pmurias, but it seems that isn't syntatical 14:58
pmurias how is the Type role defined
?
moritz_: i disabled the evalbot runtime limits for mildew 14:59
ruoso pmurias, it isn't
pmichaud_ rakudo: say 'a'..'z'
p6eval rakudo e61569: OUTPUT«abcdefghijklmnopqrstuvwxyz␤»
ruoso for now, it's just "undefined values"
moritz_ pmurias: please don't disable them entirely, in case it loops. Rather bump them up for mildew
ruoso pmichaud_, one thing i just realized... can that be locale-specific? 15:00
pmichaud_ ruoso: I don't think so... I think it's defined in terms of the unicode codepoints
pmurias locale specific? argh! :)
viklund Huzzah! 15:01
ruoso rakudo: say '𐌰'..'𐍊' 15:02
p6eval rakudo e61569: ( no output )
pmichaud_ at the moment rakudo only does the uppercase, lowercase, and ascii digit ranges
(it can be extended to do more, but we just did those to start with)
pmurias moritz_: i'm not sure which limits get violated
viklund moritz_: I'm working with you json parser 15:03
15:03 riffraff left
viklund I've managed to get it to pass one more test ;) 15:03
ruoso rakudo: say 'α'..'ω' 15:04
p6eval rakudo e61569: ( no output )
moritz_ pmurias: just do somthing that violates a limit, and I'll tell you which one 15:06
viklund: cool. Want a commit bit?
viklund moritz_: sure ;)
but I think it needs more structure tests
moritz_ sure, I was just getting started with them 15:07
viklund: done 15:11
15:11 mizioumt1 joined
viklund moritz_: thx 15:11
moritz_ viklund: please test it, I never did that before ;-) 15:12
pmurias moritz_: anything violates the limit 15:13
ruoso moritz_, the problem is that mildew runs after running STD
viklund I was thinking of actually trying it for real ;)
moritz_ ruoso: ... and? 15:14
ruoso pmurias, one thing that would ease things in mildew would be to build the mold in perl directly and simply run it, instead of generating the m0ld to then generate the C, than compile it than run
moritz_, it runs gcc from inside that same process 15:15
after using all the memory STD uses
and without releasing it
moritz_ ruoso: I still don't see the problem in figuring out a sensible resource limit for mildew, and setting it
ruoso ah... sure...
moritz_ but not having any resource limit is just suicide.
ruoso I was just explaining why it's so high
moritz_ ok 15:16
ruoso lunch &
15:16 exodist left 15:20 donaldh left, donaldh joined 15:24 mizioumt left
viklund ohh, speedup after compile 36s -> 32s... 15:24
clintongormley i realise this is a bad regex, but in perl5 it succeeds, in perl6 it just hangs: 15:29
say ' ' ~~ /\s*+/
moritz_ in perl 5 it's optimized out 15:30
clintongormley ah ok
15:31 masak joined
pmichaud_ rakudo/pge doesn't know how to handle quantified zero-length patterns 15:31
15:32 nihiliad joined
pmichaud_ it tends to get stuck in infinite loops 15:32
clintongormley it was actually a typo, and i couldn't figure out why rakudo was hanging on a simple white space match :) 15:33
masak pmichaud_: thank you for your answer to #66546. getting a data structure representing a Perl 6 AST has become something of a holy grail for me. now it again seems like I'll need to reconsider my strategy.
a shellout call to STD.pm and then parsing the YAML might be the more long-term solution anyway. 15:34
DanielC Since I can't post my blog about Rational numbers yet, today I'm making a different post: daniel.carrera.bz/2009/06/perl-6-an...s-problem/ 15:37
It's also nice, but not "cool" like the Rational numbers one.
masak DanielC: why can't you post your piece about Rational numbers? 15:39
DanielC masak: Because == won't work.
masak DanielC: did you try a workaround?
DanielC == is one of the parts that makes the post interesting.
masak: I could... but the workaround would not look as interesting. It would look like a workaround. 15:40
So it wouldn't be a great advertisement for Rakudo.
masak DanielC: for example, try defining infix:<==> for two Ints, even though in real Perl 6, that wouldn't be needed.
it could work, and it would only be one extra line.
pmichaud_ masak: you can always get Rakudo's AST format
masak perks up
pmichaud_: how?
DanielC masak: I could try that. 15:41
pmichaud_ rakudo: say Perl6::Compiler.compile('say <hello>', :target('past')).perl;
p6eval rakudo e61569: OUTPUT«Method 'perl' not found for invocant of class 'PAST;Block'␤»
pmichaud_ rakudo: say Perl6::Compiler.compile('say <hello>', :target('past')).WHAT;
p6eval rakudo e61569: OUTPUT«PAST::Block()␤»
masak DanielC: over the last year, I've repeatedly lost and re-gained that enterprising spirit that makes me go after workarounds. I much prefer when I have it; it makes for more interesting solutions. :)
15:42 scrottie joined
DanielC masak: But the blog about the Josephus is not bad. It shows that Perl 6 now has really nice OOP syntax. It slightly beats both Python and Ruby. 15:42
pmichaud_ we could see about adding a .perl to PAST nodes
masak DanielC: sorry -- when you say 'blog' you mean, like, a blog post. right?
DanielC yeah, sorry
post
masak just checking.
DanielC: 'beats'? in length, you mean? 15:43
DanielC yeah
both by line count and character count.
I was surprised that Perl 6 was shorter than Python, but it is.
TimToady seems like there would be a much shorter FP solution... 15:44
masak it actually took me a while to get used to some aspects of Perl 6 OO.
some parts of the journey hurt.
but now I really like it.
DanielC And I made every effort to implement the code the same way in all languages so no one can say that I was biased.
pmichaud_ I think that actually shows a bit of a bias in a different direction 15:45
DanielC I made no real effort to make Perl 6 compact. It came out compact and easy to read all on its own.
pmichaud_ or, at least, it *can*
(not sure it does in this case)
the blog post is really cool, btw
masak it is. DanielC++
DanielC pmichaud_: Well, I can't avoid being biased (Perl rulez!). All I can do is try to control it.
thanks :-)
masak DanielC: typo in a comment in all the code snippets: s/Kill ever/Kill every/ 15:46
DanielC thanks 15:47
15:47 hanekomu left
DanielC goes fix it 15:47
pmichaud_ I mean, how does one write the equivalent of my @a = @b >>*<< @c >>+<< @d; in Perl 5? ;-)
DanielC masak: fixed, thanks 15:48
TimToady in fact, there's probably a one line solution using [^$n] ... {}
masak pmichaud_: what's the precedence of that one? same as with * and +?
pmichaud_ masak: yes, I believe so.
the hypers maintain their precedence
masak ah, makes sense.
TimToady yes, hypes are transparent, while X force list infix 15:49
*forces
masak [] makes infix into prefix, but still doesn't change precedence?
pmichaud_ makes it a listop
DanielC rakudo: my @a=<1 1>; my @b = @a; my @c =<4 4>; (@a >>*<< @b >>+<< @c).perl.say
15:49 pmichaud_ is now known as pmichaud
p6eval rakudo e61569: OUTPUT«[5, 5]␤» 15:49
masak oh, right. listop.
DanielC rakudo: my @a=<2 2>; my @b = @a; my @c =<4 4>; (@a >>*<< @b >>+<< @c).perl.say 15:50
p6eval rakudo e61569: OUTPUT«[8, 8]␤»
DanielC Well, it behaves rationally.
15:54 alanhaggai left
masak going on a long cross-country bus trip tonight. expect to get some Perl 6 hacking done on the way. 15:56
barney Shouldn't line 1932 of action.pm simply say; $past<itype> := 'Perl6Scalar'; instead of $past<itype> := container_itype('Perl6Scalar');
'Perl6Scalar' is no Sigil
15:56 szabgab left
pmichaud looking 15:56
15:57 meppl joined, Gothmog_ joined
pmichaud I'm guessing it should actually use the sigil 15:57
masak pmichaud: I'm not sure I understood what you meant by "you can always get Rakudo's AST format" -- it sounds intriguing, though.
pmichaud masak: Rakudo compiles first to PAST (its AST format) before going on to anything else
from the command line we do this via --target=past
masak ooh. 15:58
yes, I'll try that.
15:58 mkfort_ joined
masak thanks. 15:58
pmichaud but it can be done from code by calling the compiler directly, and passing :target('past')
masak even better.
pmichaud unfortunately the PAST nodes don't know how to .perl themselves, but we could come up with something there 15:59
masak yes, this looks like what I want, and not --target=parse.
pmichaud: I'll make a try at making PAST nodes .perl themselves. 16:00
16:00 payload left
pmichaud ultimately they're just Captures (like Match objects 16:00
but they can/do have cycles in them
pmichaud decides to bring back '_dumper' for now. 16:01
16:01 synth joined
TimToady sub josephus ($n) { [^$n] ... { .push(.shift); .push(.shift); return .[0] if @$_ == 1; .shift; $_ } } 16:01
or some such
16:02 dakkar left
TimToady well, using ... is probably overkill, since it's just looping over the same array 16:04
16:04 mkfort left
masak that infix:<...> operator is scary. 16:05
anyway, going home to make food. 16:06
have a good weekend, everyone!
16:06 masak left
pmichaud japhb: I can't bump PARROT_REVISION to 39521 yet -- something in Parrot at that revision is causing t/spec/integration/99problems-11-to-20.t to fail in the spectests. 16:06
barney Is there a way to follow the references in the dumped PAST? e.g. [0] => \past
pmichaud barney: the \past occurs whenever it refers to a node that was already displayed 16:07
it's something about the dumper code that is causing it to not display the full reference properly.
(I've not quite figured out what)
probably need to look at dumper.pbc and/or the 'dump' method
TimToady rakudo: my @p = ^10; say gather while @p { take splice(@p, 2 % @p, 1) } 16:09
p6eval rakudo e61569: OUTPUT«Could not find non-existent sub splice␤»
TimToady aww
pmichaud @p.splice
lambdabot Unknown command, try @list
pmichaud rakudo: my @p = ^10; say gather while @p { take @p.splice(2 % @p, 1) } 16:10
p6eval rakudo e61569: OUTPUT«No applicable candidates found to dispatch to for 'splice'␤in Main (/tmp/Fr7ClbT5Kr:2)␤»
pmichaud o_O
rakudo: my @p = ^10; say @p.splice(2 % @p, 1);
p6eval rakudo e61569: OUTPUT«No applicable candidates found to dispatch to for 'splice'␤in Main (/tmp/wC2lIwPTR1:2)␤»
pmichaud rakudo: my @p = ^10; say @p.splice(2 % +@p, 1);
p6eval rakudo e61569: OUTPUT«No applicable candidates found to dispatch to for 'splice'␤in Main (/tmp/YnBL8QJRZO:2)␤»
pmichaud bah. 16:11
rakudo: my @p = ^10; say @p.splice(2, 1);
p6eval rakudo e61569: OUTPUT«2␤»
pmichaud rakudo: my @p = ^10; say 2 % @p;
p6eval rakudo e61569: OUTPUT«2␤»
TimToady rakudo: my @p = ^10; say gather while @p { take @p.splice(int(2 % @p), 1) }
p6eval rakudo e61569: OUTPUT«2345678901␤»
pmichaud rakudo: my @p = ^10; say (2 % @p).WHAT;
p6eval rakudo e61569: OUTPUT«Num()␤»
pmichaud I bet someone constrained it to Int
TimToady oh, that's not right
Bad Algorithm Go Home!!! 16:12
pmichaud (agreed, 2 % @p should be Int)
(not easily done just yet.)
TimToady I mean my josephus is wrong
sbp try psellus 16:13
TimToady rakudo: my @p = ^10; say gather while @p { @p.push(@p.splice(0,2); take @p.shift } 16:14
p6eval rakudo e61569: OUTPUT«Statement not terminated properly at line 2, near "(@p.splice"␤in Main (src/gen_setting.pm:0)␤»
TimToady rakudo: my @p = ^10; say gather while @p { @p.push(@p.splice(0,2)); take @p.shift }
p6eval rakudo e61569: OUTPUT«2581607493␤»
TimToady there you go 16:15
DanielC: see ^^^ for a more idiomatic josephus
ruoso DanielC++ # nice post 16:16
DanielC TimToady: I'm blind. Where? 16:17
TimToady my last rakudo command
DanielC rakudo: my @p = ^10; say gather while @p { @p.push(@p.splice(0,2)); take @p.shift }
p6eval rakudo e61569: OUTPUT«2581607493␤»
TimToady it was 50, right? 16:18
rakudo: my @p = ^50; say gather while @p { @p.push(@p.splice(0,2)); take @p.shift }
DanielC Should be 28...
p6eval rakudo e61569: OUTPUT«25»
DanielC I think
TimToady rakudo: my @p = ^50; say gather while @p { @p.push(@p.splice(0,2)); take @p.shift }
DanielC I hope it's 28! Otherwise all my versions are wrong!
p6eval rakudo e61569: OUTPUT«25811»
pmichaud TimToady: rakudo resource limits
DanielC rakudo: my @p = ^40; say gather while @p { @p.push(@p.splice(0,2)); take @p.shift }
p6eval rakudo e61569: OUTPUT«258» 16:19
pmichaud moritz_: see "resource limits"
p6eval is aborting the process early
16:19 pmurias left
DanielC I don't understand the output though. What is 258 supposed to mean? 16:19
pmichaud kill 2, then kill 5, then kill 8
DanielC ruoso: thanks
ah
pmichaud rakudo: my @p = ^40; say (gather while @p { @p.push(@p.splice(0,2)); take @p.shift }).join(',') 16:20
p6eval rakudo e61569: OUTPUT«2,5,8,11,14,17,20,23,26,29,32,35,38,1,6,10,15,19,24,28,33,37,3,9,16,22,30,36,4,13,25,34,7,21,39,18,0,31,12,27␤»
TimToady yes, it's 28
allowing for 0-based
DanielC my @p = ^40; say (gather while @p { @p.push(@p.splice(0,2)); take @p.shift }).join(','); say @p 16:21
TimToady you want while @p > 1 16:22
DanielC rakudo: my @p = ^40; say (gather while @p { @p.push(@p.splice(0,2)); take @p.shift }).join(','); say @p
p6eval rakudo e61569: OUTPUT«2,5,8,11,14,17,20,23,26,29,32,35,38,1,6,10,15,19,24,28,33,37,3,9,16,22,30,36,4,13,25,34,7,21,39,18,0,31,12,27␤␤»
DanielC is trying to get Rakudo to say "the winner is 28"
TimToady if you want the last one in @p
pmichaud rakudo: my @p = ^40; say (gather while @p { @p.push(@p.splice(0,2)); take @p.shift })[*-1]
p6eval rakudo e61569: OUTPUT«27␤»
DanielC rakudo: my @p = ^40; say (gather while @p > 1 { @p.push(@p.splice(0,2)); take @p.shift }).join(','); say @p 16:23
p6eval rakudo e61569: OUTPUT«2,5,8,11,14,17,20,23,26,29,32,35,38,1,6,10,15,19,24,28,33,37,3,9,16,22,30,36,4,13,25,34,7,21,39,18,0,31,12␤27␤»
ruoso working in another OO implementation of it... although it is certainly not a classical OOP problem, more of a functional problem instead
DanielC rakudo: my @p = ^40; (gather while @p > 1 { @p.push(@p.splice(0,2)); take @p.shift }).join(','); say @p
p6eval rakudo e61569: OUTPUT«27␤»
DanielC yay
Can you remove the "gather"? 16:24
rakudo: my @p = ^40; (while @p > 1 { @p.push(@p.splice(0,2)); take @p.shift }).join(','); say @p
p6eval rakudo e61569: OUTPUT«Died␤in Main (/tmp/HiA1rvTOnN:2)␤»
TimToady rakudo: my @p = 1..40; while @p > 1 { @p.push(@p.splice(0,2)); take @p.shift }; say @p
DanielC oops
p6eval rakudo e61569: OUTPUT«Died␤in Main (/tmp/kUtcUyHHAw:2)␤»
TimToady rakudo: my @p = 1..40; while @p > 1 { @p.push(@p.splice(0,2)); @p.shift }; say @p
p6eval rakudo e61569: OUTPUT«28␤»
DanielC Beautiful. 16:25
TimToady seems like the push/splice is wrong
16:25 Psyche^ joined
pmichaud rakudo has a push problem I think. 16:25
it tends to push references instead of values
TimToady but maybe it doesn't matter once it gets down to one element
hmm, if push returned the array, we could say @p.push(@p.splice(0,2)).shift while @p > 1 16:27
DanielC TimToady: I just added the compact solution as a comment to the blog post.
TimToady if we had rotate we could do @p.rotate.rotate.shift 16:28
16:28 M_o_C joined
TimToady or @p.rotate(2).shift 16:28
pugs_svn r27060 | pmichaud++ | [spec]: Correct buglet in example -- 16:30
r27060 | pmichaud++ | squaring elements of array is @a »**» 2, not @a [**] 2.
r27061 | pmichaud++ | [spec]: Correct typo: "fullwith" --> "fullwidth"
r27062 | lwall++ | [Containers] add @array.rotate, define push and unshift to return array 16:33
r27063 | lwall++ | [Containers] default rotation is 1 16:34
TimToady okay, the solution is now officially @p.rotate(2).shift while @p > 1 16:35
16:37 pmurias joined
Roxas Drink the pineapple juice before you make the apple juice. 16:39
DanielC @p.rotate(2).shift would be really cool. 16:41
16:41 Chillance joined
DanielC my @p = 1..40; @p.rotate(2).shift while @p > 1; say @p 16:41
16:42 Patterner left, Psyche^ is now known as Patterner
ruoso is rotate in-place modifier? or does it return a new list? 16:43
viklund One more test passed
DanielC In place would be cooler.
ruoso DanielC, nah... that would break functional mind-set 16:44
DanielC shift, pop, splice, push and unshift are all in-place.
@p.rotate($n) is just @p.push(@p.splice(0,$n))
lambdabot Unknown command, try @list
DanielC But it's easier to read. 16:45
I wonder if TimToady will add @p.rotate($n). 16:46
my @p = 1..40; @p.rotate(2).shift while @p > 1; say @p
I'd love to do that...
ruoso if it weren't in-place... it would be even prettier...
or not...
ah... forget it, misfire... 16:47
DanielC :)
pmichaud looks like it's inplace from the description
pugs_svn r27064 | lwall++ | [Containers] more clarifications
DanielC pmichaud: Description? You mean it exists?
TimToady thought about requiring .=rotate
pmichaud DanielC: TimToady just added them. 16:48
svn.pugscode.org/pugs/docs/Perl6/Sp...ainers.pod
on channel, that's what the following means
DanielC *click*
pmichaud 16:33 <pugs_svn> r27062 | lwall++ | [Containers] add @array.rotate, define push and unshift to return array
16:34 <pugs_svn> r27063 | lwall++ | [Containers] default rotation is 1
i.e., @array.rotate was added to the spec in r27062. 16:49
in r27063 it was clarified to have a default rotation of 1.
DanielC COOL
pmichaud I might even get around to adding them to Rakudo today :-P
DanielC I just showed my fiancee. She loved it :-)
I had just finished explaining to her how TimToady's code worked. 16:50
ruoso you're certainly a romantic guy 16:51
pmichaud shift, splice, and push can be really .... ummm, never mind.
DanielC She is geeky enough to appreciate a nice algorithm.
moritz_ boo, with latestet parrot t/spec/S10-packages/basic.t assertion-aborts on exit
pmichaud moritz_: yes.
moritz_: the parrot folks have been working on cleaning up context gc so it doesn't leak, and there are a few issues yet. 16:52
16:52 rfordinal joined
moritz_ and we have two passing TODOs, which I might clean up later 16:52
viklund ahh, now json passes all tests 16:56
lets see if I can add a few more
moritz_ pulls
pugs_svn r27065 | moritz++ | [t/spec] unfudge two passing tests for rakudo
moritz_ viklund: if you happen to have written a makefile, feel free to add it 16:58
(or a build script, or whatever)
dalek kudo: 9aa210f | pmichaud++ | src/setting/Array.pm:
Remove incorrect constraints on splice() arguments.
kudo: 31443c0 | pmichaud++ | src/classes/Str.pir:
Add some more ranges to Str.succ .
viklund moritz_: ok 16:59
i made one actually
but first: supper
pmichaud rakudo: say "updated?"; 17:02
p6eval rakudo e61569: OUTPUT«updated?␤»
pmichaud not yet. :-|
moritz_ triggers a pull/rebuild manually
TimToady DanielC: pge's proto is not the same as p6's proto 17:03
pge's proto is functioning at the macro level, not the grammar level 17:04
17:06 barney left 17:07 M_o_C left
pmichaud rakudo: say ('⚀⚀'..'⚅⚅').join(' '); 17:08
p6eval rakudo 31443c: OUTPUT«⚀⚀ ⚀⚁ ⚀⚂ ⚀⚃ ⚀⚄ ⚀⚅ ⚁⚀ ⚁⚁ ⚁⚂ ⚁⚃ ⚁⚄ ⚁⚅ ⚂⚀ ⚂⚁ ⚂⚂ ⚂⚃ ⚂⚄ ⚂⚅ ⚃⚀ ⚃⚁ ⚃⚂ ⚃⚃ ⚃⚄ ⚃⚅ ⚄⚀ ⚄⚁ ⚄⚂ ⚄⚃ ⚄⚄ ⚄⚅ ⚅⚀ ⚅⚁ ⚅⚂ ⚅⚃ ⚅⚄ ⚅⚅␤»
pmichaud rakudo: say ('Α'..'Ω').join(' '); 17:09
p6eval rakudo 31443c: OUTPUT«Α Β Γ Δ Ε Ζ Η Θ Ι Κ Λ Μ Ν Ξ Ο Π Ρ Σ Τ Υ Φ Χ Ψ Ω␤»
ruoso Cool!
[particle] :)
TimToady rakudo: say <⚀⚀ ⚀⚁ > ... *
p6eval rakudo 31443c: OUTPUT«Statement not terminated properly at line 2, near "... *"␤in Main (src/gen_setting.pm:0)␤»
ruoso rakudo: say '𐌰'..'𐍊'
p6eval rakudo 31443c: ( no output )
ruoso isn't there some generic way of detecting that sequences? 17:10
pmichaud what sequence is that?
(short answer: doesn't appear to be a generic way)
ruoso that is the gothic alphabet
viklund rakudo: say ('⚀⚀'..'⚅⚅').pick 17:12
p6eval rakudo 31443c: OUTPUT«⚅⚄␤»
pmichaud viklund: looks like you win!
viklund 11! yay
ruoso how can I get the description of the character? 17:13
[particle] needs a font with better unicode support
pmichaud I don't think we have a way to do that (in Rakudo, perhaps not defined in Perl 6)
ruoso I saw that here some time ago 17:14
as an output of rakudo in the evalbot
TimToady hmm, the spec is contradictory in saying increment only does \w chars and then mentioning dice, i think
ruoso the thing that returns "GOTHIC LETTER AHSA" for "𐌰" 17:15
17:15 DemoPhreak left
TimToady rakudo: say '⚀' ~~ /\w/ 17:15
p6eval rakudo 31443c: OUTPUT«␤»
pmichaud does the spec say that increment only does \w chars?
TimToady S03:431 seems to think so 17:16
pmichaud rakudo: my %dice = ( '⚀⚀'..'⚅⚅' Z ( 1..6 X+ 1..6 ) ); say %dice.perl;
p6eval rakudo 31443c: OUTPUT«{"⚀⚀" => 2, "⚀⚁" => 3, "⚀⚂" => 4, "⚀⚃" => 5, "⚀⚄" => 6, "⚀⚅" => 7, "⚁⚀" => 3, "⚁⚁" => 4, "⚁⚂" => 5, "⚁⚃" => 6, "⚁⚄" => 7, "⚁⚅" => 8, "⚂⚀" => 4, "⚂⚁" => 5, "⚂⚂" => 6, "⚂⚃" => 7, "⚂⚄" => 8, "⚂⚅" => 9, "⚃⚀" => 5, "⚃⚁" => 6, "⚃⚂" => 7, "⚃⚃" => 8, "⚃⚄" => 9, "⚃⚅" => 10, "⚄⚀" => 6,
.."⚄⚁" …
TimToady but it's probably wrong
ruoso depending on how standard the definition of "GOTHIC LETTER AHSA" is, you could probably get the "GOTHIC LETTER" part, and follow to the next one until you get a different "group" 17:17
that would be a generic way of doing it
pmichaud ruoso: there are a couple of places where we need to skip codepoints (in the greeks, for example) 17:18
ruoso hmm... that's a shame 17:19
pugs_svn r27066 | lwall++ | [S03] tweak rangechars not to be limited to \w
japhb pmichaud: sorry to hear about PARROT_REVISION to 39521 specfail. Ah well, here's hoping we can bump it to a higher number soon, as this allows Rakudo to 'use Foo:from<parrot>;' . Speaking of which, did you want to do the deprecation honors?
pmichaud japhb: what gets deprecated?
ruoso pmichaud, is there an easy way to get all characters from the "GOTHIC LETTER" group?
pmichaud japhb: and we'll have to bump it to a higher revision on Tuesday
ruoso: I don't think ICU gives me an easy way to do that, no.
japhb pmichaud: you said you wanted to immediately deprecate the import/export system, so that it was effectively marked "experimental" -- there being no direct way to do the latter. 17:20
pmichaud oh, that. Just add notes to DEPRECATED.pod
japhb: also, we'll have to bump the revision on Tuesday at the latest, so it should be addressed by then 17:21
TimToady basically Perl 6 defines it own rangechar class to be all codepoints with a known cyclical successor
pmichaud does the successor have to be at "codepoint plus 1" ?
TimToady nope
pmichaud right.
TimToady spec says to skip trailing sigma, for instance
pmichaud (which Rakudo does :-) 17:22
japhb pmichaud: My understanding of DEPRECATED.pod is that you need a TT for each entry. Do you want me to create a basic TT referencing your 'experimental' wishes?
pmichaud sure.
TimToady a TimToady? a TemplateToolkit? :P
pmichaud japhb: also, I'm currently bisecting parrot revisions to find the one that is causing the breakage, so we can report that.
I *think* it's likely a gc issue. 17:23
japhb excellent.
pmichaud TimToady: TT in parrot speak is "Trac Ticket"
TimToady yes, I know
pmichaud okay.
off to pick up daughter from Mandarin class
TimToady but till now that usage as primarily been in #parrot, not #perl6...
*has
17:26 mizioumt1 left
japhb TimToady: my apologies; sometimes I forget to switch to the other tab while talking to someone who is in both channels -- and we're talking about something that applies at the border between them. :-/ 17:28
TimToady np 17:31
just giving pm a hard time
well, and you
:)
japhb :-) 17:38
viklund moritz_: now it has a build system as well
pmichaud the fact that TimToady gives me a hard time is how I know he still cares :-) 17:44
17:44 rindolf joined
pmichaud my $dice = '⚅'; say ++$dice; 17:45
17:45 molaf joined
pmichaud rakudo: my $dice = '⚅'; say ++$dice; 17:45
p6eval rakudo 31443c: OUTPUT«⚀⚀␤»
17:46 mizioumt joined, Meldrake joined 17:52 DanielC left 17:56 bejuryu left 18:19 felipe joined
pmurias ruoso: so undefined object functioning as types is the way i should implement types in mildew? 18:21
ruoso pmurias, that's the way it is in Perl 6 at the moment 18:22
18:24 frew_ joined 18:29 M_o_C joined
moritz_ viklund++ 18:29
viklund moritz_: I think I fixed the \uXXXX parsing as well 18:34
;)
18:36 DanielC joined
moritz_ viklund: parsing or transforming? 18:40
viklund transforming
I've added two small tests 18:41
moritz_ viklund++ again
I think we need a #perl6-all channel with reporting bots of the various Perl 6 projects 18:42
it might get a bit croweded if we did that here
viklund yes, I've been thinking of writing one in p6
but then I need a json parser for github ;)
IO::Socket::INET should be enough for the rest I think 18:43
need to implement the IRC protocol though
moritz_ json parser... ;-) 18:44
ruoso TimToady, considering "my @a = 1,[2,[3,4]];" is @a[1;1;1] a valid access?
or does that only applies to captures?
18:44 frew_ left
DanielC my @a = (1, 2, 3; 4, 5, 6; 7, 8, 9); @a.perl.say 18:44
ruoso std: my @a = 1,[2,[3,4]]; say @a[1;1;1] 18:45
japhb #perl6-bot-party
p6eval std 27066: OUTPUT«ok 00:03 38m␤»
DanielC rakudo: my @a = (1, 2, 3; 4, 5, 6; 7, 8, 9); @a.perl.say
p6eval rakudo 31443c: OUTPUT«[7, 8, 9]␤»
DanielC What does ; do?
ruoso DanielC, it says there is one other level of captures 18:46
but rakudo is not implementing it correctly now
japhb It's the interdimensional rift operator
:-)
DanielC one other level of capture == ?
japhb DanielC: Capture containing captures.
viklund moritz_: the fix for the \uXXXX was changing the grammar slightly
moritz_ it's supposed to be the same as @a = \(1, 2, 3), \(4, 5, 6), \(7, 8, 9) # not entirely sure though 18:47
DanielC So, it's like (1,2,3,(4,5,6,(7,8,9))) ?
moritz_ DanielC: but if you assign it to an array, it's flattened anyway
DanielC ah, no, thanks
moritz_ DanielC: so it's the same as @a = 1..9; really
DanielC Yeah, but you could do something interesting with it, no? 18:48
ruoso moritz_, looks like ((1,2,3),(4,5,6),(7,8,9)) as well... which looks easier
DanielC Something more interesting than giving it to an array.
moritz_ yes, like assigning it to @@a or a scalar
DanielC ok
ruoso which remembers me of the ¢a idea
18:50 molaf left
viklund delving deep into RFC-land 18:55
pmurias ruoso: does the default ACCEPTS method disappear in defined punned class instances?
ruoso the common understanding is that it will behave like a multi
with a constraint in the invocant
so you can simply do a if self.defined { ... } else { ... } for bootstrapping purposed 18:56
*purposes
pmurias ruoso: and the else ... should contain what? 19:09
ruoso the code to make $signature ~~ Signature to work 19:10
since $signature ~~ Signature actually means Signature.ACCEPTS($signature)
19:11 hanekomu joined
pugs_svn r27067 | moritz++ | [t/spec] RT #61988, $.foo form did not accept arguments 19:13
19:13 masak joined
masak greetings, fellow hacktivists. 19:13
masak is moving about 100 km/h 19:14
19:14 masak left
moritz_ oh hai masak 19:14
19:14 masak joined
ruoso masak, it looks like your connection gets a bit flacky at 100km/h 19:15
;)
19:15 jantore left
masak ruoso: yes. guess you can't have everything. 19:16
I'm also a constant three meters above ground.
moritz_ masak: is it a train? 19:18
masak no, a bus.
DanielC @seen pmichaud
lambdabot pmichaud is in #perl6. I last heard pmichaud speak 1h 33m 25s ago.
ruoso a flying bus?
masak pmichaud: a closer look at PAST reveals that it's already too digested for my aims.
pmichaud lambdabot: shhhh! I'm not really here!
masak: possibly, yes.
DanielC :-) 19:19
pmichaud however, the PAST tree does contain the parse tree as well :-)
masak ruoso: yes, we have flying buses in Sweden.
DanielC pmichaud: I can't find pgc.pir in the Rakudo distribution.
pmichaud ...pgc.pir ?
DanielC pmichaud: PGE
pmichaud PGE is part of parrot
DanielC Parrot Grammar Engine?
masak pmichaud: I guess. :) I'm not interested in the parse tree as such, but the structure of the program.
19:20 donaldh left
masak DanielC: Parser Grammar Engine. 19:20
DanielC pmichaud: Yeah, but Rakudo includes parrot, no?
masak (I learned that from Wikipedia today)
DanielC in the rakudo/parrot directory.
pmichaud DanielC: sure, Rakudo is built from parrot. It'd be in the parrot/ subdirectory
more to the point, PGE is parrot/compilers/pge/
19:20 donaldh joined
pmichaud I think there might have been a pgc.pir at one time, but it's long gone 19:20
DanielC pmichaud: I must have done something wrong. I could not find it with 'find' 19:21
PGE.pir ?
[particle] wikis are always wrong.
DanielC [particle]: And slide presentations too :) I got pgc.pir from one of pmichaud's presentations. 19:22
pmichaud DanielC: what in particular are you looking for with PGE?
DanielC: from how long ago?!? ;-)
DanielC parrot pgc.pir simple.pg >simple.pir
^^^ that's what I'm looking for.
pmichaud it's now Perl6Grammar.pbc
in runtime/parrot/library/PGE
masak pmichaud: 2007, methinks. :/
sorry, I gave him that link.
viklund bah, it will take too long to read the RFC, I'll go read Net::IRC instead
DanielC thanks
pmichaud we still had pgc in 2007? wow. 19:23
moritz_ viklund: it might be even easier to read a dumb of the protocoll ;-)
masak pmichaud: no, 2006 :/ www.pmichaud.com/2006/pres/yapc-par...start.html
DanielC pmichaud: Wait... "runtime"? That's not in Rakudo, is it?
ah... parrot/runtime
not runtime/parrot :) 19:24
pmichaud parrot/runtime/parrot/library/PGE
DanielC ah, there it is.
thanks
[particle] anyone here good at setting up terminal, irssi, and %ENV set up for unicode? 19:25
DanielC pmichaud: Can I copy that file to a different directory? Or does it depend on other files in that directory?
pmichaud I think it can be copied around.
viklund moritz_: aaahh, a dump...
DanielC pmichaud: Seems to work. :-) 19:26
moritz_ viklund: right ;-)
19:26 szabgab joined
viklund yes, that might be better 19:26
19:27 masak left, masak joined
viklund masak: 3g? 19:27
masak (oops.)
pugs_svn r27068 | pmurias++ | [re-smop] start of implementing .ACCEPTS for punned classes
masak viklund: no, bus.
viklund bus wifi? 19:28
masak viklund: for the win.
viklund cool
masak swebusexpress++
japhb bus wifi is getting fairly common, it seems.
Of course, when the bus is out of cell tower range (so it can't provide the next hop), you tend to experience fail.
Muixirt pmichaud, please look at nopaste.snit.ch/16881
why is new $P20, "Int" inside the loop? 19:29
TimToady ruoso: see my latest multi-dim speculation in p6l 19:30
ruoso TimToady, you sent it just now? 19:31
or is something older?
19:32 estrabd left
viklund wireshark requires... gnome-keyring?? 19:32
TimToady oh wait, hasn't left my mqueue yet...
ruoso ah...
pmichaud Muixirt: we don't do any loop constant factoring yet
japhb viklund: that's odd. Perhaps so it can show you decrypted traffic?
masak Muixirt: told you. :) 19:33
ruoso was pressing 'refresh the imap folder' constantly
TimToady stupid isp doesn't like me forwarding spam for some reason... 19:34
anyway, proposing @a[] to replace @@a
ruoso uh?
you mean @a[] = ? 19:35
TimToady if @@ is a capture sigil, it's a no-op
but we want a coercion
ruoso we do?
Muixirt pmichaud, the assignment of the constant integer isn't a problem
masak "On #perl6, we're so up-to-date that we get confused by p6l emails that haven't even been sent yet." 19:36
pmichaud Muixirt: what "problem" are you referring to, then?
ruoso TimToady is a marketing wizard... he can even make a teaser about a mailing list post... 19:37
:P
masak "Coming soon to an inbox near you!"
Muixirt pmichaud, that while 1 { }; leaks mem
19:37 estrabd joined
pmichaud Muixirt: the problem is very likely not rakudo. 19:37
it's very likely an issue in parrot.
ruoso TimToady, btw... I really thought the capture sigil was precisely there to mean "no coercion" 19:38
pmichaud Muixirt: here, I'll give a test program that is pure parrot as a demo
ruoso actually... it wasn't still there, but it would be there...
pmichaud (building a parrot demo... one moment) 19:39
ruoso still hitting "refresh" in his IMAP folder
Muixirt pmichaud, nopaste.snit.ch/16885
pmichaud Muixirt: ...?
Muixirt that example leaks mem and is derived from the rakudo example 19:41
pmichaud right
TimToady okay, reset the network, and email went
pmichaud Muixirt: that tells me it's a parrot bug 19:42
Muixirt: as parrot shouldn't be leaking memory there
Muixirt: how are you measuring "leaked memory", ooc?
Muixirt pmichaud, via top and strace -c 19:43
pmichaud Muixirt: anyway, it's a parrot bug, not a rakudo one.
dalek kudo: de1e9f0 | moritz++ | docs/ChangeLog:
[docs] initial changelog for 2009-06 release
Muixirt pmichaud, i think there are two bugs
one of parrot for not collecting $P0
and rakudo for placing the $P0 allocation inside the loop 19:44
pmichaud Muixirt: I already said we don't do loop constant factoring
ruoso TimToady, there is one thing that gets lost in that behavior
my ¢a = (1,2,3;4,5,6;7,8,9); my @b = ¢a.rotate(1;1); 19:45
Muixirt pmichaud, so the $P0 allocation and the assignment to it cannot be split?
pmichaud Muixirt: suppose the loop was instead
while foo() { }
the call to foo() would have to be inside the loop, yes?
ruoso TimToady, unless you meant that only "in addition" to the behavior described in the above code
so the above would also work
TimToady use @b[] = 19:46
ruoso no... I really meant to flatten at that point
my point was precisely to have an unflattened structure, manipulate it, and only later flatten it
TimToady @b = flattens 19:47
lambdabot Maybe you meant: b52s babel bf bid botsnack brain bug . ? @ v
19:47 payload joined
TimToady maybe you have a brain bug, lambdabot 19:47
ruoso right.. so your email only describes new behavio
masak @botsnack
lambdabot :)
ruoso not replacing behavior ;)
TimToady correct
pmichaud Muixirt: from rakudo's perspective, a constant 1 currently results in $P0 = new 'Int'; assign $P0, 1; 19:48
19:48 Helios left
pmichaud it does not attempt to factor out all of the constant 1's that might exist into a single PMC 19:48
ruoso TimToady, I think that's cool, and solves a lot of semantic issues as well...
19:49 Helios joined
pmichaud it does not attempt to pre-allocate the PMCs before the assignment (as that could also be wrong) 19:49
what you're describing is properly an optimization, not a bug.
ruoso TimToady, and I do think that means '@@' should die... leaving room for '@%' and '¢' in its place
Muixirt pmichaud, ok thanks for the explanation 19:50
19:50 M_o_C left 19:51 masak left
ruoso TimToady, one thing that should be noted, otoh, is that @a[] = (1,2,3;4,5,6;7,8,9) would create 4 new arrays.... 19:52
pmichaud Muixirt: it could indeed make a lot of sense to move all of the constant initializations to the beginning of the sub. 19:53
ruoso in contrast to "¢a = (1,2,3;4,5,6;7,8,9)" which would simply store that capture in '¢a'
TimToady sure, but that's how you say you want mutable subarrays
pmichaud I may see about doing that in PCT. We would get a speed and memory benefit.
ruoso TimToady, right... makes sense... 19:54
pmichaud (we would get a penalty by constructing constants that might not get used, though)
TimToady a penalty that Perl 5 pays continually, with not that much ill effect :) 19:55
pmichaud TimToady: constructing constants at the beginning? 19:56
or on each reference?
TimToady pmichaud: I don't follow 19:57
pmichaud right now we construct PMCs for each constant that appears in the program
currently the PMC gets constructed at the point where the constant is actually used
so, if we have a loop: while foo() { say 1; } 19:58
then the "1" constant ends up creating a PMC on each loop iteration
TimToady why?
ruoso pmichaud, but that certainly can be optimize
pmichaud that's just the way the code generates at the moment
ruoso *optimized
pmichaud ruoso: yes, that's what I'm talking about.
TimToady seems like a waste for value semantics
pmichaud which part of "waste" are you referring to? ;-) 19:59
Muixirt and the problem is that parrots GC doesn't catch that
ruoso has a feeling that we're not talking about the same thing 20:00
TimToady you should only ever have to create a single 1 pmc, and just reuse it, meseemeth
pmichaud Muixirt: I'm after a different issue.
TimToady as long as it's sufficiently readonly
pmichaud TimToady: right. So the question is, when does that 1 PMC get created?
at the beginning of the program?
ruoso my guess is BEGIN
TimToady sometime between the time it's compiled and the time it's used
pmichaud so my question is, does Perl 5 currently construct all such constants at the beginning? 20:01
ruoso but that's one of the issues of code compilation that I don't see really solved yet
pmichaud or does it do something more lazily?
ruoso pmichaud, it stores in the optree as constant, iirc
which means that it's done in the "compiling"
pmichaud I don't presently have a way to store the already-compiled-PMC in the compiled parrot code. I have to construct it at "runtime". 20:02
I can construct it at the beginning, though, and refer to it thereafter.
(then the questions become: (1) where do I store them for easy lookup, and (2) is it more expensive to look them up than to create new ones?) 20:03
TimToady the compiler has to be able to set up data and pass it preconstructed to the run time
20:03 amoc is now known as amoc^sleep
ruoso pmichaud, that's why I think you need to be able to serialize things into the bytecode 20:03
pmichaud ruoso: I don't disagree. I'm working with what I have available.
ruoso in fact... the result of the compilation is supposed to be a serialized state of the runtime
pmichaud Right now what we do is generate the code that sets up the data as preconstructed to the run time 20:04
ruoso and I'm kinda reaching the conclusion that "use" means "static link"
pmichaud (and yes, this is part of why rakudo's startup is a bit slow)
ruoso that's what mildew is doing now too
TimToady note that the whole YOU_ARE_HERE setting thing is assuming a similar ability to dump the snapshot, just in a different spot than for normal compilation 20:05
pmichaud agreed.
I do expect most of these problems to be solved in Parrot, I just don't expect them in the near future.
Unless, of course, we can convince jnthn++ do to them. :-) 20:06
ruoso my vision for this goes like
if you "use", the "compiled" thing will contain all the "use"d modules 20:07
if you don't want that, you need to "require"
pmichaud (and yes, Parrot does have a serialization mechanism for bytecode in its design, I've just (1) had trouble getting it to work reliably and (2) have questions about how it's supposed to work when there are tons of references lying about)
ruoso so.. "use" is "static link" and "require" is "dynamic link"
TimToady it's really more of a Smalltalkian save-the-current-workspace kind of thinking
ruoso pmichaud, just make sure everything is serialized
and make sure BEGIN don't do anything bad 20:08
pmichaud right now "everything" includes the entire compile environment, I fear.
TimToady well, it has to, for eval to work right 20:09
ruoso but we already know you need that
TimToady hopefully that part doesn't get paged in unless you need it :)
pmichaud you mean also serializing copies of the builtin library subs and the like?
ruoso yes... you probably want to mmap the bytecode instead of reading it
pmichaud, yes... *everything* 20:10
TimToady except
ruoso otherwise you won't really be able to have all the semantics
TimToady again it comes down to a question of identity
pmichaud ruoso: that is likely to be huge. The bytecode image is already 3MB.
TimToady you can cheat as long as you can nail down the identity of everything so it doesn't change
ruoso right... except if your bytecode reader can assume the identity of built-in types 20:11
pmichaud right, that's what I'm saying. Parrot doesn't have support for those kinds of cheats yet.
ruoso but that's not quite easy
TimToady if it's going to grow back to that big anyway, better to have it pregrown
ruoso agreed...
pmichaud it doesn't grow back to that big
ruoso that simply means you don't need to load the setting before running the file 20:12
pmichaud the subs themselves are precompiled
ruoso since the setting is part of the file
pmichaud so we don't have to rebuild them.
we just rebuild the signature and other attributes that we don't have a way of compile-time attaching to the subs.
20:13 mberends joined
TimToady seems like a bogus dichotomy to me 20:13
ruoso the problem is that you're trying too hard to split compile-time from run-time
pmichaud TimToady: I don't entirely disagree; again, I'm kinda constrained by "what actually works in parrot"
pugs_svn r27069 | moritz++ | [t/spec] RT #65674, multi subs in classes 20:14
ruoso having the code compiled down in Perl 6 simply means: hibernate the process before INIT
TimToady well, but pmichaud++'s point is sure to apply to other VMs as well 20:15
maybe what we have is a model more like...
pmichaud ruoso: Let me give a more concrete example.
TimToady there's snapshot that has everything, and
pmichaud let's say I'm compiling the following
sub foo(Int $x) { ... }
TimToady a snapshot that has the initial load for a particular VM
and that intial load knows how to fetch anything else from the complete image as necessary 20:16
pmichaud after compilation, there's an entry '&foo' in my current namespace, that points to a Sub object, that Sub object has a $!signature attribute, that $!signature attribute references the 'Int' type object
if I attempt to serialize &foo, I'm going to get a copy of 'Int' and everything that it references in my serialization
TimToady no, not a copy, if you serialize everything 20:18
you only get one Int, and everyone refs it
pmichaud okay, so I have to serialize the entire compile-time enviornment
ruoso exactly
pmichaud my point is that a serialization of &foo includes everything that it references.
TimToady that's what I'm saying goes into the complete snapshot
the intial loader doesn't have to pull in everything though
*initial 20:19
it can be lazy
20:19 decasm left
pmichaud I'm still missing a piece here 20:19
TimToady espically about loading in the compiler, if there's no eval
sjohnson when is the next State of the Onion?
TimToady the Onion has a new state continually :) 20:20
pmichaud let's suppose I create Foo.pm, and compile it to bytecode. That bytecode compilation has a copy of the entire compiler?
then I create Bar.pm, and compile it to bytecode. It also gets a copy of the entire compile-time environment?
pmurias why not?
TimToady again, you can factor out immutables as long as the identity is fixed 20:21
pmichaud right.
which is the piece I'm claiming is missing from Parrot.
ruoso but...
TimToady which is why I keep insisting that officially installed modules change version when modified
ruoso you can have a compilation that targets "shared library" 20:22
and in that "compilation mode" you produce a specialized bytecode
20:22 Whiteknight left
ruoso but that's different from the "runnable" bytecode 20:22
pmichaud so the correct way of phrasing what Rakudo currently does is that it does the equivalent of a dynamic link when a bytecode module is loaded
ruoso yes... but "use" is "static link"
pmichaud ruoso: I'm sorry, but I really don't envision that we want every pre-compiled bytecode to have a complete copy of the Perl 6 library in it. 20:23
TimToady in the abstract, yes
ruoso that's why the Perl 6 library should be small
but it's the same with C or C++ 20:24
pmichaud the Perl 6 library is anything but small.
300+ operators pretty much guarantees that
(not to mention all of their multi variants)
ruoso you always get the entire libc or libstdc++
pugs_svn r27070 | moritz++ | [t/spec] RT #66552, Null PMC Access on user defined infix:<==>
TimToady you'll note those are dynamically linked these days...
ruoso pmichaud, one thing you're missing
pmichaud What TimToady++ just said -- you get them, but normally through dynamic linking.
Try compiling a program with static linking that uses all of the library and see how big your .o and executable gets :-) 20:25
moritz_ rakudo: say Nil ~~ undef
p6eval rakudo de1e9f: OUTPUT«1␤»
ruoso pmichaud, that might mean we need to rethink the way we're building the Perl 6 CORE 20:26
moritz_ rakudo: say Nil ~~ ()
p6eval rakudo de1e9f: OUTPUT«1␤»
moritz_ rakudo: say Nil.new ~~ ()
p6eval rakudo de1e9f: OUTPUT«1␤»
moritz_ rakudo: say Nil.new ~~ undef
p6eval rakudo de1e9f: OUTPUT«1␤»
ruoso pmichaud, btw... the thing you're missing is that you're going to load the setting anyway... that being statically linked or dinamically linked... 20:27
TimToady any compiled module that starts with the standard setting does not need to serialize anything in the standard setting, as long as it can dynamically link back to the standard setting
pmichaud ruoso: that's not at all what I'm missing.
ruoso and unless we make it use the OS shared library mechanism, it will make little difference to be statically or dinamically linked
pmichaud ruoso: I understand that we're going to load the setting anyway. Your statement "serialize the entire compile-time environment" sounds like I'm having to serialize the setting as well. 20:28
moritz_ rakudo: say (42 if 0) ~~ undef
p6eval rakudo de1e9f: OUTPUT«0␤»
moritz_ rakudo: say (42 if 0) ~~ ()
p6eval rakudo de1e9f: OUTPUT«0␤»
pmichaud and when I do "use Foo;" I've already got a copy of the setting in memory. I don't want to load Foo's copy in on top of that.
ruoso pmichaud, that's what I was talking about a specialized "compiling target" for "shared libraries" 20:29
pmichaud ruoso: which libraries aren't shared?
in a perl 6 sense?
ruoso the ones that were statically linked into your program
pugs_svn r27071 | moritz++ | RT 20:30
pmichaud anyway, the things TimToady is saying I can agree with, and it sounds similar to what we're doing now.
ruoso I mean... a runnable file is usually not a "shared library"
moritz_ dammit, that shouldn't be the whole commit message 20:31
ruoso can we guarantee that serialization of an old version can initialize a newer version of something in the setting?
[particle] ...that makes it easy to agree with...
pmichaud the notion of serializing the entire compile-time environment doesn't fit in my worldview, though. serializing everything up to the parts that we know are "fixed" makes sense to me, and providing a way to link back to them upon load. 20:32
sjohnson question: I haven't learned OOP Perl 5 yet. I have the Camel book. Should I just learn Perl 6 OOP instead?
TimToady the point is that there are boundaries through which reference happens in only one direction, and so while your program can have references into the used modules or into the setting, they cannot have references back into your code
moritz_ sjohnson: that depends on what you want to do
TimToady pmichaud: yes, I think we're in violent agreement on that
moritz_ sjohnson: if you want to program in perl 5, read the camel book.
ruoso my point is: the compilation of a "shared library" generates the bytecode that by being run will define the library...
moritz_ sjohnson: if you want to program Perl 6, learn the Perl 6 OO 20:33
ruoso the compilation of a "runnable object" is the serialization of the vm
ruoso fu**ng lag
pmichaud TimToady: I agree. AFAIK, Parrot doesn't give me a "serialize up to the fixed parts" option yet. It's pretty much "serialize everything".
sjohnson moritz_: ok cool. i have decided that i might as well learn what's "up and coming"
as I will easily like Perl 6
pmichaud regardless, we need the "link to the fixed portions" step in the initialization code, and that's part of what we generate 20:34
we also have to dynamically generate our constants, because they include hooks into the setting that have to be established
ruoso pmichaud, yeah... one thing that got lost here is that I do think we might need to rethink how we define the CORE
pmichaud ruoso: I'm fine with rethinking that.
sjohnson TimToady: have you done your own "Just Another Perl Hacker" program? 20:35
ruoso pmichaud, so it is more "shared library"-like
pmichaud sjohnson: learning the Perl 5 OO will probably be more of a distraction than help if you're wanting to mainly do Perl 6
sjohnson: if you're wanting to do Perl 5 stuff, then I'd recommend looking at Moose
TimToady sjohnson: yes, it went out and looked up all of merlyn's JAPHs on the net, and ran one of them at randome
*random
ruoso pmichaud, we might need a 'header file' kind of thing
one thing this brings up is the need of two different versions of the same module loaded by the run time 20:36
pmichaud anyway, to get back to the issue that started this all -- Rakudo's idea of constants aren't quite read-only enough yet to be able to move them around cleanly, but they'll soon be that way when I've done my rw refactor. 20:37
TimToady we need something yamlish that can not only detect duplicate copies and &1 them, but do the same for things already externally immutable and just put the external ref
so, for instance, the ::OUTER link from ::UNIT to ::SETTING is broken and stored as an identity, not as a tree 20:38
pmichaud Exactly.
TimToady so from the standpoint of the compiler, it *thinks* it's serializing the whole magilla
ruoso TimToady, even if the setting changes version?
TimToady but it isn't 20:39
THE IDENTITY INCLUDES VERSION AND AUTH!!!
20:39 alester joined
TimToady sorry 20:39
ruoso so what happens if that specific version is not in the runtime? 20:40
TimToady in my parlance, if I say "identity", I mean the name of something immutable
you don't throw away old versions of things that are still in use in P6
pugs_svn r27072 | moritz++ | [t/spec] RT #66404
20:41 kate21de joined
TimToady it's just a library GC problem 20:41
ruoso TimToady, but what if you take something compiled a while ago into a machine and sent to another machine with only the newer versions?
ruoso thinking about distributors...
moritz_ rakudo: (my @s)[0] //= 0; say @s.perl
p6eval rakudo de1e9f: OUTPUT«[0]␤»
moritz_ rakudo: (my @s)[0] //= 3; say @s.perl
p6eval rakudo de1e9f: OUTPUT«[3]␤»
TimToady then you make sure they know how to get the old version
which you can unambiguously name, because it has an identity
ruoso alright...
TimToady reconstituting some programs requires more work than other programs 20:42
japhb ... sorta like reconstituting old versions out of version control.
TimToady and, in fact, you could probably do pretty good compression from one version to another :) 20:43
DanielC I'm having trouble with 'token' vs 'rule': pastebin.com/m1e6c0850 <-- This should parse '(name Daniel)' as a list, but it doesn't. :-(
rule List { \( <Word>+ \) } <--- "rule" means that it won't care about spaces, yes? 20:44
pmichaud there's nothing there to match the space between "name" and "Daniel"
StephenPollei so you won't have have a require foo>=3.2 version .. or symbol versioning in the libraries 20:45
DanielC pmichaud: I thought that "rule" meant that I didn't need anything to match the space.
StephenPollei it's always exact match?
pmichaud DanielC: "rule" means that spaces in the rule definition correspond to spaces in the target
<Word>+ doesn't have any spaces
20:46 molaf joined
pmichaud rule List { \( <Word> <Word> \) } would match in this instance 20:46
because there's a space between the two <Word>'s
you probably want
rule List { \( [ <Word> ]+ \) }
DanielC pmichaud: But that doesn't require a space after ( an before ) ?
TimToady bbl & 20:47
pmichaud DanielC: depends on where you want to allow spaces and where they are forbidden
DanielC Why is the space between <Word> and <Word> significant while the space between \( an <Word> is not?
pmichaud the space between \( and <Word> is significant in a rule
it means that spaces can occur between the paren and the first word
DanielC Oh... so in a rule a space means \s* ? 20:48
pmichaud actually it means <.ws>
DanielC ok
I think I finally understand.
pmichaud which by default means "not in the middle of a word"
DanielC I had been really confused about token vs rule, but I think I finally get it: In a rule a space means <.ws> 20:49
pmichaud rule List { \(<Word>\) } matches "(Daniel)" but not "( Daniel )" or "(Daniel )" or "( Daniel)"
rule List { \( <Word> \) } would match all three
DanielC Thanks.
Thank you very much. This has been nagging me for ages. 20:50
DanielC tries
weird 20:51
rule List { \( [ <Word> ]+ \) } <-- should work, right?
token List { \( [ <.ws> <Word> <.ws> ]+ \) } <-- Even this fails, so I must be missing something more fundamental. 20:53
It's the parens. 20:54
Something is wrong with \( and \)
StephenPollei also if they old library has a bug like a security flaw is there anyway to fix it and have it automaticly be used 20:55
DanielC Maybe ( are considered a "word" because they match \S
20:56 icwiener joined
DanielC rakudo: if "(" ~~ / <-[(]> / { say "y" } else { say "n" } 20:56
p6eval rakudo de1e9f: OUTPUT«n␤»
DanielC rakudo: if "(" ~~ / <\S -[()]> / { say "y" } else { say "n" } 20:57
p6eval rakudo de1e9f: OUTPUT«Statement not terminated properly at line 2, near "\\S -[()]> "␤in Main (src/gen_setting.pm:0)␤»
DanielC hm
DanielC is trying to make a regex that matches non-space characters except for ( and ) 20:58
rakudo: if "(" ~~ / <-[()] -ws> / { say "y" } else { say "n" } 21:00
p6eval rakudo de1e9f: ( no output )
DanielC rakudo: if "daniel" ~~ / <-[()] -ws> / { say "y" } else { say "n" }
p6eval rakudo de1e9f: OUTPUT«y␤»
DanielC rakudo: if "( )" ~~ / <-[()] -ws> / { say "y" } else { say "n" }
p6eval rakudo de1e9f: OUTPUT«n␤»
DanielC rakudo: if "()" ~~ / <-[()] -ws> / { say "y" } else { say "n" } 21:01
p6eval rakudo de1e9f: OUTPUT«n␤» 21:02
DanielC rakudo: if "(" ~~ / <-[()] -ws> / { say "y" } else { say "n" }
p6eval rakudo de1e9f: OUTPUT«n␤»
DanielC I wonder why it was "no output" before.
japhb rakudo: my @code = {"Hello, $^a"}, {"Goodbye, $^a"}; @code[0]("World").say 21:03
p6eval rakudo de1e9f: OUTPUT«Hello, World␤»
japhb rakudo: my @code = {"Hello, $^a"}, {"Goodbye, $^a"}; @code>>.("World").say
p6eval rakudo de1e9f: OUTPUT«_block50_block61␤»
japhb How do I make that second one work?
21:05 ruoso left 21:06 molaf left 21:12 payload left
DanielC rakudo: if "(name" ~~ /<-ws>/ { $/.perl.say } else { say "no match" } 21:12
p6eval rakudo de1e9f: OUTPUT«Match.new(␤ # WARNING: this is not working perl code␤ # and for debugging purposes only␤ ast => "a",␤ Str => "a",␤ from => 2,␤ to => 3,␤)␤»
DanielC This can't be right. 21:13
It is matching the "a", not the "n"
or the "(" actually
<ws> means whitespace, so the first non-whitespace character is "("
no?
rakudo: if "(name" ~~ /^ <-ws>/ { $/.perl.say } else { say "no match" } 21:14
p6eval rakudo de1e9f: OUTPUT«no match␤»
DanielC :-P
StephenPollei rakudo: if "(name" ~~ /^<-ws>/ { $/.perl.say } else { say "no match" }
p6eval rakudo de1e9f: OUTPUT«no match␤»
StephenPollei rakudo: if "(name" ~~ /^<[\(a..z]>/ { $/.perl.say } else { say "no match" } 21:15
p6eval rakudo de1e9f: OUTPUT«Match.new(␤ # WARNING: this is not working perl code␤ # and for debugging purposes only␤ ast => "(",␤ Str => "(",␤ from => 0,␤ to => 1,␤)␤»
pmichaud I'm not sure what you would expect <-ws> to match. 21:16
DanielC Same as \S
pmichaud but <ws> isn't the same as \s 21:17
DanielC Ultimately I want to match characters that are not parens or spaces.
<-[()] -???>
I nee to figure out what to put in ??? 21:18
StephenPollei rakudo: if "(name" ~~ /^<a..z]>/ { $/.perl.say } else { say "no match" }
pmichaud rakudo: say "( name" ~~ /<-[()] -space>/
DanielC ah, thanks.
p6eval rakudo de1e9f: OUTPUT«Statement not terminated properly at line 2, near "a..z]>/ { "␤in Main (src/gen_setting.pm:0)␤»
rakudo de1e9f: OUTPUT«n␤»
pmichaud I think <space> might be deprecated, though.
DanielC :( 21:19
What's the replacement?
pmichaud no, it's not deprecated.
DanielC I hope it doesn't go away because it works
pmichaud <space> is the same as \s
DanielC ah!
good
Ah... wonderful. I'm finally making progress with my grammar. 21:20
DanielC is making a LISP grammar.
StephenPollei rakudo: if "(name" ~~ /^[<a..z]>/ { $/.perl.say } else { say "no match" }
p6eval rakudo de1e9f: OUTPUT«Syntax error at line 2, near "]>/ { $/.p"␤in Main (src/gen_setting.pm:0)␤»
pmichaud transpose [ and < 21:21
StephenPollei rakudo: if "(name" ~~ /^<[<a..z]>/ { $/.perl.say } else { say "no match" }
p6eval rakudo de1e9f: OUTPUT«no match␤»
StephenPollei rakudo: if "(name" ~~ /<[<a..z]>/ { $/.perl.say } else { say "no match" }
p6eval rakudo de1e9f: OUTPUT«Match.new(␤ # WARNING: this is not working perl code␤ # and for debugging purposes only␤ ast => "n",␤ Str => "n",␤ from => 1,␤ to => 2,␤)␤»
DanielC Yay! \o/ I have a LISP grammar. 21:23
StephenPollei www.dwheeler.com/readable/
DanielC *click* 21:24
21:25 rosch joined
DanielC Honestly, I don't care. Lisp is readable enough for what I need, and much easier to parse. 21:25
StephenPollei sure it's just something that some like, I didn't know if you wanted something more to test your perl6 grammar stuff on 21:26
DanielC ah, thanks
21:27 rosch left
DanielC Is it a rule that a lisp expression has to start with a token? Is this valid?: ((foo bar)) ? 21:29
The outer list does not start with a token. It just contains a list. 21:30
justatheory StephenPollei: Us David Wheelers are insane. Pay no mind! 21:37
See also xkcd.com/224/ 21:38
21:51 Whiteknight joined
StephenPollei justatheory, you are a wheeler as well? I don't know many here well 21:52
justatheory I'm a *David* Wheeler, in fact. :-)
Just not the dwheeler.com David Wheeler.
StephenPollei paulgraham.com/avg.html read this about lisp just the other day.. but don't want to get into language flame wars 21:54
justatheory What else is there to do on the Internet?
21:56 rewt left
StephenPollei tons of stuff but to enumerate would surely be OT besides if I get some other stuff done I might write a few more things for t/spec 21:57
justatheory :-)
22:01 rewt joined 22:04 pmurias left 22:25 nihiliad left 22:29 alester left 22:35 hanekomu left, clintongormley left 22:36 sri_kraih_ left, sri_kraih joined 22:56 rindolf left 23:01 Sunbeam joined 23:13 Roxas left 23:15 M_o_C joined 23:19 amoc^sleep is now known as amoc 23:20 donaldh left, donaldh joined 23:22 sri_kraih left
sjohnson can anyone explain to me in plain english, how the &function(); syntax differs from function; ? 23:23
i have looked everywhere, read the camel book 5 times, and still cannot understand this in simple terms that my peanut brain can understand
DanielC It doesn't. 23:25
AFAIK
rakudo: sub foo { say "hi" }; &foo(); foo(); 23:26
p6eval rakudo de1e9f: OUTPUT«hi␤hi␤»
sjohnson i use it cause it looks nicer in my VIM perl syntax colouring
rakudo: sub foo { say "hi" }; &foo(); foo;
p6eval rakudo de1e9f: OUTPUT«hi␤hi␤»
japhb sjohnson: Are you talking about the Perl 5 difference?
sjohnson yes
DanielC rakudo: sub foo { say "hi" }; &foo(); foo(); foo
p6eval rakudo de1e9f: OUTPUT«hi␤hi␤hi␤»
sjohnson i am wondering if i keep programming like the above rakudo example... if it will bite me in the ass 23:27
if i'm doing something really complicated
japhb Ah. &foo carries caller's @_ along, as I recall. Been a while.
23:27 frew left
sjohnson will the real TimToady please stand up? :) 23:27
japhb And goto &foo is a tail call of sorts.
sjohnson japhb: thanks.. i will have to "read up"
DanielC rakudo: sub foo { say "hi" }; &foo(); foo(); &foo; foo; 23:28
p6eval rakudo de1e9f: OUTPUT«hi␤hi␤hi␤»
DanielC hm
rakudo: sub foo { say "hi" }; &foo; 23:29
p6eval rakudo de1e9f: ( no output )
DanielC I wonder if that's correct.
japhb DanielC: remember, in Perl 6, &foo by itself just returns the coderef
sjohnson well, i know Mr. Wall said that perl 6 is different
DanielC japhb: So () means "run me" ?
japhb Code or Routine or whatever the heck the base type is now.
sjohnson i shuold be asking on a perl5 chan, but you guys are just so goshdarn helpful, too 23:30
japhb DanielC: yup.
"invoke"
sjohnson: we try.
DanielC japhb: But plain foo works...
Then again, foo is the real function and not a reference, based on what you said. 23:31
japhb &foo is (in my mind) special cased. STD may say that the other forms are the real special cases. But that's the one I think of as different.
DanielC But I didn't think that &foo was a reference. I thought it was the real thing.
japhb DanielC: it *is* the real thing.
23:31 LCamel joined, icwiener left
japhb But by including the sigil, you're telling it you want the Code object back, not to *call* the Code. 23:31
DanielC yeah but &foo() 23:32
A slight ambiguity there, by your description.
japhb &foo is the real code. &foo() says "call the real code", which is the same as "foo()". "&foo" is simply the name of the real code, and doesn't say to do anything with it. 23:33
rakudo: foo { say "Hello" }; bar($code) { $code() }; bar(&foo);
p6eval rakudo de1e9f: OUTPUT«Statement not terminated properly at line 2, near "{ $code() "␤in Main (src/gen_setting.pm:0)␤» 23:34
japhb rakudo: foo { say "Hello"; }; bar ($code) { $code(); }; bar(&foo);
DanielC I still think that there is some inconsistency. Either "&foo" should run or "foo" should not (pick one).
p6eval rakudo de1e9f: OUTPUT«Statement not terminated properly at line 2, near "{ $code();"␤in Main (src/gen_setting.pm:0)␤»
japhb oh heh.
rakudo: sub foo { say "Hello"; }; sub bar ($code) { $code(); }; bar(&foo);
p6eval rakudo de1e9f: OUTPUT«Hello␤»
japhb there
DanielC rakudo: sub foo { say "Hello"; }; sub bar (&code) { code(); }; bar(&foo)
p6eval rakudo de1e9f: OUTPUT«Hello␤» 23:35
japhb DanielC: "foo" is using it as a listop or so.
DanielC even better :-)
rakudo: sub foo { say "Hello"; }; sub bar (code) { code(); }; bar(&foo)
p6eval rakudo de1e9f: OUTPUT«Malformed routine definition at line 2, near "bar (code)"␤in Main (src/gen_setting.pm:0)␤»
DanielC I guess that was too much to hope for.
japhb DanielC: I was simply pointing out that the change of sigil from call point to signature didn't matter. Only the fact that at the call you asked to hand it the real object
DanielC ok 23:36
japhb Although, technically my example was ambiguous
rakudo: sub foo { "Hello"; }; sub bar ($code) { say $code(); }; bar(&foo);
p6eval rakudo de1e9f: OUTPUT«Hello␤»
japhb dammit. 23:37
DanielC I just think that if "&foo()" and "foo()" do the same thing then "&foo" and "foo" should do the same thing too (whatever that might be).
japhb oh
DanielC You see what I mean?
japhb rakudo: sub foo($who) { "Hello, $who"; }; sub bar ($code) { say $code("World"); }; bar(&foo);
p6eval rakudo de1e9f: OUTPUT«Hello, World␤»
japhb There!
That disambiguates the call order. 23:38
23:38 LCamel left
japhb I understand what you mean. I happen to disagree -- I think that's mistaken orthogonality. Perl dimensions often go off at odd angles. :-) 23:39
DanielC :-)
I think orthogonality is a good property. It makes the language more predictable. 23:40
One of the things I love about Perl 6 is its increased orthogonality and predictability. There are a lot fewer special cases.
japhb Fair enough. 23:41
I've just accepted this particular waterbed lump, I guess. :-)
DanielC Yeah. This is the sort of thing one just has to accept. 23:42
23:45 justatheory left
pugs_svn r27073 | lwall++ | [Containers] define .rotate more like a PDL slice, rely on other ops for exact semantics 23:47
TimToady verb phrases work differently from noun phrases in most languages 23:52
leaving off the noun marker puts you into verb-phrase land 23:53
and makes listops possible
otherwise you can't know whether to expect a term or an operator after "foo"
and that's fundamental to everything Perl does 23:54
japhb Man, I really need to start remembering to explain things linguistically ....
TimToady Perl knows that &foo has no arguments, and that foo might
(syntactic arguments) 23:55
the &foo object responds to .() of course
23:56 Muixirt left
japhb Which brings me back to my question from earlier: 23:56
rakudo: my @code = {"Hello, $^a"}, {"Goodbye, $^a"}; @code[0]("World").say
p6eval rakudo de1e9f: OUTPUT«Hello, World␤»
japhb rakudo: my @code = {"Hello, $^a"}, {"Goodbye, $^a"}; @code>>.("World").say
p6eval rakudo de1e9f: OUTPUT«_block50_block61␤»
viklund if foo is a multi, then I should be able to get a Perl6MultiSub object with &foo, right?
japhb What am I saying wrong in the second example above? 23:57
TimToady that's the intent, but rakudo doesn't implement it yet
viklund Is there (or will there be) anyway of declaring an anonymous multi?
TimToady &foo is the abstraction for "what would happen if I called foo() right here"
viklund yes, I know I won't get it in rakudo yes...
*yet
TimToady japhb: nothing, looks like a bug to me 23:58
japhb AH! OK. I was going insane trying to wave chickens over that gumbo. 23:59
TimToady oh, wait
you need >>.say to
*too
rakudo: my @code = {"Hello, $^a"}, {"Goodbye, $^a"}; @code>>.("World")>>.say
p6eval rakudo de1e9f: OUTPUT«_block50␤_block61␤»
TimToady but still buggy
japhb TimToady: only if I want them on separate lines.
TimToady right
japhb but yes, that was a thinko.
TimToady should work without

Powered by Cro and the Raku® Programming Language.

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

Thank you!