»ö« | perl6-projects.org/ | nopaste: paste.lisp.org/new/perl6 | evalbot usage: 'perl6: say 3;' | irclog: irc.pugscode.org/ | UTF-8 is our friend!
Set by moritz_ on 30 July 2009.
00:08 jferrero left 00:10 qp_pq joined
qp_pq hi moritz_ 00:10
I need to write something that parses logs and spits out some nice graphs 00:11
er..charts
and I saw you guys were making some progress on some modules...
for charting using SVG
they look really cool and I get the feeling of jumping in and using them 00:12
TimToady moritz_ is probably asleep at the moment 00:13
qp_pq TimToady: he has an email maybe ? 00:16
TimToady dunno offhand 00:18
presumably, or he wouldn't have a pugs commit bit
00:22 japhb left
qp_pq is there some special namespace on cpan for p6 modules ? 00:25
wayland76 qp_pq: Not at the moment. The whole "What we're doing with CPAN" thing is still not well defined 00:26
sent moritz_'s address to qp_pq privately 00:27
So as to avoid spam for moritz -- please don't post it on the channel :)
But probably just ask questions on the channel instead of e-mailing moritz. There's also a perl6-users mailing list: dev.perl.org/perl6/lists/ 00:29
00:30 PZt joined
qp_pq what should I look on for some quick svg docs , tutorial or something 00:31
I want to jump in , read the module and add some functionality to what he wrote
00:31 youwin joined
qp_pq svg-plot that is 00:31
wayland76 You mean perl6 specific, or just in general? 00:34
00:38 ruoso joined 00:43 nErVe joined
sjohnson howdy wayland76 00:50
wayland76 sjohnson: g'day :)
Hmm, I think we've discovered two more entries for the parametric roles example that greets people in their language :) 00:51
sjohnson there's a whole whack of them to use for later 00:53
rakudo: given "text" { when m/ex/ { print "found ex"} when m/(t$)/ { print "found $1"} } 00:55
p6eval rakudo 0d4fe0: OUTPUT«found ex»
sjohnson rakudo: given "1" { when '1' { print "A"; } when '1' { print "B"; } } 00:57
p6eval rakudo 0d4fe0: OUTPUT«A»
sjohnson is there a way for the given/when structure to go through all the when's in the statement? 00:58
or is it called a "do" statement
01:00 Whiteknight left
sjohnson oh, i think it's "continue" 01:00
could a name like, "do" be used in place of "when" for if you want it to just continue; at the end of the block? or am i losing my mind 01:02
01:08 nErVe left
sjohnson or maybe "try" 01:08
01:09 frew joined
colomon I've been assuming (without checking it out in detail), that each when triggered automatically jumps to the end of the block. 01:12
01:13 tak11 joined
colomon Time to check the spec... 01:13
sjohnson rakudo: given "text" { when m/ex/ { print "found ex "; continue} when m/(t$)/ { print "found $1 "} } 01:16
p6eval rakudo 0d4fe0: OUTPUT«found ex Use of uninitialized value␤found »
colomon where do you want the continue to go?
sjohnson after the first check
rakudo: given "text" { when m/(t$)/ { print "found $1 "} }
p6eval rakudo 0d4fe0: OUTPUT«Use of uninitialized value␤found »
sjohnson perhaps they don't use $1 anymore 01:17
colomon $0
sjohnson codes in too much Perl 5
colomon is the first match.
sjohnson rakudo: given "text" { when m/(t$)/ { print "found $0 "} }
p6eval rakudo 0d4fe0: OUTPUT«found t »
sjohnson thanks mate
rakudo: given "text" { when m/ex/ { print "found ex "; continue} when m/(t$)/ { print "found $0 "} }
p6eval rakudo 0d4fe0: OUTPUT«found ex found t »
sjohnson perfect
colomon I'll be darned. Didn't know that was possible.
sjohnson i'm wondering if it can be done easier... with a different word than "when" 01:18
run, try, do would be good keywords
colomon Well, "if" would do it, right? 01:19
sjohnson if i could use if and it uses $_ as the switch var, probably could be done
let's see if it breaks the compiler
colomon good plan. 01:20
(Bet it doesn't.)
sjohnson rakudo: given "text" { if (m/ex/) { print "found ex "; } when m/(t$)/ { print "found $1 "} }
p6eval rakudo 0d4fe0: OUTPUT«found ex Use of uninitialized value␤found »
sjohnson oops
rakudo: given "text" { if (m/ex/) { print "found ex "; } when m/(t$)/ { print "found $0 "} }
p6eval rakudo 0d4fe0: OUTPUT«found ex found t »
sjohnson pats colomon on the back 01:21
colomon rakudo: given "text" { if m/ex/ { print "found ex "; } when m/(t$)/ { print "found $0 "} }
p6eval rakudo 0d4fe0: OUTPUT«found ex found t »
sjohnson ya that was going to be my next text
colomon I think we're figuring this thing out.
01:21 shinobi-cl joined
sjohnson i think it's kind of a big deal this given/when thing 01:21
01:21 reppie left
sjohnson the core module in Perl 5 (switch) is very fragile 01:21
and buggy, and doesn't support regex matching in case statements 01:22
colomon I suspect the Perl 5.10 version is better? (Or is that what you're talking about? I remember the module being a disappointment.)
sjohnson yep that is the one i mean
colomon Ah. I've never really used 5.10.
sjohnson i am surprised it made it into the core module packages
well, i am using 5.8.8 but i think it might be unlikely that it is any different in the newer perl 5's 01:23
oh well, i can hold off until Perl 6
colomon I think when, in particular, is fantastic. It doesn't have to work with given, which allows a lot of pretty code.
rakudo: for <text test tex> { if m/ex/ { print "found ex "; } when m/(t$)/ { print "found $0 "} }
p6eval rakudo 0d4fe0: OUTPUT«found ex found t found t found ex »
colomon I think a lot of the classic file parsing idioms will be more elegant this way... 01:24
sjohnson rakudo: for <text test tex> { if m/ex/ { print "found ex in $_"; } when m/(t$)/ { print "found $0 in $_"} } 01:25
p6eval rakudo 0d4fe0: OUTPUT«found ex in textfound t in textfound t in testfound ex in tex»
sjohnson man i can't wait for perl 6
this simple example will make my life a lot easier
i suppose i could just start using it now
colomon I have.
sjohnson and try to break it
colomon Speed is a big issue, alas.
but for a lot of things it is already quite usable. 01:26
sjohnson my needs are quite simple, so it should be okay
i mostly just write shell utils to make my life at work easier
colomon I mostly do simple stuff as well, but simple stuff that processes a lot of data fairs badly still in Rakudo. 01:27
One piece of code I tried to write in p6, only to find that the p5 version was 5000 times faster. 01:28
01:28 shinobi-cl left
colomon Loops that repeat a lot are bad news in current Rakudo. 01:30
Other than that, I'm finding it perfectly usable.
And in lots of cases, very nice. 01:31
sjohnson cool 01:33
thanks for the tips on that given/when structure 01:34
i was trying the examples from this page: blogs.gurulabs.com/stephen/2008/12/...nwhen.html 01:36
i suppose i could write to the author and let him know that you can also use the 'if m/.../' technology 01:37
01:38 frew left 01:43 dukeleto joined 02:03 KyleHa joined, alester joined 02:05 japhb joined 02:07 cdarroch left
japhb (flaky net connection)-- 02:08
Man, sometimes it can be so cathartic to just decrement something in the presence of others. ;-) 02:09
KyleHa We're here to witness your pain, japhb. 02:10
02:12 alester left
sjohnson japhb: use GNU screen technology for your irc usage 02:12
if you don't already
japhb sjohnson, that only works if you have a box not behind said flaky net connection with which to screen. 02:13
:-)
sjohnson indeed it does 02:14
i had to ask around for a while before someone had a stable shell i could use
02:14 alester joined
Tene japhb: I have a box you can use. 02:15
japhb: ask me later tonight. AFK shopping now.
japhb Tene, thx
02:15 sjohnson sets mode: +oo japhb Tene
japhb what happened to the opbot? 02:16
sjohnson was there ever one? 02:17
japhb I thought so ...
sjohnson i've never seen any bot-like characteristics
japhb hum. may have been years ago 02:18
02:31 hercynium_ joined 02:40 meppl left 02:46 hercynium left 02:49 molaf joined 02:50 tann left 03:16 hercynium_ left 03:19 Limbic_Region left 03:20 sri_kraih left, sri_kraih joined 03:25 molaf left 03:48 carlin joined
qp_pq are the grammars capable of modifying the input at run-time ? 03:53
the very text they're parsing... 03:54
03:54 mikehh_ joined, frew joined 03:59 xomas_ left 04:10 mikehh left
zetta I'm interested in that question as well 04:11
so I looked it up in S05
:rw modifier
the way the doc reads suggests maybe no implementation has :rw yet, but I don't know 04:13
04:14 Chillance left
zetta ah, but you said grammars... not regexes... anyway... yeah. 04:15
04:26 mikehh_ is now known as mikehh
qp_pq zetta: so why are you interested ? 04:39
zetta just seemed like an interesting question, I've read up on the grammars/rules system, but I hadn't thought of using it that way. 04:40
I can sling around Perl5 s/// with the best of 'em but rules >> regexes. imo anyway. 04:42
anyway, maybe a perl6 guru can weigh in. 04:45
04:46 frew left 05:04 justatheory left 05:05 KyleHa_ joined, KyleHa left 05:06 tann joined 05:09 ruoso left 05:12 ruoso joined 05:13 Zloyrusskiy joined 05:17 ruoso left 05:20 drbean joined 05:24 hah joined, abra joined 05:39 RonOreck left 05:41 alester left
wayland76 I suspect that the intent is that they could modify the input, but just how may not yet have been specified 05:41
05:45 zamolxes left, orafu left, szabgab left, orafu joined 05:57 jaldhar left
moritz_ good morning 06:10
sjohnson morning to you, night to me :) 06:13
perl6: say 3; 06:14
p6eval elf 28016, pugs, rakudo 0d4fe0: OUTPUT«3␤»
sjohnson ahh there it goes
pmichaud good morning, #perl6 06:18
moritz_ pmichaud: isn't it *very* early morning for you? :) 06:19
sjohnson pugs: say 'good morning, #perl6'; 06:20
p6eval pugs: OUTPUT«good morning, #perl6␤»
pmichaud moritz_: yes. 01h20 here :-) 06:21
06:21 rfordinal joined
pmichaud but I mainly came online to pass the word that the 'testcommitted' tag is now available in rt :-) 06:21
moritz_ cool 06:22
and is there a moniker in the subject that will set it automatically?
pmichaud not yet.
Robrt++ says that he can do it but doesn't have time at the moment. He's left it on his to-do list, however.
anyway, I'm off to bed -- be back in a few hours 06:24
moritz_ good night
wayland76 Does anyone ( moritz_? ) have any thoughts about how Path objects should be compared? 06:26
moritz_ not
wayland76 Ie. if I want to test if two Paths are the same inode, I'd hope to do it with something like $path1 === $path2
And I was wondering if === was the right operator here
moritz_ no
eqv is, probably 06:27
wayland76 Hmm
moritz_ === asks "are these two the same object", in terms of "at the same memory location"
wayland76 Right
But isn't eqv still about comparing objects? 06:28
moritz_ eqv doesn't fit fully as well
wayland76 Ok
moritz_ it's roughly the same as $a.perl eq $b.perl
wayland76 So does that mean that nothing fits? :)
06:29 rfordinal left
moritz_ well, in Perl 6 (at least notionally) the operator determmines the operation, not the type of the arguments 06:29
so if you introduce a new type, you also need to introduce another operator
unless you re-use smartmatch, of course
wayland76 Hmm. Well, that's all very painful then :) 06:30
APL, here we come :)
moritz_ well, you *can* reuse smartmatch, no?
wayland76 Hmm, maybe we can.
Because, it will do different things for Str ~~ Path and Path ~~ Path, right? 06:31
moritz_ that's entirely up to Path.ACCEPTS to decide
wayland76 Ah, cool :) 06:32
See, I was pretty sure you could answer that question :) 06:33
moritz_ ah well, when you first asked I thought you were talking about semantics 06:34
a la "how do I determine if two paths point to the same file?"
that's something I can't answer ;-)
wayland76 No, I'm rewriting parts of S32/IO to reflect some good ideas that have come through :) 06:35
06:36 unitxt left, unitxt joined
wayland76 That new guy, in particular -- I pasted parts of his e-mail, and replaced "should" with "does" :) 06:36
One more smartmatch question -- if I want the two things I mentioned above, I put a multi method .ACCEPTS on the Path object, so that it can do Str and Path, right? 06:37
Wait, the doco implies .ACCEPTS is single dispatch 06:39
Oh, I found the bit where it says I can do multi
Don't worry :)
moritz_ doesn't worry :) 06:41
wayland76 Has another question 06:59
wayland76 has another question (sorry ) :)
Is there some way I can say "functionA() also takes all the parameters of functionB() in addition to its own"? :) 07:00
dalek kudo: 5a2aeaf | moritz++ | (17 files):
remove some trailing whitespaces
 07:01
moritz_ doesn't think so 07:02
XY problem?
wayland76 What does XY problem mean?
07:03 rfordinal joined
wayland76 Ok, finding google info :) 07:03
Found Google info 07:04
Will continue thinking. Thanks :)
07:11 krakan joined 07:21 frederico joined 07:23 qp_pq left
pugs_svn r28017 | wayland++ | [S02] Changed :io to :p and :path 07:24
r28017 | wayland++ | [S16] Documented p{/path/to/file}
r28017 | wayland++ | [S28] Made $*CWD have type Path instead of Str
r28017 | wayland++ | [S32/IO] Many changes, including:
r28017 | wayland++ | * Changed IO::FSNode into Path
r28017 | wayland++ | * Merged most of IO::DirectoryNode and all of IO::LinkNode into Path (but still need to
r28017 | wayland++ | merge IO::FileNode)
r28017 | wayland++ | * Moved remnants of IO::DirectoryNode to IO::Directory
r28017 | wayland++ | It's not finished yet, but I thought I'd commit anyway. I'll keep working on it.
wayland76 @karma wayland
lambdabot wayland has a karma of 106
wayland76 Ooh :)
07:31 abra left
frettled wayland76++ :) 07:34
07:35 tak__ joined 07:37 abra joined 07:40 tak11 left
wayland76 The one I referred to as "the new guy" is "Troels Liebe Bentsen". Can we ++ him yet? 07:40
@karma we
lambdabot we has a karma of 1
moritz_ (Troels Liebe Bentsen)++
wayland76 lambdabot: No, that should be "we have a karma of 1", unless you're a lolcat :) 07:41
frettled haha
wayland76 moritz_: Yes, but the () is not recognised
@karma Lenz)
lambdabot Lenz) has a karma of 64
wayland76 See?
frettled @karma Bentsen)
lambdabot Bentsen) has a karma of 1
frettled \o/
Whose is lambdabot anyway?
moritz_ Troels Liebe Bentsen++ 07:42
@karma Troels Liebe Bentsen
lambdabot Troels􏿽xC2 has a karma of 0
moritz_ bah
it doesn't even speak UTF-8
wayland76 Well, I'll wait for him to choose a nick, and then do it.
07:42 leedo left
wayland76 Hmm. From my limited knowledge of German, it looks like "Troll-loving Bentsen" :) 07:43
Would someone like to fix my German? :)
carlin @karma c/c
lambdabot c/c has a karma of 144
carlin heh
frettled Troels is similar to Truls.
wayland76 Yeah, I c/c-- every chance I get
moritz_ Liebe = Love, but Troels is not a German word
frettled wayland76: c/c-- is sensible
I would like to see a c/c-- languag. 07:44
+e
wayland76 Wouldn't that be a lang-gulag? :) 07:45
frettled oooh 07:47
carlin So there is no chdir function now? 07:51
moritz_ rakudo: chdir '/' 07:55
p6eval rakudo 0d4fe0: ( no output )
moritz_ rakudo: chdir '/'; say "alive"
p6eval rakudo 0d4fe0: OUTPUT«alive␤»
moritz_ it seems to work just fine locally 07:56
p6eval restricts file I/O, so you probably can't easily tell if it worked here on IRC
carlin Latest S32/IO says; "chdir FILENAME - Gone, just set $*CWD"
moritz_ doesn't trust S32::IO 07:57
07:59 mj41_ joined 08:00 mikehh_ joined 08:01 abra left, tak__ left, rfordinal left, mikehh left, PacoLinux left, synth left, elmex left, KatrinaTheLamia left, s1n left, mattp_ left, cxreg left, gabiruh left, sbp left, nsh left, mj41 left, lisppaste3 left, mj41_ is now known as mj41
carlin doesn't like the idea of manually setting $*CWD 08:01
08:01 [sbp] joined
Matt-W Morning 08:01
08:02 nsh joined
moritz_ oh hai Matt-W 08:02
08:02 mattp joined, gabiruh joined
Matt-W carlin: I think you can take that as fairly provisional 08:02
hai moritz_
08:02 cxreg joined
Matt-W can haz christmas? 08:02
08:03 elmex joined, rfordinal joined
moritz_ not quite. Bug svg bar charts with clikable bars and neatly aligned labels (thanks to masak++) 08:04
Matt-W I saw your blog entry about that 08:06
Interesting stuff
It's just a shame that SVG.pm currently is just an XML maker
08:07 elmex left, mj41 left, gabiruh left, s1n joined, abra joined, tak__ joined, mikehh joined, PacoLinux joined, synth joined, elmex joined, KatrinaTheLamia joined, lisppaste3 joined, gabiruh joined, sbp joined, mj41 joined, mattp_ joined, irc.freenode.net sets mode: +o s1n, mattp_ left, gabiruh left, gabiruh joined, abra left, elmex_ joined, KatrinaTheLamia left
moritz_ the blog post didn't include the linking stuff :-) 08:07
well, I talked to masak about that
and we agreed to make the current SVG.pm into a general XML::Writer
08:07 mj41_ joined
Matt-W good, I was thinking I might suggest that 08:07
Making XML is fairly straightforward after all 08:08
It's parsing the bloody stuff that's hard
moritz_ and add SVG specific behavior to SVG.pm
08:08 abra joined, sbp left, KatrinaTheLamia joined
Matt-W yes and SVG.pm should use XML::Writer 08:08
moritz_ exactly 08:09
Matt-W that's probably the most straightforward thought I'm going to have all day 08:10
moritz_ that sounds really depressing
Matt-W so's the work I'm doing at the moment 08:11
trying to get stuff done with the most overengineered library in the history of library writing
08:12 TimToady left, kolibrie left, kcwu left, cono left, buu left, ewilhelm left, yahooooo left, dmpk2k left, moritz_ left, ewilhelm_ joined, buu_ joined, elmex left, kolibrie_ joined, elmex_ is now known as elmex, yahooooo joined 08:13 buu_ is now known as buu 08:14 buu is now known as buu_ 08:15 buu_ is now known as buu__, kcwu joined, buu__ is now known as buu 08:16 mikehh left, mj41 left 08:17 moritz_ joined
moritz_ (net flakyness)-- 08:17
I remember reading up on a linear algebra library - to solve an equation system you'd first have to create a factory, which then instantiates a solver - at that point I stopped reading :-)
Matt-W oh dear
well this is an incredibly elaborate layered comms library 08:18
in some ways it's amazing
wonderfully flexible
but it's far more complicated than we actually need
all we needed was a set of fairly barebones classes for simulating our own network protocols for automated testing
and instead we've got this enormous timesink 08:19
08:25 dmpk2k joined 08:27 tak__ left 08:30 tak11 joined 08:31 zamolxes joined 08:33 terje_ joined, terje_ is now known as bionoid 08:36 rfordinal left 08:41 dakkar joined 08:42 kcwu left 08:51 [sbp] is now known as sbp
wayland76 carlin / carlin_ What's wrong with setting $*CWD instead of using chdir? 08:57
moritz_ wayland76: for example not being able to use relativ paths? or at least not in an intuitive way?
wayland76 And keep in mind that S32/IO is only a draft -- it will keep changing based on feedback
Matt-W The core functionality of just changing the working directory is nearly useless anyway 08:58
Have to have some framework for tracking it and moving around
moritz_ Matt-W: why?
Matt-W ponders some sort of module for that
moritz_ $*CWD tracks it, chdir() changes it
Matt-W no, $*CWD tells you where you are now
it has no idea how you got there or where you started
if you're crawling around a directory tree you might want to know that
moritz_ that doesn't have to be a problem
if you want to know it, you can still store it at the beginning 08:59
Matt-W admittedly all you need for implementing such a mechanism is $*CWD
08:59 masak joined
wayland76 Hmm. $*CWD = "../foo/path" 08:59
...relative to current path :) 09:00
masak no please no.
wayland76 Ok, it's a bad idea.
:)
moritz_ wayland76: so what's $*CWD after that/
?
Matt-W the canonical form of "$*CWD/../foo/path"?
need to sort out this path literals thing really
wayland76 Well, lets say we start with $*CWD = "/home/wayland"
moritz_ so you assign something to a variable, but after it has a different value than you assigned to it? 09:01
that's less than intuitive.
wayland76 My thought was that $*CWD = "../foo/path" would change it to "/home/wayland/foo/path"
Or maybe....
moritz_ and silently drop the '../' part?
wayland76 It could have a $*CWD.path() returns "../foo/path", and $*CWD.realpath() returns "/home/foo/path"
Oops, that comment above should've said "/home/foo/path", sorry :( 09:02
moritz_ but why the magic?
wayland76 I think my fingers know "/home/wayland" too well :)
moritz_ what's wrong with good ol' chdir()?
wayland76 As I implied above, I realised during the course of this conversation that it's probably a bad idea 09:03
moritz_ good :-)
wayland76 But I don't know where to put chdir
Ah, I know :)
On the IO::Filesystems object
(IO::Filesystems is the whole file tree -- maybe it needs a better name, though) 09:04
moritz_ I mean I was asking because there might have been a real benefit to handling it all in a variable
jnthn lolhai
moritz_ ohlolitsjnthn
jnthn :-) 09:05
09:05 tann left
wayland76 moritz_: I just thought it didn't belong where it used to be, and so I needed somewhere else to put it, and I thought "I know, we can just hide it under the hood" 09:06
But it seems not to have worked out, so maybe IO::Filesystems.chdir is the way to go
09:08 missingthepoint joined
masak jnthn: o/ 09:09
wayland76: ...is export
wayland76 Yeah, IO::Filesystems.chdir isn't how you'd use it, I'm just saying where it should live 09:10
09:11 guest_007 left 09:15 missingthepoint is now known as bpetering
masak frettled++ # that email reply was essentially "you think it's cute today..." :) 09:17
frettled masak: :D
For once, I was reading my mail more frequently than IRC. Woot. 09:19
09:20 frettled sets mode: +o moritz_, frettled sets mode: +o masak 09:22 DanielC joined
DanielC Hi all. 09:23
phenny DanielC: 04 Jul 07:46Z <mberends> tell DanielC gitorious.org/parrot-module-lib/main updated
DanielC *click*
masak DanielC: you're back!
\o/
DanielC yeah
That was a long trip. 09:24
(honey moon)
masak so I seem to recall.
moritz_ DanielC: belated congratulations!
DanielC thanks :-)
masak oh, and from me too. :)
DanielC :-)
Anything new and exiting happening in Rakudo world? 09:25
masak nah. :)
moritz_ yes, lots of introspection
and custom traits
masak and a promised release for April.
moritz_ jnthn++ I might add
DanielC Stable release for April? Stable?
moritz_ pmichaud didn't use "stable" 09:26
wayland76 No, unstable :)
masak DanielC: um, the word 'stable' has been declared too fluffy.
09:26 dukeleto left
moritz_ he said a "usable and useful subset of Perl 6" 09:26
DanielC I guess it is...
Ok.
masak all software is less than stable, with the possible exception of TeX and some space probe software.
moritz_ the biologists have a special word for stable. 09:27
wayland76 ...and any software small enough to be provably correct :)
masak promising 'stable' is essentially promising to have a multi-billion budget, or to be Donald Knuth.
wayland76 moritz_: Not *all* stable things are called that :)
moritz_ wayland76: only stable software, right :-)
09:28 dukeleto joined
wayland76 moritz_: C'mon now, I bet some people call Perl 6 stable, if that's the case :) 09:28
DanielC can't figure out how to explain to someone what makes the April release different from every other monthly release
wayland76 (for those missing the joke, the biologists word for stable is "dead")
DanielC: We're hyping it more :)
DanielC Can't call it "production" because I'm sure someone is using the current Rakudo in production somewhere.
stable = dead... he he he 09:29
wayland76 But if you look for "Rakudo Star", then you'll see what it's about :)
masak DanielC: I don't think there has to be a difference.
DanielC: what we have _now_ is pretty neat.
DanielC: but in April it'll be even neater.
DanielC What Rakudo had back in July was already very neat. 09:30
masak and outsiders know next to nothing about Rakudo.
therefore, it's a good idea to pretend that April will have a 'special' release, and then to a PR dance.
DanielC Sure.
wayland76 Like I said, we're hyping it more :)
frettled stable - a place where you put your horses when they're not currently being flogged.
masak frettled: and the horses are dead, yes? 09:31
frettled masak: a stable is essentially a black box when seen from afar, we cannot know. ;)
masak but we can make probably deductions... :)
s/ly/le/
Matt-W DanielC! 09:32
wayland76 I'd prefer a red box with shite stripes :)
s/shite/white/ 09:33
DanielC o/ Matt-W
wayland76 That was certainly unintention -- my apologies :)
DanielC Can someone remind me of the git command to update the repository? I want to get the latest Rakudo...
Ah... git pull
wayland76 Flogging dead horses: www.sucs.swan.ac.uk/~cmckenna/humou...horse.html 09:34
frettled If it's the piece I think it is, then it is excellent.
masak DanielC: by the way: new build instructions for Rakudo.
DanielC: it (and Parrot) are now meant to be installed.
DanielC masak: The instructions on the website look the same...
$ perl Configure.pl --gen-parrot && make 09:35
frettled DanielC: read the README as well. You will also need to use «make install» to remain reasonably sane.
DanielC "meant to be installed" - that's good rogress
masak aye. 09:36
many people were blocking on that, it seems.
DanielC Website: "there is no make install target"
masak then the website is outdated.
DanielC ok
09:36 jrtayloriv joined
wayland76 DanielC: Which page on the website? 09:36
DanielC $ perl Configure.pl --gen-parrot && make && make install -- is this correct?
wayland76: rakudo.org/how-to-get-rakudo 09:37
wayland76: "Get Rakudo"
wayland76 DanielC: Thanks
masak DanielC: yes; probably with 'sudo ' on the last step. 09:39
DanielC y
09:39 jrtayloriv left
wayland76 DanielC: You can build RPMs too 09:39
If you prefer
09:39 jrtayloriv joined
DanielC Debian rulez! 09:39
wayland76 With no Rakudo package? I don't think so :)
09:39 jrtayloriv left
wayland76 Hey, everyone, what would you think of the idea of the Path object supporting named paths? 09:41
DanielC I wonder if you can use alien to make a .deb or Rakudo.
wayland76 ie. $HOME 09:42
09:42 Gothmog_ left, Gothmog_ joined
wayland76 DanielC: Probably, but the .spec file isn't very complex, and should be able to be ported to Rakudo 09:42
But not just $HOME, also things like $bindir ="Program Files" == "/usr/bin", etc 09:43
jnthn has a shiny new 64-bit multi-core server. 09:45
frettled Yay or something. 09:46
I hope you don't get in too much pain from faulty drivers etc.
masak jnthn: does it feel like a Rakudo day today?
jnthn masak: No, it'll be tomorrow.
masak oki
jnthn masak: Today I need to do some things that I probably shoulda done yesterday.
masak yay, Rakudo day tomorrow \o/ 09:47
wayland76 jnthn: My new desktop that should be hardware-ready tomorrow has two multi-core CPUs :)
pugs_svn r28018 | wayland++ | [S32/IO] A few changes, including:
r28018 | wayland++ | * Revived .chdir() due to popular demand
r28018 | wayland++ | * IO::FileNode merged into Path
carlin wayland++ # for bringing back chdir 09:48
wayland76 carlin++ for pointing out that it was stupid :) 09:49
09:52 guest_007 joined, tlb joined
sjohnson nice one 09:56
sjohnson pats wayland76 on the back 09:57
i use chdir in p5 often
wayland76 Well, yes, but setting $*CWD would've been fine without the relative path issue :)
Matt-W wraps his own stuff around chdir for ease
masak "Second system syndrome done right" :D 09:58
sjohnson ahh i think $*CWD is too ugly for me
i suppose i could always say i'd like the timtowtdi way that looks more like a function call
10:00 ejs joined
carlin unless there's a benefit to changing it, better to stick with the familiar 10:00
frettled And the benefit needs to outweigh the disadvantages. 10:02
wayland76 Well, like I said, I couldn't figure out where to put it.
And I thought one less thing to remember would be good :).
10:03 meppl joined
jnthn Hmmm... 10:07
make: *** [src/gen_actions.pir] Killed
Anyone know what might be killing this?
(ulimit claims "unlimited"...)
pugs_svn r28019 | wayland++ | S32/IO: Mention that Path can be used as an array of path elements 10:16
r28019 | wayland++ | S16: Restrict filenames to POSIX
10:19 naruto joined
naruto hi 10:19
10:19 naruto is now known as Guest63713
Guest63713 hi 10:20
10:20 Guest63713 left
DanielC hi 10:20
wayland76 I guess he didn't like it here :) 10:21
Maybe we should've all gone HEYMATEYWASSUPDUDE!!!!!1!!1!!1! 10:22
DanielC heh
jnthn gives up trying to build Rakudo on his new server for now, figuring he can't procrastinate debugging that 1,000 line stored procedure forever... :-/ 10:31
Matt-W wayland76: but OH HAI is much more in keeping with the channel's culture
jnthn wishes they hadn't used "temp" in the name of a table that stores very non-temporary data...that should not go away. 10:37
Man, people do some funny things when writing software. :-/
wayland76 Maybe he left because of no hai :)
Matt-W jnthn: yes they do, I've been pondering this today
10:38 jdv79 left
Matt-W sometimes they do strange things when managing people who are writing software, too 10:38
Matt-W has wasted a week on a pile of software that has no distinctly useful purpose
moritz_ wonders what people do who manage software that writes people
jnthn Heh, the only thing that bothers me about this particular system is that it *is* used for stuff... :-)
Matt-W that bothers me about a lot of things... 10:40
ever played with a database which has three tables with >400 columns each? 10:41
something strikes me as wrong about that
moritz_ well, having "no distinctly useful purpose" is quite different from "not being used" :-)
Matt-W indeed it is 10:43
what really bugs me is that I already had some Perl scripts which did what this stuff does (more or less) in about 5% of the code
jnthn Matt-W: Not encountred anything quite *that* extreme in terms of database design fail.
Matt-W could've done all its functionality with only the same code again
but *someone* got hooked on the idea of unit tests 10:44
despite knowing that we will never, ever be able to refactor to the point where true unit tests are possible
but somehow that meant the testing system had to be written in C++
despite this being the perfect use case for Perl or another language operating at the same level
And oh look, who's on the team, the office's go-to guy for Perl 10:45
(i.e. me)
Don't know enough Perl to understand this? Give me two afternoons and I'll teach it to you
Matt-W bangs his head on the desk
Sorry, this has nothing to do with perl 6 10:46
It just all bubbles over sometimes
moritz_ knows that feeling very well
Matt-W try and do anything in anything other than C++ or Java here and everyone looks at you very oddly
almost like they'd never heard of using the right tool for the right job... 10:47
10:48 xomas_ joined
Matt-W argh and someone's mucked up the underlying structural library I wrote 10:51
aaaaaaargh
why are all the template parameter names now different and meaningless?
they used to have scriptive names, now they're 'T' 10:52
moritz_ gives Matt-W a big punching bag to channel his aggressions
Matt-W if only I could pop up to my aikido school for their lunchtime session and work it off by throwing people around 10:54
and rolling back and forth until I'm dizzy
always takes my mind off things
huf Matt-W: oh, i've seen mysql tables where you couldnt add more columns even if you wanted to 11:00
moritz_ speaking of horror stories - I read that ebay moved avoid from C++ for its midlleware because they hit maximum number of methods on a class 11:02
tlb Hey, how does mixins work in Perl 6 and how its it handle with build in types. Fx. the new Path object will need to have some utility IO functions, like splice and lines, but in a sense the object it self should not care IO as it only handles path information. 11:08
moritz_ splice is not an IO function 11:09
tlb sorry not splice, slurp
moritz_ mixins work like '$object does Role', but I don't think you need them here
delegation seems much better 11:10
class Path { has $!io handles <slurp lines get>; ... }
or something along these lines
pugs_svn r28020 | jani++ | Rephrasing for clarity, also changed "POSIX" to "portable POSIX" where appropriate 11:18
frettled Hrm, I should find a sensible way to automagically include the name of the synopsis I altered without engaging in manual typing hell. 11:19
moritz_ frettled: how often do you do that?
frettled moritz_: good question, I haven't done it often enough to have a decent data set yet. :D 11:21
But I imagine that I might spend a bit of time trying to clarify synopses, so making a svn commit wrapper for that purpose wouldn't be too bad. 11:22
I could make it in Perl 6!
…and make it depend on a feature that's not yet implemented in Rakudo, semi-automatically updating it to depend on another unimplemented feature everytime the former gets implemented by pmichaud. ;) 11:23
I don't have a gun, is it okay if I stab my foot instead?
moritz_ that sounds like cool meta programming
11:24 tak11 left
wayland76 frettled: Not even a water pistol? 11:38
You could fill it with tequila :) [NB: I don't drink :) ] 11:39
moritz_ in Perl 6 we expect you to shoot yourself with rope.
masak "enough rope to shoot yourself in the foot" 11:41
11:41 mdxi left
frettled wayland76: hee-hee 11:41
11:47 Whiteknight joined
jnthn In my experience, tequila is probably far better used for water fights than for drinking... ;-) 11:48
carlin because you'll be able to remember the water fight? 11:49
wayland76 ...and because it evaporates more quickly than water, and cools you faster :) 11:51
jnthn And won't have the taste of tequila in your mouth. :-)
Nor a tequila hangover the next day. 11:52
11:52 KyleHa_ left, KyleHa joined
moritz_ just your cloths stink 11:52
frettled Tequila is good for getting other people drunk in bars or other people's apartments. 11:55
masak frettled: your emails today are a joy to read. now you're backing up your claims with real shell runs! frettled++ 11:58
12:01 jauaor joined
frettled masak: thank you :) 12:02
masak and I learned something. :) 12:03
so $PWD isn't strictly equal to the current directory?
12:03 jaldhar joined
carlin if you manually set PWD to a dir that doesn't exist the current dir doesn't change 12:05
masak ah, only in the case when the dir doesn't exist? 12:06
12:09 qp_pq joined
carlin I think that's the only time 12:09
12:09 sri_kraih left
frettled nopes. 12:09
carlin but "doesn't exist" includes PWD=..
moritz_ anyway, we convinced wayland76 that assigning to $*CWD is a bad way to change the directory
frettled In the examples I cited, I could have also done «cd» to get back to the homedir, then «PWD=test», and «pwd» and «echo $PWD» will still be different. 12:10
wayland76 moritz_: No, just that it was bad to get rid of chdir() :) 12:11
But, what about TMTOWTDI? :)
moritz_ wayland76: ah well, that's a start :-)
frettled But note that the \w (working dir substitution thingy) in $PS1 uses $PWD instead of checking.
moritz_ wayland76: we must not exaggerate with TIMTOWTDI :-) 12:12
wayland76 moritz_: No, but I think that 2 ways to do something is not too many. So the idea should be evaluated on its merits :)
moritz_ imagines all kind of weird problems with a writable $*CWD
carlin I really don't like the idea of $*CWD = '../foo' mimicing chdir ... 12:13
moritz_ $*CWD ~~ m:rw/<some_regex>/
does every change to $/ also try to change the current directory?
masak I think if an idea is 50% good and 50% queasy, it should not be specced. 12:14
12:14 sri_kraih joined
Matt-W I don't like it 12:14
masak me neither.
Matt-W I think directory changing should involve something that looks like calling an action
masak a sub call.
Matt-W no problem with $*CWD having methods
carlin except it looks ugly ;)
Matt-W but assignment - and especially these strange string concat overloads that have been floating around - seem completely wrong
masak I have no problem with $*CWD being a vanilla Str. and chdir working as always. 12:15
Matt-W doesn't like things in the global namespace :P
masak modules can do any leet trickery.
moritz_ would even be fine with $*CWD being a Buf
masak whatever.
rakudo: ((Temporal::DateTime.new(:date(Temporal::Date.new(:year(2010), :month(4))), :time(Temporal::Time.new)).epoch - time) / 86400).ceiling.fmt("%d days left until April!").say 12:16
p6eval rakudo 0d4fe0: OUTPUT«226 days left until April!␤» 12:17
Matt-W I think if we do have a special type for filesystem paths, $*CWD should be one of them
masak sure.
carlin rakudo: $*CWD.say;
p6eval rakudo 0d4fe0: OUTPUT«/home/evalenv/rakudo␤»
frettled ahaha
rakudo: chdir(".."); $*CWD.say; 12:18
p6eval rakudo 0d4fe0: OUTPUT«/home/evalenv␤»
moritz_ wonders if there's any harm in allowing chdir in p6eval
Matt-W depends if there are any holes left in what it can do when it gets there 12:19
frettled Only if it naïvely allows shell meta characters through.
carlin we could always manually set CWD
frettled …to a shell
12:19 takadonet joined
takadonet hey all 12:19
masak takadonet: o/
frettled rakudo: chdir(".."); $*CWD.say; $*CWD="/home"; $*CWD.say;
moritz_ \o
p6eval rakudo 0d4fe0: OUTPUT«/home/evalenv␤/home␤» 12:20
frettled rakudo: $*CWD="/home"; $*CWD.say; chdir(".."); $*CWD.say;
Matt-W Actually in a way
p6eval rakudo 0d4fe0: OUTPUT«/home␤/home/evalenv␤»
moritz_ frettled: just wanted to write the same :-)
Matt-W $*CWD = absolute path isn't all that bvad...
it does have a nice symmetry to it 12:21
$*CWD ~ subdir though... uck
Matt-W oscillates through opinions as he eats lunch
wayland76 moritz_: regarding $*CWD ~~ m:rw/<some_regex>/
12:21 payload joined
moritz_ if $CWD is declared as Path where { $CWD.io ~~ :d } you'd get a type mismatch when trying to enter a non-existent directory :-) 12:22
wayland76 I don't understand it, but I would expect it to check if that was a legitimate path, and then assign
What's $CWD.io ?
frettled That would be kindof cool. 12:23
moritz_ well, m:rw does a binding, not an assignment
wayland76: $foo.io ~~ :d is the (new) way to do file tests
or was it .IO?
don't remember
Matt-W .IO
wayland76 Where $foo is what type? 12:24
moritz_ Str
or Any
wayland76 And .io is what?
moritz_ don't know for sure
Matt-W this is one of the source of the debate on a file path type
moritz_ .io is .IO
wayland76 Ok, different question -- where do I read.
wait, I'll find it
moritz_ ack -wQ .IO
12:25 szabgab joined
wayland76 Ok, but with this new Path quoting idea, the .IO should die 12:26
moritz_ not all path names are literals.
wayland76 Because Path should do that now
moritz_: example?
moritz_ my $filename = $handle.get; 12:27
$filename is a Str (or maybe Buf)
not a Path
wayland76 Hmm. If we did my Path $filename = $handle.get; would that coerce it?
moritz_ no 12:28
it would throw a "type mismatch" exception
wayland76 Ok, what about my $filename = p:qq{{$handle.get}}
Note double {{}} :)
moritz_ screams in pain 12:29
wayland76 Sorry, that was rather evil :-} 12:30
But it works, right?
:)
moritz_ the multiplicity of delimters doesn't carry any semantics (except tripple braces in quasi quotes)
wayland76 But this is a code block embedded in a path string
masak wayland76: nope.
wayland76 No? 12:31
masak wayland76: re-read S02 :)
wayland76 Whereabouts?
moritz_ wayland76: $str.Path ~~ :e might be made to work
but .IO is shorter.
masak rakudo: say qq{{special? I don't think so!}}
p6eval rakudo 0d4fe0: OUTPUT«special? I don't think so!␤»
wayland76 Ok, but if we assume that .IO returns a Path?
frettled And $foo.Io would tell us whether it was a daughter who was also a priestess.
…having been seduced by Zeus.
moritz_ slaps frettled 12:32
wayland76 protects frettled
After all, he didn't mention moons, by Jupiter! :)
Matt-W umm 12:34
if we had Path, wouldn't $handle.get return Path anyway?
moritz_ Matt-W: and how would you read non-paths from a file, then? 12:35
wayland76 Argh, you're right
Matt-W oh
sorry
.get's for reading from it
wayland76 (well, right about the {{ }} business)
Matt-W bangs head on desk
teach Path how to be assigned to from strings
masak Matt-W: please don't do that. I get concerned. :/ 12:36
moritz_ aye, poor desk
Matt-W around here nobody notices
we're all doing it
masak :P
wayland76 Well, I vote for $str.Path ~~ :e
:)
IO is admittedly shorter, but still 12:37
Matt-W I've never liked that
wayland76 Never liked which? .IO ?
carlin $str.IO.e is shortest
frettled Couldn't $str be a pipe, socket or somesuch? 12:38
wayland76 frettled: ???
moritz_ wayland76: I think you have to ask for permission from larry before changing file test operations
Matt-W no the ~~ :e business
moritz_ frettled: it would be evil to store a socket in a variable named $str
but certainly possible 12:39
frettled moritz_: yes :D
Matt-W yah someone's bound to do it
but that's their problem
frettled mm
wayland76 moritz_: Yeah, I figured that. But if we can get $str.IO or $str.Path to return a Path, it will all work out.
moritz_
.oO( semantics bound to variable names would be a cool, evil idea)
wayland76 moritz_: FORTRAN does that 12:40
frettled $luke.force is always true?
wayland76 ...well, type based on name, anyway :)
frettled while $luke.warm checks weather.com for your locale first?
moritz_ wayland76: only in F77 or older, I think
wayland76 Well, F77 is what I know :)
After all, my choice of languages when I was young was a) BASIC, b) F77, c) x86 assembler 12:41
So I dabbled in all of them, but FORTRAN the least
But my Dad started using FORTRAN in the punched card days 12:42
12:43 KyleHa left 12:45 finanalyst joined, Zloyrusskiy left 12:58 szabgab left
carlin I wonder what they argued about in the punch card days, with no tabs vs spaces and emacs vs vim to get worked up about 13:01
omega carlin: round vs square holes
lambdabot omega: You have 1 new message. '/msg lambdabot @messages' to read it. 13:02
qp_pq is there some GUI framework I can use with p6 ?
like wx/qt/gtk2 ?
Matt-W not yet
moritz_ Tene had some working binding to e17 or something similar
DanielC What is e17?
moritz_ a window manager, long name "enlightenment" 13:03
it also comes with a GUI toolkit
which Tene used
DanielC Why is enlightenment called e17?
moritz_ because it's shorter to type
rakudo: say chars('enlightenment') 13:04
Matt-W because it's different to e16
DanielC Is it version 17?
Matt-W e17 is the fabled new version
p6eval rakudo 0d4fe0: OUTPUT«13␤»
DanielC ah
Matt-W which has been in development for longer than perl 6, I believe
moritz_ it should be called e11t :-)
Matt-W no 13:05
no it shouldn't
eiro Matt-W, the core lib was reimplemented lot of times afaik
so it should :)
e17 is the window manager by default on hurd ... coming with a "duke nukem forever" theme 13:06
frettled moritz_: what Matt-W said; e11t ≠ e17 ;)
eiro++
Matt-W eiro: coming with duke nukem forever
oh wait it was cancelled 13:07
oh dear
moritz_ nobody in the Perl 6 community should make fun of other projects' release schedules.
mkelly32 at least there's publicly visible code for e17..
compared to dnf
eiro yeah .. canceled ... that's very sad
Matt-W yeah but duke nukem forever was just insane
at least we've been making progress
they seemed to keep throwing it all away and starting again
PerlJam moritz++ 13:08
eiro moritz_, what about auto derision ? 13:09
masak (Leon Timmermans)++ # p6l email 13:10
moritz_ eiro: well, irony is hard to tell on IRC
eiro sure ...
moritz_ masak: indeed 13:11
13:11 Front_slash joined
masak I think I'll side with the people who propose sanity and lack of magic in the field of files, filesystems, IO and paths from now on. 13:12
I don't grok much of the different standards, but I think being as simple as possible is the best way not to get very old very quickly. 13:13
13:13 zloyrusskiy joined
PerlJam To quote Kernighan and Pike: Simplicity. Clarity. Generality. 13:13
moritz_ I agree.
Matt-W magic can always be added with module goodness 13:14
masak: please post an astonishingly elegant argument which will convince everyone else
PerlJam I haven't really been reading the discussions ... just skimming to a get to something that makes me go "ick!" and then skip to the next message.
moritz_ the difference between multi volume systems (Windows, VMS) and single root systems (unix) alone makes it really hard to come up with a unificaton
masak Matt-W: you're overestimating me. :)
DanielC Sedulously eschew obfuscatory hyperverbosity and prolixity.
Matt-W it's an inherent problem with a cross-platform language core 13:15
moritz_ to the point where it's good to ask if you even want a unification
Matt-W yeah
local directory navigation is more or less the same
but then you get syumlinks
moritz_ AFAICT no perl 5 module (or module system) has been invented so far that deals well with all of those
Matt-W and roots
and all the different rules about what can be in a filename 13:16
moritz_ and *that* should make us think
Matt-W it's tempting to say it's all microsoft's fault, but then... VMS
I have no idea at all what VMS's filenames mean
moritz_ not allowing a file named 'COM1' sure looks like a wierd restriction :-) 13:17
wayland76 Can I just point out to everyone that I've been using E17 for months, and it appears to have fewer problems than Rakudo? :) 13:18
Matt-W aaaah the days of COM1
and PRN1 13:19
and in the early days of USB you used to get USB devices pretending to be COM5 etc.
DanielC I'm a bit lost... in Windows you can't have a file named COM1 ??
moritz_ DanielC: no, you can't
PerlJam copy con foo.txt # What happens when I have a filenamed "con"?
DanielC hasn't used Windows in more than a decade
moritz_ copy foo com1 # copies to a com port, not to a file named com1 13:20
DanielC That seems like a very stupid restriction.
13:20 jaldhar left
Matt-W yeah it's crazy 13:20
wayland76 If you try to write to a file called COM1 it sends your output to the Serial port. I'll tell you the rest later, though -- it's a serial story.
Matt-W you used to be able to say
copy con prn1 13:21
and it'd send what you typed straight to the printer
wayland76 Yah, I remember that :)
Well, I'd forgotten it, but I remember doing it now that you mention it :)
DanielC I much prefer the Unix /dev/ directory.
wayland76 Well, it does have certain advantages :) 13:22
moritz_ DanielC: well, that requires a root directory, which windows doesn't have :)
wayland76 moritz_: No, you could've had a DEV: drive :)
DanielC Oh, that's right... Windows has those stupid letters like C:
wayland76 DEV:\COM1 ...but they didn't do that
Matt-W no because drive letters can only be 1 letter 13:23
moritz_ aye
Matt-W: sure?
Matt-W and even in windows 7, your main partition is still C:
wayland76 Matt-W: I'm talking about what they *could* have done :)
Matt-W because A and B are still reserved for the floppy drives you probably haven't got
PerlJam This whole filepath discussion kind of reminds me of string encoding. Where it's assumed that we can have some canonical path representation and ways to transform it into something appropriate on a per filesystem basis.
wayland76 No, you can have two-letter drives under certain conditions -- I read it today
jnthn I've not seen anything suggesting you can have anything beyond A..Z. :-/
Matt-W PerlJam: I think it requires us to know too much about file systems
PerlJam: maybe a module :)
wayland76 jnthn: Yes you can :) 13:24
moritz_ very windows-like would be use a special letter: |:\COM1
jnthn wayland76: Heh, curious.
wayland76 p:unix{/usr/local/bin/!!!}
Or whatever :)
Matt-W o fcousre these days they try to present a single root
they try to make it look like everything stems from Desktop
or maybe from Computer
or maybe from...
it's all illusory and confusing
jnthn Yeah, but it all falls apart as soon as you dig.
Matt-W goodness knows what newbies think 13:25
Unix filesystems are bad enough
jnthn Well, newbies probably don't have to think too much about it. :-)
moritz_ yes, and Desktop itself lives under C:\Windows\UserBla\...\Desktop\
Matt-W Depends which version of Windows 13:26
since Vista it's been c:\Users\Username\Desktop
moritz_ so you have a nice circularity, which reminds me of meta object protocols
Matt-W they actually have a 'home directory' now
moritz_ please tell me that everything outside the home directory is not writeable by default 13:27
(except /tmp/ maybe)
wayland76 OTOH, Unix would've been nicer if users had a /user/<username> directory that mirrored /usr in directory structure, but had /user/<username>/data as their default directory when logging in
moritz_ wayland76: aye
that would remove the need for hiding the dotfiles 13:28
Matt-W unix's way is far from perfect
i rather like mac os x's way, with application bundles
wayland76 moritz_: I don't know if they've gone that far, but they're working on locking things down. They have to lock things down a bit at a time, and then wait for all the software to catch up. Otherwise they get a flop like W2K
Matt-W none of this mixing everythign together in /usr/bin
moritz_: there's some write protection, but it's nowhere near complete at this stage
wayland76 Or even having /home/<username>/.local/etc for config files :) 13:29
Matt-W too many apps depend on being able to write to their own directories and things
wayland76 Matt-W: There are advantages to app bundles, but they make the $PATH a lot longer
Unless you're using links from /usr/bin anyway :) 13:30
In which case, it should all come back to the package manager :)
Matt-W mmm
frettled masak++ for an elegant argument which puts things in the right perspective.
Matt-W or you have a different method for locating a binary to run...
masak frettled: I'm amazed by people's ability to bikeshed. I generally try not to disturb them, but this time, after Matt-W encouraged me, I fell for the temptation. 13:31
Matt-W :)
well eventually someone just has to cut through it and make a decision
which is what we really need in the band at work... 13:32
frettled yup
Matt-W it's not just programmers who bikeshed
wannabe musicians do it too
frettled I'm still stunned by David Green's claim that the behaviour of environment variables is unexpected, i.e. he thinks that changing an environment variable should affect the semantics of some function that normally updates the variable, and that if it doesn't, it's a bug. Or he's joking. 13:33
(and I'm just not Getting It) 13:34
I hope I'm not being too crass, though.
Matt-W I think it is a bit unexpected
because chdir() changes CWD
changing CWD should... what
dunno
probably be forbidden
since changing it manually just can't be a good idea 13:35
moritz_ agreed
I'm still unclear about another thing... 13:36
13:36 Front_slash left
Matt-W so by extension, if it goes that assigning to $*CWD doesn't change the current directory, it should cause an error 13:36
frettled Matt-W: what about $USER, $LANG, $LC_COLLATE, …?
moritz_ that is, the specification allows an interpreter/server/whatever to execute multiple perl scripts
Matt-W changing $LANG actually has an effect immediately which is useful
moritz_ and has a separate PROCESS namespace for that
Matt-W changing $USER... bad 13:37
frettled Matt-W: it has an effect on new programs that are started, because new programs inherit ENV
That is expected.
moritz_ but in unix the current directory is stored per process
so, what will chdir() do in such a situation?
Matt-W DanielC++ # for brevity 13:38
moritz_ if it just calls the chdir syscall it violates encapsulation of the separate perl scripts
should it die?
wayland76 moritz_: Well, if you pass in the "party" option, then it changes them all, but if you pass in the "solo-woodland-trek" option, it just changes itself
Matt-W moritz_: you shouldn't run perl scripts in that situation which use chdir
it's just the same as calling chdir in a thread 13:39
while your other threads are making assumptions about the CWD being constant
you just don't do it
or at least, if you do you're going to suffer for it
frettled You just don't make assumptions like that.
moritz_ Matt-W: ok, that was what I thought
Matt-W unfortunately issues of in-process concurrency are the kind of nasty problem it's very hard to insulate people from 13:40
frettled yep
Matt-W I suppose the interpreter *could* pretend to have multiple working directories for each subprocess
but that'd be rather unpleasant
frettled It's quite feasible, but then the memory gains are fewer.
13:42 azawawi joined
azawawi hi 13:42
moritz_: ping 13:43
masak awwaiid: oh hai
moritz_ azawawi: pong.
masak argh, sorry awwaiid. mistab.
azawawi: oh hai.
moritz_ a<tab> :-)
azawawi :) 13:44
masak moritz_: actually, I did aw<tab> for some reason. :/
azawawi we're removing .p6 extension for .pl in Padre Perl support
masak good.
azawawi so we need to reliably detect whether this is Perl 6 or Perl 5
'use v6;' alone?
masak azawawi: that's your safest bet, I think. 13:45
moritz_ use v6;, class, grammar, role or module declarations
all indicate Perl 6
masak azawawi: but not all scripts in the wild do that.
azawawi since 'class/role' in MooseX::Declare...
masak azawawi: Rakudo is very lax and treats everything as Perl 6.
13:45 ejs left
jnthn That's unfortunate...given the Perl 6 spec said "package is the way we'll know it's Perl 5". :-) 13:46
Or a way...
13:46 Front_slash joined
azawawi hmm 'use v6;' is the safest bet... i'll make Padre warn the programmer when use v6; is not found... 13:46
moritz_ azawawi: well, the class/role/grammar/module/v6 thing is what the spec says - you'll have to find out what works best in the wild, I guess
jnthn You may be able to take use of MooseX::Declare as a heuristic. 13:47
azawawi interesting
13:49 frew joined
jnthn That is, class/keyword/grammar/role package declaration without MooseX::Declare implies Perl 6. 13:51
Assuming MooseX::Declare is in wide enough use that it's worth making an exception for.
13:53 unitxt left
colomon rakudo: class X { has $.y is ro; multi method new ($a) { self.bless(*, y => $a); } method foo() { $.y = 10; } }; my $a = X.new(2); say $a.y; $a.foo; say $a.y; 13:54
p6eval rakudo 0d4fe0: OUTPUT«Statement not terminated properly at line 2, near "method foo"␤in Main (src/gen_setting.pm:3390)␤» 13:55
moritz_ colomon: you need a ; after the }
azawawi some thing like this is used to detect right now ... gist.github.com/169722
moritz_ rakudo: class X { has $.y is ro; multi method new ($a) { self.bless(*, y => $a); }; method foo() { $.y = 10; } }; my $a = X.new(2); say $a.y; $a.foo; say $a.y;
p6eval rakudo 0d4fe0: OUTPUT«2␤Cannot assign to readonly variable.␤in method X::foo (/tmp/81VA8o8316:2)␤called from Main (/tmp/81VA8o8316:2)␤»
colomon rakudo: class X { has $.y is ro; multi method new ($a) { self.bless(*, y => $a); } method foo() { $.y = 10; }; }; my $a = X.new(2); say $a.y; $a.foo; say $a.y; 13:56
p6eval rakudo 0d4fe0: OUTPUT«Statement not terminated properly at line 2, near "method foo"␤in Main (src/gen_setting.pm:3390)␤»
azawawi jnthn: any comments on this one gist.github.com/169722?
moritz_ rakudo: class X { has $.y is ro; multi method new ($a) { self.bless(*, y => $a); }; method foo() { $!y = 10; } }; my $a = X.new(2); say $a.y; $a.foo; say $a.y; 13:57
p6eval rakudo 0d4fe0: OUTPUT«2␤10␤»
13:57 frew left
moritz_ we *really* need to teach people that ro accessors are also or within the class 13:57
13:58 unitxt joined
colomon moritz_: ??? 13:59
moritz_ you can't assign to $.y unless $.y is declared as 'is rw'
because it's just a method call
it's short for $( self.y() )
jnthn azawawi: Looks like a good start.
# Perl 6 'use v6;' 14:00
return 1 if $text =~ /^\s*use\s+v6;/msx;
Any reason not to try that first? Is it not fairly sure-fire?
colomon moritz_: Errr... isn't that what you just did with that code? Assign to $.y?
moritz_ confused by return if $text =~ /^=head[12]/msx
colomon (I was trying to see if "is ro" was actually implemented yet....)
moritz_ colomon: 'is ro' is the default 14:01
azawawi jnthn: embedded Perl 6... in Perl 5
jnthn azawawi: Heh, true.
moritz_ azawawi: maybe another approach would be useful...
azawawi: iterate over all lines
azawawi: and use the first successful check
then when something is identified as Perl 5 somewhere, you can ignore embedded Perl 6 later on 14:02
azawawi moritz_: will think about for sure...
moritz_: will think about it for sure... :) 14:03
14:03 masak left
azawawi thanks guys 14:03
moritz_ azawawi: oh, and ^=cut is a sure sign of Perl 5
azawawi adds it... 14:05
wayland76 is almost bikeshedded out :)
carlin We need Bikeshed.pm which implements every behaviour anyone has ever asked for :) 14:06
frettled wonderful :D
wayland76 what, like class Bikeshed { foo() handles *; method foo() { say "It is done, O master" } } 14:08
moritz_ use Bikeshed :from<p6l@2009-08-08> :auth<moritz@faui2k3.org>;
wayland76 (Ok, you can't have that handles part -- how would I do that?)
14:08 bpetering left 14:09 Front_slash left
pmichaud good morning #perl6 14:09
wayland76 How do I shoot Google for finding "text" when I asked for "TeX"? :)
14:10 masak joined
PerlJam bonus nachos pm 14:10
moritz_ oh hai pmichaud
wayland76 hail to one of our Rakudo leaders!
pmichaud wayland76: Shooting Google would be like firing nukes at the Sun. :-)
moritz_ wayland76: I suggest ICBMs
jnthn morning pmichaud
wayland76 Can't afford ICBMs. How about IBMs? :) 14:11
moritz_ is IBM so much cheaper? :=)
masak pmichaud: oh hai. ROADMAP says PGE 'will be undergoing some significant refactoring in summer 2009'. is that deliberate?
Matt-W o/ pmichaud
pmichaud masak: yes.
wayland76 Well, I'm sure I can find a secondhand IBM mouse or something on eBay :) 14:12
Matt-W masak: there will be refactoring!
wayland76 masak: It's so we get LTM, etc
colomon moritz_: If 'is ro' is default, and still lets you reassign $.y in the above, what does 'is ro' actually do?
moritz_ proto regexes!
masak wayland76: oh, I know what it is. I'm just querying about the timing.
wayland76 I've been following it as closely as I can without hassling pmichaud
masak wayland76: same here.
moritz_ colomon: where did it allow it? 14:13
wayland76 Well, when pmichaud said summer 2009, I said "Nah, February 2010" :)
moritz_ colomon: note that assignment to $!y is allowed within the class, to $.y not
colomon: if you find an instance where assignment to $.y is allowed and it's not declared as 'is rw', please submit a bug report 14:14
pmichaud note that "significant refactoring in summer 2009" does not necessarily imply "ltm/protoregexes in summer 2009"
wayland76 But that was only based on how much progress had been made at that point
Ah, ok :)
I guess I knew that
But the sig. refact. has to be finished before you start LTM and protoregexes? 14:15
pmichaud there's a fair bit of refactoring that has to take place even before we get to protoregexes and ltm
right.
the deadline (and it is really a deadline of sorts for Rakudo *) is january 2010 for protoregexes/ltm
although I hope to be finished much sooner than that, because my grant income depends on it 14:16
wayland76 That would be nice :)
PerlJam refactor mercilessly.
pmichaud this refactor will indeed be merciless.
PerlJam (I have that written on one of the whiteboards in my office :)
pmichaud For one, when I started PGE I had no notion of Cursor objects
wayland76 Ooh, motivation. I like that for us, even if it's hard on you :)
pmichaud and apparently STD is completely based around the notion of Cursors (which is fundamentally more correct, imo)
moritz_ so you'll pass around cob's instead of mob's? :-) 14:17
masak pmichaud: I've been skimming PGE lately. it's deliciously small, for what it does.
wayland76 Hmm. Well, I can see that making a refactor useful :)
pmichaud masak: thank you. I'm shocked about PGE also for two reasons 14:18
(1) that it is as small as it is
wayland76 moritz_ et. al: Just for the amusement value, a "cobber" in Australia is somewhat like "mate", or "buddy", or "friend" or whatever
pmichaud (2) that it runs with any sort of usable speed at all. Granted, it's horribly slow, but it's about 100x faster than I would've predicted at this point
wayland76 So cobs are more friendly than mobs :) 14:19
14:19 szabgab joined
moritz_ :-) 14:19
moritz_ has to take a leave, TTFN folks
masak moritz_: o/
Matt-W wayland76: given what 'mob' usually means in an MMO, yes :)
colomon moritz_: (Sorry for the delay -- taking care of an 11-month-old here at the moment.) That little code sample you got to work had the method foo change the value of the 'is ro' variable. That was the point of the test, to see if it was possible. 14:20
14:24 yath joined
wayland76 Hey, everyone, I'd just like you all to note the subject line of David Green's recent tongue-in-cheek message 14:24
I think it's past my bedtime too :)
Anyway, as it's after midnight, I'm going to bed 14:25
night all o/
14:26 ruoso joined
PerlJam colomon: I'd call that a bug. 14:27
colomon I thought it might be a NYI, which is why I was checking. 14:29
PerlJam colomon: oh, it may be NYI
"is ro" may be only parsed. 14:30
What does this mean? $value := PAST::Op.new( :pirop('new PsP'), 'ObjectRef', $value); ?? what is PsP ? 14:39
14:41 azawawi left, jferrero joined
[particle] the pir op new has a sig of PsP 14:41
returns a pmc, takes a string, and a pmc 14:42
lowercase s may mean 'string or string constant'
i forget, but it should be in the past pdd iirc 14:43
PerlJam Thanks.
14:43 __ash__ joined, Psyche^ joined
PerlJam parrot can still use strings for class names? I thought there was movement towards using keys everywhere 14:44
I mean, doesn't that turn into something like $P0 = new 'ObjectRef', $P1 ? Am I wrong about the keys? 14:47
PerlJam goes with "must be wrong" for now 14:48
14:50 leedo joined
tlb Hey a little question when we read from prompt() what's encoding is set on the string and what does we look at? 14:52
and what about readdir? 14:54
14:58 KyleHa joined
KyleHa Howdy doody, #perl6. 14:59
14:59 nihiliad joined 15:01 Patterner left, Psyche^ is now known as Patterner
carlin Night everyone 15:01
15:01 carlin left, cdarroch joined
__ash__ what are the perl6 --target= options again? or where is it documented? I cant find it again 15:07
PerlJam I don't know that they are documented anywhere in particular. 15:09
pir, past, post, and parse I think are the options. 15:10
(I mean, I think they are the only options)
15:15 alester joined
masak __ash__: it's "documented" in the Parrot repo, in compilers/pct/src/PCT/HLLCompiler.pir 15:15
PerlJam gutentag alester
alester hey
__ash__ thanks PerlJam and masak, i was looking in the wrong documents then, oops 15:16
15:17 tann joined
PerlJam __ash__: oh, it's also in the parrot repo in docs/pct/gettingstarted.pod 15:17
and that doc only mentions the ones I know about, so we'll go with that's all there is :) 15:19
15:21 tlb left
ruoso just typed our &foo := something_that_returns_a_coderef() in Perl 5 15:25
15:26 justatheory joined 15:31 TimToady joined 15:37 szabgab left
takadonet What method name do you need to use to create your own constructor ? Is it BUILD? 15:38
TimToady you should call .bless
mikehh_ t/spec/S12-enums/thorough.rakudo ............................... Failed 1/71 subtests 15:39
15:39 mikehh_ is now known as mikehh
ruoso takadonet, there isn't a special "constructor" method in Perl 6, as there is in Java, for instance... 15:39
TimToady unless you know more about object's layout that lets you bypass bless/BUILDALL/BUILD 15:40
ruoso takadonet, constructing an object is a matter of blessing a candidate (as it is in Perl 5), but Perl 6 has some *default* implementations in place
TimToady note that the fact that BUILDALL and BUILD are in all caps is indicative that you shouldn't generally call them directly
ruoso or override unless it really makes sense
takadonet i thought you just wanted to shout it at people.... 15:41
TimToady it generally indicates something that is called automatically
masak ruoso: funny, I feel more ready to override BUILD than to create my own 'new' method. 15:42
takadonet masak: me too but I was not sure if it was the right thing to do
TimToady overriding BUILD is fine, just like writing your own END block is fine 15:43
it'll just get called at the appropriate moment
ruoso masak, if it really makes sense
masak it feels like a smaller change than overriding .new
ruoso masak, my point is that people usually override the constructor to do things that could be solved with default values and other stuff
mikehh rakudo (5a2aeaf) builds on parrot r40624 - make test PASS, make spectest (up to 28020) FAILs 1 test - t/spec/S12-enums/thorough.rakudo - Failed test: 2 15:44
masak ruoso: aye.
ruoso masak, my point wasn't in favor of overriding new instead ;)
masak I see. :)
takadonet Well not sure if this 'issue' can be solved without overriding BUILD 15:45
ruoso and the issue is?
PerlJam ... always simpler than people think it is ;) 15:46
ruoso lunch &
takadonet I'm rewriting bioperl libraries so I'm pretty much taking what's already written in perl5 to perl6. One object has two 'attributes' called 'description' and 'desc'. One is just a short hand for the other and they just point to the same thing. 15:47
mikehh btw the test was on Ubuntu (.04 amd64 (gcc) 15:48
9.04
takadonet The problem is on creating the object. I need to populate both 'description' and 'desc'.
TimToady if you change one, does the other change? 15:49
takadonet correct
TimToady that is, do you just want an extra accessor to a single attribute? 15:50
if so, then just write the extra accessor
takadonet .... and mutator
TimToady method desc is rw { $!descriptor }
tion
PerlJam takadonet: Just make sure you initialize the attr consistently in your constructor.
takadonet PerlJam: I do, but others may not 15:51
TimToady ah, but yes, you'll need a BUILD if they might initialize with either name
takadonet You got it
also going to be the same with 'seq' and 'sequence'
15:52 frettled sets mode: +oo masak TimToady
PerlJam this feels like something perl could do for you though. 15:52
TimToady the accessor would work if BUILD set $.desc instead of $!desc, but it doesn't generally
however, this would actually be a decent place to define new instead 15:53
PerlJam class Foo { has $.desc is aliased('description') ... } or something
TimToady to canonicalize the attribute names before calling bless
15:55 nihiliad left
s1n @seen masak 15:56
lambdabot masak is in #perl6. I last heard masak speak 11m 29s ago.
masak s1n: I'm right here.
s1n yay
TimToady possibly $!desc ::= $!description could be forced to work somehow
PerlJam ah, even better.
s1n masak: the sources listed in the Makefile.in of web are the only installed sources??
masak: what about the drafts, smartlinks, and hitomi stuff?
masak s1n: well, those don't need to be compiled. 15:57
s1n: Makefile.in contains things that need to be compiled.
s1n masak: ah, but the others are still installed?
TimToady oh, wait, it's just alternately named args
s1n masak: i ask because i can't seem to run the Configure
15:58 nihiliad joined
TimToady so you can write the BUILD, but canonicalize with :desc($!descriptor) and such 15:58
masak s1n: I'm not sure what you mean by 'installed' here.
TimToady or other way around, depending on which one is the real one
masak s1n: at this stage, Perl 6 project don't get installed.
s1n masak: well, you know what i mean 15:59
masak s1n: no, and that's my point. I don't.
s1n: what error does Configure give you?
s1n masak: okay, what do you do with the smartlinks and hitomi files? are they referred to from other modules in the project?
masak s1n: no.
TimToady likewise, a parameter for :seq($!sequence)
s1n masak: (keep in mind i haven't looked at Web in a few months) Can't find ./Configure in @*INC 16:00
masak s1n: ah. try doing `export PERL6LIB=`pwd`/lib`
argh, quote fail. 'export PERL6LIB=`pwd`/lib' 16:01
s1n masak: okay, so how should Web be "installed"? what files are necessary for an installation?
takadonet I think I understand TimToady 16:02
16:02 justatheory left
TimToady I'd still probably just write a new 16:02
with a *%_ parameter 16:03
masak s1n: Web.pm is generally not installed. neither is it 'installed'.
takadonet Any good example of new constructor in the spec?
masak s1n: but if you manage to compile it, and then grab all the *.pir files in lib/, you should be off to a good start.
s1n masak: not just Web.pm, your web.git project
TimToady then do %_<seq> ||= (%_<sequence>:delete) or some such before calling .bless 16:04
masak s1n: I think I still would be more helped by knowing what it is you're trying to do, not step 5 or so of your thought process.
16:04 tann left
s1n masak: pm 16:04
16:05 rfordinal joined
TimToady takadonet: no full examples, but use of .bless is discussed in S12/Construction and Initialization 16:05
takadonet TimToady: Already reading it :) 16:06
16:07 __ash__ left
TimToady actually, since the last change to adverbial precedence, you don't need the parens on the :delete example above 16:08
not that rakudo implements :delete yet...
16:09 payload left 16:17 kent\n left 16:18 qp_pq left 16:19 kent\n joined, nihiliad left
takadonet hmmmm I just realized that this has to be done on a role..... 16:21
takadonet hits head against the wall
masak careful!...
TimToady a role's BUILD is supposed to be integrated into the class's BUILD, so maybe that's the way to go eventually 16:22
16:22 nihiliad joined
TimToady dunno if rakudo does that right though 16:22
masak I think it does.
TimToady fortunately, I'm hard headed
16:23 mj41_ left
TimToady people think I have a swelled head, but it's all thick skull 16:23
16:24 masak left
TimToady wayland76: the point of using $*CWD would be (and would *have* to be, given how context vars work) to give each thread its own working directory, independent of the process as a whole 16:25
jnthn It'll bring the BUILD method into the class, but if the class has one of it's own then it'll clobber the one from the role. At the moment, BUILD just has normal composition semantics (and I don't recall seeing anything in the spec saying otherwise). 16:26
takadonet nuts.... 16:27
So much work just to have some attribute with aliases. 16:28
16:28 mj41 joined
jnthn Why is that harder than just method the_alias() { $!the_attr } 16:35
TimToady because build sets $!the_alias rather than $.the_alias 16:36
(or should)
jnthn Ah.
TimToady posibbly defining method !the_alias could be made to dtrt
jnthn :-S 16:37
Unlikely.
takadonet Just ignore this for now. Only thing that this limits is when creating the object, you need to use 'desc'
TimToady goes against what we decided about $!the_alias before
you can still write your own new, can't you?
16:37 molaf joined
TimToady and do the key renaming trick on %_? 16:38
16:39 __ash__ joined
TimToady or method new (:seq($sequence), :desc($description), *%_) or some such 16:39
then $class.bless(*, :$sequence, :$description, |%_)
'course, if there's lots of different classes, that could get old 16:40
16:41 finanalyst left
takadonet There is going to be LOTS of different classes in the range of 100 or so 16:41
I have only one written so far.....
TimToady well, it's an interesting problem, for which we thank you :/ 16:42
16:43 payload joined
TimToady jnthn: I remember discussing role BUILD with someone somewhere, and how they should compose, but I don't see it offhand in the specs 16:43
16:44 smash_ joined
jnthn TimToady: It's an interesting problem, but I've read the role specs in quite some detail and don't recall seeing anything along those lines. 16:44
TimToady: Question.
takadonet TimToady: Your welcome but I rather just have an answer hehe 16:45
TimToady it's possible that it's submethods in general that should just merge under composition rather than conflict
jnthn class Foo { has $.attr is my_custom_trait; }
What does the trait_mod get passed here? 16:46
16:46 DanielC left, nErVe joined
jnthn The underlying storage unit, e.g. what's in $!attr? 16:46
That is, is it the same as in my $foo is custom_trait;
TimToady hmm, well, is rw has to be able to trigger method generation 16:47
or rather
jnthn Well, we handle that in the compiler.
TimToady transfer its affections to the generated method
ruoso has $.the_alias is DynamicAlias<the_attr>;
where DynamicAlias would reference the proper attribute when used
jnthn ruoso: DynamicAlias? nah, is aliased<the_attr> :-) 16:48
;-)
But anyway, it's writable as a mdoule _provided_ there's some way of getting at the current metaclass.
ruoso jnthn, right... that should work... I was thinking about storage type
but trait should work as well
jnthn TimToady: Which is my real question. A trait on an attribute probably should be able to get at the metaclass... 16:49
TimToady well, basic principle is that if you're throwing away info too soon that you need later, then you're probably passing the wrong object to the trait
ruoso thinks at $*CLASS
TimToady or there needs to be a link somewhere
jnthn Aye.
TimToady but links don't always work under one-to-many conditions
jnthn Custom traits on attributes don't work in Rakudo yet (is rw is compiler handled...) 16:50
ruoso isn't CALLER::<$?CLASS> enough?
TimToady which often means you have to pass the other object and link many-to-one
don't know if that principle applies here though
jnthn ruoso: Probably yes.
I'm just fine with that being the answer.
TimToady well, "has" is in some sense declaring a slot, not any particular piece of that slot 16:51
so maybe we need to make the slot abstraction a bit more official
and the slot delegates to the $!name or $.method as appropriate
ruoso you mean the trait routine receives the "declaration" and not just the "declared" 16:52
?
TimToady it's still a declared, but a wrapper of sorts around other declareds
ruoso I think that's what I meant by "declaration" ;)
TimToady I like my way of thinking about it better :) 16:53
jnthn A kind of Slot object that lets you get at the declaration, but has other properties too?
erm
The declarand I mean
:-)
TimToady to me a declaration is a syntax/semantics thing
yes, a synthetic declarand
ruoso points at semantics
jnthn TimToady: So what might this look like, more concretely? 16:54
TimToady semantics isn't an object like a declarand is
semantics is a mass noun
would you please fill that jar with some semantics, please, dear?
ruoso ahehaeaehaehae
TimToady 'course, you can point at water too 16:55
but computers don't like pointers to mass nouns
ruoso .oO( void* )
anyway... jokes aside, I think we agree on the practical side 16:56
jnthn Aye.
So you think that a declaration will invoke a trait mod with something like a Slot object, with properties like .declarand, .name, etc? 16:57
[particle] padre is a mass noun
__ash__ @a.length is the number of items in the array right>
lambdabot Unknown command, try @list
TimToady jnthn: I would guess that we really just want to take one of the existing components such as $!foo and make it smart enough to act like the synthetic declarand by giving it suitable links
whichever of our components is most-instantiated already
so you end up with many-to-one links
jnthn Are you thinking this applies to just attributes, or variable declarations in general? 16:58
ruoso .oO( Is it just me or TimToady just ported Class::InsideOut to Perl 6? )
TimToady well, attributes are generic variables, basically
jnthn Right, so this is something more general.
TimToady but we're likely to run into other kinds of generic variable declarations, especially if TheDamian starts to write new talks 16:59
logic programming, constraint programming
these tend to have names that represent concepts, not just storage
jnthn Hmm, true.
ruoso wishes TheDamian had his talks available to the public in continents he doesn't visit
jnthn TimToady: Is this an area you want to be left alone to think about for a while? 17:00
Or do you think there's something I can implement now-ish?
jnthn was going to try and wrap some of this up today... 17:01
ruoso is still confused...
jnthn ruoso: Me too. :-)
TimToady for now just do the simplest thing that works :)
jnthn goes to the pub
TimToady that works
jnthn ;-)
ruoso simplest thing that works seems to be "have an abstract Slot that refers to each part of the declaration"
jnthn ruoso: What do you mean by abstract Slot here? 17:02
__ash__ jnthn: i have a question if you have a sec, about PAST's and actions.pm, well 2 really
jnthn I mean, let's take an example declaration.
ruoso an object that wraps all participants in the declaration
jnthn my $x is custom_trait;
17:03 riffraff joined
jnthn At one point, we're going to call trait_mod:<is>(XXX, :custom_trait) 17:03
ruoso thinks that Slot might be a bad name...
jnthn What is the XXX going to look like?
TimToady declarand is better
ruoso can't think of something much different than Declaration
jnthn __ash__: Sure.
ruoso right... declarand is find
so... a Declarand object, in this case a ScopeVariableDeclarand 17:04
jnthn On the one hand, the XXX could just be a Scalar. Or in the case of my @x, it's be Array, etc.
But I'm guessing instead we want something that wraps around that.
TimToady it's often just a container
ruoso that has $.visibility = 'my'; has $.name = '$x'; has $.storage_type = 'Scalar'
jnthn OK.
__ash__ jnthn: when you do --target=past, i have seen <pasttype> => "callmethod" and <pasttype> => "call" what is the difference between those? is that how it's translated into a method call at the next level down?
TimToady but it can be a container factory
ruoso container factory++ 17:05
jnthn __ash__: Difference is in the PIR that is generated.
ruoso TimToady, but not every declaration specifies a container
jnthn __ash__: call makes a 'foo'(c1, c2, c3,...), callmethod makes a c1.'foo'(c2, c3) 17:06
ruoso I think we would have an abstract Declarand type, that would have several specialized subtypes
TimToady ruoso: some containers are more abstract than others :)
17:06 nErVe left
ruoso AttributeDeclarand 17:06
MethodDeclarand
ScopeDeclarand
17:07 jhorwitz joined
ruoso using specialized subtypes will even allow you to multi-dispatch on the declaration type 17:07
jnthn Well, of routines passing the Routine itself is fine, since it contains a whole bunch of traits we can examine. 17:08
ruoso s/tion/nd/
jnthn *for routines...
TimToady a routine is a container of an indirection, which is why a routine is wrappable
jnthn Right.
ruoso jnthn, except if the trait is a modifier to the way the routine is stored, not the routine itself
yath mhm. say, if i just want to play around a bit with perl6, what would i use? rakudo? pugs? or what's the "most complete" implementation? 17:09
TimToady at the point of the declaration at least, we generally still know where it is going to be stored
jnthn Well, that's quite probably already happened before the traits get run.
17:09 justatheory joined
TimToady jnthn: yes, which is why "is also" changed to "augment" 17:10
ruoso TimToady, but the "how" it is stored is that can modfiied
jnthn yath: Rakudo is actively developed and pretty feature rich.
yath jnthn: okay, i'll check that out, thanks :)
TimToady
.oO( :carefully )
jnthn ruoso: I'd think not so much by a trait, unless it's going to retroactively go and fiddle.
pmichaud (should that be a FAQ somewhere useful? )
yath are there any "fundamental" changes between the different implementations, or is it just that one has less features than the other?
ruoso jnthn, is aliased<foo> is a good example 17:11
jnthn yath: Pugs still does a few things that Rakudo does not (e.g. laziness).
yath jnthn: but there's nothing in either pugs or rakudo that's not spec'd?
jnthn: like, say, some magic compile-specific things as in C 17:12
s/compile/&r/
pmichaud I think that by definition different implementations are likely to offer at least some items that "aren't specced".
TimToady they both provide ways of getting at lower-level primitives that aren't going to be specced
yath pmichaud: hm :(
ruoso yath, you should have a lot of fun with rakudo already...
yath TimToady: what primitives? like extending STD.pm? 17:13
TimToady yath: an implementation *must* do that, or there's no way to write Perl 6 code to implement the non-primitive bits
ruoso yath, daniel.ruoso.com/categoria/perl/dice-game-perl-6 for an example of fun
yath TimToady: maybe i don't get what you mean with "primitives" :) the NQP thing?
jnthn yath: The places where implementations do have such things is not something you're likely to run into by accident.
TimToady I mean things like inline PIR 17:14
pmichaud yath: Rakudo is built on the Parrot virtual machine. Some people will want to get access to features that are only available in Parrot, and Rakudo isn't going to forbid that from happening.
yath ruoso: ah, nice
TimToady, pmichaud: ah
so like, er, say, asm() in C ;)
TimToady that will never be in the p6 spec
precisely
pmichaud but pretty clearly we aren't going to specify that all Perl 6 implementations require Parrot.
jnthn yath: Along those lines.
__ash__ jnthn: also, what is contained in $/ ? and is there a reason why some methods have a space after ( and some done? (ie. $past.name('call') vs $past.unshift( $<variable>.ast ); ?
jnthn __ash__: $/ contains the match object 17:15
TimToady that's just formatting
jnthn __ash__: The other difference is just different people's coding preferences.
yath but if the "average" perl6 programmer writes a program, what's the chance that he accidentally uses a compiler-specific feature?
pmichaud yath: very small.
[particle] __ash__: some believe that when a simple item is contained in quotes, it should not have spaces around it
TimToady one tends to put more whitespace when it's harder to find the other end visually
__ash__ I was just wondering if there is a 'standard' i should be following since i am messing around in the core stuff 17:16
yath pmichaud: ah, good to hear
pmichaud yath: and in Rakudo's case we may add pragmas that are required in order to make use of the special features
(we just haven't done that yet)
yath that was *that* nice in perl5. if something worked in, say, perl 5.8, it worked in perl 5.8.
TimToady yath we tend to go on a forgiveness-rather-than-permission basis here
[particle] we've permitted that for quite some time 17:17
yath and i'm very concerned that in perl6 one writes code for rakudo, one for pugs, and so on
TimToady: forgiveness on what?
ruoso tries to pull back the trait issue...
jnthn yath: Unless you're doing something obscure, that's unlikely.
TimToady on where you put your optional spaces :)
yath my spaces?
jnthn __ash__: Just do what you think looks pretty. ;-) 17:18
yath am i doing python? :P
jnthn: fine :)
TimToady only in the spots where p6 requires spaces
[particle] ...perl 6 is closer to python spacing rules than you may know...
17:18 jferrero left
pmichaud peoplewhothinkthatwhitespaceisinsignificantneedtorethinkthatabit.:-) 17:18
__ash__ jnthn: alright, time to name my variables a, b, c /sarcasm
[particle] latin++ 17:19
TimToady but we don't use indentation for syntax
jnthn __ash__: That'd be an odd definition of pretty, but I guess beauty is in the eye of the beerholder.
erm, beholder.
:-)
pmichaud beerholder++
I'll have to remember that phraselet.
jnthn So anyway, getting back to traits on variable/attribute decls... 17:20
ruoso I think we have a generic Declarand type 17:21
__ash__ well, thanks for your time jnthn: i have something that almost works... it makes method calls like $a.Foo::bar() translate into a nearly identical Foo::bar($a) call, its not working (mainly i left it as callmethod rather than call) but when i do i'll try to send you a git-patch, gotta run
ruoso and specific sub-types
for attribute, method, variable etc
jnthn (This is like about the last thing I need to do before I can call my current Hague grant complete...thus my eagerness ;-))
TimToady roles, perhaps
17:21 __ash__ left
jnthn ruoso: OK, and would that have a way of actually getting at the container? 17:22
That is, the real declarand?
ruoso yeah...
something like .container
if it's a ContainerDeclarand maybe
the cool thing is that you can easily restrict where a trait can be used 17:23
jnthn So my $foo is bar; multi trait_mod:<is>(Declarand $d, :$bar!) { $d.container does bar; }
ruoso simply by setting that specific declarand subtype in the signature of the trait routine
yeah...
jnthn Right, just as you can already do now e.g. multi trait_mod:<is>(Method $m, :$thingy!) { ... } already in Rakudo will do the right thing (that is, only be applicable to methods, not routines in general). 17:24
ruoso jnthn, right... indeed... it carries just a bit more information than "Method" 17:25
jnthn TimToady: When you said roles, are you thinking we'd set this up as more like class Scalar does ContainerDeclarand { has $.name; method container { self }; ... }, but it's possible other declarations could have a completely separate object? 17:26
yath hm, won't corrent rakudo svn build against parrot 1.0.0?
err, rakudo git
jnthn yath: No.
yath hm, damn
pmichaud it's very rare that rakudo git will build against any given parrot release. Rakudo evolves too quickly for that. 17:27
jnthn yath: If you grab Rakudo from git (which I guess you've already done...)
And then just do
perl Configure.pl --gen-parrot
TimToady jnthn: I have no idea what people will want to declare 100 years from now :)
yath jnthn: yeah, but i thought i could use debian's parrot instead
jnthn It will get the right Parrot, make an install in your build directory, build Rakudo against it and make you something that Just Works. :-)
yath: Ah, OK. No, 'fraid not.
yath jnthn: but i didn't know if chaning build/PARROT_REVISION is a good idea %-)
jnthn yath: Changing it to an earlier version number? No, that'd count as a bad idea. :-) 17:28
yath well then, we're in 2009. doesn't take too much time to build parrot, i hope :)
jnthn No
yath jnthn: i changed it to 0 :P
jnthn TimToady: Heh. That isn't much help to me now. :-P 17:29
TimToady: Anyway, I can come up with something that works, that is along the lines of what ruoso++ is saying too, and I guess we can iterate from there.
yath uah, svn is *that* slow compared to git 17:30
*wait*
TimToady jnthn++
jnthn yath: Heh, the checkout will probably take almost as long as the build. ;-)
yath hehe
jnthn Thanks all. 17:31
TimToady Do you realize they can't even do karma in Haskell without a *monad*!?! :)
karma is a side effect...
szbalint Rakudo probably builds against Parrot 1.5 though, as it was just released :) 17:35
but I haven't tested it...
yath hm, i should have passed -j to make... 17:37
$ ./perl6 -e 'say "foo"' 17:43
foo
\o/
hm, is there no qw in perl6 anymore? 17:53
ruoso yath, just < >
yath ah, <>
yeah
thanks :)
yath just found Perl6::Perl5::Differences 17:54
17:56 riffraff left, Infinoid left
yath is there any difference between «» and »«? 17:57
as meta-operator
jnthn Yes.
dwimminess.
yath both DWIM :P
jnthn You point the sharp end at a side to have it auto-upgrade.
That is, if it's too short.
yath yes, in the case of «« or »»
but if i do «+» or »+«?
jnthn <<+>> means "upgrade the shorter side, it doesn't matter which is shorter" 17:58
>>+<< means "explode if they're different lenghts"
yath ah
thanks
jnthn rakudo: say (1,2,3) >>+<< 1;
p6eval rakudo 0d4fe0: OUTPUT«Non-dwimmy hyperoperator cannot be used on arrays of different sizes or dimensions.␤in Main (/tmp/dhw1sHVOKM:2)␤»
jnthn rakudo: say (1,2,3) >>+>> 1;
p6eval rakudo 0d4fe0: OUTPUT«234␤»
yath rakudo: say (1,2,3) <<+>> 1; 17:59
p6eval rakudo 0d4fe0: OUTPUT«234␤»
jnthn (just checking I got it the right way around :-))
yath hm
jnthn Right. But if you have @a and @b and you don't know which is shorter, then <<+>> becomes more useful. :-)
yath cool, even >>+<<< works :) 18:00
jnthn phenny: tell masak you wrote github.com/masak/csv in your blog but it gives me 404? 18:03
phenny jnthn: I'll pass that on when masak is around.
yath isn't [] used for array refs anymore? <foo bar>.perl gives ["foo", "bar"], althought it's an array. or a list. or whatever.
pmichaud that's a piece that is needing some upgrading
we're waiting for some spec revisions to land 18:04
18:04 eMaX_ left
pmichaud older versions of the spec indicated that lists would .perl as arrays. That's no longer true, but the new spec is incomplete on the subject. 18:04
yath ah, so the "traditional" [] and {} still stand?
pmichaud ['foo', 'bar'] is an array
('foo', 'bar') is a list
<foo bar> is also a list 18:05
yath but how would i do an array ref then?
pmichaud ['foo', 'bar'] acts like an array ref
but references are somewhat "hidden"
i.e., my $a = ['foo', 'bar']; say $a.WHAT # Array
rakudo: my $a = ['foo', 'bar']; say $a.WHAT;
p6eval rakudo 0d4fe0: OUTPUT«Array()␤» 18:06
pmichaud rakudo: my $a = ['foo', 'bar']; say $a[1];
p6eval rakudo 0d4fe0: OUTPUT«bar␤»
pmichaud note that it's _not' $a->[1]
yath hmmmmm
jnthn In general, you just don't have to think about references in Perl 6. 18:07
yath just wonderes because @a = <a b c>; @b = <d e f>; both look like array refs, but @c=(@a,@b) works as i'd expect it from perl5
s/wonderes/wondered/
pmichaud neither of those are array refs
jnthn rakudo: my $foo = [1, { x => 42 }, 3]; say $foo[1]<x>;
p6eval rakudo 0d4fe0: OUTPUT«42␤»
pmichaud @a = <a b c> is the same as @a = ('a', 'b', 'c')
lambdabot Maybe you meant: activity activity-full admin all-dicts arr ask . ? @ v
KyleHa pmichaud: Good to see you! 18:08
pmichaud KyleHa: glad to be "back"
18:09 barney joined
zloyrusskiy i wanna ask lame question: why in interactive mode i can't declare variables, i'm getting error: Symbol '@a' not predeclared in <anonymous> 18:09
in Main (src\gen_setting.pm:3343)
KyleHa Any objection to me moving the PARROT_REVISION up to today's release version? 18:10
jnthn KyleHa: I already have a patch here to do it...
KyleHa I ran the spectest with it, and it has the same failures I've come to expect.
jnthn KyleHa: Ah, I was just waiting for my run to finish.
KyleHa jnthn: In that case, I'll leave it.
jnthn zloyrusskiy: At the moment, the interactive mode treats each line as its own lexical scope. 18:11
zloyrusskiy: Which, yes, sucks.
zloyrusskiy =)
jnthn zloyrusskiy: my $a = 42; say $a; # all on one line is fine
zloyrusskiy jnthn: ok, thanks
jnthn: yeah, works 18:12
jnthn :-)
We have plans to improve it...hopefully in the not too distant future that'll get done. :-)
18:13 jferrero joined
pmichaud jnthn: I get one failure in thorough.t after bumping PARROT_REVISION 18:16
18:16 zamolxes left
jnthn pmichaud: That's my fault, don't worry about it. 18:17
pmichaud other than that, bumping PARROT_REVISION looks okay to me
pugs_svn r28021 | jnthn++ | [t/spec] Fudge a test that is failing in Rakudo. 18:18
jnthn That's thorough.t taken care of.
OK, pushed the bump. 18:19
Who's release manager this month for Rakudo?
pmichaud KyleHa++
KyleHa *bow* 18:20
I'm looking for a namer.
smash_ pmichaud: did you add me to the pool ?
pmichaud smash_: Hadn't heard from you yet for August, and KyleHa++ came up so he got August. Next available slot is December or January, if you want them
smash_ pmichaud: i can take december 18:21
dalek kudo: 6c7b8b3 | jnthn++ | build/PARROT_REVISION:
Bump us up the the current Parrot release, in preparation for the next Rakudo release.
 18:22
pmichaud smash_: okay, I'll give you december and move chromatic to january (he and I had already discussed this possibility)
note that PARROT_REVISION is now frozen until the Rakudo release on Thu
smash_ pmichaud: i can do january too, it's the same for me
pmichaud smash_: I'm trying to make sure we don't have two months in a row with the same release manager 18:23
oh, you mean instead of december. Either one, choice is yours :)
smash_ pmichaud: either way is fine with me, just schedule it as you see fir
pmichaud smash_: okay, will do, and thanks!
smash_ (s/fir/fit/) 18:24
KyleHa I'm shooting to have the release all but done in about 24 + 10 hours.
18:24 molaf left
KyleHa ...because my free time may be more restricted after that. 18:24
pmichaud KyleHa: that sounds fine
jnthn pmichaud: I'm rakudo-daying tomorrow, will mostly do bug triage. 18:25
pmichaud jnthn: works for me. I may do the same -- depends on how other events play out. I might also review/update the "make install" target a bit more tomorrow.
Of course, all bets are off if I don't have $otherjob server woes taken care of first.
jnthn nod 18:27
18:27 payload left
pmichaud (but I think that's likely to be taken care of by then) 18:27
18:27 payload joined 18:33 payload left 18:35 xomas_ left 18:42 zamolxes joined 18:50 icwiener joined 19:00 abra left 19:10 masak joined
masak TEH PARROTZ! 19:10
phenny masak: 18:03Z <jnthn> tell masak you wrote github.com/masak/csv in your blog but it gives me 404?
masak hm. 19:11
PerlJam s/csv/cvs/
you typo'ed
masak yes... in creating the repo :/ 19:12
19:12 payload joined
masak anyway, I've renamed it now. github says it'll take a couple minutes to go through. 19:12
jnthn: thanks.
PerlJam masak++ 19:13
masak apparently, getting things done/right is an iterative process. :)
19:23 cono joined
masak rakudo: role Self[Self] {} 19:27
p6eval rakudo 0d4fe0: OUTPUT«set_pmc_keyed() not implemented in class 'Integer'␤in Main (src/gen_setting.pm:1538)␤»
masak expected behaviour? discuss.
std: role Self[Self] {} 19:28
p6eval std 28021: OUTPUT«===SORRY!===␤Invalid typename Self at /tmp/1OlrJQaA7X line 1:␤------> role Self[Self⏏] {}␤FAILED 00:03 41m␤»
PerlJam it's hard to tell if they're making any progress on #parrot.
jnthn PerlJam: We are, don't worry.
masak progress with what?
jnthn Trying to prevent a Parrot clusterfuck that would cause Rakudo development problems. 19:29
TimToady translation: we now have jnthn and pmichaud worrying about it, so you don't have to :)
19:29 hah left
PerlJam jnthn: but do you, pmichuad, TimToady, allison and chromatic all agree? :) 19:29
masak oh, a clusterfuck, no less?
pmichaud PerlJam: I think that I, jnthn, and TimToady agree.
TimToady I don't think chromatic agrees yet, but he's in snark mode at the moment 19:30
he'll come around eventually
masak backlogs
masak submits rakudobug
jnthn chromatic's points are whines about language design, which ain't really applicable to the discussion right now.
pmichaud this question came up 10 months ago :-) irclog.perlgeek.de/parrot/2008-10-22#i_638371
jnthn
.oO( seems I'm in snark mode too ;-) )
pmichaud: Depressing. 19:31
pmichaud: Anyway, from what allison has said, there's only one real issue now.
Basically, yes, the CallSignature will have the invocant as the first positional.
TimToady I think it was shortly after that that we melted ruoso's brane with the same issue and made him rewrite smop :)
pmichaud TimToady: actually, that came about 4 months later, iirc 19:32
I think we melted ruoso's brane in the february time frame
TimToady that's shortly, in Perl years :)
jnthn Parrot will record something that lets us know whether the invocation has a a.$P0(b, c)
Or a $P0(a, b, c)
pmichaud jnthn: that's not how I read it.
jnthn And allison proposes that a :method will, by default, care about that.
PerlJam so much for agreement :)
19:32 jantore joined
pmichaud I read that $P0(a, b, c) throws an exception. 19:32
jnthn Yes, but separate out the issues. 19:33
That's _not_ because a is not the first positional in CallSignature.
pmichaud PerlJam: oh, looks like we agree. :)
jnthn It's because the two CallSignature's carry different extra metadata.
ruoso TimToady, pmichaud, you bet... thankfully my brain got back in place already
pmichaud jnthn: right, it's because a sub flagged as :method will carp if the first argument isn't flagged as invocant
ruoso at least I thikn it did
jnthn pmichaud: Right, which means basically the issue is making sure we have a way to say to Parrot, please don't carp about it.
TimToady well, parrotfolk are going through the branemelt now :)
jnthn pmichaud: fwiw, I think very likely is that we are going to take the CallSignature and unpack and bind it ourselves. 19:34
pmichaud jnthn: yes, I had come to that conclusion as well.
jnthn: which means I don't really care much about :lookahead 19:35
jnthn But I'm reluctant to suggest that the only way to surpress it is " .param pmc args :call_sig"
pmichaud: Right.
Me either.
pmichaud (and no, :lookahead was not my choice of name, despite what allison++ has said :)
jnthn Now that I've kinda confirmed to myself that we need to go on the self-unpacking route.
(I was quite open to both at one point. "inter_call.c refactors come later" just changed that...)
pmichaud the hard part is going to be getting PGE to conform to both the Rakudo and P6 worldviews.
jnthn ...Rakudo and P6? 19:36
pmichaud sorry, Parrot and P6 worldviews
jnthn You mean Rakudo and Parrot?
pmichaud (typo)
19:36 jhorwitz left
jnthn Right, well that's why we need to try and push for a way other than just saying "I'll unpack myself" to surpress the warning. 19:36
erm,
the exception
not warning, exception
Knowing our luck, it'll probably end up being called :submethod ;-) 19:37
TimToady just flip the resumable bit on it and resume it. :)
then feed parrot to the optimizer, and it notices that the exception is always resumed, and deletes it entirely :) 19:38
(I wish)
19:41 tann joined
masak rakudo: role Self[Foo] {} 19:42
p6eval rakudo 0d4fe0: OUTPUT«set_pmc_keyed() not implemented in class 'Integer'␤in Main (src/gen_setting.pm:1538)␤»
jnthn where the heck does the integer come from?
masak don't ask me. :) 19:44
jnthn Yeah. I should ask the fool that implemented the thing.
masak there's now an RT ticket reminding you to find out, however. :P
19:44 zloyrusskiy left
masak jnthn: I'm pretty sure 'fool' isn't very collocational with 'type system implementor'... 19:45
besides, I'm only investigating corner case failure modes, as usual. 19:46
TimToady there's a joke to be made there about the former checkoslovakia, but I'm still trying to figure out what it is
czech, bah 19:47
can't spell this week
masak looking forward to that joke, anyway. :)
jnthn will czech back later for it
19:50 barney left
masak *rimshot* 19:50
I'm glad you seem to be reaching consensus on #parrot. 19:52
19:53 payload left 19:54 xomas_ joined 19:57 hercynium joined
jnthn ...well, I almost thought we had... :-) 19:57
masak :/ 20:01
sjohnson i have a question for everyone: I want to become able to use Perl very well, and I was thinking of mastering the Camel book. But should I instead just master the Perl 6 specifications? 20:04
masak sjohnson: depends on what your short-term goals are, I'd say. 20:05
20:06 kane__ joined
masak Perl 5 is one nice really language. and it has lots of good libraries. 20:06
sjohnson i want to be able to weild perl like some of the more creative guys around here
20:06 nErVe joined
TimToady buying a Camel book puts a few cents into my bank account, while the P6 specs are a net drain... :) 20:06
takadonet hehe
masak :D
sjohnson my boss paid for the Camel Book :)
the only thing about it can't figure out for the life of it, is the difference between "my $goose" and "our $goose" 20:07
takadonet so does mine :) Any book I need, we get for free
20:07 nErVe left
PerlJam sjohnson: do you understand lexicals? 20:07
sjohnson: do you understand package vars?
sjohnson i understand the "my" scope and the exporter things from packages
PerlJam our $foo; makes the name $foo lexically available and it maps to $SomePackage::foo 20:09
20:09 evilmight joined, evilmight left
sjohnson must the $foo be declared in SomePackage/foo.pm? 20:10
Tene sjohnson: no, it's just a variable...
sjohnson back to the drawing board for me -_- 20:13
masak sjohnson: my suggestion to you is to start writing code. preferably lots of it. 20:14
discussing theory is good, but it only gets you so far.
rakudo: say <-> $n, $l = 3 { join ",", reverse gather while $n { my $s = sprintf "%03d", $n % 10**$l; $n = truncate($n / 10**$l); take $n ?? $s !! $s.subst(/^0+/, "") } }.(65536) 20:19
p6eval rakudo 0d4fe0: OUTPUT«65,536␤»
20:19 payload joined
jnthn glances at #parrot again and bangs his head against the wall 20:20
OK, it's high time for me to get dinner...be back in a bit.
masak this day has had a record amount of people banging their heads against hard, flat surfaces.
KyleHa What are the Parrot folks trying to solve? 20:22
masak also, who would have thought in 2000 that nine years later, a group of Perl 6 developers would have a hard time reaching consensus with a group of Parrot developers? from the point of view of nine years ago, the thought staggers. 20:23
takadonet cya all later 20:24
masak takadonet: o/
takadonet hopefully tomorrow I could show you guys a fasta parser!
PerlJam masak: 9 years ago it would have been "What's Parrot?" and "What's Rakudo?" :)
masak PerlJam: oh! which year was the Parrot April's Fools joke? 20:25
20:25 takadonet left
PerlJam might have been 2000, but I think it was 2001 or 2003 20:25
er, 2002
masak it was no later than 2001. oreilly.com/news/parrotstory_0401.html
20:25 Whiteknight left
masak aye, 2001, it seems. 20:26
by the way, it would still be kinda funny to make such a hybrid language.
20:30 tak11 joined, perletc joined
perletc im here, coolness 20:30
20:30 payload left
perletc perl6 looks awesome 20:30
20:30 payload1 joined
masak perletc: welcome! 20:30
rakudo: say 'perletc: so, have you tried Rakudo yet?'
p6eval rakudo 0d4fe0: OUTPUT«perletc: so, have you tried Rakudo yet?␤» 20:31
20:31 icwiener left
sjohnson wow, perletc is a cyborg 20:32
perl6 truly can accomplish anything
masak sjohnson: what makes you think perletc is a cyborg? he looks perfectly human to me.
sjohnson oh 20:33
i didn't see your "rakudo:" part
i just saw the perletc: part
thought you were executing code with him
masak sjohnson: :) 20:34
perletc: even the residents here can't tell the difference between droids and humans. that's how advanced Perl 6 is. :)
TimToady I should turn TimToady into a bot that also lets me type through...
sjohnson timtoady: print Dumper Advice.life; 20:35
masak TimToady: there's prior art. buu has a buubot...
TimToady Out of Memory
mofino haha 20:36
sjohnson BotToady
masak rakudo: my ($a is readonly) = 5; say $a; $a = 7; say $a 20:38
p6eval rakudo 0d4fe0: OUTPUT«5␤7␤» 20:39
PerlJam sjohnson: and together they are ... the Toadys :) 20:40
cono TimBot - There is more beautiful onions there :D
perletc sorry guys, im back
no havent tried Rakudo yet, im scared 20:41
masak no need to be scared.
just type 'rakudo: ' and then your first program.
rakudo: say 2 + 2
p6eval rakudo 0d4fe0: OUTPUT«4␤»
perletc why does the output look like that? 20:42
20:42 ruoso left
perletc 􏿽xAB4␤»? 20:42
masak perletc: those are french quotes on my screen.
perletc maybe its my client, im using hydrairc
cono rakudo: .say for 1..3
masak probably an encoding issue, yes. 20:43
p6eval rakudo 0d4fe0: OUTPUT«1␤2␤3␤»
20:43 RonOreck joined
mkelly32 so. sub MAIN basically works like: run script like perl5 would. then, if there is a &main::MAIN, call main::MAIN(|@*ARGV)? or, is it fancier than that? 20:43
masak rakudo: say ("OH HAI", "try me,") X~X " perletc"
p6eval rakudo 0d4fe0: OUTPUT«Could not find non-existent sub X␤»
masak oops. 20:44
rakudo: say ("OH HAI", "try me,") X~ " perletc"
p6eval rakudo 0d4fe0: OUTPUT«OH HAI perletctry me, perletc␤»
perletc what so great about perl6 (that a cool title for a blog article)
TimToady mkelly32: it also translate normal switch syntax to named args
masak rakudo: .say for ("OH HAI", "try me,") X~ " perletc"
p6eval rakudo 0d4fe0: OUTPUT«OH HAI perletc␤try me, perletc␤»
perletc does it have dot syntax now?
masak perletc: yep.
20:44 dbrock joined
masak perletc: and it's better than ever. 20:44
perletc niceness 20:45
masak perletc: look at this, for example:
rakudo: for "niceness" { .=flip; .say }
p6eval rakudo 0d4fe0: OUTPUT«ssenecin␤»
mkelly32 hrm, neat. so, is it able to be a complete replacement for Getopt::Long? or are there still a few things that Getopt does that a sub signature can't?
masak mkelly32: I think it'll be a pretty complete replacement. 20:46
perletc rakudo: package foo; sub new { return bless {}, class; } sub moo { return 'blah'; } 1; say foo.new.moo;
p6eval rakudo 0d4fe0: OUTPUT«Unable to parse class definition at line 2, near "; } sub mo"␤in Main (src/gen_setting.pm:1525)␤»
perletc im pretty sure i F'd that up 20:47
masak perletc: whoa! one step at a time. :)
perletc :)
masak perletc: you need semicolons after blocks now, if you continue writing code on the same line.
perletc ah
masak also, you generally don't need to declare your 'new' methods.
watch:
perletc niceness 20:48
masak rakudo: class A { method moo { return 'blah' } }; say A.new.moo
p6eval rakudo 0d4fe0: OUTPUT«blah␤»
sjohnson moo is a cute var name
perletc auto-instansiation
cewl
masak this also works: 20:49
perletc wait wait, dont tell me it supports prototyping too
masak rakudo: class A { method moo { return 'blah' } }; my A $a = A.new; $a.moo
p6eval rakudo 0d4fe0: ( no output )
masak oops.
rakudo: class A { method moo { return 'blah' } }; my A $a = A.new; say $a.moo
p6eval rakudo 0d4fe0: OUTPUT«blah␤»
masak perletc: it's prototype-based, yes.
the above ('my A $a = A.new') can also be shortened to just 'my A $a .= new' 20:50
perletc fa sho
so, what about method moo (my $foo, $barr) { ... }
limiting the parameters? 20:51
masak there are parameter declarations, yes.
but you don't need a 'my' there.
perletc very cewl
very very cewl 20:52
masak rakudo: sub foo($a, $b, $c = 5) { .say for $a, $b, $c }; foo(1,2)
p6eval rakudo 0d4fe0: OUTPUT«1␤2␤5␤»
perletc damn Perl is back
ok how can I help
masak :)
perletc: ok, so here's our current problem:
we need people who write applications.
perletc me 20:53
masak you guessed it.
doesn't have to be ambitious stuff.
just try things out with Rakudo and see if they work like you think.
pmichaud we could also use updates to rakudo.org/how-to-help :)
20:53 dakkar left
pmichaud I can give out edit bits for those who need them 20:53
masak blog about it, tweet about it, tell your friends. re-visit us here at #perl6 and ask all manner of questions.
perletc k 20:54
mkelly32 are heredocs still NYI?
masak perletc: we simply love people who use Rakudo for new, interesting things!
pmichaud still NYI, sadly.
mkelly32: but qq{...} and q{...} are often reasonable substitutes
mkelly32 true.
masak mkelly32: heredocs are blocking on docking with STD.
reasonable substitutes, but still worthy of an expectant 'RAKUDO: ' comment. 20:55
mkelly32 though, aren't perl6 heredocs supposed to support being indented and such?
pmichaud yes.
TimToady STD already implements that
perletc can rakudo be installed parrallel to current installation, is it backwards compatible, is there a win32 distro yet?
mkelly32 ah, ok.
masak STD++
perletc: parallel: yes. back-compatible: no. win32: yes.
pmichaud if by "backwards compatible" you mean "does it execute Perl 5 code", the answer is "not yet"
mkelly32 is there a target timeframe for using STD? (which, iirc, is the perl6 grammer written in perl6?) 20:56
pmichaud mkelly32: December 2009
mkelly32: could be sooner than that
but that's the likely target
masak wow, it's like a Christmas gift! \o/
mkelly32 is the idea that all the different perl6 implementations would use STD?
perletc is rakudo the official name?
masak perletc: Rakudo is one of the most lively Perl 6 implementations.
pmichaud mkelly32: STD will be the reference grammar 20:57
masak perletc: the long name is, I believe "Rakudo Perl 6".
pmichaud mkelly32: I don't know how many of them will use STD exactly as-is
mkelly32 ok
20:57 alester left
TimToady commuting & 20:58
perletc lastly, probably dumb, if perl6 can't execute perl5 code, are the modules from perl5 standard in perl6 distro?
does, CGI, DBI, etc work 20:59
masak perletc: you put your finger on another burning issue.
not at all a dumb question.
for the most part, we get by without those.
some modules, we ported.
pmichaud there's also not a single "perl6 distro"
(*not going to be...)
perletc ah, ok.
masak others, we'd like to see ports of.
PerlJam perl6 is a multi-headed hydra ;)
perletc :) 21:00
masak perletc: there's a project called November, a wiki engine written in Perl 6. it uses its own very simple CGI.pm.
perletc: then there's the Web.pm project, which tries to do things better than CGI.pm.
we're still waiting for some heroic figure to grab DBI by the horns. 21:01
cdarroch on this subject, btw, is it anticipated that modules would mostly be written in perl6, or would some be written for parrot, say?
PerlJam masak: better? I think Web.pm and November are apples and oranges.
cdarroch with the goal of making them run in any parrot-driven HLL?
masak PerlJam: I meant better than the CGI.pm of Perl 5. sorry, I wasn't too clear.
PerlJam ah.
masak PerlJam: but yes, Web.pm will also replace the CGI.pm of November eventually. 21:02
I've promised that, and it still seems like a good idea. dogfooding and all.
21:03 payload1 left
jnthn back 21:03
perletc rakudo: method doit ($id, $user) { say for $id $user; }; doit(1, 'guy', 'failed');
p6eval rakudo 0d4fe0: OUTPUT«say requires an argument at line 2, near "for $id $u"␤in Main (src/gen_setting.pm:2510)␤» 21:04
mkelly32 i've noticed that november seems pretty darn slow. is that just a product of there not being much optimization work done yet for parrot/rakudo? (since i figure it's more important to have it complete than do premature optimizations)
PerlJam cdarroch: it doesn't really matter as long as we can turn Perl 6 into PIR (we can) and coerce a Perl 6 compiler to run PIR (I'm fairly sure we can there too :)
mkelly32 or, is it a potential for improvement in november itself?
masak perletc: you need '.say' and a comma between $id and $user.
cdarroch so you could create a p6 DBI module, compile it down to PIR, then use in ParTcl, say? 21:05
masak perletc: also, you should probably use 'sub' outside of class declarations for now.
PerlJam cdarroch: I don't know about you though, but I'm planning on writing in Perl 6 as much as possible :)
cdarroch (or at least, that's the vision ...)
masak cdarroch: oh, definitely.
perletc ok
rakudo: sub doit ($id, $user) { .say for $id, $user; }; doit(1, 'guy', 'failed');
p6eval rakudo 0d4fe0: OUTPUT«too many arguments passed (3) - 2 params expected␤in sub doit (/tmp/S36wawvDeg:1)␤called from Main (/tmp/S36wawvDeg:2)␤»
perletc cewl, very very nice 21:06
cdarroch ok ... this is how I came to be interested in the concurrency stuff, because I studied DBI's XS when I wrote net::zookeeper recently,
which uses a C library that kicks of several pthreads to handle heartbeat msgs to the server
21:06 M_o_C joined
cdarroch it seemed like a p6 version wouldn't need to C lib at all, say -- if it could kick off real threads in p6 21:07
PerlJam cdarroch: ah, perhaps you want to look at java2perl6.googlecode.com/svn/trunk
cdarroch down the road a way, obviously :-/
masak perletc: do you want to see something I drew with Perl 6? :) 21:08
perletc would be nice to have a truely unified self-sufficient perl 21:09
* requirements for such an acheivement r ovver my head
PerlJam cdarroch: DBDI is the database driver interface. Tim Bunce seems to think that we should just use the same interface that JDBC does, so we're (loosely speaking) writing code to translate the JDBC stuff to perl 6
masak well, there will be Perl 6 distributions that come pre-packaged with different useful modules, I'm sure.
21:15 payload joined
masak PerlJam: do you know if there'll be something that people can test-drive within the next few time units? 21:19
PerlJam masak: I do not know, but it feels like there could be something useful soonish. 21:21
masak soonish++
PerlJam It would be awesome if Rakudo* could have some sort of DBI-like thing, some sort of web framework, some sort of graphics library, and a few popular parsers (HTML, XML, CSV ;) 21:22
(I mean all useable at release) 21:23
We could even make a few of the first Perl 6 dists from there (simple, but useful)
masak PerlJam: yes, I agree fully. it would be awesome. we have 226 days to realize such a vision. 21:26
and to package it nicely.
PerlJam I hope that isn't 226 until Apr 1 ;) 21:27
masak it is.
sjohnson extend the deadline to xmas 21:28
PerlJam oh no! that's an unfortunate release date. Better go with March 20 (first day of Spring)
masak PerlJam: I'm not saying Rakudo Star'll be released on April 1.
PerlJam: in fact, April is only an estimate right now.
jnthn Plus Rakudo releases are on the third Thursday of the month. ;-) 21:29
Which had better not be the 1st. ;-)
PerlJam jnthn++
masak PerlJam: but I need *some* hard figure, so I'm picking days-left-until-April.
jnthn Days left until April is a nice enough measure.
It gives an order of magnitude.
226 feels like a nice big number. Lots of time to get stuff done. ;-)
PerlJam masak++
masak (it also turns out to be a pretty nice use case for improving the Temporal spec)
jnthn oh my... 21:31
How in git can I list my pending commits?
that is, stuff I comitted but did not yet push?
masak 'git log master..origin/master', perhaps? 21:32
uh, no. the other way around. :/
'git log origin/master..master'
should work.
jnthn Thanks, that shows we the commit...
jnthn wonders if he can get a diff
masak sure, just replace 'log' with 'diff' :)
jnthn wonders if he can undo having comitted it. <embarrased look> 21:33
masak sure, is it just one commit?
jnthn yeah
PerlJam as long as you haven't pushed, it's relatively easy
masak 'git reset --hard HEAD^' in that case.
jnthn ok, thanks.
:-)
Yeah, I haven't done anything insanely stupid. Phew. :-)
masak git++
jnthn git++ is great when you know how to use it.
masak ♥ git 21:34
21:35 kane__ left
sjohnson heh 21:35
that's cute
jnthn gah, today hasn't had enough hours in it.
bbiab
masak my problem with days nowadays is that they have too many hours at the end and too few at the beginning. 21:36
21:36 Whiteknight joined
dalek kudo: b9c79c2 | jnthn++ | (3 files):
Stub in AttributeDeclarand and ContainerDeclarand, as a start to getting something in place for trait application to container declarations. Based on discussion on #perl6, and not spec yet.
 21:38
masak jnthn: also, if there was just some little thing about the commit you wanted to fix, there's always 'git commit --amend', an awesome flag. 21:39
21:55 M_o_C left 22:02 ruoso joined 22:04 kst` is now known as kst
jnthn masak: Aye, --amend is one trick I did learn. ;-) 22:06
masak jnthn: what about 'git rebase'? :D 22:08
jnthn masak: I know to try git pull --rebase ;-) 22:09
Which as I understands it removes my patches, applies the upstram ones, then re-applies mine on top of those.
22:09 cdarroch left
jnthn Or something. 22:09
masak 对. 22:10
KyleHa The spectest just said to me, "OH NOES OUT OF CHEEZBURGER".
masak :)
KyleHa I reckon my autounfudge was a wee bit too aggressive.
jnthn ETOOLITTLECHINESESKILLZ 22:11
masak jnthn: but 'git rebase' can do that for you when you've finishing a branch you started four days ago, and you want to remove the commits in your branch, catch up with master, and then re-apply the commits.
jnthn masak: OK, that sounds kinda good.
masak: So it's a good thing to do before merging a branch?
masak a very good thing.
TimToady @seen KyleHa
lambdabot KyleHa is in #perl6. I last heard KyleHa speak 1m 5s ago. 22:12
masak it makes merges much easier.
TimToady gah, I'm blind
KyleHa: in this: dies_ok { RT66886::c }, 'accessing non-value of enum dies proper-like';
that's a compile-time error, not run-time
jnthn The test is geordie like!
TimToady so would need eval_dies_okay
KyleHa TimToady: Oh, good catch. Thanks. I'll fix that. 22:13
masak jnthn: (in Mandarin, 对 is pronounced DUI4. it's a preposition ('opposite', 'towards'), but alone in a phrase it means 'correct'.)
jnthn masak: Ah, OK.
TimToady hmm "opposite" as in "facing", I imagine
jnthn Learning Mandarin would be really cool. 22:14
TimToady since facing can mean both of those things, more or less
jnthn I'd like to learn one(Mandarin, Korean, Japanese)...all have their attractions, but there's no way I can manage more than one of them and get anywhere beyond very basics...
TimToady how one gets from "facing" to "correct" is anyone's guess 22:15
masak TimToady: yes, as in 'facing'.
TimToady maybe "correct" as in "facing the right way"
pugs_svn r28022 | kyle++ | fix RT #66886 test, thanks to TimToady++ 22:16
masak possibly...
TimToady ah, more like "corresponds to" 22:17
as in "that answer corresponds to the question"
masak ah, of course.
I should have known that. :)
TimToady tit for tat, this answers that
masak also, this is a good time to say "对了" :)
meaning 'yes, now you got it right' :) 22:18
TimToady I first read that as "oppositional child", but it's not child
masak no, it's a statement-ending particle.
jnthn lol 22:19
Gah, I'm going to do the traits on containers refactor...another day.
masak should sleep 22:20
jnthn Actually it's probably best left until after the release anyway. 22:21
night masak 22:23
TimToady hmm, maybe there should be a WRONG statement prefix that tells the compiler not to complain about something it knows will blow up at run time 22:25
dies_okay WRONG { 1/0 }, ... 22:26
masak jnthn: 'night
KyleHa The test suite seems to have had a much greater focus on death lately.
masak TimToady: sounds like something better suited for a module. 22:27
22:27 masak left
TimToady well, it has to instruct the compiler at compile time somehow; run time is too late 22:28
and there are lots of spots that assume 1/0 won't complain till run time
(I suspect)
oh good, looks like they're in eval_dies_ok 22:29
22:30 frew joined
jnthn The compiler can only blow up at compile time on that one, if it knows nobody has lexically defined their own infix:</> ;-) 22:30
TimToady which is purportedly does
22:30 japhb left
KyleHa I was going to say maybe '1' or '0' got redefined. 22:30
TimToady *it
jnthn TimToady: Sure, if it's bothering to keep track of such things. ;-)
TimToady fact is, the constant folder should blow up
pugs_svn r28023 | lwall++ | [STD] parse #=[...] comments 22:33
22:34 nihiliad left 22:35 KyleHa left 22:36 kane__ joined 22:37 jferrero left 22:50 frederico left 22:52 frederico joined 22:56 kane__ left
ruoso @tell pmurias have you manage to include the smop tests in the p5 build as well? 22:59
lambdabot Consider it noted.
23:05 ihrd joined 23:06 Limbic_Region joined
Limbic_Region jnthn ping 23:07
23:10 drbean left 23:11 frew__ joined 23:16 aindilis joined
TimToady wayland76: why did you change all your :$foo parameters to $:foo? that's just wrong... 23:20
std: sub foo ($:bar) {...} 23:21
p6eval std 28023: OUTPUT«Potential difficulties:␤ Illegal to use : twigil in signature at /tmp/ra5xxhLGIi line 1:␤------> sub foo (⏏$:bar) {...}␤ok 00:02 38m␤»
jnthn Limbic_Region: pong 23:28
23:30 bpetering joined
sjohnson is that buffer datatype technology? 23:31
bpetering gooood morning #perl6! :D 23:32
sjohnson hi 23:33
23:38 xomas_ left 23:44 perletc left
sjohnson how does one cope with the unhappiness of having to program 80% PHP, 15% Javascript, and 5% Perl everyday 23:46
for their job
bpetering sjohnson: i was about to ask "how goes it", but now i know.
sjohnson heheh
anyone's advice would be greatly appreciated 23:47
bpetering sjohnson: well... 23:48
i am in a similar situation. as to how i "deal with it"...
First off, I'm grateful to have a job :)
Second, use libraries when possible - jQuery can make life a whole lot easier :) 23:49
Third, PHP may not have the Tao, but it is Turing complete, so if you learn how to abstract well, you can make your own life a lot easier 23:50
Fourth: Test. 23:51
Early, and often.
I wish I'd been exposed to Perl's testing culture before I started my current job :( 23:52
Does that help? 23:53
Attitude goes a long way... it took me a while to learn that. 23:55
... and five: sneak in Perl wherever you can ;) 23:59
23:59 payload left

Powered by Cro and the Raku® Programming Language.

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

Thank you!