6.2.11 released! | pugs.blogs.com | pugscode.org | pugs.kwiki.org | paste: sial.org/pbot/perl6 | < audreyt> my fingers juxpatosed
Set by nothingmuch on 24 February 2006.
00:32 binary42 joined 00:45 nothingmuch joined 01:06 jiing_ joined 01:12 binary42 joined 01:20 binary42 joined 01:36 FurnaceBoy_ is now known as FurnaceBoy 01:42 whiteg_fi joined, whiteg_fi left 02:18 nothingmuch joined 02:35 DesreveR joined 03:15 scook0 joined 03:22 FurnaceBoy left 03:32 scw joined 03:41 tmlsd joined 03:49 Qiang joined 03:59 Qiang joined 04:03 tmlsd left 04:06 Limbic_Region joined 04:13 javierza joined 04:16 ayrnieu joined
ayrnieu ?eval loop (my $i = 0; $i < 10; $i++) { say $i } 04:17
... 04:18
also, { chdir $noexist or ... } , { mkdir $noperm or ... } just die in haskell-space. 04:22
I guess I need try {} ... 04:23
ah, I guess that 'use fatal' is default. (although I can't 'no fatal', here) 04:24
splice does not behave like perl5's splice, contrary to documentation. 04:29
04:29 Eroick joined
Eroick How is P6 going? 04:29
Its been a while since I was last around these parts. 04:30
ayrnieu contrast:
pugs -e 'my @a = <5 6 7>; say for splice @a, 0, 1000'
perl -le 'my @a = qw/5 6 7/; print for splice @a, 0, 1000'
eroick - well, I suppose. If you've any specific questions, someone more knowledgeable might come along to answer them. 04:31
Eroick ayrnieu: :)... good point. So, last time I checked, pugs had just come out and everyone was amazed. Is pugs still comming along nicely? What about parrot? 04:32
04:33 Hamilton joined
ayrnieu people seem pleased with pugs's progress. I recently saw allison express confidence in parrot's imminent maturity. 04:33
Eroick is there any more information on when we will see perl6's own compiler come out 04:36
ayrnieu the last I heard someone speculate, which may have been a few weeks ago: this year.
my impression is that pugs will compile perl6 to something that runs on parrot, and then self-compilation will happen, and then everyone will be happy and excited :-) 04:37
Eroick i hope so :). I saw something quickly that was new. I believe it was caled ponie and it was a bridge between p5 and 6. 04:38
am I right?
ayrnieu I'm not familiar with it. Are you?
Eroick no 04:39
ayrnieu are you right, I mean.
randomly, www.perlmonks.org/?node_id=535873
04:39 Hamilton joined
Eroick www.poniecode.org/, hey, I was right :) 04:39
04:39 Hamilton left
ayrnieu the perl6 there is (not very) forward looking, in that it assumes a !broken loop and splice, and incidentally forward looking in that it assumes that perl6 is named 'perl'. 04:41
Eroick Im liking this 'everything is an object' idea Ć  la ruby. It is one thing that python is really missing.
ayrnieu I see python as missing other things, but OK.
Eroick ayrnieu: yes, so do I, but that is one of them. One reason that I prefer ruby over python. 04:42
ayrnieu perl6 still has contexts, though, which makes direct ruby comparisons a bit interesting.
but my favorite ruby example, one which I came up all by myself years ago :-) -- class Array; def rand; self[super length]; end end , translates to perl6 as: sub List::rand { @_[rand +@_] } 04:43
Eroick :) 04:44
ayrnieu (which, alas, I did not come up with all by myself years ago.)
Eroick i hope p6 brings more life back to perl. people have been drifting off to other languages.
ayrnieu *shrug*, perl5 does not seem undead to me. 04:45
Eroick its not undead, but it is less lively than it used to be
ayrnieu and of course people have been drifting off to other languages; it wasn't long ago that there were /no/ comparable languages, and even more recently thas those languages had relatively little going for them. The whole world is more lively, now. 04:46
Eroick yep.
but, perl has cpan. the very very large swiss army knife which can do /almost/ anything and is constantly growing :). 04:47
ayrnieu CPAN, yes, and all of the (sometimes human) machinery that CPAN assumes.
Eroick will parrot run on mac/*nixs/w32? 04:50
ayrnieu I'm not familiar with Win32, and 'mac' is now 'unix' for many purposes.
jisom parrot is cross platform
ayrnieu will it? Absolutely. Does it? I haven't cared enough about win32.
jisom it runs on darwin, linux, freebsd, openbsd 04:51
smoke.parrotcode.org/smoke
those are just reports in the past month
even solaris! 04:52
Eroick school computers run windows. 'nuff said :)
Supaplex slowaris!
jisom but, no dos it seems, c'est la vie
Eroick wow, its a huge high res monitor! too bad it can only display black and white!
ayrnieu eroick - hm?
Eroick the old solaris needed very highres monitors to display a console 04:53
ayrnieu eroick - how hold?
Eroick before X.
ayrnieu how old is that?
Eroick hm
04:53 hamilton_ joined
jisom isn't x from the 70's or something? 04:53
ayrnieu well, that's before Ultra2, which is the only generation I've recent experience of.
jisom i.e. really really really old
04:54 hamilton_ left
Eroick yes. very old 04:54
jisom in computer terms anyway
1950 is ancient
ayrnieu complaining about unbelievably old Solaris (why even call it 'solaris'?) systems in the context of a casual reference to Solaris is a bit odd.
anyway, I'm more impressed by the HP-UX build -- an OS which its proprietor no longer supports (who tells people to move to linux. See sun.com 'no! come over here!' propaganda ^_^) 04:56
Eroick yes, it is. I just remeber that I knew someone who held on to one and we were playing on it. One insane monitor which could only be used for console, black and white.
jisom parrot aims to support any platform with ansi c....
Supaplex I sold off my hpux box last week
*sniff*
Eroick i need to install linux on a computer. im thinking of using ubuntu, any suggestions? 04:57
jisom although MacOS 9 and below I don't think are used, ever....
Eroick, www.freebsd.org :)
Eroick jisom: i never used BSD. want to convince me? :P 04:58
ayrnieu eroick - do so?
Eroick ayrnieu: you like ubuntu?
jisom decades of development and stability?
ayrnieu eroick - I don't dislike it, but I also don't use it. I use OpenBSD and Solaris. 04:59
arcady ubuntu has a pretty good installer though
jisom install cd's include gnome and kde, there's a linux compatability layer
you know freebsd will be around in ten years
Eroick jisom: why?
jisom: and what is the big differance?
arcady I don't like freebsd for their refusal to support powerpc 05:00
jisom ports collection, standard, stable
then netbsd? :-p
I hear it runs on a toaster, although I don't want a toaster with a computer chip 05:01
arcady netbsd runs on everything...
badly
ayrnieu this in an OpenBSD/macppc system, on a mac mini.
Eroick OpenBSD vs FreeBSD... is there a big differance?
ayrnieu arcady - no, it runs well enough on sparc64.
eroick - yes.
jisom openbsd's focus is security above all else
Eroick and is it a compile everything type thing, or packaged? 05:02
ayrnieu both offer both. 05:03
jisom plus you either have to buy an openbsd or create your own iso....it's not too difficult(it's how I installed it)..but they have to make money somehow
ayrnieu jisom - that is not accurate.
jisom openbsd doesn't offer iso images
ayrnieu jisom - again, you are mistaken.
jisom last I checked, probably about a month ago, there weren't iso images 05:04
3.8
ayrnieu jisom - you checked mistakenly.
arcady wow, OpenBSD is actually distributed via AFS...
Eroick hmm. i still am not very sure why BSD varies from Linux (other than the kernel) any recomended reading? 05:05
ayrnieu OpenBSD.org does not distribute ISOs which incorporate pre-build packages and the system tarballs. It does, however, distribute perfectly normal ISOs for installation by a number of methods. The simplest way to get both packages and system tarballs and installation ISO on CD is to burn two CDs, although pages exist which explain the more involved process.
arcady Eroick the userspace is pretty different
jisom "CD-ROM: Of course, we prefer you use the Official CD-ROM set, but for special needs, you can also make your own." - openbsd.org/faq/faq4.html
arcady it uses BSD stuff, not GNU stuff 05:06
ayrnieu jisom - please stop; you're mistaken. You don't have repeat your mistaken argument. You could have restated it in a way which was not mistaken.
I generally net-install with the given cd*.iso and use the ports system, which (re: ports) I'd do anyway. 05:07
Eroick hm, I have a sandbox computer around here somewhere.
jisom Eroick, I can tell you for sure it's easy to dual boot openbsd and freebsd.....I just installed openbsd on one partition and freebsd on another and it was instantly dual booting 05:08
surprised me 05:09
Eroick i think i will install it and try it out.
ayrnieu That is the best way. 05:10
jisom Eroick, since you're not familiar with bsd's, I don't recommend installing openbsd because it's installer isn't for beginners
at least not at first
Eroick jisom: well, I have installed debian(s), gentoo, and some GUI stuff 05:11
05:11 f0rth joined
rantanplan_ Hehe. But you can read the manpage of the install tools while installing. :). 05:11
ayrnieu amusingly, I think most highly of OpenBSD's installer, of all the Linux and BSD installers I've used.
jisom ayrnieu, it can be daunting, and when key lines of help text disappeared off my screen
it's an expert installer more than a beginner installer 05:12
ayrnieu *shrug*, when I was a beginner (only redhad and (many-floppies-of:) slackware under my belt), I accidently had a very good experience with OpenBSD :-) 05:13
05:13 coumbes joined
ayrnieu anyway, ubuntu is probably also fine. 05:14
Eroick heh, i have installed slack on a cd-drive-less computer
jisom eh, I had trouble partitioning the disk to allow dual boot, it didn't seem to want to let me format to split the hard drive and ignore the others.....but was able to partition in freebsd and then install openbsd
arcady I remember installing slackware before I had internet access
many floppies that someone else had to make for me
ayrnieu jisom - ah. I've never bothered with dual booting. 05:15
jisom I use it mainly just to build and test parrot...
Eroick does NDISwrapper work on FreeBSD? I need that
jisom what is it?
arcady probably not
it's a hack to let linux use windows drivers
ayrnieu eroick - why do you need it? 05:16
Eroick a linux module that lets you use wifi cards for windows only on linux
arcady OpenBSD does claim to support rtl8180 cards
jisom the card may be supported by freebsd perhaps
Eroick ayrnieu: i have a USB wifi card
ayrnieu eroick - so do I, but mine doesn't need a windows driver.
Eroick netgear wg111
ayrnieu It helps to look at 'supported hardware' lists before buying things :-)
arcady problem is, they don't tell you what's actually inside the card 05:17
jisom silicon? :-p
arcady you know, like FooBaz Brand 5530 Ethernet Card
could actually be 2 or 3 entirely unrelated products 05:18
with completely different chips inside
Eroick prism chip inside
ayrnieu in any case, it doesn't seem that anybody knows about NDISwrapper. #FreeBSD or a similar channel may help. I'd suggest supported hardware.
arcady I am about 90% sure that NDISwrapper is a linux only thing 05:20
jisom freebsd 6.0 does have an ndis driver... 05:22
www.FreeBSD.org/cgi/man.cgi?query=n....0-RELEASE
Supaplex ndis is hacktackular :) 05:23
05:40 K joined 05:59 jisom joined 06:03 binary42 joined 06:22 szabgab left
audreyt praises freebsd-ndis (otherwise she wouldn't be using internet now) 06:40
Kattana praises your inexorable effort 06:56
audreyt for setting NDIS up? 07:06
yeah, that took a while ;)
jisom have you backed up the work? 07:07
audreyt jisom: hm?
jisom for getting ndis working for you 07:08
audreyt er no, I just kept chasing kernel cvsups until I hit a level that has a good version of ndis -- and then save if_ndis.ko 07:09
and has been using it since
though recently if_iwi begins to work too
(but is 11b-only)
jisom whatever works 07:10
audreyt (and crashes when the network drop+reconnects)
(but otherwise works fine ;))
jisom eh, how's an isp that disconnects after nine hours?
ayrnieu hash context is hard. 07:11
audreyt jisom: very bad?
jisom extremely annoying
audreyt ayrnieu: container contexts are in general difficult to reason... 07:12
ayrnieu even 'my %a = %a.reverse' takes List::reverse
audreyt ?eval method Hash::reverse (Hash %x is rw) { ()} my %a = (:x); %a.reverse 07:14
ayrnieu evalbot died a little while ago, when I asked it to splice something. 07:15
audreyt how sad. resummoning.
ayrnieu ah, no, when I asked it to loop. ?eval loop (my $i = 0; $i < 10; $i++) { say $i } 07:16
07:16 kisu joined
audreyt fglock et al: I found the perfect solution to the p6-compiled-to-p5 file extension game! 07:17
perl6 modules should be .pm
starting with "use v6-pugs;"
the compiled perl5 code should live in the same filename
with a .pmc extension
with pure perl5 in it
and lo and behold, perl5 will load the .pmc before .pm.
("perldoc -f require" for the spec)
it'd be v6.pm's duty to write the .pmc file. 07:18
isn't that handy? :)
jisom confusing if you work with parrot and it's .pmc files
audreyt perl5's .pmc predates parrot, though.
jisom yeah
07:19 Limbic_Region joined
jisom I think it's mentioned once in one of the pods, and it surprised me when I first tried parrot that it used .pmc 07:19
07:19 Qiang-zh joined
Qiang-zh ping audreyt 07:20
audreyt Qiang-zh: pong
Qiang-zh hey, you are back to tw now ?
audreyt no, still in .at
Qiang-zh oh, you think you can passby beijing while you heading back for osdc tw? :) 07:21
audreyt well, I need to visit .jp first
what's the occasion in .cn?
Qiang-zh we are preparing a Perl conference in beijing 25th this month. 07:22
audreyt oh... YAPC::Asia is 29th
and I'd need to arrive by 27th I think
Qiang-zh hehe. yeah. i had the feeling that you have a tough schedule. 07:23
audreyt indeed... any details on the conference? url?
Qiang-zh no biggie. just thought i would ask anyway.
audreyt (it's just two weeks away)
Qiang-zh here. xrl.us/kdg9
i wasn't going to ask you knowingly that you have so many conf to go. but chunzi is asking about you so i am giving it a try :) 07:24
svnbot6 r9402 | audreyt++ | * move Rule.p6 to the real Rule.pm. 07:25
audreyt Qiang-zh: ok... then I probably couldn't make it. :/ perhaps if you asked a bit earlier..
Qiang-zh audreyt: yeah. no problem. my bad that i didn't ask you earlier. well. untill YAPC::China :-)
audreyt indeed :D 07:26
07:26 evalbot_9401 joined
arcady ?eval loop (my $i = 0; $i < 10; $i++) { say $i } 07:33
07:33 evalbot_9401 is now known as evalbot_9402
evalbot_9402 Error: No such method in class Any: "&my" 07:33
arcady well at least it doesn't crash
ayrnieu ?eval my @a = <a b c>; splice @a, 0, 1000 07:34
evalbot_9402 undef
07:35 xinming joined 07:36 Barry joined
Barry Anyone know of an open source programmer's editor written in Perl besides open Perl IDE? 07:38
arcady I recommend you write one
in perl 6
: )
Barry Ok. Give me a couple of hours.
arcady actually, does perl 6 have any GUI bindings yet? 07:39
Supaplex moans
Barry Why is *** moaning? 07:40
Supaplex wth is *** suppose to mean 07:41
Barry Probably something dirty, but you're apparently done.
audreyt Barry: EPIC allows pure-perl plugins 07:42
Barry I've looked at EPIC...don't remember anything. Better look again. 07:43
Supaplex instant perl editor: perl -e 'exec($ENV{EDITOR}||"vim",@ARGV)' 07:44
arcady or pugs -e 'exec(%ENV<EDITOR>||"vim",@*ARGV)' for that matter 07:48
07:50 ghenry joined
Barry $695 for Epic...maybe I was thinking of Eclipse. 07:50
audreyt I mean epic.sf.net. 07:51
I mean
e-p-i-c.sf.net
ayrnieu class Buffer { has $.before; has $.after; method insert (Str $s) { $.before ~= $s }; method move_cursor (Int $i where $i < $.before.chars) { $.after = $.before.substr($i); $.before .= sustr($i) }; method move_cursor (Int $i where $i > $.before.chars) { $.before ~= $.after.substr(0,$i); $.after .= substr($i) }; method move_cursor (Int $i where $i == $.before.chars) {} } # mumble mumble mumble
audreyt ~V~ 07:52
# nullary hand-wavy operator
07:53 pmurias joined
Barry Yeah, I though it was an Eclipse thing. I'm looking for something that could use a graphical, web based diff plug-in. Eclipse has a lot of stuff, but I'll see. 07:53
ayrnieu multi method
Eclipse is good because Java is bad. 07:54
arcady and how do you figure that one? 07:55
ayrnieu yow, was that a controversial assertion? 07:56
arcady i mean, how does one follow from the other?
with the little I've seen of it, I don't really like eclipse as a text editor 07:57
audreyt it's probably not designed as a text editor
ayrnieu It's designed as something that makes Java feel less verbose. 07:58
So I imagine that it has all kinds of features, like autocompletion and introspection and refactoring wizards and code folding, to support this simple aim.
arcady yeah
which are great and all 07:59
but make it rather slow when all you want to do is write some java code
ayrnieu which are great and all in context of mindflaying verboseness.
but I'm neither a Java programmer nor an Eclispe user, and but the bearer of truisms, so you should ask #java for more propaganda. 08:00
oh, and Eclipse probably manages to write a surprising amount of code by itself. 08:01
Barry ...works up the courage to double click his Eclipse icon...
ayrnieu You can't have succint accessors, but you can generate comb-code very easily.
arcady so basically you end up programming in eclipse, rather than in java 08:02
thus saving a bit of time
and demonstrating that Java really wasn't meant for human consumption
I wonder if it puts in exception declarations for you
ayrnieu oh, it'd have to.
Barry I think Java was meant for human consumption and Perl was meant for programmer consumption. 08:03
arcady that was probably the thing that annoyed me most about java
no, i think java was not actually designed with users in mind
ayrnieu I think more of humans, Barry.
so, unstated XML analogy complete. 08:04
08:04 kanru joined
arcady ah yes, "human" "readable" 08:05
Barry In the last Java discussion I heard, it sounded like Java is for businesses that want to play it safe, but it's not very flexible.
arcady well, businesses use it because they see other businesses using it
svnbot6 r9403 | audreyt++ | * p6prelude.p6: parse "use v6-pugs;". 08:06
r9403 | audreyt++ | * p6compiler.pl: all diagnostics messages go to STDERR.
arcady and it's not like java doesn't have its benefits
ayrnieu arcady - Java programmers, mainly.
Barry I just wish I had learned Java in high school instead of BASIC, and that Java was built into computers like the C-64. I'd have a bit more useful knowledge today. 08:07
ayrnieu It's not kind to wish hardship upon your past selves, Barry. 08:08
arcady I think the industry needs to standardize on a better language 08:09
like, say, intercal
ayrnieu Java is not worse than intercal.
arcady I know 08:10
but it would be amusing if people were convinced to use intercal anyway
arcady goes to have some tea 08:15
svnbot6 r9404 | audreyt++ | * Pugs.Parser: parse for the magickal "use v6-pugs;" form. 08:21
Barry I opened Eclipse 2.1 after years of disuse, did an update check, and it found none, but said my preferences affect that. Couldn't find preferences, so I clicked help and Windows said Java 2 Standard Edition Binary is blocked. I chose not to unblock it. Maybe I'll uninstall Eclipse and install the latest version.
ayrnieu blocked?
Barry Yeah. Maybe I have an updated version? 08:22
svnbot6 r9405 | audreyt++ | * Pugs::Grammar::Rule - switch to 'use v6-pugs' for it; simplify .pmc generation.
ayrnieu I mean, what does it mean when windows 'blocks' something?
Barry I assumed it's not safe. That big Windows service pack update prompt you like that foe some things. 08:23
svnbot6 r9406 | audreyt++ | * rename the directory Pugs-Compiler-Rule to Pugs-Grammar-Rule to agree
r9406 | audreyt++ | with the module.
Barry If I get the prompt after installing Eclipse again, I'll unblock it. 08:24
svnbot6 r9407 | audreyt++ | * Pugs::Grammar::Rule: add the generated Rule.pmc. 08:26
Barry But Eclipse is meant for real programmers, and the plugin I'm looking to add to something isn't that powerful, so I better look for a less technicall editor. Like for HTML. 08:28
Something between Epic and TinyMCE. 08:29
ayrnieu On windows, you should probably consider Emacs or possibly UltraEdit. Unless you'd like to start by writing one. 08:31
Barry And written in Perl, and good, with good documentation. Emacs has everything, doesn't it? My plugin is a graphical diff tool. Emacs is also too good for my plugin. Maybe UltraEdit. 08:32
I'm sure I won't find exactly what I want, but that never stops me from trying.
svnbot6 r9408 | audreyt++ | * some POD sections for Pugs::Grammar::Rule 08:33
Barry Doesn't have to be for windows. I have access to a Unix server, from my website. 08:34
gaal morning 08:36
08:36 iblechbot joined 08:37 mtve joined
audreyt gaal: hey 08:38
svnbot6 r9409 | audreyt++ | * prepare to reorg Pugs::*::Rule for release 08:39
gaal lots of backlogging for very short time :)
Barry Can't believe I've been here for an hour already. 08:41
08:41 kane_ joined
svnbot6 r9410 | audreyt++ | * move the iterator engine to Pugs::Runtime::Rule 08:45
r9411 | audreyt++ | * stubs: Pugs::Compiler::Rule, Pugs::Emitter::Rule::Perl5
pmurias Barry: what do you need the editor for?
svnbot6 r9412 | audreyt++ | * rename Pugs-Grammar-Rule back to Pugs-Compiler-Rule. 08:47
r9412 | audreyt++ | yes, I'm practicing the art of changing-my-mind-faster-than-larry.
Barry I created a web-based graphical diff application that I want to turn into a plugin for something. A remote plugin...it will run from my server, so the tool it's plugged into has to be used with an internet connection. I like to be different.
gaal do we have the beginning of grammars for Perl 6? Apart from Rules themselves? 08:48
audreyt Pugs::Grammar::Perl6
gaal: a very primitive beginning at misc/pX/Common/lrep/p6prelude.p6
08:49 tcliou joined
pmurias Barry: try vim it supports perl plugins(and python and ruby ones) 08:49
audreyt Barry: got a demo somewhere?
Barry I'd love to do it for Vim. I was scared that there's something better for Vim, but I'll look into it.
audreyt: www.polisource.com/diffnote 08:50
pmurias if you want it fit for a statisitc win32-er you can turn the modality of in vim 08:51
gaal audreyt: thanks 08:53
Barry I installed gVim (graphical) a couple of years ago. I'll dust it off. I still use Windows Word Pad. Never felt like learning anything fancier.
ayrnieu vim also supports tcl extensions, or used to.
gaal is Pugs::Grammar::Perl6 in the source tree? where does it live? 08:56
pmurias any one here using the wmii window manager(www.wmii.de) ?
audreyt gaal: it lives in the future where the Common/lrep/ things are moved into perl5/ 08:57
svnbot6 r9413 | audreyt++ | * chase the name change of Pugs::Runtime::Rule.
audreyt gaal: for example, perl5/Pugs-Compiler-Perl6
ayrnieu pmurias - no, but why? 08:58
audreyt gaal: but I'm getting the Rules part factored out and released first
gaal gotcha 09:00
pmurias just wanted to mention it as a counter balance to Word Pad, it would be cool to be able to extend it with perl6
though
ayrnieu ah.
audreyt gaal: I'm thinking that the v6-pugs pragma can be the catalyst of piecemeal migration 09:01
gaal: i.e. we should work on rationalizing VCode and Pad structure etc
and for Emit.Perl5 to output perl5 code (at this time more complete than the primitive lrep-based grammar)
so that we can write more complex parsing logic inside {return...} blocks for Pugs::Compiler::Perl6 (in perl6) 09:02
and let v6.pm invoke pugs.exe to compile it to Perl6.pmc
until a day where it can parse itself
then let v6.pm switch to that
and all code will remain working as-is 09:03
meanwhile, make a DrIFT.Perl6Class or something like that
to convert Hs data structures verbatim into p6 land 09:04
gaal I have to learn to use these new tools... /me goes to follow fglock++'s email to p6-l
audreyt and as such the yaml dump of intermediate structures (parse tree, objs) etc
can be shared among p5/p6/hs/pir/whatever lands
end of digression. back to coding
gaal oh, that should be relatively straightforward (I think?)
audreyt yup 09:05
for someone who knows how to use DrIFT, that is
:)
miyagawa hi audreyt
audreyt hi miyagawa-san!
miyagawa helllllo
did you sort out the accommodation stuff for YAPC? :)
audreyt miyagawa: looking at shigetsu now 09:07
miyagawa oh, okay
audreyt still a bit expensive :/ 09:08
miyagawa right :/ 09:09
audreyt so I think still Dan's
miyagawa ok 09:10
er, Japanese Sutie Y 26k 09:11
that's expensive
audreyt wonders if "use v6-pugs;" will be ready in time for YAPC::Asia 09:12
09:29 tuxxboxx joined 09:30 xinming joined 09:51 Barry left 09:54 nothingmuch joined 10:00 tuxxboxx joined
arcady wow, what exactly is misc/pX/Common/lrep/p6object_space.p6 supposed to be? 10:01
wishful thinking that someone will come along and fill it in? 10:02
10:04 tuxxboxx joined
audreyt arcady: yeah, mostly a hint for stevan et al to fill in, iirc 10:06
brb
10:16 turrepurre joined 10:37 larsen joined 10:43 iblechbot joined 10:45 tuxxboxx joined 10:50 ruoso joined 10:57 elmex joined 11:07 clkao joined 11:16 hlen joined
ruoso wonders if there is a xemacs init file for the code standards in pugs... 11:17
11:24 Aankhen`` joined
svnbot6 r9414 | ruoso++ | now support reading a file from STDIN 11:37
ruoso now working on how p6compiler.p6 would looks like 11:39
11:43 chris2 joined
ruoso will use a trick to bootstrap p6compiler.p6. Will use as many p5 modules as possible. So they can be rewritten in perl6 later. 11:49
audreyt yes! 11:50
pull in the whole CPAN!
(that way we will collect those module's authors as new pokemons) 11:51
I mean committers
ruoso audreyt, I'll use IO::File, so there will be no need to implement the rules for open or anything... 11:54
audreyt nod 11:55
ruoso audreyt, what do you think? 11:56
audreyt I think the more cpan modules the merrier
ruoso :) 11:57
gaal DrIFT.Perl6Class - representation of e.g. our ASTs? Is there a good way to represent Hs tagged unions in p6? 12:02
integral ["tag", $data]; given $foo.[0] { when "tag1" { ... } when "tag2" { ... } } 12:03
audreyt gaal: case classes
12:04 TimToady joined
gaal wants a way to patmatch those 12:04
audreyt gaal: S06 gives you a way.
using :() in rules
as well as multis
gaal: VInt VStr are classes; they all "does Val"
Val in turn does Exp
pmurias ruso: if you need any help in convincing fglock to increase the dependency could I'll be back in ~15min 12:05
:)
audreyt gaal: basically: make constructor into classes, and make types into roles
gaal that'd work, but it looks a little bass awkward
audreyt how so? surely VInt can become a Val? 12:06
gaal since in Hs a data type doesn't have to know some *other* type has it in a variant
audreyt er no, sorry for the punning
data Moose = M Oo | S Ee 12:07
M and S are classes; Moose is role
Oo and Ee has no knowledge of anything.
gaal okay, how do you represent
data Alces = Moose Moose {- depun -} | Caribou Caribou 12:08
audreyt we use role/class duality
nothingmuch bangs head on table
gaal read "Exp" for Alces and "Val" for Moose
audreyt Moose is now a class that can construct anything that does Moose into somthing that does Alces. 12:09
nothingmuch wrapper mc parsec represents class
gaal nothingmuch: MC really stands for Moose Caribou
nothingmuch www.seeklyrics.com/lyrics/Lil-Jon/P...Remix.html
there's lots of representation in there 12:10
12:10 TimToady joined
nothingmuch (googled for 'yo represent' ;-) 12:10
nothingmuch goes back to fixing his address book
gaal so are we going to rap in YAPC::NA or what?
nothingmuch moral of this weekend: nnever EVER EVER EVER use a "cool" address book interface without first backing your address book up
gaal moral, golfed: 'grep'. 12:11
audreyt: so, data constructors are factory methods in a Role? 12:12
(is dev.perl.org/perl6/doc/design/syn/S06.html up to date or should I always use the svn version?) 12:13
audreyt gaal: factory methods in a design
s/design/class/
or rather, actually, just classes with infix:<as>
remember that
Foo(1,2,3)
is
infix:<as>(\(1,2,3), ::Foo) 12:14
i.e. the same as
(1,2,3) as Foo 12:15
12:16 DesreveR is now known as r0nny
gaal whoa. that's pretty deep 12:16
ruoso is "$arg1 = shift;" still a valid way of getting the command line arguments?
gaal ruoso: look in @*ARGS 12:17
audreyt unspecced, so currently not
gaal ruoso: there's a Getopt::Std module in ext/ you can use for basic stuff. 12:18
migo and ran are working on a much better thing
audreyt: so basically, (1, 2, 3) would have been Foo automatically if we had an inferencer; since we don't, we need to coerce manually 12:20
?
(there's a missing application there)
ruoso gaal, I'll leave that to a second moment, for now I'll receive "inputfile" "outputfile" as parameters
audreyt gaal: no, Foo(1,2,3) is just a call to &Foo 12:21
gaal: it's just "class Foo" defines a ::Foo _and_ a &Foo
&Foo handles coercion toward Foo
in fact, maybe the default infix<as> will fallback to that.
gaal ruoso: sure, all I'm saying is that FWIW, you can have -i and -o too if you like.
ruoso: not really advocating that :)
audreyt class Val { multi infix:<as> (VInt, Val) { ... } } 12:22
this may be generated as a factory
from DrIFT, or from p6land
ruoso gaal, I will, just not right now... 12:23
gaal the content of that multi being new :tag<VInt> 12:24
oops
new(:tag<VInt> :VInt<$x>)? 12:25
or whatever representation we use?
audreyt yah. 12:27
gaal what representation lends itself to pattern matching a la integral's [$tag, $data] example above? 12:29
has $tag; has $data; # and has nothing else? 12:31
audreyt thinking 12:32
12:32 kane_ joined
audreyt I think it should be arglists all the way down 12:33
that is
pmurias is xgrammar2p6 for a tool that converts the xquery grammar from xml to p6? 12:34
audreyt data T = Nul | Pos A B | Rec { l1 :: X, l2 :: Y }
pmurias s/6 f/6 a good name f/ 12:35
audreyt Rec(l1=>$x, l2=>$y) should just preserve the arglist
into "has X $.l1, has Y $.l2" 12:36
which can be matched using
multi f (Rec $r ( $l1, $l2 )) {} 12:37
12:37 ruz joined
audreyt see "Unpacking tree node parameters" 12:37
S06
gaal okay. do we have fancy arglists in pugs already?
audreyt no. but this will be a great motivation to hack them in.
gaal (
("Argument list binding") 12:38
k
audreyt you can start by modelling it as Hs data structures :)
now for Pos, we can settle on traditional numbering
$a $b $c $d
and match them likewise
gaal how do you model Nul? dummy var?
s/model/match 12:39
svnbot6 r9415 | pmurias++ | a tool to convert the xquery grammar to p6 12:40
12:41 iblechbot joined
audreyt class Nul {} 12:42
multi f (Nul) {}
will do
multi f (Rec ($l1, $l2)) {}
multi f (Pos ($a, $b)) {}
er I mean 12:43
mulfi f (Rec:($l1, $l2)) {}
(because we omit $r)
gaal class Nul does T
audreyt yup 12:44
gaal class T does what?
audreyt T is not a constructor
hence not a class
gaal HsishClass?
audreyt it's just dummy role
role T {}
gaal Role T is Hsish
audreyt yeah
gaal is TaggedUnion or something
audreyt right. but we don't need them at the moment 12:45
if the type names are not conflicting, we can pun
class Pos { has $.a; has $.b } 12:46
class Pos { has A $.a; has B $.b }
if there are two As
class Pos { has A $.a1; has A $.a2 }
perhaps.
makes sense? 12:47
gaal uh, where's the pun here?
audreyt data T = Con Int Int String Val 12:48
class Con { has Int $.int1; has Int $.int2; has String $.string; has Val $.val } 12:49
that's the pun
not sure if it works better than $.a $.b etc in practice or not.
or even $.0 $.1.
gaal ah, okay
audreyt has Int $.0; # now that's a charming thought
gaal it's less confusing than the Hs puns
sufficiently less confusing that I was confused about where the pun was :) 12:50
audreyt :D
I take that as a compliment
gaal drafting up a simple transliteration of Exp 12:51
audreyt gaal++ 12:52
I think modeling the core structures in Hs for the short term is still going to be a win
over modeling it in p5
and as such making Hs as the upstream and simply carrying the modelled result to p6/p5/pir would be a win indeed
gaal I forget, do we have nullable types? 12:53
(Maybe Exp) => ?
audreyt in p6 everything is nullable 12:54
so we are forced to do
class Just { has Any $.a } 12:55
class Nothing {}
which really sucks :/
gaal class App { has Exp $.func; has Exp $.invocant; has Exp @.args; } # wrong, because of $.invocant 12:56
audreyt otoh, maybe we can ignore Maybe for now. 12:58
has Maybe[Exp] $.invocant; # correct translation, requires parameterized roles
wow, Rails runs on YARV.
YARV got there fast. 12:59
that's like, Catalyst runs on Parrot :)
gaal brb
nothingmuch is diff(ruby2.0, ruby) big?
audreyt nothingmuch: ruby2 is less well defined than perl6
nothingmuch: so nobody knows ;) 13:00
nothingmuch ah
audreyt try JS2 for some well-defined stuff :)
nothingmuch i thought YARV was for ruby 2
audreyt it is
but it's also for ruby 1.
because ruby people care about migration 13:01
:)
(also because its metamodel is considerably easier to port than perl5's.)
nothingmuch aye
chris2 nothingmuch: the changes will be far less than p5->p6, but significant and require fixing code 13:02
audreyt that reminds me. I need to invite sasada over to the YAPC::Asia hackathon. 13:03
now that matz and takahashi will already be there.
audreyt goes mail-writing 13:04
chris2 wow
nothingmuch ciao
nothingmuch goes to amuse his sister
audreyt: please tell balas that the Amazing Toy For 21st Century is Bringing [my sister] Sci Fi Fun! 13:15
audreyt nothingmuch: done. we are glad to hear it! 13:19
nothingmuch our cat scratched it though 13:23
so it's now taped
and slightly leaky
(has to be refilled daily) 13:24
audreyt which adds to the fun?
nothingmuch uh 13:25
not really ;-)
audreyt aw ;)
nothingmuch it's life expectancy is not that big, i guess
nothingmuch was hoping to go to the local cheap-crap store and filling it with helium
oh 13:26
bah
she just showed me a hole the size of a foot
=(
okay, please tell balas that the amazing toy for 21st century is no longer going to bring my sister sci fi fun 13:27
audreyt ...done.
though you can perhaps still patch it up
but with ongoing cat presence
maybe ultimately futile
nothingmuch hole size of foot == no go
will upset the balance
and it's hard to find a good adhesive
audreyt nothingmuch: balas would like to remark that she is glad that your _cat_ had sci fi fun.
nothingmuch heh
it was more territorial than sci fi ;-) 13:28
his personality is very much like foo's
seen bsb? 13:33
oh shucks
audreyt starts SEE hacking with ingy on pmc.pmc
(the module that enables v6.pm magic) 13:34
13:34 whiteg_xen joined
LeTo audreyt: .pmc is confusing (for parrot folks at least:) 13:34
audreyt LeTo: it's in use since 1999 :) 13:35
ingy in non-use since 1999 :p
er... in no since 1999 :D
13:36 saorge joined
audreyt LeTo: after this session I'll go find you and hack on pugs.pmc 13:36
the _other_ pmc
ingy whiteg_xen: did you install postgres on xen1?
whiteg_xen ingy: no, I didn't do anything on xen1 13:37
whiteg_xen is using **** gprs
nothingmuch hola ingy
ingy whiteg_xen: ok 13:38
audreyt ( SEE session at see://freepan.org/ ) 13:41
nothingmuch joins for fun 13:43
13:46 dolmen joined
nothingmuch continues Blondie in Haskell 13:47
audreyt have fun :) 13:48
nothingmuch danke
ooh 13:51
dad thinks he fixed the tear
xinming wiki.freepan.org/ 13:53
How can the spammer work so hard? :-S 13:54
s/hard/fast/
nothingmuch fast, hard - it's all the same to those spammers ;-) 13:55
13:58 fglock joined
stevan loves the fact that all the part of Perl 6 will end up on CPAN 13:59
fglock audreyt: I've added some things to Pugs::Grammar but I don't have time to merge right now :( 14:00
audreyt fglock: that's ok
nothingmuch i wish the world had a volume button 14:01
fglock audreyt: I merged p6rule_lib into the grammar, it is no longer needed
nothingmuch crap 14:02
i just tried to use it too
fglock later... &
nothingmuch turned down his laptop's volume
14:02 fglock left
stevan audreyt++ ingy++ # the smell of crack smoke is in the air 14:08
ingy :p
stevan pmc++ very cool (your's too LeTo :) 14:09
stevan has to go fetch bagels or his children will revolt
14:09 rantanplan_ joined
stevan audreyt: I will start doodling some in p6object_space.p6 sometime today 14:09
audreyt stevan: excellent 14:10
stevan ingy: my Moose will beat up your Spiffy,.. (ask nothingmuch for more info)
ingy :)
stevan dashes off &
nothingmuch Spiffy Moose 14:12
audreyt: i was losing my concentration 14:16
so instead i tried to make a cleaned up parse tree thing for fglock 14:17
ingy nothingmuch: m00z3 14:27
nothingmuch ingy: oh come on, i was not serious
i was actually going to make a real comment
ingy neither was I ;)
nothingmuch retires from conv 14:37
not enough concentration-fu
pmurias fglock: It there anything I could do help the move to the new layout?(which dosn't require much project vision) 14:45
/exit
wolverian um..
audreyt wow
14:45 pmurias joined
wolverian pokes pmurias 14:46
14:46 Khisanth joined
pmurias sorry, the irc client mefunctioned 14:46
again :) 14:47
nothingmuch audreyt: ? 14:48
audreyt hm? 14:50
nothingmuch what was the wow aabout?
actually, neverkmind
i can't bring myself into listening to anything
off to drown myself in the shower
audreyt considers the slogan "deploying source filters without the user knowing about it" for pmc.pm 14:53
audreyt ponders real macros for p5 (via MAD or PPI) using a filter supporting q:code and {{{}}} 14:54
nothingmuch the problem is that they're parse trees 14:55
14:55 beppu_ joined
nothingmuch == more special cases to deal with 14:55
14:56 szabgab joined
audreyt MAD is actually the optree 14:56
nothingmuch oh, right
with syntax annotations, right? 14:57
audreyt yup
szabgab ppl: what do you think about setting up a beginner mailing list for perl6 on learn.perl.org/ ? 14:58
audreyt very good idea! 14:59
nothingmuch szabgab: i'll help out 15:00
15:00 pmurias joined
szabgab is any of the perl.org admins here or shall I send a message to Ask ? 15:00
audreyt don't ask, just ask ask :) 15:01
szabgab err I did mean to Ask :)
what name do you suggest for the mailing list ?
integral perl6-beginners? 15:02
szabgab integral: that's probably the right name
nothingmuch FUCK!!!!!!!!!!! 15:03
fucking vim
at least it has .swp files
audreyt just lost a year's worth of work? :) 15:04
nothingmuch no
just almost lost an hours worth of non work
my brain is not functioning 15:05
and despite almost nil progress I still felt like i couldn't redo it
pasteling "nothingmuch" at 82.81.247.180 pasted "# some desugaring for a simple" (155 lines, 4.8K) at sial.org/pbot/16254
nothingmuch eek
why does pbotutil take the first line instead of the filename as a desc
anyway
this is the beginings of a non parse level AST for Perl 6 programs
please comment somehow 15:06
but i won't be here to read /msgs etc
15:10 hlen_ joined 15:45 dolmen joined 15:46 audreyt joined 15:52 avar joined 16:14 justatheory joined 16:15 nothingmuch joined
nothingmuch moosen 16:16
ingy !
nothingmuch SpĆøĆøne
SpĆørk SpĆørk SpĆørk!
ingy dƶt
nothingmuch wonders if his torrent picked up 16:19
and how risky it is to watch BSG in the bathtub
TimToady PĆøĢˆ 16:21
PĆøĢˆrl 16:22
ingy hi TimToady
TimToady likes combining characters...
ingy audreyt has started me down a path sufficiently evil... 16:23
TimToady That should have come out as a slashed o with an umlaut.
nothingmuch isn't the order ĀØĆø ?
hmm
TimToady combining characters always follow in Unicode. 16:24
nothingmuch ah, didn't know
TimToady my evil thought for the day:
nothingmuch so MacOS's convention of combining first is kinda silly?
TimToady sub main ($a, $b, :$switch, *@files);
note semicolon. 16:25
nothingmuch TimToady: see also Migo and Ran's work on Getopt
TimToady nothingmuch: both conventions are silly.
audreyt TimToady: we already started on a sigs-based getopt
TimToady Yes, but do you have a syntax for turning the entire file into a sub? 16:26
nothingmuch TimToady: should that really be a special case?
TimToady I was just looking at some code with @*ARGS and thinking how ugly that was...
nothingmuch could we just write a macro to specialize $?SUB in the main body?
TimToady not a special case if sub is in the same category as module or class.
audreyt TimToady: yes I do 16:27
# in main.p6
nothingmuch even so, that dirties up the macro i think
i think that much mroe useful would be a meta api to get at the code ref to main()
whatever it may be
audreyt module Main;
is signature($a, $b);
done!
nothingmuch and then add a signature to that
err, like that =)
TimToady multi main (:$foo, :$bar); 16:28
(just call some main that matches my args...)
nothingmuch TimToady: see also Getopt::Process which can coerce an argument list into subroutine arguments
but not consolidate nicely
that logic is already written =) 16:29
isn't that multi *main, btw?
TimToady I spose.
audreyt TimToady: why "main"? 16:30
why not just the module itself?
TimToady hysterical raisins
nothingmuch i have an idea
every package has a special named subroutine
that is just it's body
audreyt nothingmuch: the name is &::('') in my mind
TimToady sub (*%switches); 16:31
nothingmuch i think so too
either that or ::main
16:31 jdv79 joined
nothingmuch or __main 16:31
if we're going for that feel ;-)
jdv79 if i would like to use p6 rules in p5 - what of it?
TimToady s/__/{die}/
nothingmuch then we &*main ::= &::('');
and we get DWIM semantics for that bit
audreyt jdv79: wait a few hours and start using re::override-perl6 16:32
jdv79 fabulous!
will do, thanks:)
audreyt see perl5/Pugs-Compiler-Rule and misc/pX/Common/regexp_engine_demo.pl for now
they need to be packaged a bit
nothingmuch swears that no couch worth less than $1000 will ever enter his house when he grows up 16:33
TimToady I don't want someone to have to lean a different syntax for processing args to the main program than to a normal binding, at least for ordinary arglist processing. $*ARGS is always available.
*learn
nothingmuch we have a crap sofa and I can't comfortably sit on it for than 15 minutes
okay, in that case isn't it 'proto' ? 16:34
proto *main (...) after &*main ::= Main::('') gives you that, right?
TimToady Are you suggesting that mere mortals have to write that? 16:35
ingy nothingmuch: you really planning on growing up?
nothingmuch ingy: yes, in about 30 years
actually, pretty soon
ingy sad :(
nothingmuch is moving abroad 16:36
sick of israel
audreyt TimToady: ok, so any module can serve as a main program as long as it has some sort of &main defined
ingy let's pair up
nothingmuch i don't think i'm crazy enough for .tw ;-)
TimToady maybe. It'd be nice to serve mains up from that end too.
nothingmuch audreyt: i think a cleaner variant is to add multi variants to Main::('') 16:37
audreyt: i think a cleaner variant is to add multi variants to &Main::('')
audreyt nothingmuch: right, and the logical syntax is
multi module Main (...) {
}
multi module Main (...) {
}
;)
nothingmuch oh dear me 16:38
Juerd
.oO( multi grammar )
nothingmuch that mushes up two modules, and combines all their subs as variants
audreyt er no, not really
nothingmuch Juerd: it's very simple, actually... all the methods of a grammar are just things operating on $?SELF as a sort of curer
audreyt it just let the main body sub take args.
nothingmuch audreyt: that was a question, actually
TimToady no, I think it's just multi main in Main package, and multi Main::main elsewhere.
nothingmuch just missed the question mark
ah
Juerd: then the only args they normally get are rules 16:39
so you could in theory make a multimethod +
audreyt TimToady: I agree with you, I'm just trying to show the logical end of nothingmuch's line of thought and how strange it is
nothingmuch that optimizes simple subrules
and a general one for any rule
and implement that with multi rule &whatever_category_its_in:<+> 16:40
Juerd nothingmuch: Please - discard my remark :)
nothingmuch Juerd: it's actually very practical =)
and useful
think
audreyt TimToady: in Haskell, you can have a test.hs that goes
TimToady so 'sub main' and 'proto main' would be valid variants in Main.
nothingmuch a+
audreyt module Test where
main = print "Hello"
nothingmuch vs <something_hairy>+
audreyt and if it's run as the main program
nothingmuch we know how to make 'a+' more efficient
audreyt it will use Test.main instead of Main.main as the entry point
TimToady module Test :main{ say 'howdy' }; 16:41
audreyt TimToady: but then, hs doesn't have the .pm/.pl distinction
so maybe Main is still good (for .pl) and .pm are named as non-main
TimToady well, at least we know whether we were execed or used. 16:42
nothingmuch audreyt: also, haskell doesn't have any exprs that aren't bindings/decls at the top level
audreyt yeah... but this concerns the Main::main generation. if I run Test.pm as my main program 16:43
Test::ok doesn't somehow become Main::ok
nothingmuch ofcourse not
TimToady I think if you run a module Test we can look for Test::main to run...
audreyt but Main::main will dispatch to Test::main
nothingmuch ponders
audreyt and failing that, &('Test::')
nothingmuch where do we use the name 'main'
event loops
objects with one action 16:44
bad code (C written in Perl)
Juerd nothingmuch: do_one_loop
Oh, you had event loops already
TimToady A lot of P5 code currently has 'sub main'
audreyt TimToady: so
# test.p6
integral TimToady: not necessarily *good* P5 code that should be copied
audreyt sub main { say "Hi" }
say "ho!"; # never said?
TimToady have to translate it though. end up with 'sub main_' or some such...
Juerd TimToady: I speculate that most of that is because they like main-before-subs, where you need a block to keep lexicals from leaking into the rest. 16:45
And if you have a block, it's good style to name it. This is only rarely done with labels.
TimToady We could detect if the only thing outside the sub is the call, and optimize it away.
Juerd I think claiming "main" is a bad idea.
There must be a better word
Or even symbol?
audreyt Main ;)
since we are already renaming ::main to ::Main ;) 16:46
Juerd Main is better
But maybe just _?
nothingmuch TimToady: yes, but 99% of the time it's along side 'main(@ARGV)'
i'm tryhing to think where this might conflict
with the 99% approach
test... test....
audreyt Juerd: no, _() is gettext()
Juerd _ is weird in Perl 5 anyway.
nothingmuch bah
network nearly died
TimToady magically called sub should be MAIN...
Juerd audreyt: Bah; But true.
TimToady: ++
AUTO? ;)
audreyt TimToady: so if I have
# Test.pm 16:47
module Test;
say "hi";
does Test::MAIN automatically get bound?
as {say "hi"}
TimToady why not?
audreyt well
Juerd Ahh, goto &MAIN; # like basic's run
audreyt module Test { say "hi" }
module Test { say "ho" }
nothingmuch audreyt: any comment about the nopaste, btw?
audreyt concatenated together?
(the "multi module" problem) 16:48
Juerd audreyt: sub MAIN { { say "hi" } { say "ho" } }
audreyt Juerd: that's what I mena by concat
TimToady error unless "is mumble"...
audreyt yeah, "is reopen"
"is extended"
but the semantic is to concat?
Juerd TimToady: But open classes etc were default, right?
TimToady yeah, whatever it was...
but you still have to say that you're reopening it, to prevent accidents. 16:49
Juerd I see
audreyt Juerd: they were. under ./pugs -o (which closes classes), "is reopen" would be fatal
I mean -O
-Ofin # finalize classes
TimToady there's also 'is instead' to say that you're replacing it.
Juerd "is okay("Just do it")"
;)
TimToady: Why not use self-documenting "is replacement"? 16:50
audreyt TimToady: "is also"?
Juerd audreyt++ # I like is also
audreyt sub MAIN is also { say "run this after the body" |
TimToady Yeah, that seems like the one I like most often..
yikes.
audreyt TimToady: btw, :COMPILING turning off hygeine worksforme. should I just commit to S06? 16:51
nothingmuch so the is also trait is like adding a wrapper
that basically does 'call; ....'
Juerd audreyt: END == sub MAIN is also?
audreyt Juerd: nope; END runs even if you dies
different control step
Juerd Oh, right
nothingmuch what's program global UNDO then?
audreyt same thing
TimToady just a sec--i have some damian edits to commit first.. 16:52
audreyt TimToady: ok
nothingmuch UNDO is per block though
audreyt the main program is a block
nothingmuch right
but you don't always get access to that
audreyt why not?
nothingmuch if you have a module, that is creating a structure
that needs to be cleaned up
well, what do you do?
audreyt to be cleaned up... when?
at program exit?
nothingmuch &*MAIN.meta.pad.get('@*UNDO').push(sub { }); 16:53
yes
due to die
or rather fail
16:53 bsb joined
audreyt well, easily 16:53
sub Main::MAIN is also { UNDO { ... } }
nothingmuch bsb: messaged you on perlmonks
oi
TimToady audreyt: k, s6 free
nothingmuch right
audreyt actually, bsb is here... 16:54
bsb hey nothingmuch
gaal remoose
bsb what did I do?
Juerd Is it okay to use "is instead" on the *first* definition?
TimToady 'sub is also' looks an awful lot like what p5 is doing internally with CHECK blocks, etc. except that we need 'is prealso'.
nothingmuch bsb: your resume's capoeira link goes to a "free porn" link farm
audreyt bsb: have cycles to create a patch to S06 that says hygeine is assumed (implicit scope around q:code) unless you say :COMPILING ? 16:55
that's the saner option I have for hygiene
the insane option is to encourage use of macro-scoped lexicals as containers ;)
macro f { my $x; q:code/ $x = ... / } 16:56
Juerd TimToady: "is also :before"?
audreyt "is butfirst" of course ;)
TimToady I guess most of the BEGINish blocks are 'is (pre|post)also'
audreyt </troll>
nothingmuch audreyt: the lexical scopes of 'is also' are not fudged though, right?
TimToady 'is first' is taken...
nothingmuch s/fudged/mashed/; 16:57
Juerd Is there a specific reason for having "before" in rules, but "pre" outside them?
nothingmuch &*MAIN.=wrap(sub { .... }); ?
Juerd (Maybe that "before" is an assertion, while "pre" is not?)
audreyt sub foo { BEFORE { ... }; AFTER { ... } }
just doesn't quite read correctly
nothingmuch is that what the trait is doing?
Juerd audreyt: I think it does. I read it just the same as PRE/POST
nothingmuch or is it really combining the bodies?
Juerd nothingmuch: Maybe a sub is really an array of subs :) 16:58
audreyt sub foo { BEFORE { this; } AFTER { that; } body; }
Juerd If so, we should be able to splice it. Oh my.
nothingmuch Juerd: you mean a matrix....
audreyt Juerd: what's the sequence? that-body-this?
Juerd nothingmuch: Perhaps.
audreyt or this-body-that?
Juerd audreyt: this-body-that, but your point is a very good one. 16:59
"before" and "after" beg for a grammatical object
TimToady with most of our BEGINish blocks, the block itself determines whether it should be pre'ed or post'ed. So maybe 'is also' should just tell the block and let the block decide.
Juerd "pre" and "post" don't have this, because they're not English.
TimToady can always .wrap if you want better control.
Juerd They don't feel natural anyway, and that's the property that is (ab?)used here.
Okay :)
audreyt stevan: any/bit/int/num/str are in parrot now 17:00
TimToady Every CHECK/INIT/FIRST/ENTER/UNDO is 'is also'
stevan audreyt: nice
audreyt nil/sub/seq/map/obj next
stevan cool
audreyt parrot is very annoying in that it thinks all values are also containers
Juerd TimToady: is CHECK/INIT/... an argument for "also"? (is also("CHECK"))
audreyt so I need to disable all mutators
otherwise
1++; # always succeeds 17:01
# and changes the idea of 1 elsewhere!
stevan audreyt: I am fudging a p6metamodel_bootstrap.p6 right now
Juerd audreyt: readonly
audreyt considers it utterly broken
Juerd audreyt: In Perl 5, a "value" is also always a container, but with the readonly flag.
stevan :D
nothingmuch in python does 1++ cause all 1s everywhere to be a 2?
;-)
Juerd audreyt: Interesting things happen if you make "undef" writable, changing the definition of undef.
bsb audreyt: I'll try to write a hygiene note of S06
Juerd Undef can be defined.
audreyt Juerd: bsb danke
Juerd (What's bsb?) 17:02
audreyt er
bsb: danke
Juerd:
TimToady ohh a ghostghostwriter.
audreyt yeah, parrot just took the SV idea
metaghostwriter
Juerd: I think the SV idea is unsound really, because it prevents "1" to be shared.
Juerd audreyt: The SV idea is useful when you *want to* override literals. Though you could quite easily argue that that is always a bad thing. 17:03
audreyt when you autobox "1" twice, you get two SVs
stevan audreyt: do you have a moment to discuss object space primatives for lrep?
audreyt Juerd: there are cleaner way to override literals _lexically_
Juerd audreyt: I guess so.
audreyt macro term:<1>
is the way
nothingmuch audreyt: but if it's unshared then the value can have a pointer to it's box cached, right?
Juerd That's nice
audreyt nothingmuch: well yes, but neither p5 nor parrot can do that 17:04
17:04 GabrielVieira joined
nothingmuch right 17:04
audreyt also it results in different
1.id
nothingmuch just a semantics question
audreyt values
across program runs
Juerd userfriendly.org/ # Applicable to Perl 6? ;) 17:05
audreyt compare ruby, which always has 1.id at 3 :)
TimToady "Everything is an object" :)
audreyt Everything.isa(Object)
TimToady given P6 Everything.does(Object)
Juerd audreyt: As long as two different $foo and $bar that happen to both contain 1 have their .id equal, it's fine with me :) 17:06
TimToady given P6 { Everything.does(Object) }
stevan Object.isa(Everything) # metacirculatiry :)
Juerd Just please don't have Python's foo = 2; bar = 1+1; foo == bar
nothingmuch Juerd: but their IDs don't equal
audreyt Juerd: yeah, that is not quite easy to do in parrot. I'm still thinking about it
Juerd eh, foo.id == bar.id
nothingmuch i mean, it makes no sense
they're not the same
Juerd Because containers are incredibly useful :)
nothingmuch: Agreed. Tell Python, though :)
nothingmuch they just map to the same value
audreyt Juerd: well, variable($foo).id and variable($bar).id would be different 17:07
nothingmuch Juerd> audreyt: As long as two different $foo and $bar that happen to both contain 1 have their .id equal, it's fine with me :)
didn't you just contradict yourself?
audreyt but larry has this axiom that $scalar.method never dispatches to the container.
Juerd audreyt: Eh, hm, ...
I have to think about that.
TimToady There are times you want identical values to be distinguishable, and times you don't...
audreyt since you can always get the container with variable($foo)
nothingmuch err... in that case how do &infix:<=>, &infix:<:=> etc work? or are these back to being special cases?
audreyt which is btw special form.
nothingmuch: they desugar with variable() semantics 17:08
$foo = 3
nothingmuch how can I declare a variable() desugarer?
audreyt variable($foo).infix:<=>(3)
Juerd nothingmuch: I meant unequal :)
nothingmuch Juerd: oh
audreyt nothingmuch: you use the "is rw" trait.
nothingmuch =)
audreyt: oh
so same thing ;-)
audreyt subtly different though :)
Juerd Can we rename variable to container?
Variable has always been an exeedingly confusing word. 17:09
nothingmuch how do implicit variable($x) and $x compete in multimethods?
TimToady The point of eqv^H^H^H=== is to give you the 'values equal for value types, ids equal for object types' semantics.
use =:= if you want the other thing.
Juerd TimToady: Wasn't everything an object?
nothingmuch i thought that ids equal for object types unless objects compare themselves in a silly way?
TimToady Juerd: you missed the quotes... :) 17:10
Juerd TimToady: No, I just don't get the difference between value and object...
TimToady and the s/is/does/ :)
A value is an object that wants to pretend that it's a value.
Juerd TimToady: values are unreferenced?
TimToady Like the 3 object.
Juerd (In perl 5 jargon)
nothingmuch as in, multi &infix:<===> (Str, Str) and Num Num, and whatever are just defined in the prelude to not work on .id
and &infix:<===> (Object $x, Object $y) { $x.id === $y.id } 17:11
pasteling "gaal" at 192.115.25.249 pasted "Example transliteration of Exp" (50 lines, 1.5K) at sial.org/pbot/16255
TimToady One could say that the value types are the ones that define infix:<===>.
Juerd Ah
Do I understand it correctly if I think that Str, Int, Num are value types, but LWP::UserAgent would be an object type? 17:12
TimToady I think infix:<===> is also related to the mystery of what you can use as an "untyped" hash key.
gaal oh, another question re: transliteration: do we use qualified names? Pugs::AST::Internals::Exp, etc.?
Juerd And which would Array be?
fsck; afk.
TimToady except historically hashes have forced to type Str. 17:13
nothingmuch TimToady: i think that can't work except as a convention
TimToady There are, like, how many things in P6 that aren't conventions? 15 or so?
nothingmuch heh
TimToady It's sugar turtles most of the way down.
wolverian was Juerd right with the LWP == object, Str == value thing? and what is an Array? 17:14
gaal mmm sugar turtles
nothingmuch in that case we also want: class Object { method canonical_key_for_hash_with_better_name { self.id } }
imho Array is an object
TimToady Str is probably an object too. Too much Unicode floating around to do anything else... 17:15
And we have 'eq' for forcing string comparisons.
wolverian then what is a value? an object without methods? :)
17:15 elmex joined
nothingmuch multi method &infix:<eq> ( $x, $y ) { $x as Str === $y as Str } ? 17:16
wolverian Str $x === Str $y # :)
TimToady A value is something you don't have to clone if you want another one of it.
nothingmuch err, right
wolverian TimToady, ah. so really really primitive.
nothingmuch wow, that's a nice distinction
TimToady I suspect Str has to define === though... 17:17
gaal audreyt: so, which types are high targets for DrIFT.Perl6? Exp?
nothingmuch TimToady: i thought it already does ;-)
TimToady which would make it a value type. Maybe it just can't decide...
nothingmuch oh
i'd expect $str === $other to be a "stringish equality" thing
wolverian is tempted to put the value/object distinction into the topic
TimToady === always fails for differing types.
nothingmuch that converts to the "best" encoding and compares
does MMD let it coerce? 17:18
TimToady could
nothingmuch 2 === "2" ?
if it matches Object, Object instead it's on .id which is fail
TimToady === always fails for differing types (in the absence of MMD)
nothingmuch if it matches Str, Str or Num Num (depending on which coercion is chosen, and whether Num "2" fails) 17:19
but it won't
since coercion happens after proto
right?
TimToady don't follow...
nothingmuch ponders a new coercion semantic:
17:19 saorge_ joined
nothingmuch mutli &infix:<===> (Object $x, Object $y) { }; # this is " 17:19
"more specialized" than 17:20
multi &infix:<===> ( Str, Str ) # on a 2, "2",
right?
repharse:
TimToady 2,"2" doesn't even see Str,Str 17:21
audreyt gaal: start with PIL.Native.Types.Native, actually :)
nothingmuch TimToady: until it tries to coerce, right?
but it will never coerce, because Object, Object will match it, right?
anyway, new semantic for 'as':
17:21 hcarty joined
nothingmuch $x = mmd_distance( Object, Num ); 17:21
TimToady we haven't really nailed down MMD distance yet, except to nail up Manhattan...
nothingmuch $y = mmd_distance( Str, Str ); 17:22
$y += mmd_distance( Num, Str ); # for 'as'
i'm having trouble saying this
but if the distance in the same variant is bigger than the distance through an 'as', maybe we prefer the 'as'?
audreyt nothingmuch: you grokked docs/notes/multimethods.pod ?
TimToady probably needs a slightly more formal proposal.
nothingmuch which incarnation?
audreyt: i think so 17:23
audreyt nothingmuch: there's only one incarnation
nothingmuch there around 3-4 mmd docs, IIRC
two by luke, one by me
gaal audreyt: that's a fairly simple structure, if you don't descend to its children... 17:28
audreyt gaal: but suppose I do... 17:29
gaal ...but stop at the type level, surely?
audreyt surely. 17:30
17:30 pdcawley joined
nothingmuch audreyt: does my as-in-distance proposal disagree with multimethods.pod? 17:30
17:30 hcarty left
gaal okay, so NativeObj is another style of thing to emit 17:31
nothingmuch gaal: err... remember the blondie conclusions
this is all a generalization of linking
gaal "this"?
undefined symbol
nothingmuch emitting "native" constants, native ops, etc
XS type behavior 17:32
FFI
except that some of this linking is done in the emitting stage
audreyt nothingmuch: yes, because fallbacking-to-coercion yields a different type
nothingmuch and you're not always linking with symbols
audreyt that has no subtyping relationship with the original type
nothingmuch: I proposed "sub f ($x as Dog)" as a encouraging-coercion form 17:33
nothingmuch audreyt: so how *do* i say Num is closer to Str than it is to Obj
audreyt TimToady said he's thinking about having "sub f ($x does Dog)" to mean the same
gaal nothingmuch: I have no idea what you are telling me
nothingmuch gaal: when you are emitting '1 + 2'
audreyt nothingmuch: "Num does Str", clearly.
(according to the encouraging-coercion theory)
nothingmuch i meant the other way
numify a string 17:34
that is not well described as a role
much better as a coercion
audreyt well, it's closer to a Num because they share a nearer common ancestor
let's call it Val
nothingmuch i guess method f ($x as Num) is a good way to go
audreyt than Object's common ancestor, let's call it Any 17:35
TimToady sub f (Dog $x) already means 'does', so 'as' is probably better for coercion.
Unless we mean 'to'.
audreyt TimToady: as in JS2 "to"? isn't that just the perl6 "as"? 17:36
(user-defined casting)
TimToady Yeah, and 'does' says 'can you do this without changing yourself?'
nothingmuch changing? isn't that more of a "deriving a copy"? 17:37
TimToady nothingmuch: yes 17:38
audreyt ($x as Num) then. I like the punning
nothingmuch *phew*
okay
what about from the other direction?
audreyt it's asif you said f($x as Num) at caller site.
nothingmuch yes
17:38 netanya_goy joined
nothingmuch wait 17:38
what other direction....
nothingmuch tries to remember
oh, yes
how do i declare in role Str that Str is *always* as Num before it's does Object? 17:39
err, isa Object
we have 3 "locations" which control the outcome of MMD, the call site, the class declaration, and the method declaration, but this gives only call and method decl
audreyt nothingmuch: well, Str and Num share common ancestry, let's call it Val, so you can say that everybody does Val with various coercion forms 17:40
automagically
that is, Num isa Val and Val does Num
nothingmuch i thought that was a nono in luke's proposal
audreyt this is bizzare though.
nothingmuch very bizzare 17:41
audreyt it is a nono, because of circularity
say, if all our common builtins are declared with "as"
then this problem does not arise
nothingmuch hmm
audreyt numeric ones, at least
nothingmuch i think we want a class level 'as' thing
audreyt class Dog as Cat { ... }
nothingmuch i'd also like to say that MyUriObject as URI even though it doesn't inherit it
and that URI isn't a role in any logical way 17:42
audreyt and you don't want to say does
nothingmuch i think that works
audreyt because you can't emulate its api
TimToady C++ handles all that with constructor/coercion
nothingmuch right
audreyt you can only, on demand, construct a URI for it
nothingmuch exactly
nothingmuch is getting the hackathon feeling
when my brain is too clogged to actually work i start bugging audreyt and TimToady 17:43
;-)
audreyt and all this just so you can omit explicit coercions.
(in call site or function site.)
nothingmuch exactly
because they are orthogonal
you might have control only over one of them
audreyt certain languages has special method designed for value classes ;)
as_num
etc 17:44
but class-level "as URI" may be useful.
TimToady That's why I originally had coerce:<as> make both coercion and constructor semantics, if I recall...
audreyt TimToady: I don't think coerce:<as> is defined anywhere in the docs... 17:45
TimToady Maybe the constructor autogen is just an "is deep" variant.
it was historical
Anyway, I'm pretty comfy with "as" allowing a single direct coercion on MMD 'as' args, as long as we don't get into coercion chains like C++. 17:46
17:46 mako132_ joined
TimToady I presume it's just infix:<as> these days, but as I said, we could have an 'is deep' variant to autogen the corresponding constructor. 17:47
nothingmuch the usefulness of class level as is that sometimes the class writer has no control of either the call site or the method def site
and wants to have "correct" semantics" 17:48
this can also simplify MMD
audreyt nothingmuch: I already agreed. now write a S12 patch?
nothingmuch in that 'as URI' could short circuit the MMD dispatch much earlier
17:48 fglock joined
nothingmuch i'm off to take that bath I was talking about 17:48
audreyt oh ok :)
nothingmuch i think i'll do it later 17:49
i am too blocked mentally to write something coherent
TimToady Actually, given MMD there's really no constructor/coercion distinction, so nevermind the 'is dep'
*deep
audreyt (and given structural unpacking in MMD, coercion can be 1)deep 2)automatic 3)fun) 17:50
TimToady agreed
nothingmuch oh wait:
how do parametrized types work with this?
TimToady brain melts 17:51
nothingmuch does 'as' affinity on the parameter type leak up to the parametrized type?
e.g. if Str as Num, then does Array[Str] as Array[Num] wwrt MMD?
audreyt don't think so. 17:52
nothingmuch how come, actually?
audreyt because of contravariance. Str as Num, but when you store a Num into a Array[Str]
no matter how Str-as-Num you are 17:53
TimToady But perhaps a generic 'as' could use that info to autogenerate the specific one.
audreyt it can affect only existing Array[Str] lements
but not affecting the STORE call
i.e. it'd only work on FETCH.
TimToady feels like a co/contravariant conflict of some sort. 17:54
fglock audreyt: I'll try to merge the Pugs::Grammar code now - I was distracted testing optimizations - none implemented, but there could be much improvement with inlining - 10x?
audreyt fglock: cool! 17:55
fglock: ingy just pulled together Filter::PMC
17:55 avar joined
audreyt v6.pm is trivial based on it 17:55
so instead of a .sh or Makefile.PL rules
Rule.pm -> Rule.pmc can happen reliably and automagically
17:55 FurnaceBoy joined
fglock audreyt: cool - will it use lrep in the beginning? or maybe pugs - is it possible already? 17:56
audreyt fglock: pugs is possible already -- parses much more stuff than lrep -- but you can switch
use v6-lrep;
use v6-pugs;
fglock audreyt: ws can send the pmc files to cpan, so people don't have to compile 17:57
audreyt fglock: that is the whole plan.
so, yes! :)
and that can really encourage people starting new cpan modules with Perl6
(or with Spiffy, or with some other source-filterish-technologies that will no longer be recompiled every run)
and because Filter::PMC can do nesting and ordering 17:58
the dreaded no-two-filters-at-once bug can at least be partially solved
TimToady audreyt: I wonder if you could STORE.wrap something that redispatches MMDly to proper constructor before calling "real" STORE...
fglock ruoso, pmurias: no problem with CPAN anymore (the initial plan was 100% perl6)
audreyt TimToady: but it needs the other direction's "as"
ok, I need to write a rant about the parrot container system. a sec :) 17:59
TimToady well, it doesn't seem all that important to nail down, so don't let me distract you.
audreyt TimToady: yeah, I'll just delegate it to the metaghost protocol 18:00
TimToady are those sugar ghosts?
or turtle ghosts?
or sugar turtle ghosts?
audreyt cute tutle ninja ghosts, more likely
turtle ninja, even
# news.bbc.co.uk/olmedia/705000/image...les300.jpg 18:01
TimToady what kind of a rant is that? :-)
audreyt TimToady: parrot has no idea of SV_readonly 18:02
18:02 Limbic_Region joined
audreyt TimToady: and hence, no distinction between Val and Ref 18:02
TimToady: and hence, "$x is readonly" is not practical
TimToady: and hence, reference semantics are broken; 1.id != 1.id, and you can modify how 1 works by calling 1++.
TimToady Yes, I saw your earlier rant. 18:03
thought maybe you had a better one. :)
audreyt I think the workaround is simply to have a "readonly" interface to PMCs in addition to the current "scalar"
fglock programs always have big TODO lists
18:04 GabrielVieira left
audreyt TimToady: the default f(*@_) thing 18:04
you can't say @_[0]=1, but can you say @_.push(3)?
(two conceptual levels)
TimToady I think @_.push probably works by default. The "readonly" distributes along with the *. 18:05
audreyt so we don't get "locked hash" style things by default 18:06
that's good, because it's far easier to explain :)
TimToady sub f(constant *@_) {...}
audreyt right. parrot doesn't do that either. hm 18:07
audreyt adds another rant item
TimToady constant %hash = (...) is a locked hash, presumably.
audreyt nod. 18:08
TimToady though maybe constant locks the values as well as keys.
audreyt so it's basically "is readonly is locked"
or some such.
thought parrot calls it "fixed" 18:09
TimToady okay to lock the values if you only change values thru existing ref.
audreyt er, "fixed_size", and maybe it reads better than "locked"
nod.
LeTo fixed_sized is still r/w but not extendible 18:10
TimToady I think constant should lock values too by default, since constant hashes are good -O fodder.
audreyt LeTo: yes, "readonly" and "fixed_size" are orthogonal concepts
and "constant" is them together
TimToady constant is more than that. It's a compile-time thing in P6. 18:11
audreyt nod.
though persumably:
sub f (*@_ is constant)
TimToady That's why it's a declarator unlike "readonly".
audreyt or is it more
sub f (*@_ is readonly is fixed_size) ?
TimToady nod 18:12
LeTo well, constant in Parrot terms means that storage is from a different object arena, which isn't scanned by GC - its for very long-lived objects
audreyt LeTo: rant sent 18:35
18:35 GabrielVieira joined
fglock GabrielVieira: hi! 18:36
svnbot6 r9416 | fglock++ | Pugs-Compiler-Rule: merged new tests, TODO
GabrielVieira GabrielVieira hi ;D 18:37
fglock where can i find some topics about what is being doing and where can i help (or try to)? :) 18:38
fglock GabrielVieira: want a commit bit? so you can start coding
GabrielVieira humm
yep 18:39
fglock where can i find?
audreyt GabrielVieira: your email addr? 18:40
fglock GabrielVieira: you can start by making a local copy of pugs with svn
GabrielVieira [email@hidden.address]
audreyt GabrielVieira: commit bit sent. add yourself to AUTHORS. welcome aboard!
GabrielVieira tkz ;) 18:41
audreyt dinner, bbiab &
GabrielVieira i'll check
fglock GabrielVieira: you can get the current p6 specification by installing Perl6::Bible from CPAN - then type p6bible ; if you are in debian, you can install pugs/unstable 18:43
18:43 nnunley joined
fglock or compile it yourself 18:43
GabrielVieira hum
i'll see here wicht one is the best for me 18:44
i thing ill install fbsd here just for it
:P
fglock also, talk with ruoso (Daniel Ruoso from cascavel-pm) - he is here too
GabrielVieira yeah 18:45
i thought about
:)
he is away right now
GabrielVieira is going to lunch right now.. so.. <- when i come back i'll get involved with it ;) 18:47
fglock GabrielVieira: ok!
18:49 fglock joined
pmurias fglock:is it ok to move emit.pl to perl5/Pugs-Compiler-Rule/lib/Pugs/Emitter/Perl5.pm? 18:52
18:53 Daveman joined
pasteling "bsb" at 192.115.25.249 pasted "S06 q:code scoping patch" (53 lines, 2.5K) at sial.org/pbot/16257 18:53
fglock pmurias: I'm trying to do that :) if you can help, plese do 18:55
I'll work in the parser side now
pmurias: wait! 18:56
pmurias: Compiler-Rule is the rule compiler - the emitter code is p6rule.pl 18:57
18:57 Daveman left
fglock pmurias: you can remove the non-emitter code in the first part of the file 18:57
pmurias: I'm moving that code to the grammar 18:58
18:58 FurnaceBoy_ joined
fglock pmurias: the Rule compiler will be completely separate from the p6 compiler - 2 reasons: 18:59
it's a simple language, can have a faster compiler;
and Rule can be used separately, in p5 programs
19:02 fglock joined
FurnaceBoy_ audreyt: I didn't know there were female ninja turtles. thx. 19:05
fglock where is v6-pugs.pm ? 19:09
audreyt fglock: it will bein lib/v6.pm 19:10
the -pugs is just a import switch 19:11
fglock audreyt: I need it to use the grammar...
audreyt fglock: the Makefile.PL should gen it for you for now
in this hour
19:12 amnesiac joined
audreyt p6i rant: www.nntp.perl.org/group/perl.perl6....nals/33240 19:12
fglock audreyt: will it be a aeparate cpan package? 19:13
audreyt fglock: v6.pm? very likely, yeah
fglock: but not really needed
fglock: the .pmc will contain the magic and v6 will not be needed on user site 19:14
you only need v6.pm if you modify the .pm file and need to regen pmc
in which case you probably want to have a pugs tree anyway
fglock audreyt: it didn't work for me: Error: Can't locate v6.pm in @INC
audreyt fglock: did you run MAkefile.PL? 19:15
which file is this?
fglock Pugs::Grammar::Rule (I didn't run makefile.PL - will retry)
audreyt: String found where operator expected at Makefile.PL line 19, near "name 'Pugs-Compiler-Rule'" 19:17
audreyt fglock: hm, upgrade your Module::Install?
actually nvm 19:18
I can fix that
fglock audreyt: thanks
audreyt fixed 19:19
svnbot6 r9418 | audreyt++ | * reuse M::I for P::C::R
fglock audreyt: "... Pugs-Compiler-Rule/inc/Module/Install.pm does not exist!" 19:23
audreyt sec 19:24
audreyt curses the author of M::I
tail re-cursing
try again 19:25
svnbot6 r9419 | audreyt++ | * really use Pugs's own M::I.
19:29 dduncan joined
TimToady obra: planetsix seems to have syndicated a huge mudball of your old bloggings... 19:29
fglock audreyt: no errors in Makefile.PL, but still no v6.pm
audreyt: how about just send v6.pm to CPAN? 19:30
= less hacks
audreyt fglock: you are doing ./lib/Pugs/Grammar/Rule.pm?
fglock: we'll certainly do that, but it doesn't quite help users who has no pugs /lrep on her machine
fglock: it worksforme 19:31
$ perl -Ilib t/01-iterator_engine.t
ok 1 - use Pugs::Runtime::Rule;
# etc
19:31 rgs joined
audreyt fglock: try that in perl5/Pugs-Compiler-Rule ? 19:31
fglock audreyt: I'm running make in perl5/Pugs-Compiler-Rule - it installs to C:\PXPerl\site\lib
audreyt: try t/02-
Limbic_Region fwiw - I am currently unable to compile on Win32 19:32
audreyt # Error: Undefined subroutine &ruleop::alternation called at lib/Pugs/Grammar/Rule.pm line 267.
Limbic_Region if not a known issue and someone wants to investigate - /msg me
Limbic_Region wanders back off
audreyt fglock: well, because it doesn't load Pugs::Runtime::Rule
and hence no ruleop::alternation 19:33
if you say
use_ok( 'Pugs::Runtime::Rule' );
use_ok( 'Pugs::Grammar::Rule' );
then it works
solution: compile "use perl5:" statements?
fglock audreyt: one sec
audreyt and put a "use perl5:Pugs::Runtime::Rule" in Pugs::Grammar::Rule body perhaps
or just forget about "use perl5:" now and compile straight "use" 19:34
pmurias fglock:is it ok to move emit.pl to perl5/Pugs-Compiler-Rule/lib/Pugs/Emitter/Perl5.pm?
fglock audreyt: my problem is "can't find v6"
pmurias: you mean p6rule.pl - the rule emitter 19:35
pmurias: and Pugs/Emitter/Rule/Perl5 - we are not moving the p6 compiler yet, only rules 19:36
audreyt fglock: I wonder why "can't find v6" happens - can you paste the error here? 19:37
Limbic_Region: nopaste the log?
fglock audreyt: maybe my perl5 can't find .pmc files?
what's the nopaste cmd? 19:38
pasteling "fglock" at 200.102.180.83 pasted "v6 error" (13 lines, 649B) at sial.org/pbot/16258 19:41
19:41 nothingmuch joined
audreyt mm 19:44
fglock: do you have ib/Pugs/Grammar/Rule.pmc ?
lib
and is its modification time newer than Rule.pm? 19:45
fglock audreyt: will 'touch' it...
audreyt: works! - Makefile.PL will need to touch Rule.pmc when installing from CPAN 19:47
audreyt ok 19:48
Limbic_Region perlbot nopaste
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
pasteling "Limbic_Region" at 24.35.57.240 pasted "snippet of compile failures on Win32" (87 lines, 4.1K) at sial.org/pbot/16259
audreyt fglock: I hacked that together 19:52
svnbot6 r9420 | audreyt++ | * Makefile.PL for Pugs::Compiler::Rule should always touch pmc
r9419 | audreyt++ | * really use Pugs's own M::I.
audreyt (touching)
Limbic_Region: we just fixed this
should be fine now
pmurias fglock:where should match::{get,str} live?
fglock pmurias: Match will be a separate class (needs to be written) 19:54
19:56 fglock left, fglock joined
Limbic_Region audreyt - good to know 19:59
will svk pull and try again
fwiw - I am only lurking here so apologies if I should have known that 20:00
audreyt np, no apologies needed :)
fglock audreyt: perl5 will have ::Match, ::Rule, ::Str classes? 20:01
audreyt fglock: Pugs::Runtime will provide them.
using the MOP to map them to p5 packages 20:02
/methods/etc
nothingmuch Darjeeling FTGFOP 1, 294 + 298
eep
pretty slick: svn.code-host.net/project.html 20:03
fglock audreyt: the rule match will be: Pugs::??
Limbic_Region Cannot write blib/lib/Pugs/Grammar/Rule.pmc:Permission denied at C:/Perl/lib/ExtUtils/Command.pm line 127 20:05
still blowing up audreyt
audreyt fglock: "the rule match" ?
Limbic_Region unless of course I should do a make realclean and start over (again)
audreyt Limbic_Region: nono
a sec, fixing that
Limbic_Region heh - thanks
obra TimToady: this is a bug in use.perl/planetplanet integration :/ planet perl gets justin mason's old postings every week :/
Limbic_Region but this machine has resources (not like the dinosaur at work) 20:06
fglock what class will be the object returned by running the object returned by Pugs::Emit::Rule :)
audreyt fglock: Match. I think.
(yes, simply "Match")
fglock ok
svnbot6 r9421 | audreyt++ | * for win32, try to chmod 644 on .pmc 20:10
buu Hey Limbic_Region 20:11
fglock pmurias: are you ok if I commit the version of Emitter::Rule::Perl5 I was working on? 20:12
20:13 fglock joined
Limbic_Region hey buu 20:14
pmurias fglock:np
svnbot6 r9422 | fglock++ | Pugs::Emitter::Rule::Perl5 - work in progress 20:19
nothingmuch fglock: i have an AST brain dump for you 20:20
it's more of an "operation" level than a parse level one 20:21
it's after desugaring
and it's more consistified
but it's still before folding, linking, and that stuff
and it's definately not an intermediate language (loops etc are not "flattenned")
fglock nothingmuch: I've seen the paste 20:22
nothingmuch okay 20:23
it's not a very complete example
and I think it is a *tad* too dumb down
e.g. functional calls should be different from object calls
even though they are "really" the same thing under the surface (prototype multimethod crap ;-)
but i think it's a balanced place to aim to as the step right after parsing
note, ofcourse that Immediate is *immediately* (during the construction of the *PARSE* tree) translated in to this tree, and then compiled and executed into an AST to be put in that expression 20:24
also, i should note that we need a Container note 20:25
node
or something like that
sinc Syn('&fact') reduces to it's container
and then only becomes the value right before dispatch if the parameter it's being bound to is not rw
fglock nothingmuch: the current work is to modularize the compiler a lot, so that we can test these new ideas - you'll be able to write this into a parl module and try it out :) 20:27
nothingmuch i won't today
or this week
maybe this coming weekend
(my weekend is fri,sat)
so i just want to toss around food for thought
on the other hand, I am quitting my job
so in one month from tomorrow I should have a lot of free time
fglock s/parl/perl/ 20:28
nothingmuch so, until i can hop on this wagon i still want you to casually review my ideas, or at least pretend you did ;-)
fglock nothingmuch: in one month I'll completely without time :) YAPC, studies, work...
nothingmuch so you should work on my ideas now and then i will replace you later ;-) 20:29
fglock nothingmuch: sure
nothingmuch seriously though, this is just food for thought
fglock nothingmuch: I'm working in the rule language as a separate compiler - audreyt and stevan go on with p6 - porting pugs to p6/p5 20:30
and gaal, and TimToady, and... 20:31
nothingmuch: we've discussed how MOP could be a plug-in-like thing - that's one example of how things can be modularized 20:32
Steve_p seen TimToady?
Hmmm...no infobot :-/ 20:33
Limbic_Region [13:13] <TimToady> nod
my local time is now 15:34
ooops, missed something more recent 20:34
[14:29] <TimToady> obra: planetsix seems to have syndicated a huge mudball of your old bloggings...
Steve_p OK, so a few hours ago
20:35 K joined
Steve_p Ahh...I may have it figured out myself 20:36
fglock nothingmuch: ping?
Juerd rafl: You also left a cacert form in my car. You lost it almost 7 years too early ;) 20:39
TimToady Limbic_Region: you rang? 20:41
s/Limbic_Region/Steve_p/
Steve_p TimToady, yes, I found something in the MAD code, but I've got it figured out. 20:42
Just testing now 20:43
TimToady Steve_p: glad one of us has it figured out. :-) 20:44
Steve_p It was just a variable initialization out of place. Hopefully, I'm not breaking something else :-) 20:46
Steve_p would rather no add days to Nicholas's sentence^Wgrant time 20:47
;)
TimToady All sorts of things are out of place in MAD. It's usually a hard choice between fixing it up in C or letting nomad move it back to the "right" place.
But I assume you mean a C variable init out of place...
Steve_p Yes, a C variable. 20:49
TimToady The whole MAD thing was predicated on the notion of being minimally invasive, so if you have a choice between breaking the mainline or making nomad's life harder, definitely make nomad's life harder.
But occasionally you can refactor and get a win-win. 20:51
20:51 proefgb joined
Steve_p No worries, it's just a little tweak :-) 20:52
TimToady That sound's like the kind of quote you put at the front of the next chapter in that famous novel "Jenga Book". 20:53
s/'//
TimToady wanders off to take an old-fogey-style nap. 20:56
Steve_p heh 20:58
21:03 FurnaceBoy_ is now known as FB|afk
svnbot6 r9423 | fglock++ | Pugs/Grammar/Rule - resolved dependencies 21:10
fglock pmurias: re "return_block_hack" - any ideas how to fix it? 21:14
audreyt: ping
audreyt fglock: pong 21:21
fglock audreyt: Pugs::Grammar::Rule is mostly in place, I'm trying to fix some hacks, but I'm out of ideas 21:23
audreyt: how does the return block "returns"? 21:24
audreyt: how does it has access to the whole match?
audreyt: it's currently a hack - it gets compiled after the whole rule is compiled 21:25
audreyt: that is, the return is moved outside
audreyt a sec 21:27
it doesn't have acess to whole match 21:28
it has access to a partially constructed match obj
with the succeeding match before it
fglock audreyt: ok - I've got an idea how to do this 21:35
audreyt cool! 21:37
fglock audreyt: I'll make the ruleop abort, returning the subref in a flag - the subref will be called in the root op, which knows what the match is 21:39
audreyt: this way we can have multiple returns (is it allowed?) 21:40
rafl Juerd: D'OH.. how many? 21:45
fglock audreyt: the 'abort' documentation says: "the match was stopped by a { return } or a fail()" :) 21:46
audreyt: I wrote that, and then forgot it
Juerd rafl: 1 21:57
rafl: Sven's 21:58
rafl Juerd: That's useless anyway. I didn't assure him yet and he already has 150 points.
Juerd rafl: Shall I shred it or send it to you? 21:59
audreyt fglock: multiple returns makes as much sense as multiple returns inside a sub
i.e. makes no sense :)
rafl Juerd: Well. If it's not a problem for you i'd like to have it back, but you don't need to send it to me right now. 22:01
Juerd: Maybe if you're sending something else anyway (PBP, gadgets, ..) or when we meet the next time.
Juerd: Easter Hegg maybe? 22:02
fglock audreyt: what if you have an alternation, and for each case you return a different thing? (just a thought)
Juerd rafl: What's Easter Hegg? 22:03
rafl Juerd: A hacker meeting 22:04
Juerd: eh.cngw.org/
audreyt fglock: then you return as soon as one alternation matches 22:05
Juerd rafl: I've had enough of far travelling for a while, tbh
audreyt and maybe you use : for backtrack control
nothingmuch temp/local grudge: 22:06
local $i--;
someone find a nice way to make that DWIM but not be "dirty"
good night
audreyt fglock: perlcabal.org/~gaal/peek/slide37.html
Juerd nothingmuch: (local $i)-- :)
Though temp.
nothingmuch in perl 5 it won't work
Juerd Sure, but that's because local sucks ;)
nothingmuch did semantics change such that bare 'local $x' retains the outer value? 22:07
audreyt stevan: all PIL^N native types become PMCs now
Juerd I hope so.
Though temp.
rafl Juerd: OK. How about the ultrabay?
Juerd rafl: Haven't ordered yet. Will confirm the price to you first.
rafl: Please remind me Monday
rafl: What kind of drive do you want in it?
audreyt I used S11 names, so Seq is now Tuple, and I call Map "Mapping"
fglock audreyt: :) - I didn't mean to return twice
rafl Juerd: OK, thanks.
Juerd: What are the possibilities?
audreyt stevan: tomorrow I'll see if Bootstrap.pil can be made to run on parrot 22:08
Juerd rafl: CDRW/DVD, CDRW/DVDRW, battery
rafl: Or nothing :)
nothingmuch good night
Juerd: if you feel like being a darling check if the synopses say that? if not i'll try to remember tomorrow ;-) 22:09
Juerd nothingmuch: No, sorry.
nothingmuch okies
nothingmuch writes todo
rafl Juerd: I'd like to know the prices for DVD and DVDRW.
Juerd I'm already in hypermultitasking mode :( 22:10
rafl Juerd: Is the battery stuff useful? How long does it run with a battery that fits into a cd drive slot?
Juerd rafl: Not long.
And it's very expensive.
rafl Juerd: OK, then it's not useful.
Juerd The "nothing" option is cheapest.
But puts very little extra value in having the ultrabay :) 22:11
OTOH, if you're going to use it only on the road, I'd suggest getting a USB optical drive instead. It's much cheaper.
The ultrabay is mostly valuable for its dockingstation-ness
rafl Juerd: Thought so.
Juerd: Anyway. Knowing the prices first would be nice. 22:12
Juerd Oh, I should still have the order confirmation for mine
Let's see
160 for ultrabay, 210 for cdrw/dvd
Excl. VAT 22:13
22:19 SamB joined 22:28 FB|afk is now known as FurnaceBoy 22:32 uzair_ joined, hexmode` joined
audreyt I'd like to stay up for longer but can't anymore :) 22:34
Juerd Good night, audreyt
audreyt see you tomorrow :) &
22:35 uzair_ is now known as uzair
audreyt ...and then I spent 30mins journaling ;) 23:01
sleep for real &
obra hah 23:02
sleep well, audrey
audreyt LeTo++ ingy++ # doing all the hard work on PMCs for Pugs and PMCs for Pugs :) 23:03
23:12 larsen joined 23:16 justatheory joined 23:19 nnunley joined 23:21 Nouk joined 23:57 |mjk| joined

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!