Perl 6 - the future is here, just unevenly distributed

IRC log for #perl6, 2006-01-28

Perl 6 | Reference Documentation | Rakudo

| Channels | #perl6 index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:00 putter I dont suggest the current approach cease working and being used.  simply that the alternative become permissible.
00:01 putter ah well.  just thought I'd raise the idea.
00:02 theorbtwo IIRC, ghc allows you to s|/|.| as much or as little as you like.
00:02 putter found the mono directory listing _vastly_ clearer than its p5 equivalent would have been.
00:02 theorbtwo Er, swap that.
00:02 putter oo, intriguing.  :)
00:07 * putter ponders whether one could do a  macro infix:<.> ($lhs,$rhs) { $lhs ~~ /^[A-Z]w+$/ && .. ?? "{$lhs}::{$rhs}" !! "$lhs.$rhs" }   so   class Foo.Bar.Hee { ... } worked in p6 too...
00:12 putter Oh, I'll also flag the issue I tripped over several times, of string returning macros wanting to have some control over when the returned string is macro processed.  generally you want it to be, but that precludes defining a macro which overlays a construct defined elsewhere, to provide an ancillary service.
00:14 putter Juerd: there's no reason that macro couldn't be infix:</> instead of infix:<.> ;)
00:15 putter p6 is soo going to need coding standards...
00:15 theorbtwo putter: Yes, there is; it'd conflict /badly/ with your basic ability to divide.
00:17 putter ah, but how often do you divide with a camel-case function on the rhs, and a camel-case something on the left?
00:18 wolverian next you'll want to enforce camel case package names..
00:19 putter na.  just if your going to   use Esthetics::Juerd::SlashSeparator;  you have to follow its documented constraints.
00:21 putter though if there were a hook in the parser for a parse fail, maybe you could wait till it failed, thus not conflicting with anything preexisting, zip in, patch the ast and parser state, and have the parser continue... :-P
00:23 putter sigh.  let's not let things go so long before a release again.  this debugging exercise is a real pain.
00:31 jvoorhis joined perl6
00:31 jvoorhis hi
00:33 rafl Hello Juerd
00:33 rafl Err..
00:33 rafl Hello jvoorhis
00:33 jvoorhis :)
00:33 jvoorhis i managed to build parrot and pugs, but i'm not sure what is going on now
00:34 jvoorhis should i be able to define classes interactively in pugs?
00:34 audreyt putter: totally agreed. it seems that I won't have this 3-month life crisis repeated anytime soon :)
00:34 putter lol ;)
00:34 audreyt joepurl: yes. run "pugs" and define classes as usual
00:34 jvoorhis or is that as of yet unimplemented, or did i build it wrong? :)
00:35 jvoorhis http://rafb.net/paste/results/JWVOMi25.html
00:35 audreyt $ pugs
00:35 audreyt pugs> class Foo { has $.bar };
00:35 audreyt pugs> Foo.new(:bar(3)).bar
00:35 audreyt 3
00:35 jvoorhis hmmmm
00:35 audreyt jvoorhis: oh. we use readline and it's line-delimiated
00:36 audreyt instead of waiting until a semicolon
00:36 audreyt that can change
00:36 jvoorhis ah
00:36 jvoorhis audreyt: so as of right now, i have to fit it all on one line?
00:37 audreyt jvoorhis: yeah. better ideas welcome
00:37 audreyt detect unfinished opening braces / parsefil
00:37 jvoorhis audreyt: i've never implemented anything like an interactive interpreter
00:37 audreyt and wait until closing brace?
00:37 jvoorhis but i'd like it to work like ruby's irb
00:37 jvoorhis yeah
00:37 jvoorhis exactly
00:38 jvoorhis i think python works similarly too - it just waits for... err... fewer spaces :)
00:38 audreyt heh
00:38 audreyt not really
00:38 audreyt irb is semantic
00:38 audreyt if you say
00:38 audreyt 1+
00:38 audreyt then it waits for you to continue
00:38 jvoorhis yeah that is true
00:38 audreyt while in py
00:38 audreyt 1+
00:38 jvoorhis it's statement or scope oriented
00:38 audreyt is a syntax error
00:38 jvoorhis :)
00:39 jvoorhis i haven't done enough python to know that for a fact
00:39 audreyt naively, it seems irb is vastly superior
00:39 * jvoorhis spends *lots* of time in irb
00:39 wolverian is it possible to know that a statement is not terminated in perl6?
00:39 audreyt wolverian: sure
00:39 audreyt you parse it and get a parsefail ;)
00:39 putter pilrun used a blank line
00:39 wolverian audreyt, I guess. ;)
00:40 putter python's approach is a pain if you are trying to cut and paste
00:40 audreyt indeed
00:40 wolverian that opens up the possibility of side effects, though
00:40 audreyt python is not a very cutandpastable language
00:40 jvoorhis yeah
00:40 putter (every line prefixed with "...")
00:40 jvoorhis heheh
00:40 theorbtwo ...and accidentally finishing early, but we can call that user error.
00:40 jvoorhis isn't every statement in perl ended by a ;?
00:40 jvoorhis or is the semi optional
00:41 audreyt it's optional on the last statement
00:41 jvoorhis it's been a long time since i've used perl
00:41 wolverian it's also optional for statement-level functions
00:41 jvoorhis ok
00:41 jvoorhis that makes sense
00:41 jvoorhis thank you
00:41 wolverian i.e. for, if, etc.
00:41 audreyt jvoorhis: ok, I think we go the irb route. I think.
00:42 jvoorhis audreyt: i don't know how well that approach maps to perl, but i find it very usable
00:42 putter or use a common-lisp or scheme as a prototype ;)
00:42 jvoorhis also
00:42 lypanov audreyt: irb has its own custom parser. it messes up on occasions
00:42 jvoorhis ruby has a line continuation character
00:42 jvoorhis i assume perl does as well?
00:43 jvoorhis example:
00:43 audreyt lypanov: I think we'll use the same parser
00:43 jvoorhis 2 + \
00:43 jvoorhis 2
00:43 wolverian audreyt, 1+ is a syntax error in a python file, too, so are you sure it's not semantic?
00:43 audreyt jvoorhis: perl has no layout
00:43 audreyt wolverian: the idea is in irb you can do
00:43 audreyt irb(main):001:0> 1 +
00:43 audreyt irb(main):002:0* 2
00:43 audreyt => 3
00:43 jvoorhis ah, ok
00:43 * lypanov really loves the irb behaviour
00:43 * jvoorhis does too
00:43 audreyt and I've got it implemented
00:43 audreyt fixing some corner cases.
00:44 audreyt (brb)
00:44 jvoorhis :)
00:44 wolverian audreyt, yes, at least python's repl does the same thing as python itself. it'd be weird if it was valid in the repl but an error in a .py file
00:44 lypanov yeah
00:44 jvoorhis so it seems there are other rubyists here
00:44 lypanov ruby's parser is a tad messed up
00:44 jvoorhis are roles similar to modules in ruby?
00:44 Blicero_ joined perl6
00:45 * lypanov isn't sure what roles are alas
00:45 wolverian jvoorhis, http://dev.perl.org/perl6/doc/design/syn/S12.html if you haven't read it yet
00:45 jvoorhis wolverian: i've read most of those, but i haven't actually *used* the language yet
00:46 jvoorhis ah
00:46 jvoorhis so it is interface + implementation, but you cannot instantiate a role
00:46 Juerd It's impossible to have a nice philosophical discussion with a mathematician.
00:46 jvoorhis can you add a role to a class at runtime?
00:47 wolverian Juerd, there are nice philosophical discussions?
00:47 wolverian jvoorhis, yes.
00:47 Juerd wolverian: Yes. Are you a mathematician? :)
00:47 wolverian er. to an object, yes. I don't know about classes
00:47 Juerd Because if so: then no.
00:47 wolverian Juerd, I'm half mathematician.
00:47 Juerd Then you can only get half-nice philosophical discussions :)
00:47 jvoorhis wolverian: ah ok
00:47 wolverian I can switch off that part at will.
00:48 Juerd (I'm trying to phrase this in a way a mathematician, even a half one, will understand :P)
00:48 Juerd wolverian: Then you can get nice philosophical discussions :)
00:48 jvoorhis i've been using ruby for nearly a year now and finally comfortable with metaprogramming ruby, and perl6 has been arousing my curiousity
00:48 wolverian mm, arousal..
00:48 Juerd The key point is to accept that it's impossible for every definitino to be right.
00:49 wolverian I can smell it
00:49 Juerd definition
00:49 wolverian Juerd, mathematicians also tend to walk out of the room if someone uses philosophical induction
00:49 Juerd And if you can mathematically prove something, that may be a side-effect of some other definition also being wrong.
00:50 Juerd wolverian: Yep
00:52 putter the sbcl repl looks like:
00:53 putter * (+ 1
00:53 putter 2)
00:53 putter 3
00:53 putter note that you can now cut and paste the (+ ...2).  something you can't do with the irb repl.
00:54 * Odin- <3 lispy REPL.
00:54 Odin- P6 *need* something like that. :)
00:54 putter :)
00:57 jvoorhis Odin-: perl syntax isn't quite sexp ;)
00:57 Odin- jvoorhis: That's a temporary condition. ;)
00:57 jvoorhis oh?
00:57 putter it looks like the pil2run backend is completely dead at this point.  but it had a similar repl.  no cruft on secondary lines, and input terminated by a blank line.  easy enough to write your own repl in p6.
00:58 putter lol
00:59 dduncan okay, my smoke of r8819 haskell/darwin is now uploaded
00:59 dduncan says 91% okay
01:00 putter Odin-: there was recently quite a bit of blog discussion (which I couldnt refind when I went looking for it) on common-lisp vs ruby.  one discussion point ended up being the contrast between lisp macros and ruby parse_tree based ones.  sexps being much clearer than explicit ast tags.  given that p6 with have ast macros,
01:01 putter I was wondering if the ast might be abstracted over to get sexp-y like macro writing... ?
01:02 putter dduncan: ok, that's what I got too.  I'm running a smoke-js now.  then maybe a smoke-pir.
01:02 audreyt jvoorhis: how would you like to be listed in the AUTHORS file?
01:02 audreyt (name, etc)
01:03 audreyt jvoorhis: alternately, if you let me know of your email, I can send you a commit bit now so you can add it in yourself :)
01:03 jvoorhis what did i write? :)
01:03 Odin- putter: Hm. Maybe I read it wrong, but when I read the macro parts of the spec, it sounds quite a bit like CL-ish macros.
01:03 audreyt "These people have either submitted
01:03 audreyt patches or suggestions, or their bug reports or comments have inspired"
01:03 jvoorhis aye
01:03 jvoorhis [email@hidden.address]
01:03 dduncan audreyt, is now the start of your day, or did you get up in the middle of sleep?
01:04 jvoorhis full name is Jeremy Voorhis, and that would suit me fine
01:04 jvoorhis and thanks! you're trying to suck me into the community aren't you
01:04 Southen joined perl6
01:04 mlh_ joined perl6
01:04 dduncan no, its not suck, its blow
01:04 Odin- putter: In any case, macros are supposed to be able to return an arbitrary string which is then parsed. There's no possible way CL macros can be more powerful than P6...
01:04 audreyt jvoorhis: well, it's nothing personal
01:05 dduncan there's no such thing as sucking
01:05 Odin- Ease-of-use may be an issue, though...
01:05 audreyt lypanov: you got a commit bit yet? :)
01:06 jvoorhis audreyt: understood - i've been learning about languages lately and i might just spend a good bit of time with p6
01:06 jvoorhis the rules and grammars are especially exciting to me right now
01:06 putter Odin-: but (App (Fun (Var (Varname "f"))) (Args [(Lit (Val (Int 3))]) is a lot harder to deal with than (f 3).  and that wasnt even the full ast. ;)
01:07 putter so could one deal with the ast as if it simply said f(3).  somehow...
01:07 Odin- putter: Point. But you could do a transformation, couldn't you?
01:07 Odin- putter: But ... erh ... couldn't you pass a grammar match object?
01:08 Odin- Operate on that in the macro ... return it?
01:08 putter or macros set you up for a ~~ structure match or something
01:08 * Odin- not very much in on details, to note.
01:09 putter you mean using a grammar's underspecced ability to match arbitrary objects?  that's a nice idea
01:09 audreyt ...and why am I still implementing features when I should be triaging bugs? ;)
01:09 putter lol
01:09 audreyt it's implemented though.
01:09 audreyt committing
01:10 putter the lack a of a pithy catch phrase?  "pre pre flight"?
01:11 Blicero_ joined perl6
01:11 Odin- putter: Well, since we're talking about the miserably underspecced, but quite clearly planned, macro feature of P6, I don't see why we can't rely on other underspecced parts. ;)
01:12 Odin- Basically the only place which says something less vague than "we'll have macros" is a tiny part of Apoc 6, AFAIK... :/
01:12 svnbot6 r8820 | audreyt++ | * First cut a irb-style line continuation in the interactive
01:12 svnbot6 r8820 | audreyt++ |   Pugs shell, by waiting for more lines whenever we encounter a
01:12 svnbot6 r8820 | audreyt++ |   parsefail.  Suggested by jvoorhis.
01:13 putter indeed.  pause.  this seems like one instance of a general problem.  you have a big hairy graph of data structures and/or objects, and you want to be able to match against it, _and edit/add to it_, using much less hairy code.  how can one do this?
01:13 * theorbtwo thinks audreyt meant jvoorhis++
01:13 audreyt jvoorhis++
01:13 theorbtwo putter: We need to leave something for CPAN modules.
01:13 svnbot6 r8821 | audreyt++ | * add Jemermy "jvoorhis" Voorhis to AUTHORS.
01:14 putter lol
01:14 audreyt pugs> "Hello,
01:14 audreyt ....> world!"
01:14 audreyt "Hello,\nworld!"
01:14 audreyt irb changes the prompt based on what's expected
01:14 audreyt that's trivial to do given parsec's "expected" token group
01:14 wolverian hmm, what about 'say \n "foo";'
01:14 audreyt but I think I have to stop implementing them now :)
01:15 audreyt wolverian: same as irb
01:15 audreyt "puts" and "say" both runs asis
01:15 audreyt instead of waiting for stuff
01:15 wolverian right. if the line is not a parsefail, it's a statement. otherwise we'd need to wait for ;
01:16 wolverian thanks for the awesome feature :) audreyt++
01:16 theorbtwo ...and say() is perfectly valid (though somewhat silly) code.
01:16 jvoorhis audreyt++ for listening
01:16 audreyt it's the idea that counts... it's just a 10 line patch :)
01:16 jvoorhis sure
01:16 jvoorhis well, i do spend a lot of time in ruby's repl
01:16 theorbtwo Hm, if you write something that's simply invalid code, and press enter, how can you ever get back to a real promopt?
01:17 jvoorhis so i might be able to make some usability suggestions if i feel any more friction
01:17 audreyt theorbtwo: ^D
01:17 theorbtwo WFM.
01:17 theorbtwo (Should be documented, but that can come much later.)
01:17 audreyt :)
01:20 audreyt putter: thanks for a productive shift :)
01:20 * audreyt praises putter's alt ctrl shift
01:20 putter *groan*
01:20 theorbtwo Oy.
01:20 theorbtwo putter++
01:22 jvoorhis what does alt ctrl shift do
01:23 putter waits for the next key...
01:23 putter tewk++
01:23 theorbtwo jvoorhis: Controls when the normal control shift is sleeping, of course.
01:24 audreyt jvoorhis: I live in GMT +800, so when I'm about to sleep, putter takes the "night" shift (from my TZ point of view) to work on release engineering, so he assumes alternate control
01:24 Amnesiac joined perl6
01:24 jvoorhis aye
01:24 audreyt you may have heard that pugs is optimized for pun :)
01:24 * jvoorhis updates and goes golfing :P
01:25 nnunley putter: Slate's prototype system and design is still getting shaken out, so it's still hard to latch on to where to develop from -- I know that most symbols are bound to the lobby, like self, but...  The rest is different enough from (but still close to) smalltalk and self to be a pain.
01:25 * nnunley groans at audreyt.  "On that note."
01:26 audreyt jvoorhis: oh btw, commit bit sent. welcome aboard!
01:26 jvoorhis thank you
01:26 audreyt now I need to get ready for the new year's eve family gathering and stuff. be back in a few hours
01:26 jvoorhis ah
01:26 * jvoorhis gets back to ruby hacking until work is over
01:27 theorbtwo Have fun, Audrey.
01:27 putter little red envelopes?
01:28 Odin- So, putter is Commander to audreyt's Captain?
01:28 putter enjoy &
01:28 Odin- Or are they at higher ranks? :>
01:28 audreyt yeah, and as I'm still at the receiving end, red envelopes are good news
01:28 audreyt :) *wave* &
01:28 * putter is the clown that keeps keeps the crowd from becoming restless during intermission
01:28 audreyt Odin-: my rank is meta-control-shift, I think
01:29 putter old lispm keys... Alt-Control-Shift-Meta-Hyper
01:29 Odin- audreyt: Owh. That means I can't use "aye aye cap'n". :(
01:29 putter talk like a pirate day is but once a year
01:30 Odin- putter: Not if you're a naval officer, I understand... :D
01:30 Odin- But I digress.
01:30 putter lol
01:31 putter oh, and Super.
01:31 jvoorhis actually... no more ruby for today
01:31 jvoorhis it's the weekend now ;)
01:34 putter weekend == python?
01:34 Limbic_Region joined perl6
01:37 theorbtwo G'night, all.
01:38 Amnesiac joined perl6
01:41 putter Hey, emacs and X apparently still support Super Hyper.  And you can use your "Windows" key as Super.  
01:42 putter good night theorbtwo &
01:42 putter http://world.std.com/~jdostale/kbd/SpaceCadet.html
01:45 jvoorhis putter: not python ;)
01:45 jvoorhis possibly pugs
01:45 jvoorhis not rebuilding pugs until i get back to the apt as it is a little intense
01:47 putter ah :)
01:49 putter nnunley: thanks.  # slate feedback
01:49 Blicero_ left perl6
01:51 putter nnunley: it shouldnt be too hard to create an (unspecced) postmultifix to permit smalltalk style function calls.
01:52 putter sub multifix:<foo bar hee> ($a,$b,$c) {42}     say foo 3 bar 4 hee 5;
01:54 putter the grammar engine is apparently specced as a  top-down/bottom-up/top-down-defining-tokens  sandwitch
02:08 Cryptic_K joined perl6
02:12 putter has anyone been successful running embedded parrot lately?
02:18 drbean joined perl6
02:20 beppu joined perl6
02:20 raar joined perl6
03:05 huaehu joined perl6
03:18 renormalist joined perl6
03:20 renormalist Someone awake to help me convert a little perl5 script into pugs/perl6?
03:23 stocke2 joined perl6
03:23 stocke2 hello
03:25 huaehu when will perl6 be released? before 2010? :P
03:26 stocke2 does anyone know if perl 6 is going to clean up OO in perl?
03:28 dduncan it sure is!
03:29 stocke2 would be nice is kind of a confusing hack right now ;)
03:30 dduncan it should be better in the subsequent release 6.28.0 ... the refactor onto PIL should make objects and other things much cleaner than the current pugs
03:32 stocke2 i wonder when perl 6 will be done
03:33 dduncan define "done"
03:33 dduncan perl 6 will always be done, for different definitions of done
03:35 stocke2 lol
03:35 stocke2 kinda like perl 5.8 is done
03:36 stocke2 usable, to the point they start working on the next version
03:37 dduncan well, 6.28.0, perhaps 1 month away, would be a significant milestone, when I see that it can start being seriously used
03:37 stocke2 really?
03:38 dduncan yes
03:38 stocke2 once it is ready for serious use we need new editions of the oreilly books
03:38 dduncan that would use the new PIL2 as its default foundation, which is much better for building on
03:38 huaehu is there a beta version of perl6? or something to start writting/testing perl6 code?
03:38 dduncan they are currently coming out on an annual basis
03:38 dduncan the perl 6 one that is
03:39 dduncan I think
04:30 azuroth joined perl6
04:38 Amnesiac joined perl6
04:57 putter fyi, there are now r8818 smokes of pugs, js, and pir.
04:57 putter good night &
04:57 dduncan good night
05:17 dduncan anyone here?
05:23 azuroth I am
05:23 azuroth did you need me here to say good night to you? :-p
05:27 dduncan actually, my "good night" was a reply to putter's
05:27 dduncan I will be staying up for a few more hours
05:28 dduncan I was going to ask the opinion of who was here ...
05:28 dduncan I'm going through failing ext/ now with the goal to making them not fail their tests
05:28 dduncan and not die
05:29 dduncan however, in the short term, one way to do this is to remove functionality from them, such as converting private methods into public ones
05:29 dduncan does that sound fine to you?
05:30 dduncan separately, does 'catfile' sound like something that should be built into perl?
05:30 dduncan Config::Tiny's test invokes it, but Config::Tiny doesn't declare it
05:30 azuroth hmm. I don't really know. I think private -> public sounds fine
05:32 azuroth if there was anyone else around, I'm sure they would be more helpful
05:37 dduncan in the case of Config::Tiny, I did a diff between the 6.2.10 distro and the current release, which would have passed all tests, and the current version, showing no changes of significance
05:37 dduncan I think
05:38 dduncan which suggests that catfile was a pugs function before, and now isn't
05:38 dduncan because a test containing it worked before
05:41 dduncan okay, cat_file was declared in ext/File-Spec, and that distro is currently failing some tests ... so fixing File-Spec may fix some other ext/ modules too
05:41 dduncan like Config::Tiny
05:43 dduncan and a quick search shows almost 30 files use 'catfile' ... or did in 6.2.10
05:45 azuroth hmm, okay
05:46 dduncan also the case in the repository
05:46 azuroth what does catfile do?
05:46 dduncan so I'm now focusing on ext/File-Spec
05:47 dduncan I think it just reads in a file, like a combination of open/slurp/close
05:47 dduncan its name derives from the 'cat' unix utility, which was used to pump a disk file into a piped operation
05:48 dduncan short for 'concatenate'
05:48 azuroth oh, cool
05:49 dduncan okay, File-Spec is choking on the absense of &devnull
05:49 dduncan and &cwd
05:51 dduncan I see the first test doesn't have the use_ok workaround that some other tests do, so I'll add it
05:51 dduncan oh wait
05:52 dduncan was looking at the 6.2.10 version ...
05:56 dduncan I think I found the problem
05:56 azuroth excellent :-)
05:57 dduncan essentially, File::Spec itself requires either ::Win32 or ::Unix
05:57 dduncan those latter modules declare and export functions that code using File::Spec wants to use
05:58 dduncan however, the way these are used, said functions are only exported into the File::Spec module namespace itself, not the namespace of the using code, hence the user can't find it
05:58 dduncan it seems that the module was relying on a bug in Pugs that no longer exists, where export made things visible everywhere, rather than just to the user
05:59 dduncan so the fix is to add explicit wrapper functions into File::Spec itself
06:00 dduncan I also notice this comment: # The "require" lines above is deliberately not part of the File::Spec package, because we'd like to import on behalf of the caller. Again, a horrible hack.
06:00 dduncan I'm led to wonder if such a thing ever worked in the perl 5 version
06:00 dduncan or is supposed to work in the perl 6 one
06:07 azuroth ahh
06:11 dduncan I see the perl 5 version did a subclass arrangement, dynamically
06:28 azuroth very off topic, but in terms of database design... say you were logging bunches of chatrooms to databases, would you have a different table for each room..?
06:32 pdcawley joined perl6
06:41 dduncan probably not
06:41 dduncan assuming that each room is handled the same way by your code
06:41 dduncan you just have a table column containing a room id or name to say what room you're dealing with, and the other fields are per room
06:43 dduncan this is conceptually like you have a class for handling a chat room, and you have a separate object for each room
06:44 dduncan you don't make a separate class for each room
06:44 dduncan azuroth
06:44 azuroth hmm, that's a good point
06:45 dduncan generally, you make decisions about separate vs combined tables like you do separate vs combined classes
06:47 azuroth that's actually a lot cooler. I could have the user as an id pointing to the jabber ids table
06:49 Khisanth joined perl6
06:50 azuroth oh, and I can use prepared statements properly
06:55 dduncan yes
06:55 dduncan parameters in prepared statements can only go where a literal can go
06:56 azuroth thanks :-)
06:56 dduncan ah, and File::Spec now seems to be fixed
06:56 dduncan I basically rewrote its main file
06:57 dduncan of course, if it is possible for a subclass of another class to export its methods, then that is probably what the old version was aiming for
06:57 dduncan meanwhile, my version keeps them all as modules, and File::Spec wraps the other worker class ... in an elegant fashion ... I think
07:06 drbean joined perl6
07:11 xern joined perl6
07:20 xern joined perl6
07:20 dduncan push in progress ...
07:20 dduncan slowly ...
07:22 svnbot6 r8822 | Darren_Duncan++ |  r2167@darren-duncans-power-mac-g4:  darrenduncan | 2006-01-27 23:18:30 -0800
07:22 svnbot6 r8822 | Darren_Duncan++ |  fixed File::Spec in ext/ (rewrote a lot of it) so it works (details also added to ChangeLog), which has a side-effect of making several other ext/ modules pass their tests; the problem mainly concerned the namespaces to which 'is export' exported to
07:24 dduncan so all the ext/File-Spec tests were directly fixed by that
07:24 dduncan also, the ext/Config-Tiny tests were indirectly fixed by that
07:27 dduncan fyi, I don't put things like ext/Config-Tiny working in the ChangeLog since they worked in the last release, which the ChangeLog is a delta record of, and any problems it had weren't in the module itself
07:27 azuroth you should. then you'd get more karma :-)
07:28 azuroth but that makes sense
07:28 dduncan the thing is, I see the 'ChangeLog' file as being for end users ...
07:28 dduncan who want a summary of, in this case, what's different between 6.2.10 and 6.2.11
07:28 dduncan things that were changed, and then undone, in the interim don't appear there
07:29 dduncan things that stopped working and then started working again without direct changes also don't go there
07:30 dduncan details of breaks and fixes in the interim go just in the subversion commit log, for developers that want to know about inter-release revisions
07:30 azuroth ahh yeah
07:31 dduncan the failures in ext/FA-DFA were not affected by my fix
07:35 dduncan also, ext/FindBin also works after my commit
07:36 azuroth I'm playing with jabberd, it's pretty nifty
07:42 dduncan okay, all ext//Pod-Event-Parser now work following my change
07:42 svnbot6 r8823 | Darren_Duncan++ |  r2169@darren-duncans-power-mac-g4:  darrenduncan | 2006-01-27 23:40:44 -0800
07:42 svnbot6 r8823 | Darren_Duncan++ |  add self to author of File-Spec
07:47 dduncan in total, around 4-5 ext/ distros appear to have been fixed by r8822
07:49 dduncan starting smoke of r8823 ...
07:50 dduncan should be done in 2 hours
07:52 dduncan tomorrow I should be able to fix most of the other ext/ deaths, which I would imagine are the same kind of problem ... is export not doing what it used to ... or maybe something about private subs, but more likely the other given few people use privates afaik
07:53 dduncan audreyt, putter, and whomever, you should focus on any t/ failures ... I'll handle most ext/ failures, since they're probably not due to problems with pugs itself
07:56 Alias_ joined perl6
07:56 Alias_ audreyt: ping?
07:56 Alias_ Generalised question, are the names of the module schemes for use documented somewhere
07:58 xinming joined perl6
08:03 kane_ joined perl6
08:10 Alias_ kane_: ping?
08:18 iblechbot joined perl6
08:21 svnbot6 r8824 | audreyt++ | * Various bug triaging, too many to list.
08:23 _bernhard joined perl6
08:48 elmex joined perl6
08:51 avar joined perl6
09:01 G2 joined perl6
09:03 Cryptic_K joined perl6
09:17 lisppaste3 joined perl6
09:17 Alias_ joined perl6
09:23 avarab joined perl6
10:12 wilx joined perl6
10:18 iblechbot joined perl6
10:38 nnunley joined perl6
10:40 j0sephi joined perl6
11:02 dduncan smoke of r8823 is done and uploaded ... now I to bed at 3am
11:02 lypanov night dduncan
11:03 dduncan and a smoke comparison shows 23 fewer test failures
11:26 chris2 joined perl6
11:27 lisppaste3 joined perl6
11:42 xinming joined perl6
11:51 kane_ Alias_: pong
11:51 Alias_ kane_: Is 6PAN dealing at any level with the whole auth naming scheme in use xxxx-xxxx-xxx
11:52 Alias_ Are you providing auths for more than one?
11:52 kane_ alias: i'm not sure what you mean...
11:52 Alias_ erm
11:52 Alias_ So use has use perl5:something right?
11:52 kane_ yeah, it's outlined in one of the synopsis'
11:52 kane_ but that's just the language implementation of loading the right file
11:53 kane_ that's not 6pan's 'problem' if that makes sense ;)
11:53 Alias_ well, 6pan is being the auth for the perl6 scheme or something right?
11:53 Alias_ So are you planning to make it also host others?
11:53 Alias_ Like a pure parrot one etc etc
11:53 kane_ sure.. the 6pan one is actually language independant
11:54 kane_ let me grab you a link, sec
11:54 Alias_ Did my comments to Audrey about the file layout get back to you btw?
11:54 kane_ i dont think so
11:54 Alias_ The problem with pregenerating all those index .gz files
11:55 kane_ http://svn.pugscode.org/pugs/docs/AES/S22draft.pod
11:56 kane_ the section just before the __END__ marker addresses 'use' statemetns
11:56 Alias_ yeah that was the one
11:56 Alias_ That index structure is a problem
11:57 Alias_ I have a better alternative :)
11:57 kane_ you mean the section under '=head1 Repositories' ?
11:57 Alias_ yep
11:57 elmex joined perl6
11:58 kane_ ok, by all means explain what you maen -- i believe rafl got this from the debian layout, and was writing some code to test if this was A Good Way To Go
11:58 Alias_ I need to go for a bit, and my time is really limited this weekend, but can we talk later?
11:58 kane_ what are you seeing as a problme for htis?
11:58 kane_ alias_: perhaps, not sure if i'll be here -- it's weekend here too, and the girlfriend might have some ideas ;)
11:58 Alias_ I don't really have time to go into details, but basically the problem is at a conceptual level
11:59 Alias_ To do with naming scheme
11:59 Alias_ There are some MUCH better ways to handle it, similar to the work on my PhotoN thing...
11:59 kane_ ok, nothing's set in stone yet, so might be good to get the 3 of us together and look this over
12:00 Alias_ ok, as long as you don't start building before I get a chance to layout an alternative :)
12:00 Alias_ (I much better one) :)
12:00 Alias_ (A ...
12:00 kane_ you're so modest too ;)
12:00 kane_ btw, it's rafl who was actually working on this, just FYI ;0)
12:00 Alias_ I finally kicked audreyt off cpan ladder 1st position, I get to be confident today
12:01 Alias_ of course, only because I stole Module::Install :)
12:01 Alias_ When she does the next release I go back to 2
12:01 kane_ off for food for me too -- i'll catch you later alias :)
12:01 Alias_ later
12:04 r0nny joined perl6
12:06 rafl Alias_: Why do you update Module::Install just a few hours after I updated it in Debian? :-P
12:07 Alias_ heh
12:07 Alias_ sorry
12:07 Alias_ There was a problem with install_share
12:07 Alias_ it was installing files into arch, not lib
12:07 Alias_ And so file::sharedir and (now) file;:userconfig were getting confused
12:08 Alias_ Shouldn't effect anything that doesn't use File::ShareDir
12:08 * rafl prepares a new upload..
12:08 Alias_ You don't have it automated?
12:09 rafl I have pretty much everything automated that's worth being automated.
12:09 Alias_ ok
12:09 rafl But you can't do everything in an automated fashion.
12:10 rafl I look up the changelog for every release and see if it concerns packaging stuff, new dependencies, etc..
12:10 lypanov bah. just a simple ai
12:14 Alias_ fair enough though
12:14 Alias_ About the only thing I don't have automated for my development now is the actual upload
12:14 Alias_ just in case :)
12:16 Alias_ I can see how there's similar issues with M:I
12:38 merlyn joined perl6
12:39 * merlyn stares at his future {grin}
12:39 Juerd Hi merlyn
12:47 merlyn hey
12:47 merlyn Just seeing if I remembered my freenode password after 48 hours. :)
12:47 merlyn woke up with that "omg... did I forget it again?" feeling. :)
12:49 Juerd Freenode sucks immensely because of the nickserv crap
12:49 Juerd The people who declared private messages a great risk, and think this is a solution, should be punished.
12:52 chris2 joined perl6
13:04 lypanov Juerd: just having NickServ reply when you privmsg would already help so damn much
13:04 theorbtwo Hm, I could of sworn he did, but he doesn't seem to now...
13:06 theorbtwo Oh, yes, he does, my irssi just didn't highlight the window correctly.
13:06 lypanov i don't mean a notify, i mean he should privmsg you
13:07 theorbtwo Oh!  You mean nickserv should message you when you try to /msg anybody when not identified.
13:07 theorbtwo Yes, that would be very useful.
13:09 * lypanov nods
13:12 * lypanov notes that his local moroccan baker produces some seriously addictive cakes
13:28 kanru joined perl6
13:50 kanru joined perl6
13:51 avar joined perl6
14:14 svnbot6 r8825 | audreyt++ | * Pugs.Shell: disable readline's sigwinch and signal handling,
14:14 svnbot6 r8825 | audreyt++ |   so the interactive shell no longer randomly segfaults under
14:14 svnbot6 r8825 | audreyt++ |   threaded builds.
14:15 clkao audreyt: happy new year
14:16 audreyt happy new year too!
14:16 audreyt probably not going to stay awake
14:17 clkao will be around taipei next week, hack-p?
14:17 audreyt for the chronosaur to appear
14:17 clkao lol
14:22 pdcawley joined perl6
14:22 audreyt hi pdcawley
14:24 pdcawley Hello audrey. How goes?
14:26 audreyt quite well. doing some journaling before sleep
14:27 pdcawley Ah... I look forward to some fascinating reading.
14:28 audreyt :)
14:29 obra 'morning
14:31 audreyt yo
14:31 obra Halos got...close on the plane. I hurt myself by not having cached css and js references on the laptop
14:32 obra also, good tailwinds got us in very early ;)
14:33 audreyt I parsed is as something about solar wind halos attacking the plane and you narrowedly escaped death but injured yourself
14:33 audreyt I'm glad it was a parsefail
14:33 obra heh
14:34 obra audreyt++ # active imagination
14:34 audreyt I get karma by being insane?
14:34 audreyt hm, that's actually the norm here, to think about it
14:35 verve_ joined perl6
14:35 obra indeed it is
14:35 obra constructive insanity
14:40 loggygreg joined perl6
14:48 loggygreg left perl6
14:52 kanru joined perl6
14:54 rantanplan_ joined perl6
15:08 Debolaz joined perl6
15:18 Limbic_Region joined perl6
15:23 Debolaz2 joined perl6
15:27 audreyt whew. two journals in a day ;)
15:27 * audreyt waves. good night! &
15:29 obra night
15:41 Debolaz joined perl6
16:20 dakkar joined perl6
16:27 dakkar I'm reading S06, and I find the "splat" way to pass named arguments from pairs or hashes potentially confusing...
16:28 dakkar I mean: foo(*%hash) is a splat; foo(%*hash) is a 1-parameter passing a global hash...
16:28 dakkar I think I'll overstress C-t while writing p6 code :/
16:30 pdcawley_ joined perl6
16:32 dakkar still from S06: «To pass pairs out of hash without their being interpreted as named parameters, use %hash<a>:p instead»
16:33 dakkar why? are not pairs detected «at the comma level», so that «doit $pair,1,2,3;   # always a positional arg» ?
16:33 dakkar why should it be diffirent if the pair is coming from a hash subscript expression, instead that from a scalar ref?
16:34 * dakkar is trying to understand...
16:35 dakkar what is «say *pop(@array);»? is *pop the global sub named 'pop', or a splat of the result?
16:36 dakkar joined perl6
16:37 * dakkar hit the close button by accident
17:13 Odin-LAP joined perl6
17:45 feng123 joined perl6
18:37 lypanov_ joined perl6
18:42 penk joined perl6
19:14 nothingmuch joined perl6
19:34 lisppaste3 joined perl6
19:40 avar joined perl6
19:51 elmex joined perl6
20:07 * nothingmuch wonders if perl could do with a prototype object system
20:07 nothingmuch that is just really well hidden
20:08 nothingmuch since prototype models non prototype well, but not vice versa (given enough sugar), and SMD is a subset of MMD functionality, this can really simplify the specs
20:08 nothingmuch all the other crap can be pushed to the Prelude
20:08 nothingmuch and since we assume the prelude can be uptimized in larger chunks we can simply have implementations of those features in PIR for parrot, etc
20:08 nothingmuch like, for speed
20:09 * nothingmuch thinks Perl 6 ought to be a smaller language than it is, and these kinds of savings could really help
20:10 tewk Iteresting, where can i read about "prototype models non prototype well, but not vice versa (given enough sugar), and SMD is a subset of MMD functionality"
20:11 nothingmuch you sound like Eliza
20:11 nothingmuch p6l archives
20:12 nothingmuch and, err, just common sense really
20:12 nothingmuch our object space stuff could be implemented as a prototype system
20:12 nothingmuch eliminating circularity
20:12 nothingmuch but getting a very close effect
20:13 nothingmuch and SMD isa MMD - well, it just is
20:13 nothingmuch if you only have one MMD variant, it's as if it's single method dispatch
20:13 nothingmuch since there's no alternative
20:15 nothingmuch the plus side is that we get built in prototype OO
20:15 nothingmuch which is good for some things
20:15 nothingmuch and the MMD/SMD implementation details are simpler to spec
20:30 meppl gute nacht
20:41 nothingmuch ciao
20:46 meppl ;)
20:54 Cryptic_K joined perl6
20:56 penk left perl6
21:01 dduncan joined perl6
21:05 dduncan now I'm getting a build failure, following the last update ...
21:06 dduncan Compiling Pugs.Shell       ( src/Pugs/Shell.hs, dist/build/src/Pugs/Shell.o ) /tmp/ghc12129.hc: In function 'PugsziShell_readline_entry': /tmp/ghc12129.hc:121: error: 'rl_catch_signals' undeclared (first use in this function) /tmp/ghc12129.hc:121: error: (Each undeclared identifier is reported only once /tmp/ghc12129.hc:121: error: for each function it appears in.) /tmp/ghc12129.hc:122: error: 'rl_catch_sigwinch' undeclared (first use i
21:06 dduncan I'll try a make clean again
21:06 dduncan since that seemed to work on a similar problem yesterday
21:12 Cryptic_K joined perl6
21:24 dead joined perl6
21:33 Khisanth joined perl6
21:35 SamB joined perl6
21:40 dduncan that didn't work
21:40 dduncan following 'make realclean'
21:40 dduncan I get the same error in build
21:40 dduncan someone else, please confirm
21:41 dduncan if this is working fine for others, I may try a full checkout directory delete and revert
21:42 tewk dduncan: building now
21:51 dduncan thank you ...
21:57 Khisanth audreyt: you have a typo in your Dec 28 blog entry :)
22:12 tewk dduncan: build just fine.
22:43 vytautas joined perl6
22:45 dduncan okay
22:46 dduncan left perl6
22:54 dduncan joined perl6
22:55 dduncan okay, so I've now done the delete and revert thing with the whole pugs checkout dir
22:55 dduncan now trying 'make' again ...
22:58 integral that looks related to your libreadline...
23:03 pdcawley joined perl6
23:10 pdcawley joined perl6
23:35 pdcawley joined perl6
23:47 dduncan that didn't help ... same error
23:47 dduncan does Pugs have some new dependency that I have to install?
23:47 dduncan or a dependency that differs between platforms?
23:47 dduncan I'm using stock Mac OS X 10.4.4 (Jan, 2006), save for installing the svk 1.06 bundle, ghc 6.4.1
23:48 dduncan several 'rl_catch_*' symbols undeclared
23:49 dduncan in function 'PugsziShell_readline_entry'
23:49 pdcawley joined perl6
23:49 dduncan the only other thing I have yet to try is reverting to an older pugs and see if the problem goes away
23:50 dduncan besides trying that or installing something, all attempts to 'make' fail when compiling Pugs.Shell
23:50 integral ah, r8825 changes Pugs.Shell
23:51 * dduncan back in a bit
23:52 integral hrm, the change shouldn't have been significant
23:58 pdcawley joined perl6

| Channels | #perl6 index | Today | | Search | Google Search | Plain-Text | summary

Perl 6 | Reference Documentation | Rakudo