pugs.blogs.com/dist/Perl6-Pugs-6.2.13.tar.gz released! | run.pugscode.org | spec.pugscode.org | paste: sial.org/pbot/perl6 | pugs.blogs.com
Set by audreyt on 17 October 2006.
svnbot6 r14546 | fglock++ | v6 - implemented prefix:<\> (t/var/autoref.t) 00:03
r14546 | fglock++ | - autoref.t passes 34/57 (mostly because "Pair" not finished yet)
00:30 AzureBOT joined 00:43 xerox- joined 00:44 avarab joined 00:45 avarab is now known as avar 00:58 xie joined 00:59 xie joined 01:02 buubot joined
eviltwin_b mrf. command line test DTRT, t/operators/filetest.t does something weird 01:03
01:07 xerox- left 01:12 marmic joined 01:15 f0rth joined 01:17 lisppaste3 joined 01:19 dmq joined 01:20 cmeyer joined 01:25 buubot joined 01:31 buubot joined 01:39 buubot joined 01:43 buubot joined 01:46 mdiep joined 01:58 buubot joined 02:16 weinig is now known as weinig|zZz
eviltwin_b finally understands... and doesn't 02:28
02:28 mjk joined
eviltwin_b $*BASETIME got written to the prelude, and is overriding me 02:28
which I don't quite understand since it's a constant
nor why command line works but the test script's $*BASETIME is overridden 02:39
02:41 Lorn_ joined
Lorn_ audreyt: hi 02:42
audreyt Lorn_: greetings
I'm about to sleep again :) thanks for picking me up from airport, btw :)
Lorn_ audreyt: hehe, you are welcome, its all ok in hotel? 02:43
audreyt all is fine 02:44
though jetlag sucks :)
02:45 Achilles14 joined
Lorn_ audreyt: jetlag? the internet is slow? 02:45
Achilles14 hi audreyt 02:46
do you still have the "cpanmods" ppt?
audreyt Lorn_: no, internet is just fine
time zone difference 02:47
Lorn_ audreyt: oh, ok, i understand now :) do you know, the sun adjust you time zone, automaticaly ?
audreyt Achilles14: 'fraid I lost it 02:48
Lorn_: I do
Lorn_ audreyt: its everything ok, im going to sleep now, bye 02:52
audreyt cool. yup. g'nite! 02:53
Lorn_ audreyt: g'nite too 02:55
audreyt eviltwin_b: the IO::Dir puzzle is solved 02:59
1. IO::Dir::close was declared "is export" where it shouldn't be
or maybe it should. hm. but at least IO::close and Socket::close should, too
2. the Prim.hs parser was parsing the type of IO::Dir as simply IO
anyway, both fixed 03:00
committing
svnbot6 r14547 | audreyt++ | * Pugs.Prim: Also export IO::close and Socket::close so that 03:08
r14547 | audreyt++ | close $fh;
r14547 | audreyt++ | close $dh;
r14547 | audreyt++ | close $sh;
r14547 | audreyt++ | all DWIM.
r14547 | audreyt++ | * Also fix the parser error that mistook "(IO::Dir:)" in the
r14547 | audreyt++ | Prim.hs decl table as "(IO : Dir)".
eviltwin_b yay 03:09
now I have to work out my own little issue (now I know why, need to figure out fixing it)
...but I thought you were going to bed :)
audreyt oh right. 03:10
audreyt vanishes in a puff of logick
svnbot6 r14548 | audreyt++ | * span.t and set-infinite.t: Disable infinite looping tests till tomorrow 03:11
03:13 mako132_ joined 03:18 ashleyb joined 03:24 AzureBOT joined 03:32 mako132_ joined
eviltwin_b yep, know what's wrong, have no idea how to fix. oh well. :/ 03:38
03:48 diotalevi joined, Achilles14 left 03:50 lisppaste3 joined 04:13 nicodaemos joined 04:16 buubot joined 05:54 xie joined 06:01 cmeyer joined 06:47 marmic joined 07:01 drrho joined 07:05 DHGE joined 07:06 cjeris left
gaal @tell integral yes, Pugs.Val is the new stuff. See also Pugs.Val.Base and Pugs.Val.Code and for new AST, Pugs.Exp 07:13
lambdabot Consider it noted. 07:14
07:36 lumi_ is now known as lumi
integral thanks gaal. 07:46
lambdabot integral: You have 1 new message. '/msg lambdabot @messages' to read it.
08:03 elmex joined 08:27 kane-xs joined 09:22 ank joined 09:23 mr_ank joined 09:24 ank is now known as mr_ank 09:25 kensanata joined 09:35 buetow joined 09:54 bonsai joined 09:55 andara_ joined 10:11 nusgnaf joined 10:18 chris2 joined 10:20 skew joined
skew Hi, I was wondering when audreyt is usually around. 10:23
audreyt skew: from this hour to ~15hr+ 11:02
for the next week or so
Debolaz ponders just using YAML for data storage. 11:08
11:11 kanru joined 11:13 ruoso joined
skew audreyt: I was wondering if you had looked at that gradual typing stuff. I was just reminded by some mail from Jeremy Siek about it. 11:14
audreyt I had, but that was a while ago. I'm also wondering between strictly local inference vs program wide HM inferencing
seems strictly local has the advantage of least surprise 11:15
at the expense of maybe marking more typeslots unknown
skew One problem is that I haven't really used dynamically typed languages that much for a while, so I don't really know how I'd like type inference and dynamics to fit together 11:17
audreyt I think the idea is if there's no annotation we don't infer anything
even if we know builtin function's signatures 11:18
we can arguably have two modes
under -O, where you can't augment to add new overloaded methods at runtime and classes are closed, we can perform nonlocal inference
as your prototype does
but normally, we only infer locally
the design space is large though. 11:19
skew by locally you mean inside a function?
audreyt I mean lexically, yes
skew The area where I don't have enough intuition is what to do with annotations that say "please infer some concrete type" 11:21
audreyt ah.
you can always take all the constraints and create a concrete union type... 11:22
skew specifically, where dynamics can come up in the inferred type.
audreyt in that case, infer as dynamic (defeating user's request) and throw a warning 11:23
sounds correct to me
Debolaz I don't suppose anybody has tried POE to see if that specific module works with pugs? 11:25
audreyt no, not yet 11:26
happy to see more t/perl5/
t/perl5/POE/moose.t
11:26 fglock joined
skew audreyt: it's not quite that simple. Say you give some argument x a "figure out the type" annotation, use it a few places as int, and once as a dynamic 11:27
audreyt sub f ($x is inferred) { for 1..$x { say $x.moose } } 11:28
# concrete example
in the 1..$x case we know from signature of infix:<..> that $x is int 11:29
but .moose is a dynamic method invocation
something like that?
skew probably you want inference to decide x is an int, and insert one cast where it's used as a dynamic
audreyt but in this case suppose there's no Int.moose method 11:30
then the cast is arguably better in the ..$x position
and leave $x as dynamic with a warning
and if the user has misspelled the method name 11:31
they actually mean .abs but misspelled as .moose
then the warning will be sufficient to recover
11:32 b_jonas joined
audreyt the neccessary info the fix the typo 11:32
I think. not terribly sure
skew oh, casting between types is different - the gradual typing stuff just thinks of concrete and a Dynamic wrapper, rather than stuff like casting around in a class hierarchy, or converting record types. 11:33
hmm, I don't know about that at all
audreyt I talked with Wadler a bit on this
with the author of gradual typing paper nearby
Debolaz audreyt: When perl 5 modules can be used for inheritance, what is the prospect of Catalyst working? :) 11:34
audreyt the consensus seems to be the infer-everything regime, with or without Dynamic, falls down in the face of "duck" record subtyping
and restricting the inference to be either soft (as in warning/advice only), or local (as in punt to Dynamic without lexical annotation info), seems to be prudent 11:35
Debolaz: write tests and we'll know :)
skew how is that different from type inference for extensible records, which seems to work pretty well in e.g. PolyR?
audreyt runtime augmentation? 11:36
can you reopen classes in PolyR at runtime?
skew ah, mutable values would make things interesting
Debolaz audreyt: Buzz me when p5 inheritance works then. :) 11:37
audreyt Debolaz: have you written p5 inheritance tests?
please do so so I can buzz you :)
fglock audreyt: welcome to Brasil :) 11:38
Debolaz Only under perl 5. :-/
audreyt fglock: heya. you coming over today?
Debolaz: please write a t/perl5/inheritance.t
Debolaz audreyt: Expect requests for review of it.
fglock audreyt: unfortunately no - $work until tomorrow, I'll get there on wednesday 11:39
audreyt Debolaz: sure :)
fglock: okie
skew: I wonder if we can find an encoding of MLPolyR into ATs 11:41
svnbot6 r14549 | fglock++ | v6 - implemented Pair autoref
audreyt the web-based MLPR evaluator is borken
skew audreyt: I think the HList stuff already includes a record system
audreyt and there's no code that I can find, hmm.
skew: oh they do (and even infinite records)
skew and it seems pretty hard to get nice inference for anything built out of type classes
audreyt ATs are supposed to make this more tractable 11:42
but I confess I have ~0 experience with them
skew last I heard, the problem with ATs is figuring out how to do type inference if you allow them
at least for associated type synonyms 11:43
audreyt have you seen sulzmann's ATFunDep paper? 11:44
"Associated Functional Dependencies" 11:45
skew is it new?
audreyt '06 I think
skew I haven't seen that one. Do you know where to find it? 11:46
audreyt www.comp.nus.edu.sg/~sulzmann/manuscript/afds.ps
lambdabot tinyurl.com/y82cru
audreyt the unified inferrence in particular sounds promising to me. 11:47
(which is why I had this vague hope of embedding MLPolyR) 11:48
skew I think maybe it will make more sense after I learn more Perl. 11:50
audreyt or JavaScript with Prototype.js
or IO, but yeah :)
svnbot6 r14550 | fglock++ | v6 - start autoref'ing lists 12:08
12:09 elmex joined 12:11 cmarcelo joined
cmarcelo hello 12:12
audreyt cmarcelo: greetings lambdacamel
cmarcelo audreyt: hey. already adapted to .br timezone? 12:14
12:15 fglock left
audreyt hopefully, yeah 12:15
where are you in meatspace?
cmarcelo "meatspace" meaning Earth? 12:16
audreyt meaning not cyberspace 12:17
yeah
cmarcelo heh. i'm in Campinas, a city near Sao Paulo. Probably Wed I'll be in my parents house in SP... but no network there, though :( 12:18
also looking for a friend here to lend me a notebook for yapc hacking sessions.. 12:20
audreyt that will be very helpful... 12:27
bbiab 12:30
cmarcelo audreyt: should I keep "porting" MO->MOH the way(s) I was or new ideas are coming up? 12:40
audreyt basic.hs looks about right. 12:50
12:51 Psyche^ joined 13:02 Limbic_Region joined, Psyche^ is now known as Patterner 13:16 Ken joined 13:40 fglock joined
Limbic_Region audreyt ping 13:45
14:05 polettix joined 14:15 the_dormant joined 14:20 TimToady joined
cmarcelo bbl & 14:22
14:26 buu joined
svnbot6 r14551 | fglock++ | v6 - fixed autoref of list, anon-array; t/var/autoref.t passes 50/57 14:30
14:30 zdeq1 joined, zdeq1 is now known as zdeqb
svnbot6 r14552 | fglock++ | v6 - fixed autoref of anon-hash; t/var/autoref.t passes 52/57; 14:39
r14552 | fglock++ | the remaining tests depend on porting the old runtime
fglock is t/var/autoderef.t obsolete? this looks strange 'ok tied($y).isa("Ref"), "tied with 'fake refs' (2)";' 14:43
Limbic_Region fglock - is it smartlinked? 14:45
fglock Limbic_Region: where should I look? 14:46
anyway, it is marked as "more-discussion-needed", so not a problem 14:47
14:48 penk joined
Limbic_Region fglock - there should be special POD above the test if it is smartlinked 14:48
are you not aware of spec.pugscode.org/ ?
lambdabot Title: Official Perl 6 Documentation 14:49
Limbic_Region hrm - that doesn't appear to be what I thought it was
fglock Limbic_Region: I've been using a local copy of t/, so I didn't notice smartlinks 14:50
Limbic_Region ah
anyway - if it was smartlinked to an S, you could have checked that what it was linked to is still valid 14:51
in any case, I can't seem to locate where the smartlink stuff is ATM
[particle] smoke. 14:53
pugscode.org (slipped) 14:54
14:54 vel joined 14:56 cjeris joined
svnbot6 r14553 | fglock++ | v6 - t/var/autoderef.t passes 20/22 14:59
fglock forget it - that test was commented out, but v6.pm misparsed the comment
a '=begin' without an '=end' should be marked '=begin END', right? 15:02
15:05 rlb3 joined
svnbot6 r14554 | fglock++ | v6 - t/var/autoderef.t passes 21/22 15:12
15:12 timbunce joined
fglock lunch & 15:14
15:14 fglock left 15:24 lisppaste3 joined 15:39 timbunce joined 15:41 gene9 joined 15:45 weinig|zZz is now known as weinig 15:57 jferrero joined 16:02 weinig_ joined 16:04 fglock joined 16:05 ofer1 joined
svnbot6 r14555 | rnhainsworth++ | test to check that perl5 evals access lexical pad 16:12
16:14 penk joined 16:25 ashleyb joined 16:26 Pomin joined 16:42 weinig_ is now known as weinig|bbl 16:46 penk left 16:51 thepler joined 16:59 esammer joined 17:01 Gothmog_ joined 17:07 badalex joined 17:12 timbunce joined 17:16 rblackwe left 17:21 fglock joined 17:23 justatheory joined 17:34 kanru joined 17:44 elmex joined 17:45 bernhard joined, elmex joined 18:03 hexmode joined
svnbot6 r14556 | fglock++ | v6/emitter - more array/hash fixes 18:10
18:11 Gothmog_ joined 18:24 Odin- joined 18:25 b_jonas joined
Limbic_Region b_jonas = ambrus ? 18:26
Limbic_Region 's memory isn't what it should be 18:27
b_jonas L_R: yes 18:31
Limbic_Region thought so - I haven't played with your code but is it only intended to rank combinations of a uniform size or the entire powerset? 18:32
b_jonas yes 18:33
if you want to rank all combinations
that would be much simpler:
Limbic_Region b_jonas - that's why I gave you the example in /msg I did
b_jonas in that case, you just have to use binary numbers 18:34
Limbic_Region for my $subset ( powerset(1..4) ) { # rank $subset } # where you don't have to worry about powerset()
Limbic_Region notes that we should take this to /msg land
b_jonas ok
18:46 buubot joined 18:49 foo\ joined 18:54 SubStack joined 18:56 Eimi joined 18:59 elmex joined 19:03 buetow joined 19:06 SubStack joined 19:13 elmex_ joined, chris2 joined 19:14 timbunce joined 19:28 elmex joined 19:45 hexmode joined
stevan anyone know the location of the MOH code? 19:49
audreyt perlcabal.org/~cmarcelo/moh/ 19:50
lambdabot Title: Index of /~cmarcelo/moh
stevan cool
audreyt: I finally got around to reading Algorithms
audreyt will merge into src/ this week
ooh! how do you like it? :)
stevan I like it
Haskell is like all the other functional languages I have read about, but totally different at the same time 19:51
I am finally starting to grok the syntax, which was always the part I had the most trouble with
audreyt it's quite ergonomic
in a strange way
stevan yes
Limbic_Region audreyt - the hiveminder items you recently closed - WRT the web server ones hanging - did you close because you couldn't reproduce or because you fixed something?
stevan audreyt: but I thought I would test my new found understanding by reading the MOH stuff :) 19:52
audreyt Limbic_Region: yes, r14544
stevan: oh. uhm, not extracly sure it'sbest idea
MOH uses audreyt-style records :)
eviltwin_b I still haven't quite grokked the fullness of do expressions --- I understand in general but still misapply it in specific and it takes a few revisions before I get it right
audreyt and some quite advanced Haskell syntax
Limbic_Region ok, going to smoke without deleting those tests 19:53
stevan audreyt: I been looking at your Haskell code for over a year now, I am probably more used to your style then the canonical
audreyt true, but all the feature MOH uses is GHC 6.6 only...
which means it's all new :)
stevan ah
audreyt harrorth may make more sense, actually.
stevan nah,..i read that a while ago,.. it made sense then 19:54
nothingmuch's explinations helped of course
audreyt: I will just skim and ask questions when I have them
audreyt sure 19:55
stevan: if you pull, "make" will actually compile now :) 19:59
stevan nice
19:59 Aankhen`` joined
stevan nothingmuch and were talking about bringing the "transformations" from MO over to Class::MOP 20:00
as a new way to do the immutable
audreyt oh, MO/C3.hs is nicely self-contained that will test your Algorithms-fu
stevan which will avoid the need for making custom immutable classes for each custom metaclass actually
Limbic_Region stevan - for immutable you just need to use Matrix;
</bad_joke>
stevan yeah I waas reading C3.hs actually :)
20:15 [particle] joined
fglock I merged back some parts of v6/emitter(2) to v6.pm - so many array/hash/scalar annoyances are now fixed 20:19
audreyt fglock++ 20:20
fglock in order to support types we need to fix the references problem in Data::Bind 20:22
audreyt *nod*
fglock or change to pure-perl5 binding 20:23
audreyt how would that work? I mean pure p5 binding
fglock one sec - looking for link
20:25 dduncan joined
fglock svn.openfoundry.org/pugs/misc/pX/Co...example.pl 20:25
lambdabot tinyurl.com/y858bf
audreyt oh the x= thing 20:26
fglock yes, implemented in v6/emitter(2)
audreyt that's a fair plan
but then that means access is always $$
never $
right?
fglock yes, and @$ and %$ - depends on the container type, and whether you want raw access or typed 20:27
I also did some experiments with coro and multis on this directory 20:28
audreyt but that means interop with normal p5 code
fglock we'd need a bridge
audreyt is um, not very flexible
nod. better to design this in person with pen and paper
this Wednesday you say? :)
fglock pen and paper would be nice - not sure if I'll get a notebook 20:29
wednesday about 10:00 in the hotel 20:30
audreyt you can get a wireless notebook with optical fiber display
made from pure natural organic material
in a local bookstore :)
kolibrie those kind cost less than lunch 20:34
fglock interesting how paper is better for high-level things 20:36
audreyt *nod* 20:42
fglock home & 20:52
20:53 fglock left 21:09 ofer0 joined 21:16 cmarcelo joined, weinig|bbl is now known as weinig
Limbic_Region audreyt - the web hanging fixed worked like a champ 21:24
audreyt good, as expected
Limbic_Region smokes take an awfully long time though 21:25
> 60 minutes on a machine that normally doesn't take > 40
lambdabot Not in scope: `doesn't'
audreyt hrm. 21:26
Limbic_Region I can't say for sure when this started happening because since you have been traveling commits have dropped off to nothing 21:27
and I haven't smoked twice a day as a result 21:28
audreyt *nod* 21:29
I'll see what I can do
TimToady I lost 20-30% about the time the hyper and junction parallelizations went in, I think, though it might not have been those.
audreyt that is plausible
maybe we should probe for RTS=1
and then not try to parallelize
Limbic_Region oh, I have intentionally not tried to take advantage of any of the neat new doodads available 21:30
or rather - when I do that I don't upload the smoke
audreyt k
Limbic_Region that way I have a "control" on the smokeserver
audreyt TimToady: I think it is exactly those
but that seems too drastic
20% that is
since we don't use either that much 21:31
in Test.pm certainly not
audreyt ponders
nothingmuch to clarify: esteban-san talked about hypothetical transformations
i've only thought of them
Limbic_Region in any case - I am heading homewards and I doubt I will be logging in later since we carve pumpkings tonight - er pumpkins that is
nothingmuch not impl
TimToady well, I used to average about 30 minutes, and now it's 36-38 21:32
21:34 Narcisse joined 21:43 mauke joined, jferrero joined 21:45 buubot joined
audreyt dinner at Figueira, bbl :) 21:55
cmarcelo nothingmuch: ping ? 21:57
nothingmuch cmarcelo: pong 22:13
cmarcelo nothingmuch: little question about MO, lib/MO/Run/ResponderInterface/MethodTable.pm expects its methods to have a $body, ie. to be like Method::Simple::Compiled folks? 22:20
22:27 elmex_ joined
cmarcelo nothingmuch: and... MO::Compile::Method role requires a compile function, this compile should return a MO::C::Method::Compiled object? 22:29
nothingmuch yes 22:38
it expect a Method::*::Compiled
precisely
->compile gives the method "raw" method enough info (if needed) to be smart about the entire class hierarchy it's applying to 22:39
a method definition happens once per object model
but it gets ->compiled call for every class which consumes/inherits it
23:05 AzureBOT joined 23:42 gnuvince joined
gnuvince Will Perl 6 have docstrings? 23:46
Juerd gnuvince: Do you know Perl 5, and POD? 23:49
gnuvince Juerd: yes 23:50

Powered by Cro and the Raku® Programming Language.

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

Thank you!