Run Perl6 Now! & pugscode.org ('Overview', 'Journal') & pugs.kwiki.org & channel log: xrl.us/e98m & < autrijus> damn, my journal keeps getting obsoleted by #perl6.
Set by Juerd_ on 12 March 2005.
lightstep i seem to get the types more and more right 00:15
i need people to check out the new extension
but it would take some more time 00:16
how is "my" parsed? or where? 00:24
done! 00:54
kungfuftr moo 01:02
Khisanth Juerd: don't you mean arguments? 01:06
kungfuftr mugwump: yargh 01:08
mugwump :-p
kungfuftr mugwump: so when will we see Class::Tangram in p6? 01:09
mugwump Class::Tangram is entirely unnecessary in p6
kungfuftr *blink* though i would have thought relationship stuff would still be somewhat needed 01:10
mugwump Class::Tangram can be summarised as explicit (Run time type information | introspection | reflection)
kungfuftr ah, far does 01:11
kungfuftr too tired to think 01:12
mugwump Hopefully, any "extra" information that Tangram needs to be able to represent the structure in the database can be tacked on with traits 01:13
(I hope I used the right term there - I'm referring to *Class* properties)
That way, Tangram will be able to "just work" without really specifying a schema (unless you want to override the default mapping) 01:14
After all, the "Class" objects are themselves a schema.
bd_ hmm, return inside a while loop seems broken 01:58
pugs> sub foo { while (1) { return 24; } return 42; } foo() 01:59
(42)
mugwump you should be exiting the loop nicely! return jihad! 02:10
bd_ XD
well, the whole point of the module I'm writing is to be a hacky implementation of sort-of-regexes for until pugs gets rules... sooo, for now I've added yet another hack 02:11
but it'd be nice if that worked properly :)
though, I have a working next; now :)
mugwump I still think it would be a great idea to allow the Perl5 regex syntax in pugs via pcre... 02:13
svn.openfoundry.org/pugs/ext/Kwid/l...dspec.kwid # feedback, please...
Khisanth that would be annoying 02:15
mugwump what would? the pcre thing?
Khisanth yes 02:16
mugwump why? 02:17
Khisanth the not quite compatible part 02:18
unless you mean to use it as a temporary solution
mugwump I don't know exactly which features are missing, but AIUI none of them are really awful being missing. Not as awful as having POSIX regexes or no regexes at all 02:19
perhaps the (positive|negative) zero-width look(ahead/behind) assertions would be sorely missed if they're not in PCRE 02:20
Khisanth implement the rules! :P
I think those are in, the look behind might actually be better ... 02:21
ingy mugwump++ 02:41
mugwump hopefully, the implications of that perlkwidspec will be very minor to the current implementation... 03:05
05:22 Aankh|Clone is now known as Aankhen``
autrijus last() fixed; true() implemented; @array[-12345] = 2 is now an error; GHC 6.4 warnings really inhibitied; some other misc fixes. 06:44
gotta run. bbiab &
obra hi autrijus 06:47
nothingmuch autrijus: please look at r744 07:18
which i think should pass, but i'm not 100% sure
it's got some failing tests
and i'd actually rather they fail in the way that they do =)
gaal: you forgot to attach undef.t 07:26
and in either case you should really just be a comitter
gaal did not!
nothingmuch perlbot nopaste 07:27
perlbot Paste your code here and #perl will be able to view it: sial.org/pbot/perl
pasteling "nothingmuch" at 212.143.91.217 pasted "no attachment" (19 lines, 569B) at sial.org/pbot/8199
nothingmuch see?
anyway, ride will be here shortly 07:28
gaal grr, you're right, it did manage to get lost somehow although i *remember* attaching it. mailers should follow Nat's idea and confirm sending a mail that contains the word "attachment" but which don't have one. 07:29
pasteling "gaal" at 192.115.25.249 pasted "t/base/undef.t" (134 lines, 3.3K) at sial.org/pbot/8200 07:30
nothingmuch svn add, svn ci =) 07:31
don't be shy
gaal todo_ok($self{"commit"}, "need one to ci"); 07:32
nothingmuch (autrijus^obra).invite(gaal)
obra gaal: do you have a foundry id? 07:33
gaal this is where i do the "oh, no, yet another source control system to learn" thing :) 07:34
obra gaal: email address?
gaal obra: let me read a faq about it, sec
nothingmuch obra: remember t/op/precedence.t?
obra is about to head to bed
nothingmuch: yes
gaal hurries up
nothingmuch gaal: openfoundry ~~ sf.net but not commercial
obra run for free by the taiwanese government to promote opensource in taiwan 07:35
nothingmuch obra: if you like dreaming about precedence attrocities at the design level, look at the bottom of it, there's assignment vs. , and yes
obra hah
Don't wanna
malaire I was just implementing index, and found this strange behaviour from perl5: 07:36
index("Hello", "", 999) == 5 # Should perl6 behave similarly or not?
I'd expect -1 here, not 5.
gaal anyway obra the main svn site seems to be down, i won't keep you
obra gaal: huh? it's up to the best of my knowledge 07:37
nothingmuch malaire: if it isn't a bug it's a design bug
IMHO
obra svn ls svn.openfoundry.org/pugs
gaal subversion.tigris.org ?
obra that's the dist site for subversion. but you can likely obtain it from your os vendor 07:38
gaal i have a client. i'm just looking to read about what a foundry id is.
obra oh. rt.openfoundry.org
nothingmuch gaal: err, a foundry == rt + svn server + trove
obra it's just the project hosting site pugs uses
gaal ...so why do *i* need one? 07:39
obra so you can commit to svn
nothingmuch you don't
it's just the svn you need
it hosts the server
you have a login/passwd
and then you can commit
obra access to pugs svn is managed through rt.openfoundry.org
gaal has been granted commit rights 07:42
Alias_ Welcome to the pokemon
gaal the terms of use are (i suppose) all Hanzi. I'm not signing off my house, am i? :)
obra you are not.
autrijus has a translation up somewhere
hcchien I can give you a tranlated page. :)
obra hcchien: can you get the translated page linked from the main one 07:43
?
gaal i think i trust you guys. :p
obra foundry is becoming international and this keeps coming up
hcchien wagner.elixus.org/~hcchien/termtouse.html 07:44
obra: yes, I will ask them publish it officially later. :)
obra yay
ok. really now sleep
hcchien nice dream. :) 07:45
gaal okay, thanks all. 07:47
lightstep how do i update MANIFEST? 07:48
some files were deleted in r735, and no wMANIFEST is not updated 07:49
malaire just edit manually, and the svn ci ?
lightstep there are lots of files 07:51
gaal try cp MANIFEST MANIFEST.old; make manifest, then merge the to with your favorite tool. 07:52
lightstep thanks 07:53
err, it only adds files to MANIFEST. i'll just delete the missing files manually 07:55
gaal s/cp/mv/ above and retry :) 07:56
i need to co a new tree now that i'm not anonymous, right? what's the magic incantation? 07:59
autrijus nope. just ci. 08:00
it will then ask for your username just then.
gaal clever :)
autrijus all hail DAV ;)
gaal ok to put undef.t in base/, yes?
autrijus sure. 08:01
gaal not *quite* smooth; it assumed my unix login was my remote one (not the case). how do i say what my svn user is explicitly? 08:02
autrijus just hit enter
and it will ask another username.
gaal did that; ok, will it now configure the tree with tihs username for next time round? 08:03
autrijus I think so. 08:04
it does that for me.
gaal thanks then, ci'ed. think i'll wait a bit before adding it to the makefile and manifest since they seem to be under lightstep's luv right now? 08:05
lightstep Eval.hs is very complicated 08:07
autrijus hmm, some perlmonk there? "dvergin" reported a compile error in his use.perl journal 08:09
use.perl.org/~dvergin/journal/
I attempted a fix in trunk now, but has no idea how to tell him to try again and/or turn on comments in his use.perl journal :)
lightstep: it is. care to refactor it to Eval/App.hs etc for me? 08:10
or do you think that won't help? :)
lightstep i''m trying to figure out the evaluation process, in hope i'll be able to make `my' work in the prompt 08:11
and then i'll add require, and you can just paste test code, and it'll work 08:12
autrijus ok. what needs to happen is that you need to add a kind of Val in AST.hs
mm, that's probably too kluge
lightstep i just found the correct branch of reduceStatements
autrijus the trick is to somehow trick the end of reduceStatements to return its env.
so we can use that env for the next line around. 08:13
currently it just returns a Val
so you can add a VEnv in AST.hs.
lightstep yes
autrijus as part of Val.
on the other hand,
lightstep maybe i'll change the processing of top-level local variables?
autrijus maybe you can do it be passing in a callCC
and trick the env away by feeding it to the callCC.
but that may be nontrivial.
lightstep: sure, but change how? 08:14
my original idea is for the REPL to parse the line
and add an extra statement after it
lightstep perhaps a new `reduce' for shell expressions, that handles `my' and `require' specially
your idea seems simpler, though 08:15
autrijus yeah, I'd say so.
if we can freeze the env as part of a Prim / Statement
then that's one step toward serializable continuations.
lightstep hmmm, that requires some `inline expression' which is isomorphic to Eval Val 08:20
autrijus basically Show for Eval Val.
instance Show (Eval Val) where ... 08:21
lightstep i mean, adding an expression to the end of the input line
autrijus oh. yeah.
line 181.
lightstep the expression would have to run arbitrary Eval actions
autrijus instead of retVal, return a Env/Val tuple. 08:22
I think that's all there is to it.
the Env/Val tuple of course needs to be made a Val.
Val VEnv (Env, Val)
something like that.
lightstep maybe there is a place in the Env structure to do things after the end of evaluation? 08:23
autrijus and with the final pad?
gaal my Array $a; ... should $a be defined now?
Darren_Duncan FYI, the bulk of the Algorithm::Dependency port is now done. There are still some parts that are known to be bad, but they can be addressed later.
autrijus Darren_Duncan++
Darren_Duncan If any of you spot something that stands out as wrong, tell me or fix it.
gaal wow
lightstep i don't really know what a pad is
nothingmuch morning 08:24
Darren_Duncan++
Darren_Duncan Autrijus, about your changing the /modules file types ...
nothingmuch lightstep: where lexical variables are stored
lightstep perhaps i'll add an envCaller
autrijus lightstep: there is an envCaller.
Darren_Duncan I did a few compares and they showed that nothing actually changed in the files
autrijus lightstep: it refers to the pad of the dynamic scope's caller.
nothingmuch env ~~ pad?
or is env a sort of unification 08:25
of pad + dynamic + global?
autrijus nothingmuch: env's envLexical is the pad.
Darren_Duncan mind you, I only used the OSSF web compare
autrijus Darren_Duncan: right, because on unix that file types means nothing
Darren_Duncan I assume the update is more like a 'touch' in this case
scw nothingmuch: hello, there are some tests in t/base/operator.t (line 28 ~ 36)
autrijus it's on win32 where they made difference.
nothingmuch: see AST line 487
Darren_Duncan It was just that when I did an update, it said all those files were updated, though their contents didn't change
scw nothingmuch: that would be putted to t/op/precedence.t ?
Darren_Duncan anyway, no concern
autrijus lightstep: so, to answer your question, I'm inclining toward something even simpler 08:26
nothingmuch scw: i think so
Darren_Duncan If anyone's interested, tomorrow I'm speaking at my local Perl Mongers about what I've just been doing here, writing Perl 6.
scw nothingmuch: Those are added by me so feel free to move them :)
Darren_Duncan Are all of you in a local group? 08:27
or 'any
lightstep autrijus, adding another caller at the top seems simple enough
autrijus lightstep: ok, then go ahead :) 08:28
lightstep: I'll comment when I see your code :)
nothingmuch scw: i'll get to that err, later
very busy at work now
there is a test case that tests their interaction 08:29
t/op/precedence.t does everything like this:
given two operator precedence levels
and and ?? :: are pretty far apart
adjacent ones
but it looks like it's a test for a bug
am i right?
autrijus lightstep: anyway, I'll elaborate my last pont
I think a Syn "dump"
Darren_Duncan Question re porting ...
lightstep what is App name invs args ? what is invs, actually
Darren_Duncan Is there any preference, when a Perl 6 module is meant to match a Perl 5 module ... 08:30
autrijus that is specially taken care in reduceStatements
that feeds the lastVal and env back
lightstep autrijus, that seems right
autrijus is simplest. but feel free to pursue other ways that seem better to you
Darren_Duncan Should I make the version numbers the same as the Perl 5 version, or start them over at 0.0.1?
I've been doing option 2 for now
autrijus lightstep: invs is invocants; args is non-invocant args.
lightstep ok
no
autrijus $inc.method($arg);
lightstep so why many invocants? 08:31
autrijus because that's what MMD is for.
builtins use multimethod dispatch to determine which functions to call.
lightstep ok
Darren_Duncan Q: which Synopsis doc talks about file-handles, remind me
autrijus Darren_Duncan: I prefer 0.0.1 just for sanity 08:32
i.e. avoid expectation that it will match bug-for-bug with p5 counterpart
but you as the author gets to decide :)
Darren_Duncan autrijus - okay, I'll leave it that way while modules are in the Pugs distro - authors that publish separately can use whatever system they want 08:33
autrijus yup.
Darren_Duncan One of the main reasons I put it at 0.0.1 is that the Perl 6 version is very raw and untested ...
While a higher number like the Perl 5 one implies it is well tested and complete
lightstep autrijus, my current idea is this: add a new Syn, "dump", which dumps local variables to global variables
Darren_Duncan Here's what I can do ... 08:34
1. use 0.0.1 for brand new and untested ports
2. when all the tests pass and the module is at feature parity with v5, it will zoom up to match the Perl 5 mod's version number 08:35
so unless there's objection, that's my new pattern
is it reasonable to add something about version numbers like this to a general Pugs FAQ somewhere? 08:36
autrijus lightstep: that worksforme as the first cut.
Darren_Duncan: you can certainly add it to the kwiki :)
Darren_Duncan does Pugs support version numbers yet? 08:40
... eg, module long names? 08:41
I'm thinking of sticking a version number on Sample-Module to encourage better coding practices
scw nothingmuch: Oh, were you talking to me? Away before. 08:43
nothingmuch: yes, those three are bugs, in fact, parsing bugs.
Darren_Duncan I did it anyway
nothingmuch scw: ok... I think maybe they belong in operators.t if they are parsing bugs 08:44
or maybe in t/syntax
if they really test that even though op a and op b work ok
but they might not have distinct precedence, then that's what t/op/precedence.t is about
i don't mind changing this, but i think the location is good enough at the moment 08:45
scw But I think the main problems on those three tests are the parser doesn't 08:46
make ??:: have higher precedence than '=', 'and' or 'or'
According to S03, ??:: should have higher precedence than others. 08:47
autrijus correct. 08:48
Darren_Duncan ping Ingy 08:53
how long have you been planning this freepan thing?
I discovered you had 3 domain names registered, one back in 2002
autrijus hey Darren_Duncan. I think the author part in module.pm is not yet parsed 08:54
Darren_Duncan: he's been working on freepan for years.
Darren_Duncan: but it's a matter of copying the block in Parser.hs line 294. 08:55
Darren_Duncan autrijus, that's okay, I'm not using the Author part yet
autrijus hm. can I have the author part without the version part
or the version part without the author part
according to S11?
Darren_Duncan afaik, no
version always comes straight after the base name
I can look back at S11 ... 08:56
gaal isn't "use Dog-(Any)-(Any);" real usage?
autrijus gaal: mm? 08:57
Darren_Duncan according to S11, you can't leave the version out and use author, but you can explicitly say 'Any' for version and specify an author
autrijus oh. I mean in the module line
not the use line.
gaal ah, ok
Darren_Duncan so saying nothing in any slot is the same as explicitly saying 'Any' for its effect
in the module line, version seems to be required if you are going to give an author 08:58
still, I don't see a problem with that
version is used a lot more often
autrijus okie. 09:00
Darren_Duncan Also, I haven't yet decided which form of Author is best to use yet, so I'm simply not using it at all
gaal Pugs Porting Program?
Darren_Duncan it's a great feature, though
some people on p6l are planning to do one 09:01
make a program to port perl to perl
autrijus yeah, larry is working on it, and alias here has another technology that may be an alternate way to do it
gaal hmm, but "PPP" would make it p3; we're three Ps short.
lightstep what is the second argument ofo reduceStatements (the Exp) ? 09:02
autrijus lightstep: lastVal?
the value yieleded from the last reduce 09:03
it's there in the anticipation that some day we may need to refer to the last value.
lightstep so should i pass it on if i don't use it? 09:04
Darren_Duncan there is already a mature-ish CPAN module for parsing perl 5 into a syntax tree -- I suppose that would be ideal for generating equivalent perl 6 from
it has a really short name
nothingmuch Darren_Duncan: PPI 09:05
Darren_Duncan sounds like it
nothingmuch alias is it's maintainer
=)
(i think)
Darren_Duncan I just looked that up by name, it is what I'm referring to 09:06
PPI
search.cpan.org/dist/PPI/ 09:07
if 'alias' equals 'Adam Kennedy' equals 'adamk' then yes 09:08
nothingmuch i think that expr is true 09:09
Darren_Duncan heck, even The Perl Foundation funded completion of PPI, so it must be big enough to use for something like this -- second phrase sayeth the POD 09:10
seeya 09:11
one more thing for those on Mac OS X ... I just , earlier today, installed SubEthaEdit on my machine ... so should there be a future collaboration effort that I might be able to contribute to, I have the means 09:29
apparently it was used to great effect yesterday
gaal anyone have any luck with subclipse and pugs? i can't seem to co the whole project, just subdirs from it?
Darren_Duncan that is all
gaal e.g. www.geotools.org/Subclipse+Checkout <- don't have a "trunk" dir in the structure 09:30
kungfuftr komodo++ # automagically synced with cvs
autrijus gaal: it works if you just give it the url 09:31
instead of opening from a repository..
it is perhaps simpler
if you just quick eclipse
goto workspace
check out pugs
and add "pugs" as a project. 09:32
it will automagically work that way
gaal ok, and import from filesystem?
autrijus no, just create a new proj 09:34
eclipse is smart enough that if there is already a directory with the name in the workspace 09:35
it will just assume its settings from there
Juerd Khisanth: No
Khisanth: true is not a list operator. It is "not", but negated. Or well, "not" is "true", but negated.
webmind #define true !false ? 09:36
Juerd no
webmind sounds like dos code
i know
Juerd It's not a value
webmind ok 09:37
Juerd "true $foo" return true if $foo is true, false if $foo is not true
The *values* for true and false are bool::true and bool::false
Made hard to type because you should never.
webmind ok 09:39
#define 0 !1
Juerd sighs
This isn't C
webmind I know 09:40
sorry
Juerd And true and false aren't to be used as values
webmind yes yes
I got that
Juerd So their values are highly irrelevant and uninteresting.
They could be "yes" and "no" for all we're supposed to care
autrijus so maybe "true" should be spelled "indeed"
;)
Juerd autrijus: I was looking for a better name, you found it.
autrijus: "true" is a terrible name, violating the least surprise thing. Or possibly: raping it. 09:41
autrijus haskell has it as "otherwise".
but anyway "true" is a bad bad bad name.
oh nvm, it's not otherwise in haskell.
Juerd Yes, indeed that's true. :)
autrijus it's called "id" in haskell ;)
Khisanth Juerd: but S03 lists it as a list op :/ 09:43
Juerd Khisanth: That's weird.
I can't imagine what it should do with the extra arguments. 09:44
And it having low precedence doesn't help making it a listop either
autrijus p6l! p6l! change it to "indeed" or "id"! :) 09:45
we can always argue that "id" is shorthand for "indeed"
Juerd It says not is also a list op.
I don't know how to interpret that.
idd
autrijus indd
caddadaddadr
Juerd idd is what Dutch people use for "inderdaad" (indeed), and what every english speaking person also understands as "indeed" :))
Khisanth maybe it acts a bit like undef :)
turns everything in the list to true 09:46
Juerd How useful is that? 09:48
We have @foo ōæ½xBB= 1 for that
gaal ghc 6.2.2 is out on windows/mingw, right? 09:50
(waiting for the 47mb 6.4.0 do dl...)
Juerd autrijus: sent 09:52
I had to read my own message three times before I understood what I had written :)
autrijus Juerd: danke
gaal checked in undef.t in manifest + some typo fixes 09:54
autrijus gaal++ 09:55
nothingmuch gaal: i just got a plan error in undef.t 09:56
is it fixed in this update?
gaal should be 26
shouldn't there be a way to autoplan? which is better than no_plan, but still lets you not moose aroudn with explicit numbers 09:57
nothingmuch autoplan how?
gaal anyway nothingmuch, please do try again this time 09:58
nothingmuch you could just read $?FILE
but see t/base/given.t for how to break that
gaal nuffin, i once saw someone doing push @tests, sub { a test } .... lots of times, then planning, then running.
nothingmuch well, that someone was obviously drunk or something ;-) 09:59
i would never stoop so low ;-)
either way, i think with the current state of Test.pm in pugs it's better to either plan or no_plan
DrHyde does anyone have a recent enough ghc binary package for debian stable (x86) or OS X 10.2? 10:00
nothingmuch just update the plan when you change things
DrHyde getting it to build is proving to be a right bugger
nothingmuch it's supposed to be a human-updated sanity check anyway
gaal okay, still gettgin basic compilation errors with ghc 6.4 on mingw:
Chasing modules from: src/Main.hs
Could not find module `System.Posix.Unistd':
use -v to see a list of the files searched for
(imported from src/Posix.hs)
make: *** [pugs.exe] Error
what, windows isn't posix compliant? *feigns shock* 10:01
autrijus ... 10:04
src/config.h needs to say that it doesn't have POSIX I think.
did it work with ghc 6.4 on activeperl?
(it always workforme so I can't quite tell)
gaal dunno, i have mingw and cygwin on this box 10:05
nothingmuch autrijus: err, remember t/op/yen.t false-negatives due to prec? 10:06
autrijus nothingmuch: yes?
nothingmuch now the same issues are biting me in t/op/prec.t
but the other way around
please look at the test
and the spec
and the impl
and tell me which of the 3 is insane or wrong
i believe it's s03
(insane)
autrijus I believe S03 is intentional
nothingmuch well, regardless 10:07
tests based on it are not passing
autrijus nod. ok.
nothingmuch btw, we have some tests that look like they should be todo 10:08
inc.t, subroutine_named_params
should I butcher them?
autrijus bbiab. 10:19
nothingmuch ciao
autrijus nothingmuch: yes, go ahead.
10:28 Odin-FOO is now known as Odin-
jabbot pugs - 754 - typo fixes 10:33
pugs - 753 - undef.t updates
pugs - 752 - op3 "index" -- simplified the code
pugs - 751 - index works now -- but the code is quite
pugs - 750 - Fixing t/op/index.t
pugs - 749 - added version numbers to Sample-Module
pugs - 748 - Another precedence problem on '='
pugs - 747 - AlDep updates - changed all 'foreach' to
pugs - 746 - * dvergin reported that -isrc needs to b
pugs - 745 - First shot at tests for undef.
nothingmuch autrijus: ping 11:03
jabbot pugs - 756 - Global var test case fixed 11:12
pugs - 755 - Todoify things that look like unimplemen
pugs - 757 - Fudge things into todo state, make it co 11:42
theorbtwo Morning, guys. 11:44
nothingmuch morning theorbtwo
Juerd perlmonks.org/?node_id=439594 11:46
Let this please kill this useless debate.
s/useless/fruitless/
11:56 sbkhh is now known as Odin-
theorbtwo Sigh. Juerd, you left out a critical fact that was in your power to give: print $fh ~~"Some string"; is, in fact, recommended syntax in a CPAN module: i18n. 12:03
It's certianly not a show-stopper in my mind -- I think it should go in. 12:04
But I'm a big believer in full disclosure.
You might also consider running a set of PM obfus through a B::Concise both before and after your patch, and comparing them. 12:05
nothingmuch print $fh ~~"some string"; # wtf is that?
theorbtwo It runs "some string" through bitwise negation twice, and prints the result out to $fh. 12:06
(The result being "some string" except in the presence of strange overloads.)
Juerd nothingmuch: print $fh ~~localtime is a better example
It's the lazy way to write print $fh scalar localtime
nothingmuch ah 12:07
oi vey
theorbtwo Yes, but nobody recommends actually using that syntax. 12:08
Except golfers, who I agree can probably be ignored.
Juerd theorbtwo: And when would you print an i18n'ed string to a fh?
theorbtwo Huh? Well, me, rarely, because I think everybody should just learn English, dammit.
People in general? All the time.
Producing a localized report.
In a CGI. 12:09
Juerd theorbtwo: And still, it's not ambiguous in the case of print ~~localtime or $fh->print(localtime) or print$fh+~~localtime or print$fh~+~localtime or print $fh~ ~localtime or print$fh(localtime)
theorbtwo: With i18n?! 12:10
theorbtwo I know, Juerd! I'm not arguing that it's a showstopper!
Juerd Possible but not likely.
theorbtwo No, not with i18n.
crysflame hi, juerd
Juerd Hello, crysflame
theorbtwo I'd more likely use Locale::MakeText, which looks very sane.
Juerd And Perl 5 has a special disambiguator noop for this ;)
+
So instead of print $fh ~~localtime, use print $fh+~~localtime, it's not even more characters 12:11
theorbtwo I KNOW!
You're being overly defensive.
Juerd (Though the space between $fh and ~~localtime is optional, I think anyone not using it is wrong anyway)
theorbtwo: I'm trying to be very, very clear
That there is no problem.
theorbtwo You're wrong; there is a problem, but all indications is that it is a very very very small problem. 12:12
Juerd Before I believe that there is indeed a problem, I want to see this syntax used in actual code
It's contrived
I don't believe *anyone* actually has this in existing pre-today code. 12:13
And all the golfs continue to work anyway, unless golf prints to an fh, which I don't believe either.
In other words: if there is a problem in theory, but not in practice, the theory is wrong :) 12:14
12:21 wilx` is now known as wilx
theorbtwo Juerd, would you like a random sample of the code bits of N obfu off of PM? 12:21
Or, for that matter, N random nodes?
crysflame purl: mahna 12:24
purl: mahna mahna
argh, hi #perl 12:25
12:25 theorbtwo is now known as hash-perl6
hash-perl6 Hi, crysflame. 12:25
12:25 hash-perl6 is now known as theorbtwo
theorbtwo Mornin, lightstep. 12:26
lightstep good afternoon
theorbtwo Good morning, hlen. 12:27
Why do people always join a bunch at a time?
lightstep i joined 'cause i committed
which is because lunch is over
theorbtwo wonders if anybody fixed his assert_ghc thing. 12:28
lightstep autrijus did
tests are starting to fail all over the place 12:29
nothingmuch lightstep: i have a loop, which watches svn up and pops up a window on test fails 12:31
afaict it's no different than what it was several hours ago
except some fixes i made
jabbot pugs - 758 - more work on the command shell: it's now 12:32
lightstep nothingmuch, i'm not on that scale. i remember that once all tests passed, now some fail 12:33
nothingmuch ah 12:34
Juerd theorbtwo: Why?
nothingmuch well, undef.t is sort of ok to fail
gaal hasn't decided whether it's bugs or todo
autrijus can prolly veto it
i also committed a test that looks like it uncovers some bugs
it's causing 3 fails
lightstep all rest are todos?
kungfuftr me-- # bah 12:35
lightstep ?
nothingmuch and scw also submitted a test case for a bug
kungfuftr inc.t foulup
nothingmuch lightstep: there should be 9 subtests failing as of head revision
kungfuftr: nevermind, i cleaned it up
kungfuftr =0) 12:36
nothingmuch it wasn't that bad, btw ;-)
kungfuftr shouldn't commit stuff when he's stressed out
lightstep yes, 9 fail
gaal you guys get a syntax error on t/syntax/subroutine_named_params too? 12:37
nothingmuch lightstep: they have all been investigated
gaal: i fixed that
lightstep kungfuftr, i did it yesterday. fucked up some part of Main.hs
nothingmuch lightstep: apart from undef.t where I don't know if they should fail, all should fail because they are bugs
gaal i was actually waiting for autrijus to say if the undef.t fails are ok or should be moved to todo
nothingmuch ok, so lets leave it as fail 12:38
gaal "todone" :)
i have 2 new trivial cases, but it's not worth a commit. also, there are about 20 tests i imagine waiting to be written in undef but they are majorly todo (rules, subs etc).. should i write them in eval? 12:40
theorbtwo Sounds good.
lightstep is the parsing of n-ary operators working? 12:41
theorbtwo N-ary prefix operators? Should be.
gaal so: i say "my Array $an_ary;" . Should $an_array be defined now (my guess is not)? 12:42
lightstep i mean, is $a Y 4b Y 4c different from ($a Y $b) Y $s as they should be?
kungfuftr lightstep: yar, i'm a perfectionist though, so it doesn't bode well for me
theorbtwo Oh, that'd be a listop (All listops are unary, at least as far as Prim.hs is concerned.) 12:43
lightstep looks at the screen
the way i wrote the expressions, they don't seems alike at all 12:44
Juerd theorbtwo: Why? 12:45
theorbtwo I have no idea.
Juerd Then why ask?
lightstep Juerd, they get a list of arguments as a single parameter
theorbtwo Why ask what?
Juerd No, re PM obfus
theorbtwo Oh. 12:46
Because it's a selection of perl code that's not written to standards.
I thought you were asking about listops.
Juerd Sure
Do you have a screen scraper for it?
nothingmuch gaal: you should use eval everywhere it doesn't parse or is fatal to run, but other than that, as little as possible, IMHO
lightstep: infix nary works now 12:47
but (@a Y @b) Y @c is broken as a result
look at t/op/zip.t
autrijus said it will take a lot of work
theorbtwo I'm a god, Juerd; I can find ways of getting what I want. 12:48
Juerd Ah
I'd like a sample, yes
afk # need to go
nothingmuch ciao Juerd 12:49
lightstep the whole parsing module looks totaly irrational (or maybe unhaskellish)
nothingmuch could be ;-) 12:50
gaal nothing, clearly, but the question was whether to go ahead with wiriting tests that knowingly need the eval. i decided yes
nothingmuch either way, i know that nary and listop are sort of half-baked solutions, which should be better implemented with MMD
theorbtwo I can't think of many reasons why I'd not want a correct test, and all of them are bad reasons.
nothingmuch gaal: provided that the test is correct 12:59
gaal well, i hope it is, but it's kinda hard to say :)
nothingmuch i.e., when the feature is there it will work both with and without eval 13:00
well, 80% of the work has been done, in theory
gaal classes? types? subs? rules? 13:01
lightstep aren't subs done? 13:02
nothingmuch lightstep: some 13:03
but it's a big thing
gaal autrijus said no, yesterday at least.
nothingmuch syntax/subroutine_named_params tests struff
unspecced/cont* does other things
but i reckon there's lots more 13:04
err, objects should be fun
but i reckon it won't work till it all works, for the most part
start by trying to test moderately complex examples 13:05
like a class with delegation
see what you hit on the way
(class defs, opaque type, isa_ok, accessors, instance variables, etc)
and test that too
lightstep how should i document modules? 13:06
nothingmuch lightstep: theoretically in kwid
but pod is probably also ok
lightstep is there an introduction to kwid? 13:07
nothingmuch to kwid i'm not sure
let me look at some irc logs
i'm not sure whether it's a spec draft, or an intro 13:08
to get a conceptual you should read Juerd's pod tutorial
kwid and pod are similar enough for it to be beneficial 13:09
gaal anyone get tests to run in Eclipse, please say how in the wiki
nothingmuch ./ext/Pugs-Documentation/perlkwid.kwid 13:10
./ext/Pugs-Documentation/perlkwidspec.kwid
oh, he left
decay is kwid as slow as kwiki? 13:13
nothingmuch decay: ingy should probably be up in a few hours 13:14
he should be able to answer that, i guess
stevan morning nothingmuch 13:29
theorbtwo Allo, fayland, b6s 13:32
And stevan, I suppose.
fayland hellp theorbtwo
what;s the meaning of b6s
stevan morning orbtwo
fayland o ,i see . a man 13:33
iblech Hi, there seems to be sth. wrong with (implicit) numifying 13:39
./pugs -e 'say int +"1.01"' ==> 1 (correct) 13:40
./pugs -e 'say int "1.01"' ==> 0 (wrong)
nothingmuch morning stevan 13:41
iblech The problem doesn't seem to be related to "int": ./pugs -e 'my @a = <a b c>; say @a["1.01"]' ==> a (wrong, correct is b)
theorbtwo BTW, iblech, no reason to have your bot log the channel; there's already logs at xrl.us/e98m 13:42
iblech theorbtwo: Jup, know, but my webinterface is better :) see m19s28.vlinux.de/iblech/iblechbot-2.png 13:44
theorbtwo Hm, does look quite slick. 13:46
iblech theorbtwo: Thanks :) 13:47
scw Seams because in ghc, reads "1.0" becomes [] 13:52
nothingmuch is anyone writing a test for this? 14:07
autrijus greetings! 14:09
nothingmuch hola autrijus
autrijus yo. what's up? 14:10
theorbtwo G'morning, autrijus.
Looks like ibelch found a bug: ./pugs -e 'say int "1.01"' ==> 0
nothingmuch is about to write a test, I think. ;)
pasteling "iblech" at 80.81.8.11 pasted "Test for bug regarding numifying of Strings" (4 lines, 93B) at sial.org/pbot/8207 14:11
autrijus iiinteresting.
nothingmuch =)
gaal hey. r760 adds more TODOs to undef
autrijus nice
gaal autrijus, there are three failures i'm not sure are bugs or TODOs
autrijus sure. what are them?
jabbot pugs - 760 - More tests (many TODOs) 14:12
pugs - 759 - * add back 'eval' for ghci debugging; 'm
Juerd That SubEthaEdit thing is so cool, I might get a mac mini just to see people hack :)
nothingmuch Juerd: save money
gaal aggregates: how do they work with undef and defined()?
nothingmuch pearpc
Juerd pearpc?
nothingmuch it's a ppc/chrp emulated for i386
theorbtwo PPC emulator. 14:13
nothingmuch or maybe not only for i386
Juerd That measn I need an OS X license that I can never get OS-less hardware for
nothingmuch think bochs, only the other way around
theorbtwo It's what I use for SEEing, but it's a bit on the bugy side.
gaal if i say undef @ary, shouldn't it be !defined()?
autrijus I'd say so, yeah. 14:14
theorbtwo No; definedness is a property of a scalar, not an array.
nothingmuch err, $100 for osx vs $500 for mac mini
=P
autrijus uh.
I thought undef is a value and undef anything returns that.
mattc Err, the need to define sub before they're called... is that just a pugs implementation issue? (just starting with perl6/pugs - is this the right place for these kinds of questions??)
autrijus mattc: that is a pugs TOOD.
err, TODO.
gaal ok. oh, also, %hash{"bar"} ; ok(!defined %hash{"bar"}) does not obtain. 14:15
mattc ah
cool
autrijus mostly because we had not had BEGIN support.
theorbtwo That'd be a bug, gaal.
autrijus and that's mostly because $real_work :-/
gaal: that is definitely a bug.
theorbtwo Right, it returns an undefined value, which can be stuck in an element of an array.
Juerd nothingmuch: But when you look at it that way, that makes the mac mini a $400 machine
mattc ok
Juerd nothingmuch: Which runs Linux and is thus by itself much worth it. 14:16
nothingmuch Juerd: well, sure
Juerd doesn't have a lot of money, but likes spending it wisely
autrijus oh btw.
int '0x123'
Juerd And that doesn't always mean choosing the cheapest option
gaal well, the same happens with delete %hash{"bar"}.
autrijus is 291
nothingmuch autrijus: bottom of t/op/inc.t - are those parse bugs? misimplementations? or what?
mattc where is the TODO - is it in the distro somewhere?
autrijus not sure if it's a bug.
I'd argue it's a feature ;)
theorbtwo I'm not sure -- ask p6l?
Juerd autrijus: IMO, strings should numify as they would if they were literal numbers
nothingmuch Juerd: /me just bought $200 earphones and argues it's one of the best buys he ever did 14:17
autrijus Juerd: that is what pugs do.
Juerd This has afaik not yet been discussed on p6l, but I have been meaning to
autrijus Juerd: raise it? I hope to get a quick "ok" from larry
Juerd But perhaps someone else should start the thread
autrijus: I've raised too much already, I think.
decay Juerd: btw: vim.dindinx.net/orig/html/remote.txt.php :)
nothingmuch decay: subetha is more than that
it's collaborative
decay i know 14:18
nothingmuch vim can do that too?
Juerd decay: What about that?
Oh, it's about the document 14:19
decay: Looks very complex
mattc autrijus: where is the authoritative TODO list? 14:20
Juerd nothingmuch: I should note that I have been looking for excuses to buy a mac, and the mini made it affordable
nothingmuch mattc: grep -r todo_ t
mattc oh - I see
nothingmuch Juerd: it is a very good platform, albeit with some drawbacks
Juerd nothingmuch: Also, I have been looking for a smaller enclosure for my PC
mattc thanks
Juerd I don't know if I'd like OS X 14:21
I'd have to use it for a few weeks first
nothingmuch 's reasons for using it are that it mostly works
for every day use it's less config savvy
Juerd One thing that scares me is that you can't install a 2nd harddisk in the Mini
Which means no RAID 1
Which is scary.
nothingmuch but i don't like configuring everything
Juerd I have the same problem with my laptop.
nothingmuch given a good enough toolset and something that works good enough out of the box, i'm happy 14:22
and osx has that for me
Juerd KDE has that for me mostly
nothingmuch what if you yank out the cdrom?
Juerd But I'm willing to give OSX a try
nothingmuch: That means I don't have an optical drive anymore.
decay Juerd: you can raid to an usb disk
webmind Juerd, usb/fw drive
Juerd decay: Won't that be very slow?
gaal uhhhh, this gives an error:
%hash{"bar"} = "baz";
nothingmuch usb2 or fw - shouldn't notice 14:23
gaal when that key already exists
decay Juerd: if it's usb2 or firewire it'll probably be acceptable
nothingmuch gaal: test test test!
decay Juerd: the builtin disk is 2.5" anyway
gaal it kills the test, nm
webmind Juerd, replace the optical drive with a disk and use a fw/usb optical drive ?
nothingmuch gaal: even in an eval? 14:24
webmind hmm, /me has very bad experiences with usb2 and duplex
pasteling "gaal" at 192.115.25.249 pasted "can't reassign to hash elem?" (217 lines, 5.5K) at sial.org/pbot/8209
Juerd webmind: I don't know 14:25
webmind Juerd, what ?
jabbot webmind: what is gugod?!
Juerd webmind: The small size is practically gone with an external optical drive
gaal nuffin, please confirm at sial.org/pbot/8209
webmind jabbot, sorry ?
jabbot webmind: Tell me more about that.
webmind jabbot, about what ?
jabbot webmind: Tell me more about that.
gaal no, not in an eval
webmind darnit 14:26
gaal sorry, i lagged
nothingmuch then test in an eval =)
Juerd webmind: It's a /bot$/
webmind Juerd, yeah.. got that :)
Juerd, yeah.. probably the same if you want an external drive for raid ?
gaal so, uh, yeah, even in an eval! 14:29
nothingmuch oi
Juerd webmind: No, that'd be 3.5"
nothingmuch bug autrijus
gaal autrijus: ping
webmind Juerd, portable slimdrive isn't that much bigger in volume ?
Juerd Does anyone have experience with 1.2 versus 1.4 GHz PPCs? 14:30
Is the .2 GHz worth $100?
gaal looks like serious breakage in hash -- or maybe i'm misusing the syntax?
Juerd webmind: But much more expensive than cheap harddisk + enclosure
nothingmuch ppcs usually don't need much mhz
gaal well, i'm ci'ing this then.
Juerd nothingmuch: Which means 200 MHz is a big improvement? 14:31
nothingmuch no
webmind Juerd, true.. but I'm not so sure about the performance for usb disks for raid
no dma..
gaal r762
nothingmuch r761 - "1.01"
Juerd Ah well, I'll go for the expensive option I think 14:32
jabbot pugs - 762 - Seems to expose fatal bug in hashes
pugs - 761 - int("1.01") eq 0 fails
Juerd It takes them 2 weeks to deliver one. That's LONG.
webmind Juerd, 17% speed increase
nothingmuch webmind: not bloody likely 14:33
webmind Juerd, your going to do raid over usb/fw ?
nothingmuch the bus is probably the same speed
webmind nothingmuch, uhuh
Juerd webmind: No, no raid
External backups
nothingmuch and ram/disk are usually the limiting factors
webmind ah k
Juerd I'm getting 512 MB anyway
256 is very little
webmind yes
I heard memmory @ macshops is expensive though 14:34
nothingmuch it is
but i'd invest in 1gb instead of 200mhz more
gaal i've no experience with macs, but i'd not buy a computer with less than 1gb ram today
Juerd nothingmuch: That's $250
Not going to do that
Maybe I'll upgrade later
gaal check if you can buy non-apple maybe?
Juerd But $250 extra is too much for now
webmind if the bus speed is staying the same.. I don't the the 0.2ghz is going to matter much ? 14:35
Juerd gaal: I want a small box and I've been wanting to try an Apple for a while
gaal for PCs, 512MB is about $100 where i live
webmind well more heat maybe
Juerd gaal: No PC enclosure is this small and this silent with this processing power.
gaal no i mean
only the RAM
Juerd gaal: Oh, right. Probably.
webmind gaal, elcheapo here is 60euro
gaal they probably OEM it from kingston anyway
Juerd It's SO-DIMM though
Which is not as cheap as normal dimm
webmind Juerd, don't buy the mem from mac ? 14:36
Juerd How about: get 512 MB now, save for another 512 and then add that
webmind could do that
Juerd There's no reason to rush things
webmind uhuh
512mb is enough usually
Juerd Except I find 2 weeks very long
Aankhen`` How about: donate the money to Aankhen``? :-D
Juerd In 2 weeks, pugs will have all of Perl 6 implemented 14:37
And by then I can no longer see SEE in action :)
gaal just make sure you're not screwed with the wrong config -- ie that you have a free slot and all that
Juerd gaal: Apple's very clear about that. The option explicitly says it's a 1 dimm config with 1 free
webmind Juerd, find a retailer with them in stock ?
Juerd Oh, hm, no
Just that it's 1 dimm
I think it'll have 2 slots though
And if not, then my current choice doesn't matter anyway :) 14:38
gaal :)
Juerd webmind: Good idea. Apple retailers in our country?!
webmind Juerd, sure
Aankhen`` Have rules been implemented?
webmind Juerd, but now to find one with mini's on stock :)
Juerd, afaik there are 2 in the hague
Juerd webmind: Dutch people think apples are fruits. Where are these apple retailers you think exist?
webmind vt100 sells apple I think.. but they don't have any on stock 14:39
and there's another one in the hague
that might have stock
Juerd url?
webmind and there is one in amsterdam I think..
lukkie bought his mini there afaik
Juerd, good one :) 14:40
Juerd, goudengids ?
gaal needs a break. see you later.
webmind Juerd, sounds bad.. more people complaining about backorders 14:42
Juerd What does? 14:43
nothingmuch Juerd: i'd be nitpicky and equip it with both wifi and bluetooth 14:44
webmind Juerd, www.ljs.nl/ 14:46
Juerd nothingmuch: Doesn't it support USB dongles at all then?
The airport is expensive
nothingmuch airport is good
decay hm, the bluetooth adapter is overpriced 14:47
nothingmuch bluetooth for $30 more is about as expensive as a dongle
at least now
Juerd webmind: baggersite :)
nothingmuch if it's bundled
Juerd nothingmuch: What's so good about airports?
nothingmuch Juerd: laptop fun
Juerd It's not a laptop
nothingmuch it can be a base station 14:48
Juerd (It's smaller... :P)
nothingmuch it can be away from cables if needed
webmind Juerd, true
Juerd I have an access point. In fact, I have three.
nothingmuch ah
in that case, don't bother ;-)
Juerd And ethernet cables everywhere :)
practically
"Macs met ervaring zijn ideaal voor:
- Windows gebruikers die op zoek zijn naar een echte computer"
nothingmuch would never ever want to have more than 2-3 computers 14:49
Juerd ("Experienced (used) macs are ideal for: - Windows users looking for a real computer"
)
webmind nothingmuch, why not ?
nothingmuch too much trouble, configuring, etc
Juerd nothingmuch: I have two computers: one workstation at home and a laptop, which is my workstation at the office
nothingmuch i want one puter to do the lifting
one puter to call my own
Aankhen`` I need three compooterz.
theorbtwo We've got six in our appartment, for two users. 14:50
webmind has 2 wireless computers/routers, a laptop, a laptop from work and a pda
nothingmuch and maybe another one for playing around, or being a sort of backups
webmind most stuff just works
plan on building a development machine
Juerd apple-- # wont ship USD minis outside the US
webmind and one for hosting
Juerd And they use 1:1 EUR:USD
webmind Juerd, uhuh
theorbtwo Wow. That's harsh.
Juerd Smart, though. :)
Oh, I need the dutch one 14:51
For the BTW
(VAT)
theorbtwo You could get somebody in the US to recieve it, and re-ship it.
webmind Juerd, the inc btw price is also the same between germany and the netherlands
theorbtwo Might still be cheaper.
Juerd webmind: Then .nl is better :)
Cheaper for me
theorbtwo No import duties if you mark it "gift".
webmind uhuh
theorbtwo, right
theorbtwo, doesn't work
Juerd theorbtwo: Enter dutch customs :(
They're actually good at what they do 14:52
webmind theorbtwo, if they estimate the value above somethling 20 euros
theorbtwo Oh, I thought that was an EU-wide thing.
webmind there's nothing EU wide but fucked up laws
and eu subsidies
lightstep copyright policy?
Juerd apple++ # they manage to convey lack of keyboard and mouse as something positive
webmind Juerd, how ? 14:53
nothingmuch Juerd: don't forget that in the US they don't include taxes etc
theorbtwo Oh, yeah, forgot about that.
nothingmuch because law doesn't require it
theorbtwo Well, that's not true.
They need to pay sales tax if they have a presence in the state in question.
webmind nothingmuch, neither do you have to pay taxes in certain states or if you live in a different state 14:54
theorbtwo But I bet they're real crafty about that.
Juerd nothingmuch: Customs will add them
nothingmuch webmind: true, but err, stuff usually compilcates onwards
webmind Juerd, but you call/visit the retailer ?
Juerd webmind: Their site sucks
webmind nothingmuch, never had issues with it
nothingmuch when ordering from abroad i always add a 30% overhead
webmind Juerd, yes.. but I hear the shop is good 14:55
they're not a webshop
Juerd The ljs thing?
theorbtwo Last time I got something shipped internationally that wasn't marked as a gift, I ended up paying >100% duties on it.
webmind yes
Juerd Do you know where it is?
webmind yes.. the hague
Juerd ... :) 14:56
webmind www.ljs.nl/Resources/kaartje.jpeg
map
Juerd Are they likely to have minis in stok?
Ah, I see the route page now
webmind the thick black lines are the trainrails
Juerd, dunno.. call 'm ?
autrijus rehi. 14:59
I've threw in a cheap hack to get 1.0 parsed as int.
nothingmuch hola autrijus
autrijus reruns the tests to see what's failing 15:00
jabbot pugs - 763 - * cheap hack to get '1.0' to VIntify 15:02
nothingmuch autrijus: t/base/num.t passes now?
iblech nothingmuch: t/base/num..........................ok
lightstep autrijus, did your fix cancel the whole non-integer numbers? 15:04
nothingmuch forgot if it's todo_is or is
lightstep s/the whole/all the/
iblech nothingmuch: they're isa_ok and is 15:05
nothingmuch yeah i see now
lightstep err 15:06
webmind Juerd, tried machouse in rotterdam ? 15:07
autrijus lightstep: my fix just takesWhile (/= '.') 15:08
lightstep: again, cheap hack
ingy hola
nothingmuch autrijus: lumi asks if Ā„ is lazy 15:09
i assume all listops are lazy, in a way?
theorbtwo Allo, ingy.
autrijus that is my interpretation also.
theorbtwo Sorry if I was overly harsh the other day. 15:10
ingy hi theorbtwo
theorbtwo: no worries. I should just get stuff done instead of spouting about it. 15:11
nothingmuch Juerd: when you do buy a mac, i have some power user tips 15:13
like harnessing Kicker
and doing funny vt100 codes that make background compilation a breeze
autrijus scw: I'm re-reading your ??:: writeup now. it's helpful. 15:14
ingy autrijus: I've never seen MSWin32 use C:\usr\lib for privlib 15:15
it can't be an ActivePerl
autrijus ingy: AP can change install path. 15:18
which is likely what happened
Juerd webmind: No
autrijus or maybe no AP, but MinGWPerl
but the idea is that we shouldn't bork out...
Juerd nothingmuch: Kicker?
nothingmuch: And 'breeze'?
webmind Juerd, it's near rdam CS, so not too far
nothingmuch breese? 15:19
breeze <-- righter
Juerd Ah
nothingmuch easy, err, fun
Juerd nothingmuch: What is it? :)
Why easier/more fun?
nothingmuch Kicker is a nice, sort of hidden feature
webmind Juerd, nm.. macmini temp. sold out
ingy autrijus: definitely. I just put that in there to see if it ever would happen. I got my answer. fixing...
autrijus danke!
nothingmuch and another thing is some coolness (homepage.mac.com/uucee/macosx/termi...ints.html) to make your terminal effective 15:20
i have a script, for example, that runs our build system
it docks the window, and then raises it if it sees '***' in stderr, and raises it when it's done
same thing for the test harness
mui fun
nothingmuch writes a growl based rss reader thingy 15:30
XML::RAI + XML::RSS::Timing + Mac::Growl
anyone intersted, bug me in a while
ingy autrijus: did not you say yesterday that you can assign to %Config? 15:31
ie $Config{archlib} = 'C:\usr\lib';
nothingmuch hmm... maybe there was a misunderstanding 15:32
autrijus: is it per runtime?
or is it actually persistent?
autrijus ingy: I meant that you write write to Config.pm.
s/write/can/
nothingmuch: persistent.
nothingmuch but not via the tie interface?
ingy %Config::Config is read-only 15:33
autrijus not unless you override STORE.
which you can
just won't be pretty.
nothingmuch why not?
it doesn't need to be efficient or anything
autrijus I'd do with overlaying %Config instead.
nothingmuch it can just rewrite Config.pm
autrijus use Config '%Config'; my %c = %Config; untie %Config; %Config = %c;
nothingmuch and keep %c in a storable? 15:34
autrijus ingy: it's read only because it's tied... just untie it
something like that.
ingy our %Config = $Config::Config; 15:35
nothingmuch and for lookups etc, at runtime, perl internally goes to %Config?
ingy Odd number of elements in hash assignment at util/PugsConfig.pm line 6.
autrijus ingy: s'$'%'
nothingmuch tr/$/%/;
ingy ack
I'm not awake yet
nothingmuch =D
autrijus: our %Config = $Config::Config;%'
ingy nothingmuch: :P 15:36
autrijus yeah whatever :-p
Juerd webmind: hoe weet je dat? 15:44
ah 15:47
*overal* uitgeverkocht
gaal autrijus, there seems to be a fatal error inside an eval in the latest undef.t; can you take a look? 15:53
autrijus greetings pmichaud-san
pmichaud: we have pugscc now :D
pmichaud autrijus: wow
coolness
autrijus gaal: I'm fixing ??:: now. give me 5 min?
pmichaud ??:: is a real pain
autrijus it is!
I'm currently making them both binary leftSyn 15:54
and add a special rule for :: to eat ??
and report a syntax error when it's not seen
same for the lone ??
the problem of course is that they have to be alone in that prec level
which they are.
but I don't want to think about user-defined ternary at other prec levels.
autrijus frowns
pmichaud: let me know if you've found a solution to that in Rule mode 15:55
gaal sure, autrijus, i just needed som sort of ack :)
pmichaud I don't think we're allowing user-defined ternary
autrijus gaal: ack
pmichaud certainly not in 6.0.0
autrijus oh. whew.
what about user-defined operators that occupies equiv prec level
as ??:: ?
that will also throw many things off.
can we outlaw them too? ;)
pmichaud I don't even think there's going to be a name for ??:: that one could use
autrijus I thought it's &ternary:<?? ::> 15:56
but I might be imagining things
pmichaud lemme look
I hadn't seen a ternary: category yet
autrijus good, then.
is there a plan for statement:<> ?
pmichaud well, it's entirely possible I just overlooked it
lightstep so what are op3 and op4?
autrijus I've read lwall's post 15:57
Juerd autrijus: Instead of referring to yen as "the unicode form of Y", you can use &yen; in HTML, for use.perl.org, iirc
pmichaud I'm still wondering about statement:<>
autrijus but that is still fuzzy in my mind
Juerd: danke!
pmichaud I have a few ideas about statement:<>, but that's part of what was holding me back on a few items as well
my response back to lwall on that post (when he sent it to me a month or so ago) was that we'd probably use top-down parsing for everything down to expressions, then bottom-up parsing for expressions 15:58
autrijus right, I remember you saying that.
which makes sense.
pmichaud so statement-level items would tend to be handled via rec-descent
autrijus user-defined statement:<>?
how does user define a statement:<> ?
in particular the signature will look weird.
because it's not call-by-value or call-by-ref 15:59
it's call-by-ast
in order to make things like elsif{} work
pmichaud yes. One of my questions to luke was how else/elsif would work
autrijus crosses finger for an answer
pmichaud and I think the quick answer is that we're not going to worry about statement:<> too much, at least in 6.0.0
autrijus oh. ok.
a bit disappointing, but much more relief 16:00
ok, then I think pugs's parser is actually in a fine shape :)
pmichaud well, I'm not sure how well that particular aspect of the design has been thought through
autrijus nod nod
pmichaud we know that we want people to be able to modify the grammar, and statement:<> may have been an idea of an approach that could be used to do it
autrijus but the param type is so weird I can't quite understand it.
lightstep doesn't pugss' parser contain rip offs from other packages? 16:01
autrijus I mean, I can understand if we make the whole AST into objects
pmichaud but there may be better ways to do it, and it's better to just start playing with simpler implementations to figure that out than to try to spec every last detail in advance :-)
autrijus lightstep: rip offs? in README and LICENSE and Parser.hs I clearly say it's based on parsec
err, Rule.hs
lightstep why not import parsec instead?
lightstep doesn't read documentation&license 16:02
autrijus lightstep: because before pugs 6.2.0 I'm going to move Parsec into Eval.
lightstep: i.e. make it a monad transformer
operating on IORef
jabbot pugs - 764 - Fix for weird Win2k config
autrijus (or on ReaderT)
instead of a monad it itself
lightstep: as it stands we can't eval stuff inside parsec and retain the env.
or the callCC
but perl6 rules is going to require that.
and more immediately:
my $x = BEGIN { 2 + 3 }; 16:03
needs that as well.
pmichaud yeah, I'm doing pge work today/tomorrow
I'm hoping to have the new pge ready by tomorrow evening
I've also cleared the weekend for perl6 work
autrijus so I forked parsec into Rule.hs to work on it more. SyntaxNinja also did some cleanup. everything will go upstream when it makes sense to do so
pmichaud++ # nice!
pmichaud no kids, no girlfriend, no other jobs :-)
autrijus ooh.
pmichaud: so, PGE won't target parrot AST, right? 16:04
just pasm
ingy is that rule5 or rule6?
autrijus and/or pir
ingy ;)
autrijus ingy: rule6
;)
pmichaud pge is targeting pir, because we don't have a parrot ast :-)
autrijus right. but assume we have :)
pmichaud but pge is in bootstrap mode
autrijus (actually we have, it's just 1) not working 2) not documented)
pmichaud pge is written in c/pir at the moment, and will stay there until other things are far enough along that I can write pge in perl 6 16:05
or, more likely, until someone else decides to rewrite pge in perl 6 :-)
autrijus you know, if you write pge in perl 6 now, I can run it in pugs :)
ingy autrijus: nope. 5:
5) Be Obsessed. Great software is created by obsessed people. Integrate your projects into your life. Breathe them. Too hectic for your family oriented life style? Maybe you aren't cut out for this.
autrijus oh. that rule.
ingy haha
pmichaud ingy: don't worry, I'm obsessed enough :-)
ingy got flamed for that 16:06
pmichaud autrijus: I'm keeping the pge via pugs route open at the moment, yes :-)
autrijus :)
ingy pmichaud: you are in a fortunate position.
pmichaud ingy: that's for sure.
autrijus ok, I think I've fixed ?? :: 16:15
testing.
Limbic_Region autrijus - do you ever get a chance to keep up with use.perl journals relating to Pugs? 16:25
I ask only because I would be willing to keep you aprised of journal entries and/or posts at the Monastery if you don't have time 16:26
autrijus Limbic_Region: that'd be lovely.
Limbic_Region since I am not contributing to Pugs in any other meaningful way
autrijus lwall++
I'm going to make it so.
so it so good.
err, I mean, so is so good.
Limbic_Region use.perl.org/~drhyde/journal/23658 # how to get ghc on debian or osX
autrijus perceives endless punning possibilities 16:27
Limbic_Region use.perl.org/~dvergin/journal/23656 # pugs 6.0.11 and ghc 6.4 not working on windows
autrijus I fixed dvergin's bug
it's just I can't seem to comment on his 16:28
journal and I don't know how to contact him.
I'd be happy if you can help.
Limbic_Region by default, use.perl journal entries don't have comments enabled
autrijus which is sad.
Limbic_Region he is also a monk - I will /msg him
autrijus danke
just say it should be fixed in trunk and please try trunk.
gaal oh, speaking of journals, www.livejournal.com/users/autrijus_rss/ for the LJ users out there.
Limbic_Region bitte - will also let him know he needs to enable comments (should be the default I think) 16:29
autrijus yuppers.
theorbtwo Ah, ja, thanks for pointing that out to me the other day in a /msg.
autrijus hrm, ??:: hates me :-/
autrijus hates back
1 ?? $x = 3 :: $x = 4 16:31
this is legal, even if = is lower prec than ??::
jabbot pugs - 765 - * so() is the new true(). 16:32
autrijus 3 ?? 4 and 4 :: 5 and 5
but this is a syntax error.
what is the difference between the "and" level and the other levels? 16:33
Limbic_Region autrijus - when you get time - www.perlmonks.org/index.pl?node_id=434969
theorbtwo so('this' is christmas)
Limbic_Region there are also several pugs/windows questions - I will report that it is fixed in the latest trunk and for people to try that
www.perlmonks.org/index.pl?node_id=439183 for instance 16:34
Makefile(400) : fatal error U1087: cannot have : and :: dependents for same target
autrijus that's fixed.
it's caused by space in path
and makemaker's inability to deal with them. 16:35
'fraid I can't help.
pmichaud: perl5 thinks that "3e100" is a number. 16:36
pmichaud autrijus: yeah, I know, and I've always thought that's a bug
:-)
autrijus and "3e9999" is Inf.
pmichaud well, it is a bug :-)
I mean, "3e100" is a number, the question is which number
:-)
i.e., is it 3 or 3e100? ;-)
autrijus you think it should be 3? 16:37
I happen to think that also.
pmichaud I'd prefer it to be just 3
autrijus for great consistency.
theorbtwo In p5, I think it should be just 3.
autrijus theorbtwo: be surprised.
theorbtwo In p6, I'm not all that sure.
pmichaud yeah, there are a couple of surprises like that lurking in p5
autrijus perl -e "print int('3e3')
3000
Coke (perl6 testing) instead of inventing a new test API, why not just implement the p5 test API?
autrijus Coke: we are following the p5 test API, if you are talking about harness TAP 16:38
and we're working with a close approximation between Test.pm and Test::More.
leaning toward Test::More more.
pmichaud coke: also, part of the advantage of having a new language is that we have new constructs and paradigms to work with in that language
autrijus pmichaud: ok, so barring a larry ruling, I'll make '3e100' parse to 3. :) 16:39
pmichaud what good is perl6 if we just carry all of perl5's idioms into it? This isn't to say we summarily reject any interfaces that p5 has developed, but we do often need to re-evaluate them with a fresh eye and say "well, is this the way we think it should work in p6"?
lightstep so will p6 not have scientific notation? 16:40
pmichaud lightstep: my post was in response to coke's question about the testing, sorry
autrijus lightstep: no, 3e3 is still 3000. "3e3" however is not.
at least today.
Coke pmii - as you yourself have mentioned, that's not a job for the compiler authors.
pmichaud autrijus: I think we'll definitely need a ruling from p6l on numification of strings
autrijus Coke: uh. eh. why is it not? 16:41
lightstep the sensible thing is to use the same lexer
autrijus lightstep: except people will then write "Infamy" and parse to Inf.
Coke ... y'all are not the language designers, neh? you're the language implementors.
sorry, "all y'all". =-)
pmichaud coke: is the testing framework part of the language?
Coke It's a standard module. 16:42
pmichaud coke: clearly we can implement *a* testing framework
autrijus Coke: I consider it my sacred duty as a CPAN author to design the language. :)
pmichaud and we can implement one that makes it easiest to do what we need to do
Coke to sum up: "this argument should be occuring in p6l, not p6i", that's all.
gaal This day will be remembered in infinty and beyond.
Coke er, "not p6c"
autrijus is Test.pm brought up on p6c? 16:43
oh, test logging.
lightstep autrijus, i'd consider such things to be nonsense. truncation is good, but accepting any prefix is bad
autrijus lightstep: perl has a very long tradition to parse "3blah" into 3.
lightstep: 'fraid that's not going to go away.
in light of that, better make it consistent.
(also, I'm not saying it's particularly sane, but it's part of the perl mindset, afimc.) 16:44
pmichaud coke: ah, I see where you're coming from.
lightstep lack of knowledge of perl makes me incomprehenful. but i accept that it is the way
pjcj When reading from a file, is the input a string? ie, will echo 3e3 | perl -ple '$_ += 2' behave as in perl5? 16:45
autrijus lightstep: thanks for your understanding. :)
pmichaud coke: I think the answer is that the compiler implementors may write many things that help us along the way to build the compiler that will be thrown away in the long run
i.e., bootstrapping
autrijus pjcj: it is a string, so it will be 2, not 3002, in this case.
pjcj: but it's not set in stone (yet). 16:46
pmichaud coke: and yes, many people will be tempted to mold the language in their particular direction by being first to implement a module. But we have to trust that things will shake out in the long run
autrijus Coke: also bear in mind that perl6 will probably have no core modules.
Coke pmi: I'm just annoyed at the amount of non compiler traffic I've been "forced" to endure.
pmichaud coke: If we have to wait for approval/discussion of every item on p6l before we see an implementation, then we'll never have one. :-)
coke: well, whenever you detect something that seems non-compilerish, just send me a private note and I'll push it to p6l if it's appropriate 16:47
autrijus mmm perl6-modules
autrijus ponders the need for another ML
pmichaud I was out of town wed-sun and didn't have good email connectivity
coke: I agree with you that we ended up with some non-compiler traffic on p6c, but we know there's going to be some fuzziness as things go along 16:48
Coke ... pmi, that's what I'm doing.
autrijus Coke: I agree that it clearly doesn't belong in p6c. but I'm not sure all p6 module discussion belong in p6l either.
pmichaud ah, there we go :-)
Coke and now I run away to go work on Tcl, on the off chance I'll beat you to parrot.
Coke . o O (yah, right.)
pmichaud coke -- wait
Coke yessssssssssssssssssssssss? 16:49
pmichaud btw, I wanted to thank you very much for all of your effort in getting the strings code into parrot
that is a *huge* help to me
Coke ... Thank *cvs*.
pjcj perl6-stdlib exists
gaal ah, autrijus, since i know you're using it -- repeating a question from before -- any hints on running make test / prove from eclipse?
pmichaud yeah, but it still needed someone who knew a bit more about it pushing, and you were the primary one pushing to get it done. So, thanks.
Coke 95% of what I did was a cvs merge. the rest was simple fixup. that just got it to "compile". leo did all the heavy lifting after that.
<-- can't take a compliment. 16:50
autrijus gaal: I don't do that, so I don't know
Coke and even if I *did* get it to work, I was still waiting for dan to say "what next".... but I think leo was paying attention there.
pmichaud anyway, I'm doing PGE this week and hope to have a fairly complete implementation by this time next week
autrijus pjcj: reallly.
Coke Cool. I look forward to your p5 implementation so I can use it in Tcl. =-)
autrijus moving things to perl6-stdlib looks sensible.
gaal ah ok :) anyway we don't have perl6 support in epic yet and all that so the Problems view isn't going to be very useful anyway. 16:51
autrijus if not now, certainly some time after now
gaal: aye.
pmichaud anyway, all, I have to go to meet with my financial advsors... bbl
autrijus but I set my perl into pugs.exe
Coke laters.
autrijus and it at least can tell me if it compiles.
gaal autrijus: yes, i got that far :) i think i'll go back to vim, since i'm not hacking the haskell parts anyway and that's just dandy for perl (even 6) 16:52
autrijus nod.
lightstep one more thing about interpolating numbers: if " 3e3 " is 3, will " { 3e3 } " also be 3 ? 16:54
s/interpolating/parsing/
nothingmuch channel splits are silly 16:58
theorbtwo IRC is a simple tree topology; there's no redundancy whatsoever. 16:59
You get infinite loops if there are.
autrijus purrs. 17:00
gaal so does my cat!
autrijus kosimus in #haskell gave me the spark of light to do ??:: correctly. 17:01
namely, treat "?? ... ::" as an infix operator.
that's all it will take
autrijus hacks
ninereasons I don't understand the use of the .perl method. May I have an example? 17:07
autrijus when you see .perl, think Data::Dumper::Dump
ninereasons ah - ok. That's helpful. 17:08
ingy when you see .yaml ... 17:12
autrijus heh. 17:13
($str3 eq $str4) ?? $bar = 1 :: $bar = 0;
pugs correctly parsed it as
(($str3 eq $str4) ?? $bar = 1 :: $bar) = 0; 17:14
so I'm fixing operator.t.
to say
($str3 eq $str4) ?? ($bar = 1) :: ($bar = 0);
instead.
nothingmuch we have ?? :: as lvalue? sweet!
autrijus and now, lo and behold, all passes.
nothingmuch: not yet :D
nothingmuch or is it just parsed that way?
autrijus it's parsed that way 17:15
nothingmuch =/
autrijus which is correct.
nothingmuch naughty autrijus
won't work 24/7
autrijus but not what the tester was testing.
lol.
24/7 is 3.4hrs per day.
nothingmuch 24 * 7 then
autrijus I think I spend more time than that on pugs :)
yay I see my favourite message again! 17:16
UNEXPECTEDLY SUCCEEDED
autrijus goes unTODOing stuff
nothingmuch =D
jabbot pugs - 766 - * ?? :: now works correctly, thanks to k 17:22
pugs - 767 - * add kosmikus to AUTHORS; remove old te 17:32
gaal uh how do i take a reference to a sub? 17:43
autrijus \&sub
or just &sub
actually just &sub.
gaal thanks
&undef_sub should die, yes? 17:50
undefined
autrijus it should die because there's no such sub. 17:51
same as $undefined_var
gaal k
via the symbol table, is this the syntax i need? (it doesn'y work): %{ $?PACKAGE }::{"&a_sub"} 17:52
autrijus is there still a symbol table? 17:54
I thought globs are gone
I don't know if symtable is left.
gaal S11 mentions it.
autrijus grep around a bit and ask on p6l if it's not there?
oh. sorry.
gaal though not for subs
autrijus where?
jabbot autrijus: where is current TODO?
gaal dev.perl.org/perl6/synopsis/S10.html#___top last line 17:55
urr, wrong link 17:56
last line in section "packages" in that page
autrijus oh. S10.
gaal oops, yes, i am typlectic.
autrijus ok. so there is still a symbol table implemented as a hash. 17:57
obra yay
autrijus very well, we will do that.
write tests please.
gaal anyway, if there are no symbol tables, how does one dispatch a sub by name?
obra "can I tie the symbol table"?
gaal hahaha
autrijus gaal: of course there is a symtable.
gaal: it's just how we expose it on the language front.
gaal autrijus: goodie, i was lookging for something to do :)
autrijus we probably do it as an IType.
obra: and yes, that means you get to tie it. 17:58
or add pre/post FETCH/STORE hooks.
imagine the sickitude.
gaal heh, me, I just wanted symmetry for ok(defined(&sub_that_exists)) :-) 17:59
autrijus oh. defined() as special form.
gaal so autrijus, did i get the syntax right?
ok?
s/k/h/ 18:00
autrijus I always thought that the defined(&x) form is weird.
and it looks like it may go away in p6, because &x is now just a variable.
theorbtwo Right, but is it just a variable that is undef when no such sub exists?
obra ok (&main::x) ?
theorbtwo (Consider the case of taking a reference to a sub that will get AUTOLOADed.)
obra how do you not call thesub? 18:01
Juerd It takes 6 weeks before a mac mini arrives
They're too popular
autrijus theorbtwo: so why only &x gets that treatment and not $x?
Juerd By amazing and fortunate coincidence, I can pick up mine tomorrow.
theorbtwo &foo doesn't call the sub anymore in p6.
Juerd :)
autrijus obra: &foo never calls the sub.
theorbtwo defined(&::foo), then? 18:02
jabbot pugs - 768 - * per pmichaud's suggestion, intifier no
obra wait. &foo doesn't call subs now?
autrijus gaal: %{"$?PACKAGE\::"}{'&a_sub'}
obra aiee
I mean, I'll live.
autrijus &foo is just a Code object. 18:03
&foo() # still calls.
the idea is to make \ go away.
gaal: or, using better quoting:
%<<$?PACKAGE\::>><&a_sub>;
gaal autrijus, thanks; should this work now, or is this todo?
autrijus that is todo. symtable is not yet exposed. 18:04
please write tests.
gaal (ACK.) x2
autrijus (thx) xx 2
gaal how do you pronounce "autrijus" btw?
autrijus autrijus.org/autrijus.mp3 18:05
crysflame heh!
gaal lol
obra you really need "My name is Autrijus Tang and I pronounce Autrijus as Autrijus" 18:06
in English, German and Chinese.
also, lojban.
autrijus also esperanto.
and tengwar.
gaal i'm not the only one who hears it as "outrageous", right?
autrijus it's remarkably close in sound to "outrageous" 18:07
crysflame that makes sense, phonetically
obra++
autrijus++ # tengwar
obra discovers autrijus has a livejournal
autrijus the "i" in "tri" is more "ee" though
a deadjournal. 18:08
48 commits today.
that's 2x the daily output rate I remembered
what's wrong with you guys? :)
obra autrijus: get clkao to set up metrics 18:09
daily churn
autrijus or gugod.
obra nods 18:10
or really, for every project on foundry
theorbtwo I've been told it's something like the beginning of "ought", then "re" as in reply, and "jus" somewhat like "juice". 18:11
(By leira.)
autrijus that is also true.
but "jus" is shorter.
theorbtwo Hm, "justice" works better. 18:12
gaal don't forget the tonality, folks. :)
autrijus and the resonance field
gaal anyone try running it backwards yet? 18:13
theorbtwo The resonance field is very important; you need to tune the modalities of the main deflector dish.
gaal this is failing. should it? 18:16
sub a_sub { pass "sub"; 1 }
a_sub || fail "a_sub";
autrijus short circuit? 18:17
gaal doesn't || do that?
autrijus yeah, short circuitting is borken.
gaal ah ok
autrijus tests tests :)
(not sure if there's already one for that tho) 18:18
obra what ended up happening with named params?
gaal i'll ditch this test from undef.t (carried it off perl5's, no idea why it was trheere.)
don't work yet obra last i tried them
autrijus obra: oh, I discovered that the names were passed in as part of pair
which is a bug
kept meaning to fix it. maybe I'll do it now.
obra apparently, the foo<bar> syntax is also not parsable 18:19
autrijus i.e. if you pass (name => "value")
then $name gets set into ('name', 'value')
which is why it's mostly borken now.
obra ah
autrijus and yes, :foo<bar> is not yet parsed.
Juerd WTH is splat.t +x? 18:21
18:22 metaperl__ is now known as metaperl_
jabbot pugs - 769 - Revamped perlkwid to latest understandin 18:22
gaal Juerd, don't say that word!
Juerd Which?
jabbot Juerd: Which is not as cheap as normal dimm
gaal the one that caused irc to join "", split //, "split" 18:23
Juerd The netsplit happened before that
obra what's the convention for parsefailing tests?
Juerd Your client (irssi?) probably buffered the quits
gaal not from my POV
trillian here
theorbtwo is(eval 'yadda', 'whatever', 'name of test')
Juerd Does it show many quits, or one informative line, gaal?
gaal many. 18:24
Juerd Then weird.
gaal obviously your region of spacetime was warped, warped by the disruption.
nothingmuch *sigh* 18:32
why does sleep -1 cause infinity?
nothingmuch wants p6 to sleep properly
gaal maybe it causes infamy?
Khisanth -1? what is that suppose to do?
crysflame how do you test for that, too 18:33
18:33 rgs_ is now known as rgs
autrijus nothingmuch: it cause infinity error? 18:34
nothingmuch no 18:35
it just sleep to infinity
i'd like to reliably say sleep($next - time)
but i can't, i have to say
autrijus I think in p5 18:36
sleep -1
is also infinity.
so failing to see the problem
nothingmuch if (my $s = sleep($next - time)){
sleep $s;
i'm complaining about p5
}
since p6 has Inf, it should be used instead of -1
i'd like p6 to be different
quick, off the top of your head
fastest way to make html into plaintext
theorbtwo sleep(Inf) sounds good to me. I'd expect sleep(-1) to be an error.
"Sorry, can't reverse time." 18:37
lynx --print, something like that.
gaal yeah, there's iliaz's optimization to get_tomorrow_date {}....
autrijus w3m -dump
theorbtwo Or HTML::TreeBuilder->new_from_content("<html><head...</html")->as_text. 18:38
autrijus notes that p5 also has Inf.
nothingmuch theorbtwo: i'd rather it be zero
but perhaps return undef
and die on use fatal;
gaal open in browser and copy paste?
autrijus it's just written as "1e99999"
nothingmuch damnit: Scalar value @queue[0] better written as $queue[0]
p6 is getting to me
theorbtwo: thanks, that will do the trick
autrijus damn it, I was tabbing to see where is this "damnit" person you are talking to. 18:39
irc is getting to me.
theorbtwo (Note that'll leak, unless you save the HTML::Element you get back, and $tree->delete it.)
Khisanth autrijus: of you need sleep :) 18:40
autrijus I do, I do.
nothingmuch theorbtwo: ah, that's important to know 18:41
lightstep what happens with `:l' in the prompt? there is code for it in Main.hs, but the parsing is disabled at Shell.hs? 18:45
autrijus lightstep: Main.hs is just stub. 18:46
I think :l should just be dispatched to eval "require".
i.e.
:l Test
err
:l Test.pm
should be translated to "require 'Test.pm'"
and evaluated.
lightstep i'll do it?
autrijus yay! 18:47
lightstep in a few hours
friend &
autrijus :)
gaal (in hugs and probably ghci, either of ":l Mod.hs" and ":l Mod" work, so may as well make bothw rok heretoo?
theorbtwo Why :| ?
gaal "load"?
autrijus :l
not :|
try Andale mono or Anonymous :) 18:48
gaal: sure, that'd be cool also. maybe just test ifFileExists
theorbtwo Oh, :L
gaal :L.lc
theorbtwo The shell is case-sensitive? 18:49
Oh well, I got persistancy with :r, I'm happy.
ingy Juerd: ping
Juerd pong
But not for long
ingy I added a table example to perlkwid.kwid 18:50
Juerd I'll have a look
Have you found problems with my spec yet?
ingy not yet
Juerd (It's been in my head for a while already, so I hope there are none)
ingy but I haven't even had coffee
Juerd I have energy drink for you if you want 18:51
We got 20 trays of 24 free
So we have plenty
gaal man, that's a lot of drinks.
Juerd Yes.
gaal maybe barter them for RAM?
Juerd One fridge plus enough to fill another.
gaal: They're unlabeled, so they may not leave the building unempty. 18:52
ingy Juerd: sure, bring one over
Juerd No, they can't leave the building
You'll have to come and drink it here
gaal maybe invent a way of turning them into ram?
see, if einstein had it only off-by-one (in ASCII) -> e=mb (squared!) 18:53
nothingmuch woot, rssgrowl done, in about 10 minutes 18:58
fun fun fun
theorbtwo rssgrowl? 18:59
(HTML::TreeBuilder)++, BTW. 19:00
See also utils/fixauthors.pl, or whatever I called it.
nothingmuch sorry 19:08
rssgrowl is an rss aggregator
which uses growl (growl.info)
ingy fyi: I started #kwid for non pugs related kwid discussion 19:57
theorbtwo Hm, I wonder if YAS should note that message from lilo. 19:59
nothingmuch w00t 20:11
nothingmuch finally has an RSS reader he likes 20:12
now all i need is to find some feeds ;-)
gaal fwiw LJ is both an aggregator and a source of many feed. 20:13
lightstep take the commit feed for pugs :)
gaal feeds
nothingmuch lightstep: that's partially the reason why i wrote it
gaal: how about a link for yours?
theorbtwo www.livejournal.com/user/<username>/rss 20:14
nothingmuch wonders if there is a human readable dbm
theorbtwo gaal's is, unsurprsingly, 'gaal'.
nothingmuch =) 20:15
gaal just add /data/rss/ to anyone's 'j
but if you join lj you get to read my l33t friends-only entries.
nothingmuch never ever ever
we discussed it already
my login is taken
gaal yeah, but there was no way i'd pass on an opportunity to mention that again :) 20:16
nothingmuch =P
theorbtwo nuffinmuch isn't.
nothingmuch i'll sign up just to be a friend, if you promise to never try to make me blog
gaal hereby publically promised!
nothingmuch theorbtwo: i'd actually prefer nuffin for consistency with PAUSE given choice 20:17
gaal (another pro is to get email notification of comment replies)
theorbtwo nuffin is taken, though.
nothingmuch it is too?! 20:18
oh my
i thought self deprecation was really unique at the time
everyone was hot_girl12
or coolGuy17
or overlord
gaal woobling is free... 20:19
theorbtwo I thought odd was taken, but since I needed to become "theorbtwo" because of too many "theorb"s, apparently not.
nothingmuch woobling is my domain, not my identity
theorbtwo With 6,456,598 accounts, the flat namespace can be a bit annoying at times.
gaal oh, consider how kings in shakespeare's english were named by their domains. 20:20
e.g. in Hamlet, "I am to England"
nothingmuch =P
i'll go with 'nuffinmuch', i guess
gaal (meaning the English king, though obviously he also boarded a ship to England)
theorbtwo Not just kings. Consider also Romeo and Juliet's parents.
gaal true!
fathers, actually.
theorbtwo Er, right.
gaal well, ruling-parent-male-by-default. 20:21
nothingmuch MORE MORE MORE!
gaal more anal discussion of elizebethan language? okay. 20:22
nothingmuch no
gaal did you know that "thee" and "thou" weren't high language at all in S's time?
nothingmuch rss feeds =)
gaal ah :)
nothingmuch beh
openfoundry.org/Foundry/Project/Sou.../pugs/rss/ <-- XML::RAI::Item->identifier != consistent! 20:23
it changes every time, appearantly
gaal www.ucomics.com/calvinandhobbes/
nothingmuch gaal: i have all the books, and i know most by heart =) 20:24
theorbtwo Oh! And watchyamacallit-bot should be fixed to use description instead of title.
nothingmuch switches to $item->link 20:26
gaal www.hackaday.com/ is sometimes cute 20:27
www.kimbly.com/blog/ is interesting
theorbtwo Commercialized comics don't have a RSS feed on the page, generally, but indy ones sometimes do, and if you look a bit, you can find RSS feeds of things that don't have them pretty easily. 20:28
Also, try checking people's friends pages, and stealing any feeds they have that you find interesting. 20:29
livejournal.com/users/<username>/friends/
nothingmuch uGGG!!! 20:49
asshole
my ride forgot me at work!
damnit!
ingy nothingmuch: yow! 20:53
nothingmuch indeed
especially with it being 23:00 20:54
theorbtwo Ugh.
How far are you?
ingy work all night!
nothingmuch to home?
20 minute drive
ingy: they wish
ingy Perl6 needs you!
nothingmuch i haven't been working for the last 2 hours anyway
at home, at home...
ingy you can crash at my place 20:55
nothingmuch guess whose taking a vacation tomorrow?
theorbtwo And how long of a walk?
lightstep nothingmuch, where do you work? 20 minutes is a long distance at this hour
nothingmuch theorbtwo: 20 km, approx
but rough terrain 20:56
theorbtwo Oh. Sounds not-fun.
nothingmuch lightstep: omer industrial park
nuh uh
damnit, now none of the cab drivers are answering! 20:57
this means i might not get a refund, if i go by unofficial means
phew 20:59
ingy: YAML quickie: 21:03
--- #YAML:1.0
? |-
abridgegame.org/cgi-bin/darcs.cgi/darcs/?c=rss
:
? ! >-
abridgegame.org/cgi-bin/darcs.cgi/d...e8c106f.gz
err, nopaste
(sorry)
anywho, what are the silly ? | - and ? ! > - stuff? 21:04
compare with:
www.livejournal.com/users/gaal/data/rss:
www.livejournal.com/users/gaal/152214.html: 1
ingy nothingmuch: paste it in the pugs wiki and I will explain
nothingmuch: but basically they are silly artifacts of an inplemantation that needs fixing 21:05
nothingmuch ah 21:06
ingy the yaml is legal but highly annoying
nothingmuch pugs.kwiki.org/?YuvalKogman 21:07
lightstep does svn commit also update the local mirror?
nothingmuch lightstep: what do you mean by local mirror? 21:08
nothingmuch chose YAML for that file since it's, err, sometimes human readable 21:09
bad ingy! for not being perfect!
btw, how is the .4 branch coming along? 21:10
ingy .4?
nothingmuch there's YAML .3x
and .4x 21:11
oh, there's been an update to .3x
goody
how does it eat overloaded objects now?
ingy mmm. I forget
try it
it really needs to go both ways
nothingmuch no time now, cab here soon 21:12
ingy and really on a per object basis if desired
ta
expect a perl6 yaml soon!
in haskell
nothingmuch heh
lightstep what is the best syntax for :l at the prompt? 21:50
theorbtwo Just replace the :l with a 'require ', and eval. 21:51
...because later it should get changed to "use", which has a wider syntax. 21:52
lightstep not semantics, syntax
:l works now
theorbtwo Oh. :l whitespace module name EOL.
lightstep module name can contain whitespace? 21:53
(right now it's required to be a valid haskell string)
theorbtwo No. 21:54
Don't think so, leastways.
mugwump using ::($expr) you should be able to make any package name at all 21:55
see S10
theorbtwo I'm not sure it's smart to try to make :L do everything under the sun. 21:56
mugwump Packages with embedded "::"'s are hard to refer to :)
theorbtwo For that matter, the real parser doesn't even handle Foo::Bar from the look of it.
mugwump maybe something like ::{"::"} will refer to the top level package called "::" 21:57
theorbtwo ruleRequireDeclaration in Parser.hs suggests it should be identifer `sepBy1` string "::"
Then concat (intersperse "/" names) ++ ".pm" 21:58
mugwump Yeah, if specified directly. I'm just mentioning a silly corner case :)
theorbtwo
.oO(OK, it will allow Foo::Bar, as I read more.)
21:59
23:04 theorbtwo is now known as theorbtwo_elsewh
pasteling "artol" at 66.93.197.227 pasted "subroutine bug" (26 lines, 628B) at sial.org/pbot/8222 23:23