Feather users: perlcabal.org/~autrijus/ | pugscode.org/ <Overview Journal Logs> | pugs.kwiki.org
Set by autrijus on 26 May 2005.
autrijus evalExp (App "&meth" (Just $ Val $ VObject obj) []) 00:00
clkao i'm sorry to play the insane part
autrijus I'm happy to play the game part
chromatic Ahhh, right. I had seen that before.
clkao qukc it's 1am
stevan let me rephrase
clkao quick
got your commit
stevan some insane C vs. insane Haskell game :)
autrijus clkao: gah there's no callsv form 00:01
clkao: we're doomed
clkao: can you fix the api for me?
svnbot6 r3936, autrijus++ | * obj invocation now carries env around too
autrijus clkao: add a perl5_call_sv form 00:02
the same as _call but gets a SV*sub
you do the C side. I'll handle FFI
clkao eek
autrijus quick, it's just copy/paste
(or refactor if you'd like)
clkao I TOLD you yuesterady we need a flag to say call_method and call_sv
autrijus yeah and I FORGOT that 00:03
clkao i'm fixing other stuff
mkvalref is not giving env
autrijus alright. I'll handle callsv 00:04
chromatic Hm, is src/Pugs/AST/Internals.hs the place for this? It defines the VObject constructor, but evalExp is out of scope there.
autrijus chromatic: you probably want Prim
chromatic: err, no
just define it in AST.hs
evalExp is in scope there
and you see AST.Internals symbols too
chromatic Yeah, I have to edit Prim for something else :)
autrijus alternatively move evalExp into Internals 00:05
but don't do that unless you have too
internals.hs is already too huge
chromatic I don't think I have to.
clkao pugs: user error (cannot fetch $pugs::env) 00:06
fubar
autrijus eh. 00:07
00:07 jannic_ is now known as kelan
autrijus clkao: I'll make the api as such 00:07
perl5_call(SV *sub, SV *inv, SV** argsSV *env, int cxt)
doing so now
clkao alright
autrijus come to think about it
clkao *sigh*
autrijus I'll call it perl5_apply
clkao ekk
autrijus: merge my chagnes 00:08
what's the requirement for 6.28 btw? 00:10
autrijus clkao: OO 00:11
clkao oh we do have. plus legacy perl5 oo :)
autrijus ;) 00:12
svnbot6 r3937, clkao++ | Apply more ducktapes.
clkao quick, can't wait
autrijus almost there 00:16
merging yours 00:17
crysflame blinks 00:20
clkao: porting svk?
clkao crysflame: not yet 00:21
autrijus but close
clkao autrijus: shut up and hack
crysflame clkao++ # haha
clkao XD
crysflame hands clkao a pompom with horns 00:22
mrborisguy did anybody ever figure out the "Could not find module `System.Eval':" problem?
autrijus VCode casted
mrborisguy: you probably need a new hs-plugins
clkao not terribly useful
autrijus and "make install ; make register"
mrborisguy newer than 20050501?
chromatic Ugh, addFinalizer wants an IO ().
autrijus chromatic: runEvalIO 00:24
chromatic runEvalIO App (...) [] ?
autrijus Out of memory during "large" request for 536875008 bytes, total sbrk() is 268541952 bytes.
chromatic: something like that. a sec
clkao: I'll commit borken code and you'll debug 00:25
clkao ok. what's obrken?
autrijus I don't know
it looks fine
but it complains out of memory
clkao gah
autrijus maybe I've written some bad C
committed
clkao getting
meppl gute nacht 00:26
autrijus meppl: 'nite
meppl ;)
autrijus chromatic: ok. what you need to do is to preserve the env context during destroy time
chromatic: that is actually nontrivial... hm
since the env may have already passed away 00:27
chromatic Let me diff and nopaste what I have so far.
autrijus so you need to write to some IORef
chromatic nopaste?
autrijus as a DoD queue
and have the runloop sweep the queue very now and then
I think that's the way to go
perlbot: nopaste
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
pasteling "chromatic" at 63.105.17.30 pasted "adding DESTROYALL support" (54 lines, 2.1K) at sial.org/pbot/10461 00:29
svnbot6 r3938, autrijus++ | * all should work except this memory overflow thing
clkao lol
arg != NULL
shouldbe *arg != NULL
autrijus arg 00:30
autrijus waits clkao to fix 00:31
clkao hate
Bus error (core dumped)
chromatic Gah, I'm in over my head here. :) 00:32
autrijus chromatic: ok 00:33
chromatic: you do this
env <- ask
SamB why does it say "Bus error"? wouldn't the kernel be just about the only thing that would have those?
chromatic Yeah, I have that.
now.
autrijus just before addFinaliser
now you curry it
(objectFinaliser env)
objectFinalizer :: Env -> VObject -> IO () 00:34
chromatic Ohh, I was doing it on the definition of objectFinalizer.
autrijus objectFinalizer env obj = runEvalIO env ...
that will work
clkao autrijus: same
pugs: user error (cannot fetch $pugs::env)
autrijus because "ask" only work in the eval monad
clkao: commit first
clkao committed 00:35
chromatic Right, and currying there makes it available later.
autrijus yeah. isn't currying fun?
chromatic I've always liked it. 00:36
autrijus clkao: so everything returned from eval_perl5 is a CODE now? wtf?
clkao ??
autrijus oh nvm
it's just debug msg
;)
chromatic I hate the "The last statement in a 'do' construct must be an expression" message though.
clkao autrijus: od you run method.t for testing? 00:37
autrijus clkao:
$ ./pugs -e '(eval_perl5 q[sub { print 1 }]).()'
1
I think that means something.
namely VCode deref works
clkao hang on
autrijus chromatic: that message says "you forgot to return something"
clkao ok 7 - invoking p5 coderef 00:38
autrijus clkao: it compiles! ship it!
chromatic It's complaining about addFinalizer (objectFinalizer env) obj
clkao no
autrijus clkao: also take the debug msg away
clkao fix the eng crap
env
mauke do {} is just syntactic sugar for chained >>= calls
autrijus chromatic: liftIO it 00:39
svnbot6 r3939, clkao++ | autrijus++ needs to learn how to use pointers.
r3940, clkao++ | test for invoking p5 coderef.
chromatic Wait, I had a bad pattern match in objectFinalizer.
autrijus I won't want to learn how to use pointers ;)
the compiler should tell me ;)
clkao it did 00:40
you didn't listen
autrijus then it should forbid it
autrijus mumbles something about weakly typed languages
clkao quick
oh sure, cc --pedantic
mauke gcc -W -Wall -Wundef -Wendif-labels -Wshadow -Wpointer-arith -Wcast-align -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -Wnested-externs -Winline -Wdisabled-optimization -std=c89 -pedantic -O2
FOR GREAT JUSTICE
clkao autrijus: qucik , ifx that env and i can go sleep
autrijus yeah yeah
fixing
SamB autrijus: what did you do? 00:41
clkao shssh, don't talk! 00:42
autrijus SamB: hm?
theorbtwo Hmm, clkao? 00:43
SamB autrijus: what bad thing did you do that the compiler should have refused to go along with?
clkao SamB: read my commit
autrijus SamB: using pointers to pointers as pointers
SamB oh
clkao or actually the other way around. so he doesn't really know what he is tlaking about 00:44
"don't shoot! I am a table"
SamB well, you know how most C programmers are. most don't even know a pointer from an int!
svnbot6 r3941, clkao++ | use less 'debug';
SamB its really quite appalling.
although on the other hand it would be handy if they were defined to be bijectively castable... 00:45
and storage compatible
clkao autrijus: if i pass a hash to p5something funny will happen
SamB today I learned that you aren't even really assured of the safety of casting function pointers to void* and back... 00:46
autrijus clkao: I got the bug. 00:47
finally.
autrijus hates tracing XS
clkao autrijus++
autrijus fixing
clkao you can tell me where things are and i can fix
autrijus clkao: doubleboxing
the env got back was a sv boxing val boxing env
and deref is just one layer
it's complicated. I'll fix 00:48
clkao k
autrijus I have this feeling I'm going to find parrot's C code much more appealing after this 00:49
clkao of course
SamB why don't these commits ever seem to have descriptions?
chromatic Parrot nicer than Perl 5? Nah.
The lack of macros can't possibly help!
clkao because autrijus will summarize later 00:50
clkao use more 'tea'; 00:51
SamB clkao: but then I get a bunch of patches named r3939, r3940, r3941...
clkao ?
SamB uses darcs
clkao poor you
SamB yeah, I guess it wasn't designed for projects with this much code... 00:52
autrijus clkao: fixed... but it segfaults XD 00:53
clkao *sigh*
autrijus looking
clkao then what does it mean by 'fixed' ?
commit it?
autrijus k, in a bit 00:54
"fixed" means it's the right place
clkao if it's right enought it hsouldn't segfault :P
chromatic Ugh, objectFinalizer needs to return an IO (), but I'm returning an IO Val. 00:55
theorbtwo Odd... Am I the only one getting a compile error?
src/Pugs/AST/Internals.hs:318:29: Couldn't match `Maybe PerlSV' against `PerlSV' 00:56
svn diff -rHEAD doesn't show any difference to explain it.
(Note that I am /not/ using perl5 embedding.)
clkao maybe some ifdef was missing
SamB I get that too 00:57
I was just thinking of asking that question, only without the "Odd..." 00:58
theorbtwo Who k-lined perlbot?
autrijus theorbtwo: fixed 00:59
SamB tries changing nullSV to (Just nullSV)
autrijus the prototype was wrong
clkao: I got to bottom of this 01:00
clkao k
svnbot6 r3942, autrijus++ | * fix prototype for theorbtwo
autrijus executive summary is that we can't use $pugs::env.
have to use PL_modglobal.
fixing
clkao GAH
autrijus don't blame me, I didn't design this jenga thing! 01:01
clkao ok, let's keep playing jenga
autrijus pulls another block away
clkao (for the very last times)
SamB Is it just me or is Pugs.AST.Internals awfully big? 01:03
going by compile time 01:04
autrijus implemented. 01:07
theorbtwo autrijus: Thanks. 01:09
clkao autrijus++
meanwhile i've implement svn.p6 using svn::client
pastebot? 01:10
clkao.eat('fish') 01:11
chromatic Grr, stumped again.
mugwump clkao, pastebot is at nopaste.snit.ch:8002/, but it isn't joined to this channel
SamB: it's certainly memory hungry, too. check you're not swapping.. 01:12
pasteling "chromatic" at 63.105.17.30 pasted "Almost there with DESTROYALL" (59 lines, 2.3K) at sial.org/pbot/10463
chromatic I *think* I'm on the right track, but I'm really not sure it's even happy enough to give addFinalizer what it wants. 01:13
mugwump oh, there you go clkao, there's the real pastebot at sial.org/pbot/
autrijus clkao: commited. 01:15
svnbot6 r3943, autrijus++ | * switch to use PL_modglobal. 01:16
01:16 theorbtw1 is now known as theorbtwo
clkao compiles 01:16
chromatic Ahh, it can't curry that because ... nuts. 01:17
Odin-LAP throws a squirrel at the nuts. 01:20
clkao what defer dfer and segfault?
autrijus "DEREF #3" 01:21
Segmentation fault
that's args. 01:22
investigating
did you null-pad it?
clkao yes 01:24
wait which?
autrijus no.
you didn't.
pugs_guts_invoke
clkao bah
fixing
chromatic I need something like runEvalIO :: Env -> Eval Val -> IO () 01:25
clkao same
autrijus chromatic: hm? just slap a (return undef)
clkao: WORKED
ok, I'll commit
$ ./pugs -e 'eval_perl5(q|sub {$_[0]->()}|).({ say 123})' 01:28
123
clkao not ok 01:29
i haven't handle the return value yet
autrijus heh.
not my problem
autrijus fixes some more env marshalling 01:30
clkao only return one value?
what's the context?
autrijus ah right. you should be able to specify cxt too 01:32
missed that in the pugs callconv
by default it inherits the caller's
svnbot6 r3944, autrijus++ | * various callconv fix.
autrijus but yes
clkao k
autrijus it always only return on Val
regardless of cxt
(because multi retval is represented as VList)
ok, env marshalling solved 01:35
need to fake local() semantic too
hate
clkao return value became CODE
autrijus oh?
clkao weird
theorbtwo autrijus: I'd just document that... 01:36
autrijus be sure to pull and merge
theorbtwo: nice!
clkao i did
oh not yet
try method.t
autrijus I tried 01:37
only 1 fail
clkao ya. thats the return value thing
autrijus commit whatever you got
clkao i returned it now but became code.
commit
it hsould be a plain pv 01:39
svnbot6 r3945, autrijus++ | * env is nor marshalled.
r3946, clkao++ | propogate return value from pugs
autrijus so you just throw it to ST? 01:41
after ValToSV it?
clkao *nod8 01:42
autrijus ok. I'm changing callconv 01:43
clkao the return value in the last test should be called with newsvpv.. but never
autrijus callconv changed. _Apply now returns sv 01:44
hm, instant segfault. fun 01:45
svnbot6 r3947, autrijus++ | * PerlSV now does Hash and Array
autrijus bah. that doesn't work. 01:47
clkao returngs svn? 01:48
sv 01:49
i changed things ito assume it as sv but seems to get warnings
autrijus right, can't do that
have to ret vl
I think your stack manip is bogus
clkao ok 01:50
revert that
in haskel i do Sohw(ptr) to debug? 01:52
autrijus print ptr 01:54
clkao i want to see what's actually returned in pugs land in valToSv 01:55
any dump function?
autrijus sv_dump 01:57
you can add it to conv
clkao i mean in the sense of pugsval of course 01:58
autrijus but I think I fixed everything already. so don't bother ;) 02:01
clkao is that isa prboelm? 02:02
autrijus lots of problems. 02:03
all fixed.
r3948!
clkao you++
autrijus in particular
+ sv_setsv(ST(0), ret);
instead of
- ST(0) = ret;
I don't know why really.
clkao mm 02:04
svnbot6 r3948, autrijus++ | * method.t all pass!
clkao yay! 02:05
autrijus hm, nvm, you don't need ret inbetween free
ST(0) = pugs_Apply(val, inv, stack);
thinks works just fine. r3949
whew!
now port svk over!
clkao :))) 02:06
not yet. more tests coming
autrijus come to think about it, it's a good idea to add cxt into callconv now 02:07
pugs_Apply will take one more GIMME
that good w/ you?
SamB src/Pugs/AST/Internals.hs:317:27: Not in scope: `mkVal' 02:08
SamB impersonates a buildbot
autrijus SamB: fixed. 02:09
svnbot6 r3949, autrijus++ | * remove debug from method.t
r3950, autrijus++ | * save a tmp.
SamB autrijus: it doesn't look like it! 02:10
okay now its earlier
src/Pugs/Embed/Perl5.hs:38:29: Not in scope: type constructor or class `PugsVal'
clkao autrijus: in obj invoke, what should the val forapply be? 02:11
&methodname ?
autrijus clkao: right, "&method" 02:13
casted from pv
if you feed it a val, it may be able deref later too 02:14
svnbot6 r3951, autrijus++ | * fix nonembed
r3952, autrijus++ | * typo.
autrijus but don't rely on it
SamB: better?
SamB src/Pugs/Embed/Perl5.hs:39:15: 02:16
Couldn't match `IO PugsVal' against `()'
Expected type: IO PugsVal
Inferred type: ()
In the first argument of `return', namely `()'
In the definition of `mkVal': mkVal = return ()
src/Pugs/Embed/Perl5.hs:42:18:
Couldn't match `IO PerlSV' against `()'
Expected type: IO PerlSV
Inferred type: ()
In the first argument of `return', namely `()'
In the definition of `mkValRef': mkValRef = return ()
clkao autrijus: huh? so first arg of piugs_apply can also be cstring?
autrijus clkao: no, you need to cast pv to it 02:17
clkao yes, pv to cstring or to val ?
autrijus to val. always to val
cstring is same as pv!
type CString = Ptr CChar 02:18
SamB: I _think_ it's fixed ;)
autrijus is no longer sure
clkao: so in callconv apply, you throw GIMME_V usually
but if you want specific context, use G_SCALAR etc. 02:19
mrborisguy karma you
jabbot mrborisguy: you has karma of 4
svnbot6 r3953, autrijus++ | * restore old context properly. 02:20
r3954, autrijus++ | * even more nonembed fix.
mrborisguy it was worth a shot
Limbic_Region mrborisguy == borisz ? 02:22
clkao getting prelude mepty list
autrijus $ ./pugs -e 'eval_perl5(q|sub {print while $_ = $_[0]->()}|).(coro {yield 1; yield 2})' 02:23
121212121212
clkao: that usually means the inv slot is empty during a meth call 02:24
Limbic_Region autrijus++
SamB well everything in and before Pugs.AST.Internals seems to be fine
Limbic_Region now if only it worked on Win32
SamB at least, it hasn't given me any scope errors
autrijus Limbic_Region: if you have a mingwperl
it may even work
I think rgs's fix makes mingwperl fine now
clkao method call from p5. doesn't work. 02:25
committed
svnbot6 r3955, clkao++ | pugs method call from p5.
r3954, autrijus++ | * even more nonembed fix.
Limbic_Region ok - I am not going to build it tonight
but I have built mingwperl in the past
clkao autrijus: take a look? there's new test 02:26
SamB on the other hand, it is taking forever to build!
autrijus SamB: "make unoptimised" 02:28
Limbic_Region question - should Pugs be failing to build currently on Win32 even without p5 embbeding?
autrijus Limbic_Region: should not.
Limbic_Region ok - I see several recent commits
I will sync up and try again but it isn't looking good 02:29
clkao $ ab -clkao- [~/work/pugs] ./pugs svn.p6
1.3.0
,,3955,clkao,2005-05-27T02:25:14.124482Z,pugs method call from p5.,_p_apr_pool_t=SCALAR(0x83a3c58)
Limbic_Region I'm at 3952
pasteling "clkao" at 82.35.81.242 pasted "svn now" (10 lines, 213B) at sial.org/pbot/10464
SamB well it is only taking forever in Pugs.AST.Internals
clkao autrijus: come on, method call 02:31
autrijus GIMME fixed
or rather implemented
Limbic_Region still blowing up autrijus 02:32
autrijus Limbic_Region: nopasre the err?
pasteling "Limbic_Region" at 24.35.57.240 pasted "Win32 build blowing up" (17 lines, 651B) at sial.org/pbot/10465
autrijus Limbic_Region: that's fixed as of r3954 02:33
clkao autrijus: callback to pugs with sub having param binding fails 02:34
Limbic_Region I synced though
clkao er, hangs.. use up all memory
Limbic_Region let me double check
SamB sticks {-# OPTIONS_GHC -v #-} at the top 02:35
Limbic_Region ok - svk log is yelling about revision being too large 02:36
svnbot6 r3956, autrijus++ | * new callconv -- pugs apply now takes Cxt
Limbic_Region hmm
clkao Limbic_Region: maybe you are using something too old
SamB adds -dverbose-core2core
Limbic_Region clkao - by something you mean svk? 02:37
SamB takes it out again
clkao Limbic_Region: ya. or you actually gave it too large a revision
SamB I think that module should be split into pieces if possible
Limbic_Region *shrug* - I will beat it into submission - brb 02:38
SamB decides on {-# OPTIONS_GHC -v -O0 #-}
clkao autrijus: chained invocation uses all memory 02:40
autrijus clkao: so you define mex
and call me
what gives?
clkao mex?
autrijus method mex
SamB hmm, -O0 doesn't actually seem to work
Limbic_Region autrijus - no sure what is b0rked about my svk 02:41
but after convincing it to see things my way
no more problems
at least not with embed
Limbic_Region gives up and goes to bed
SamB okay, it finally made it into cc1 02:42
autrijus clkao: you forgot &
clkao: since this is so common an error I'll remedy it in callconv
SamB cc1 was using a lot of RAM
its in the evil mangler now...
clkao autrijus: no, that doewn't work yet. 02:43
autrijus: and what i was saying is that chainged p5 mehtod call from p6 doesn't work
SamB finished building
(of Pugs.AST.Internals) 02:44
autrijus clkao: is there a test?
clkao autrijus: which one? 02:45
p5 method call from p6, not yet 02:46
but use my svn script
chain the svn::client.new.log
it dies
autrijus clkao: this worksforme
use Digest::SHA1--perl5;
my $x = Digest::SHA1.new;
say $x.add("X").hexdigest;
but yes, if you do
.new.add
then it dies
clkao mimics autrijus 02:47
svnbot6 r3957, autrijus++ | * allow omitting '&' from pugs_apply callconv.
r3958, autrijus++ | * fix typo
clkao journal up, sleep&
autrijus roflmao
clkao++ # nice work
clkao autrijus++ # doesn't work without you
method.t now passes?
autrijus no.
clkao it spins 02:48
clkao really sleeps & 02:50
svnbot6 r3959, autrijus++ | * remove the test to despin it 02:53
r3960, autrijus++ | * remove sv_dump
SamB hmm, I want to send a couple of darcs patches... 02:54
autrijus sure, send them to [email@hidden.address] 02:56
I should set the darcs sendto addr thing to there
but I forgot how to set it
SamB cool 02:57
I think it is _darcs/prefs/email in the exposed repo 02:59
autrijus set. try it? 03:01
SamB it picks it up fine ;-) 03:04
I figured it out by looking at _darcs/prefs/ in lambdabot's repo (by directly visiting the url in links) 03:06
autrijus nice 03:07
SamB perl.org must have a really wierd email setup 03:13
its probably extensively scripted in perl
what is Pugs.AST.Internals supposed to do? 03:18
autrijus SamB: read PA02 yet? 03:20
docs/02Internals.pod
perl 5 hash and array refs are now accessible from pugs. 03:21
autrijus starts writing some DBI apps
SamB that doesn't seem to mention Pugs/AST/Internals 03:22
autrijus ah. it's part of AST.hs 03:23
the convention is that we aggressively factor out bigger modules
to sub namespaces
SamB well, yeah.
autrijus there's also haddock :)
# nothingmuch.woobling.org/pugs_test_...rnals.html
svnbot6 r3961, autrijus++ | * Perl 5 Array and Hash are now accessible from Perl 6.
SamB there don't appear to be any section headings in the export list 03:24
hmm, is it possible that those are stale? 03:28
hmm, is it just me or does Pugs.Prim also take a long time to compile? 03:32
well its started mangling it now... 03:35
autrijus "make unoptimised" will be faster :) 03:36
for development it's probably best.
SamB fine... 03:37
pasteling "autrijus" at 220.132.132.105 pasted "DBI test works!" (16 lines, 452B) at sial.org/pbot/10468
nothingmuch morning!
autrijus hey nothingmuch. see nopaste :)
nothingmuch yes, i was just "holy crapping" to some of the "wow! cool" people 03:38
autrijus clkao also got callback working so you can even pass coro{} into it
nothingmuch what else is there to do?
autrijus and lazy{} too
oh, lots of things
cross inheritance, shared runloop 03:39
lots of syntax level details as well
support for tied variables
nothingmuch what does shared run loop mean?
autrijus it means that you can yield() from perl5
or return() from perl 6 on behalf of perl5's caller
i.e.
nothingmuch ah
autrijus $CALLER::x
can see perl5's lexical
and vice versa
SamB linking also seems slow and memory-hungry :-( 03:40
nothingmuch continues documenting code
i want to throw this at CPAN 03:41
SamB nothingmuch: what are you documenting? 03:42
nothingmuch SamB: it's a one of those top level namespace thingies 03:45
specifically a framework for doing make(1) like things in perl
without the shell mindset getting in the way,
ijoyce like ant? 03:52
SamB doesn't ant have XML get in the way instead? 03:54
nothingmuch_ did i miss some context? 03:55
SamB <nothingmuch> without the shell mindset getting in the way, 03:57
--> revdiablo (diablo@revdiablo.user) has joined #perl6
<ijoyce> like ant?
<SamB> doesn't ant have XML get in the way instead?
nothingmuch_ ijoyce: not really like ant
less specific
SamB: i think i agree about XML getting in the way with ant, but I can't really say, haven't used atn 03:58
SamB it gets in the way in just about every other application where you write it by hand...
luqui despises ant 03:59
you shouldn't have to install a build module for every builder program you use
SamB I mean XSLT is crazy, and Relax NG would be unusable without a Compact form (which has actual syntax!)
luqui I think the shell mindset is rather appropriate for make 04:00
you just have to get around the platform dependent syntax somehow
nothingmuch_ luqui: for make yes 04:01
since you run commands
luqui i mean for make-like utilities
nothingmuch_ but this is to run perl code
luqui hmm
SamB nothingmuch_: oh
why is it to run perl code?
luqui what sorts of perl code?
nothingmuch_ well, basically, you give it a wad of objects 04:02
and you tell it how they relate
deps and stuff
SamB I don't think having to install a module is so bad, unless installing a module is a pain...
nothingmuch_ and then the dispatcher calls the appropriate methods in order
luqui ohhhh
so you want the graph deps stuff without the command running stuff
that should be fairly easy 04:03
SamB which, for Java, it is.
nothingmuch_ not only that
it's already done, and it was pretty easy =)
anyway, enough talk, code speaks much better, especially with docs
SamB for things like Python and Perl it isn't that bad. I mean it isn't much harder than installing the build tool itself...
nothingmuch_ so i'll go back to working on that, and hopefully it will be CPAN worthy soon
joepurl hello formalin14 04:12
formalin14 hello
is there some edit support perl6?
editor
ijoyce not yet 04:13
afaik
SamB is anybody signed up to do the emacs mode?
at least for ASCII stuff?
formalin14 is there some article about how to deploy freestand program that develop by perl6? 04:16
SamB are you kidding? perl6 itself is not freestanding yet!
nothingmuch_ formalin14: there's a vimmode for perl6, in util 04:18
autrijus formalin14: you can make freestanding programs with pugscc
install pugs, "perldoc pugscc" for details
formalin14 I think it is possible to tar all the necessary file(include pugs,perl5,lib) and then untar on another mechine and run.
thinks autrijus 04:19
nothingmuch_ formalin14: don't forget the GHC runtime
autrijus GHC runtime isn't a shared library :)
pugs is always static
nothingmuch_ and parrot if pugs was linked to it
ah, i didn't know that
autrijus and parrot is also static.
and you can build your perl5 to be static as well.
so. :)
nothingmuch_ the darcs build I use on OSX requires GHC to be install for the runtime 04:20
formalin14 ok, that is good, so I can do product use perl6.
autrijus theoretically yes. :) nobody has really done end-user products in perl6
SamB if you want dynamically linked haskell code with GHC, you need to use either GHCi or hs-plugins...
autrijus for my $work it's all server side where I have better control.
SamB: pugs only supports hs-plugins 04:21
I can't wait for libGHC to be available in GHC 6.6 :)
SamB I wasn't talking about pugs
I was talking about GHC
nothingmuch_ formalin14: the problem is that pugs changes so fast that you would feel bad 2-3 days after deploying a build
formalin14 Is there a word in English: green software, that means unzip to anywhere then is can run?
autrijus ah. yes then. hs-plugins or DynamicLoader
nothingmuch_ autrijus: libGHC will make Inline::GHC easy? 04:22
autrijus nothingmuch_: like, trivial
nothingmuch_ coolness
how soon? =)
SamB is DynamicLoader what hs-plugins was based on?
autrijus no idea :)
SamB: I don't know if there's shared code
need to ask dons 04:23
but probably not
SamB formalin14: self-contained?
formalin14: freestanding?
formalin14: standalone?
formalin14 yes
nothingmuch_ formalin14: get autrijus to make PAR support pugs
autrijus use PAR--perl5; 04:24
next question?
(and pugscc already handles dependencies) 04:25
SamB I think Pugs.Prim's use of pattern-matching on strings is likely part of why it compiles slowly... 04:26
autrijus SamB: that may be the case.
SamB hmm. is it possible that GNU ld does not scale well? 04:28
nothingmuch_ what's "pattern-matching on strings"?
autrijus nothingmuch_: op1 "" 04:30
nothingmuch_ oh
why did i read that as parser?
prim.hs is not that slow to compile, is it?
nothingmuch_ ought to shower & sleep 04:31
SamB well maybe not with 1 GiB RAM and 2 GHz, but with 256 MiB and 450 MHz...
autrijus ...you really want "make unoptimised". 04:32
SamB okay, so it takes a respectable 25 seconds or so with make uinoptimized. but then linking takes forever and thrashes so much I can't get x-chat in the foreground 04:35
granted it would seem a lot less like forever if x-chat would move to the foreground and paint and work...
hmm, it isn't doing it this time quite... 04:36
maybe that doesn't happen if I bring xchat back before lots of its stuff gets swapped out... 04:37
or maybe it is all this initSyms stuff... 04:42
what are all these #includes in Pugs.AST.Internals for? 04:44
autrijus: ? 04:51
jql when I'm faced with a slow computer and a high load, I ship out my root privileges and renice -10 my shell 05:03
the one NOT running the compiler, of course
SamB Okay, I think I've figured out why all those files were included... just trying to make a single one a module is a complete mess... 05:49
chromatic Anyone here know much about runEvalIO? 05:56
scook0 chromatic: what are you trying to figure out? 05:57
SamB all I know is that Pugs.AST is a mess!
I think I will give up for now. 05:58
maybe try again tomarrow.
chromatic I'm trying to call DESTROYALL when Haskell GC claims a VObject.
DESTROYALL is a Perl 6 method, I mean.
scook0 SamB: trying to split up AST? 05:59
SamB scook0: I was.
I feer it is about ten times harder than I thought...
er, fear. 06:00
scook0 SamB: I take it you were looking at AST/Internals
SamB yes, I was
scook0 hmm, I'll give it a look
it /is/ rather big
SamB especially if you consider that it includes Plugs/Types/*.hs 06:01
chromatic The types are pretty short though.
They're mostly just constructors, aren't they?
SamB basically all you have to do is figure out how to untangle it without getting totally lost... 06:02
SamB reverts his tree
I wish there was an easy way to visualize the lexical dependencies of things 06:03
chromatic Yeah, I haven't figured out Haskell's scoping rules either.
scook0, are you pretty lambda-savvy? 06:04
SamB I understand the rules, its just that I would like to see the dependencies in some kind of visual form
what about the rules don't you understand?
chromatic It's the visibility of functions. 06:05
I'm not quite clear on what's visible to what where and when.
scook0 chromatic: I started about a month ago, but I'm progressing
chromatic If I pasted my current (failing) patch, is there a chance you could help me figure out what's wrong? 06:06
(or anyone reading this in the summaries or backlogs in the future)
SamB there is indeed
especially if you also pasted the error message 06:07
formalin14 can perl6 call java jar now ?
pasteling "chromatic" at 63.105.17.30 pasted "DESTROYALL Patch (third attempt) and compilation error" (142 lines, 7.2K) at sial.org/pbot/10470 06:08
chromatic Ahh, now the error is clearer to me. 06:09
SamB that isn't even a scope error!
chromatic No, I fixed my scope errors much earlier. :)
scook0 chromatic: so it can't reconcile an (Eval a) with an (IO a) 06:10
chromatic The problem is that addFinalizer wants an IO (), while runEvalIO produces an IO Val.
If I turn objectFinalizer into a one-liner without the do and the return, it's a lot clearer with the error.
SamB I don't think so 06:11
svnbot6 r3962, scook0++ | Haddock tweaks
chromatic Hm, you're right. That gives two errors, not one. 06:12
SamB I think you need to add 'liftIO $ ' before 'addFinalizer obj (objectFinalizer env obj)'
scook0 chromatic: could you chain the IO Val into a return using >> ?
chromatic: ultimately, you're trying to turn an Eval Val into an IO (), right? 06:13
chromatic Yes, that's right.
It's certainly taking longer to fail with the liftIO...
Ah, it compiled that one. Nifty.
SamB are you building with 'make unoptimized'?
chromatic Yes.
I don't understand the $ syntax, but I'm still fuzzy on the . syntax too. 06:14
scook0 chromatic: foo $ bar x is like foo(bar x)
SamB Prelude> :i $
$ :: (a -> b) -> a -> b -- Imported from GHC.Base
infixr 0 $
kelan $ just helps reduce parens 06:15
scook0 chromatic: (foo . bar) x is like foo(bar(x))
chromatic Ah, it's associativity hints for the parser.
scook0 basically, if you write "foo bar x", it thinks 'bar' and 'x' are args to 'foo' 06:16
whereas if you write "foo $ bar x", it evaluates 'bar x' and gives the result to 'foo'
kelan unless foo is defined as only taking one arg, tho, right?
SamB we say that $ has the loosest possible associativity
scook0 kelan: AFAIK no -- that just gives you a type error (not sure though) 06:17
chromatic I can see that being useful when you want to curry one of them too.
SamB kelan: even then!
kelan ok
SamB it might or might not be a type error
chromatic Hm, t/oo/destruction.t still fails.
Is that because there's no guarantee of when GC will collect unused objects?
SamB chromatic: this is the hard part.
chromatic: it might be 06:18
chromatic I'm willing to blame my bugs on non-determinism.
SamB look at the output of 'ghc +RTS --help' 06:19
if you prod it enough it may collect your objects 06:20
chromatic Shouldn't it collect them at the end of the program anyway?
Or does it just to a POSIX-like _exit? 06:21
f0rth is pugscc broken?
SamB it still might not manage to get them all, I saw something in GHC.IOBase or someplace like that...
gaal is & (quantum) tighter than || (classical)? 06:24
$mode = "rw" if $rw || $r & $w; # correct?
scook0 gaal: S03 says yes 06:27
gaal good. :)
thanks!
chromatic Ahh, it worked. 06:29
I added a loop to t/oo/destruction.t to create 10,000 objects.
*That* triggered GC, at least to collect 9918 of them.
ingy chromatic!
chromatic ingy-pu
ingy chromie: you never answered my email :( 06:30
nothingmuch i need some perl 6 module help
chromatic I don't call my mother often enough either.
nothingmuch I want to write a publisher/subscriber thing
chromatic ingy, I remember thinking "Yeah, that sounds good".
nothingmuch i'm thinking Class::Publisher is a role
Class::Subscriber is maybe a role 06:31
ingy chromatic: so i should just send you an article?
nothingmuch Class::Events::Subscription
Class::Events::Notification is dispatched by looking at subcriptions
how would I organize it in files? 06:32
svnbot6 r3963, scook0++ | Oops, I broke Haddock
chromatic Let me pull up the message again.
SamB chromatic: did you look at the RTS options?
chromatic ingy, it looks good. Write it!
ingy ok 06:33
chromatic SamB, I did, but I'm not sure how to set any of those from the test suite.
SamB chromatic: oh, right.
chromatic ingy, if you go to St. Louis, you won't see my new house. 06:34
Curtis and Josh H. and the Walls have, but Schwern hasn't. 06:35
DESTROYALL fires in the wrong order, I think, but that's a fix for another time. 06:43
ingy chromatic: I'll be at oscon
chromatic Yeah, I saw. 06:44
ingy maybe i can see it then
chromatic Someone said I should have a big barbecue here. I'll have to mow my lawn. blah
svnbot6 r3964, chromatic++ | Add DESTROYALL support (with help from autrijus++, scook0++, and samB++). 07:03
r3964, chromatic++ | Make t/oo/destruction.t trigger GC to fire destruction (and untodo).
r3964, chromatic++ | Make ext/Test-Builder/t/010_Test_Builder.t trigger GC.
ingy yay. Test::Chunks just got to 300 tests (in 55 files) 07:13
gaal what's the Right Way to make sure the what's in src/perl6/ ends up in blib6/lib? add a it to build_subdirs in Makefile.PL and give it a makefile of its own? 07:18
pasteling Someone at 212.143.91.217 pasted "Class::Events" (89 lines, 2.4K) at sial.org/pbot/10471 07:20
nothingmuch s/Someone/nothingmuch/
pasteling "nothingmuch" at 212.143.91.217 pasted "use v6; # the object model is" (111 lines, 3K) at sial.org/pbot/10472 07:30
nothingmuch oi, that's silly 07:31
Aankhen`` Does that code work in Pugs?
nothingmuch pbotutil.pl is confusing
didn't try it yet
i'd like someone to comment on the design 07:32
oh, i think i see how 07:39
Aankhen`` I'm thinking the replacement for CGI.pm could start with porting the HTTP::* modules from LWP. 07:47
Any comments?
nothingmuch i think that's a very good idea
do you know Catalyst/Maypole etc? 07:48
the model for request handling is very consistent there
all based around the HTTP::* stuff from LWP
Aankhen`` Not really.
Ah.
I had one question which I was hoping to answer through discussion here though: what should the package be called?
nothingmuch oi vey
Aankhen`` Should it continue to be LWP?
nothingmuch i think that literally it's the WWW lib for perl 07:49
it covers HTTP, and UA abstraction, and what not
as long as the modules are independant enough, i don't care what the package is
Aankhen`` I'm not planning to port LWP::* at first, I just want to get the HTTP::* modules into P6.
nothingmuch if they are all heaped together: URI, HTTP::Headers, HTTP::Request, HTTP::Response, all under LWP
or each with their own package, i don't really mind 07:50
Aankhen`` As far as the name goes, I'm only asking because I need a dir to throw it all under. :-)
nothingmuch they will probably be std in the distro of perl i will use
i'd say keep LWP for now
Aankhen`` A'right.
nothingmuch or maybe if you want to be more impressive, make one for each
and then if someone gets annoyed merge them
Aankhen`` Heh.
nothingmuch that would look like there are many more modules for perl 6 already ;-)
Aankhen`` I think I'll just put them under ext/LWP/.
LOL. 07:51
nothingmuch++
nothingmuch okay, i think i got the Class::Events thing now
by doing it with roles 07:52
err, by appending to roles
Aankhen`` Heh.
nothingmuch tired, many hours sans sleep
Aankhen`` They Do It With Roles.
nothingmuch =)
this is really fun stuff 07:53
Aankhen`` tries to remember what his other question was.
Aankhen`` fails miserably. 07:54
autrijus, you around? 07:55
nothingmuch okay, adding this dummy into modules 07:56
Aankhen`` It works? 07:57
nothingmuch damnit, i hate recursive svn add
Aankhen`` Heh. 07:58
nothingmuch no
starting with tests now
Aankhen`` OK.
nothingmuch s/no/no clue/
actually i'll do tests when I wake up 07:59
Aankhen`` Going to sleep?
nothingmuch yes
11AM
Aankhen`` Sleep well. :-)
nothingmuch it's a holiday
where we make fires, and stay up all night
and then pretend we can last all day long 08:00
Aankhen`` LOL.
nothingmuch and thus not screw up our sleep cycle
now, i'm a very bad case
because work celebrated it the day before yesterday
and friends celebrated it yesterday
Aankhen`` Owa.
nothingmuch so in the last 40 hours or so i've had 4 hours of sleep 08:01
Aankhen`` That sounds ever so slightly tiring.
nothingmuch but I soooo don't want to ruin my weekend by sleeping all day
and staying up another night
Aankhen`` Ah.
nothingmuch but anyway, i'll sleep some now 08:02
good luck with HTTP foo
Aankhen`` Thanks. :-)
Sweet dreams. ;-)
nothingmuch IMHO, btw, the current modules are pretty good
all they will need right now is named params in addition to positional ones
and i think they will be perl6ish enough
svnbot6 r3965, nothingmuch++ | Initial design draft of Class::Events
Aankhen`` I shall massage them.
nothingmuch because they are very lean already
ciao!
Aankhen`` Ciao. 08:03
svnbot6 r3966, nothingmuch++ | Make the useless abtstract Class::Events baseclasses have a semi-useful default behavior. 08:13
Aankh|Clone Lots of unexpected successes today. 08:14
08:33 Aankh|Clone is now known as Aankhen``
svnbot6 r3967, nothingmuch++ | Class::Events makes use of delegation 08:33
r3968, nothingmuch++ | Delegation typo fixed
r3969, nothingmuch++ | multimethods in Class::Events were utter crap
Aankhen`` What was that nonsense about sleeping, nothingmuch? :-P
clkao *yawn* 08:34
Aankhen`` stuffs spinach down clkao's throat. 08:35
kungfuftr shit... i actually have to write a talk now 08:47
nothingmuch nothingmuch.woobling.org/yapc-slide...start.html 08:49
shyte, wrong window
no one look, it's a surprise
plus i don't have enough BW
hey! someone from fotango is cheating! 08:52
nothingmuch makes log-watcher pose 08:53
Aankhen`` How come t/oo/class/nested_use.t is showing everything as TODO?
nothingmuch oh crap, people, seriously my server can't hanlde it
Aankhen`` Uh. 08:54
lives_ok means TODO?
clkao autrijus: you didn't fix method.t? and chained p5 object invocation now doesn't work at all 08:59
Aankhen`` This doesn't seem to work: my @foo = <a b c>; my %bar = @foo.map:{ uc($1) => 1; } 09:26
However, it works when you put another statement before `uc(...)`: my %bar = @foo.map:{ 1; uc($1) => 1; }
Is that supposed to happen? 09:27
Do we have tr///? 09:43
Juerd Have you tried it? 09:49
Aankhen`` I'm not sure of the syntax. 09:51
Will it also use ~~?
And it seems to give an error which I don't really understand. 09:52
Juerd I don't know. 09:53
clkao hmmmm 10:18
svnbot6 r3970, nothingmuch++ | `role Class::Events::Publisher is extended`
clkao ./pugs -Iext/Test/lib -Iext/Benchmark/lib fight.t
hyper: 1013.5884711203811660260956460597711094242747 / s
normal: 369.7574631384163029023557432855280071866061 / s
p5: 327.7763980212401072576151716943578176948974 / s
nothingmuch are units of time that small even measurable in theory? 10:19
or is the speed of light insufficient to travel a measurable distance, say smaller than a quark, in order to notice the granuality? 10:20
clkao autrijus: btw, roundtrip now fails 10:21
nothingmuch clkao: nopaste fight.t?
pasteling "clkao" at 80.169.162.72 pasted "fight.t" (33 lines, 529B) at sial.org/pbot/10473 10:22
clkao but i have to remove the overhead stuff in benchmark, otherwise it becomes negaitive value for perl5 ;) 10:23
nothingmuch how fast is that in "real" perl 5? 10:24
i.e., without the invocation overhead
because I reckon timethese is a tight loop much like the loop you're testing is 10:25
clkao Benchmark: timing 100000 iterations of me...
me: 1 wallclock secs ( 0.59 usr + 0.00 sys = 0.59 CPU) @ 170666.67/s (n=100000)
XD
10:32 TheBigBear is now known as BigBear
svnbot6 r3971, nothingmuch++ | Fix yet another typo. nothingmuch-- 10:42
Aankhen`` How do I assign an arbitrary property "bar" to an object "$self" from within a method such that I can access it using $self.bar? 11:20
Juerd $self.bar = 15; 11:21
broquaint $self.bar = 'stuff';
Aankhen`` That gives me something about not finding a compatible subroutine.
Juerd (Incidentally, this SUCKS because you get no typo checking at all this way...)
Aankhen``: That's lack of feature then, probably
Aankhen`` Ah.
dakkar $self but :bar(stuff) ?? 11:22
Aankhen`` And if the property name is contained in a variable?
Juerd $self.$var = 15 11:23
BUT - obviously, that's very bad style.
Aankhen`` What are the alternatives?
Juerd Use a hash.
$self.foo{$var} = 15
Aankhen`` Hrm, nevermind, actually, I forgot that those properties ought to have accessors anyway.
I'm working on porting HTTP::Headers; any suggestions for the name of the hash? I'm a little hesitant to just use "headers" since then you'll access it using: $headers.headers<foo> # assuming $headers is an HTTP::Headers object 11:24
Juerd Ideally, the object itself will in the future behave like a hash, making this a non-issue 11:25
Aankhen`` Alright.
I'll stick it in $object.data for now, then.
Juerd And you're forgetting that you CANNOT use a simple hash for this
Because there can be several of a given header. 11:26
Aankhen`` I'm planning to store arrayrefs in each key.
Juerd I wonder if that isn't needlessly inefficient
Aankhen`` What would you suggest?
Juerd Storing it as a single value until a second enters 11:27
Aankhen`` That seemed a little inconsistent to me, which is why I was planning to do it for all the headers.
Ah well.
Juerd And have $headers.header('Set-Cookie') be smart about context
Aankhen`` Yeh.
Juerd (in scalar context: any(@cookies).pick, in list context: @cookies) 11:28
(Random in scalar context to discourage actually using it, without making it useless.)
Aankhen`` OK.
Juerd And have these arrays do the same in non-array context :) 11:30
Hm
Aankhen``: The headers can't be in a hash, by the way, because order can be important! 11:31
(Officially, order must be ignored, but practically, it matters sometimes)
Aankhen`` Well, the P5 HTTP::Headers puts them in a hash.
It has this array which contains the headers in "Good Practice" order. 11:32
Or rather, the names of the headers.
Juerd Fortunately, Perl 6 should make creating a hasharray or arrayhash trivial :)
Aankhen`` That's nice to know. :-D
So how do you do it? :-D
Juerd That I don't know
Aankhen`` Heh. 11:33
Juerd I can imagine several ways.
wolverian I don't quite understand the whole Proxy.new business.
Aankhen`` Or are you talking about taking a normal array or hash and using a Proxy interface?
Juerd Probably class HashArray does Hash does Array { ... } 11:34
Alias_ ugh
Juerd class HTTP::Headers does HashArray { ... }
And you're practically done. 11:35
Alias_ A HashArray is an ordered associative array? 11:36
Juerd class HeaderValue does Str does Hash does Array { ... }
Alias_: Yes
Alias_ neato
I've been waiting for this for ages
broquaint Reckon we could get Hugs on feather, Juerd? And what's the best way to get modules installed, ask you or do it ourselves?
Alias_ I use something similar in a object persistance layer named OKVP
Juerd What is hugs?
Alias_ For "Ordered Key-Value Pairs" 11:37
Juerd And re modules: if there are debian package, ask me. Otherwise, install locally.
Aankhen`` It's like Tie::IndexedHash.
broquaint Haskell interpreter, but I forgot about it already has ghc, so no matter.
Alias_ But having something to inherit from directly would be great
Aankhen`` (or something like that, I can't quite remember the name)
Juerd Aankhen``: IxHash
Aankhen`` No, IxHash is different.
A different module, atleast.
broquaint K, thanks again, Juerd. 11:39
Aankhen`` Juerd >> So should I just use a "HashArray" class for it right now? 11:45
Juerd If it exists
Aankhen`` I'll create it, of course. :-)
Aankhen`` fires up pugs and checks it out. 11:46
Darn, it doesn't work.
11:50 Aragone is now known as Arathorn 11:59 TheBigBear is now known as BigBear
Aankhen`` How do I get the class of an object so I can create a new one? 12:24
(I don't want to clone it since that would copy existing attributes)
s/attributes/properties
s|$|/| # stupid keyboard
Ah, $obj.ref works. 12:27
castaway darn, shoulda suggested it when I thought of it ,) 12:29
Aankhen`` Heh. 12:31
I got confused at first and didn't quite understand what was happening.
All well now though. :-)
castaway goodgood 12:34
Limbic_Region Juerd - perlmonks.org/index.pl?node_id=461041 12:38
oops meant that as a message
Juerd installs almost all available debian cpan module packages on feather 12:49
606 MB including dependencies
Alias_ Are you going to do a comparison of what percentage that is?
Juerd No 12:50
It's just so that many modules will be installed and ready to use :)
broquaint Juerd++ # installing modules 12:52
Juerd broquaint: I saw you busy installing lots of modules, and thought I could preempt on that ;)
broquaint Thanks :) 12:53
Juerd debian++ 12:54
done. 12:56
broquaint Hurrah! 12:57
nothingmuch sigh 13:08
i feel warnocked again
Limbic_Region does 10 .. 1 work in p6 or do you still need reverse 1..10 ? 13:26
Alias_ I hope not 13:27
might cause unexpected behaviour in something like 1 .. $x for low values of $x
broquaint How do you do backtick-esque stuff in perl6? 13:29
Khisanth Alias_: how about 1..10:by(-1) ? 13:44
Limbic_Region Khisanth - BASIC rules 13:45
Khisanth eh? I am not that fond of BASIC
Arathorn L~R is alluding to FOR I=1 TO 10 STEP -1 13:46
i think
or whatever the syntax was :)
10 TO 1 STEP -1, even
Khisanth can't really decide what might be correct with that 13:47
Aankhen`` STEP -1, I think.
It basically adds the step.
Khisanth I meant for perl6! 13:48
Aankhen`` I was talking about BASIC, heh.
Arathorn grins
Khisanth 1..10 to get 10 to 1 just doesn't feel right though :p
Aankhen`` Yep. 13:49
Which is why it ought to be 10..1:by(-1).
10 + (-1) = 10 - 1 = 9 and so on, in case anyone here forgot their addition. :-D
Khisanth yeah ... seems better but that :by(-1) should be required
Arathorn isn't so sure 13:50
Khisanth hrm actually no
Arathorn is .. defined for fractional numbers?
Aankhen`` I'd think that the basic mechanism of the range operator would make it required -- it just keeps adding one till it gets to the end, right?
Arathorn why not just keep adding sign($b-$a)? (in the case of $a..$b, whatever the basic-style sign() function is in perl6) 13:51
(until you hit the endstop)
Aankhen`` Noooooooo! 13:52
Evvviiillll!
EVIL I SAY
Aankhen`` coughs.
Yes, that sounds perfectly reasonable.
Arathorn why, though?
<Alias_> might cause unexpected behaviour in something like 1 .. $x for low values of $x
how low? $x<0 or 0<$x<1 ?
Aankhen`` Anyway.
BBL.
defitro Hey hey, good morning all. Is there, or are there plans for a Data::Dumper in perl6? Thought I'd seen mention of a dump() somewhere... 13:59
Arathorn you just do .perl() on whatever you want to dump, i believe
(although i think that's a pugs-proprietary-extension)
defitro excellent! Also, is there full socket support in pugs as of yet? 14:01
Arathorn there's certainly enough to write IRC bots 14:02
c.f. search.cpan.org/src/AUTRIJUS/Perl6-.../IRC-OO.pm 14:03
defitro doh, I ran into that the other day, didn't put two and two together.. 14:04
svnbot6 r3972, autrijus++ | * guarantee a GC run before and after INIT and END. 14:11
14:12 cfranks is now known as fireartist
gaal i added a Prim but it's not getting recognized. any ideas how to debug this? 14:15
Alias_ Plus we have CPAN support, so I imagine one could use POE anyway?
gaal ..actually i'm reviving a Prim that I'd already added and which used to work... 14:16
the interesting lines are
op2 "Pugs::Internals::openFile" = \x y -> do
\\n IO pre Pugs::Internals::openFile (Str,Str)\
anyone see something wrong with that?
also: is there currently a way from p6 code to see all the available builtins? sorta like peeking at %:: in p5, only... for builtins ;-) 14:18
fireartist Hi folks, recently joined 'pugs commiters' - just dropping by to see what's happening
gaal i gotta run now, i'll backlog. & 14:19
hlafarge hi everyone, i'm getting "Could not find module `System.Eval'" error when trying to compile pugs 14:20
anything obvious i'm missing ? (i can provide more info if needed)
first, i'm compiling under "export PUGS_EMBED=parrot" control 14:21
gaal hlafarge, try unregistering hs-plugins and rebuilding
i'm out for real now. see ya. & 14:22
Alias_ Arathorn: Currently, 3 .. 0 does nothing, which is good for things like "Do something to all of the 4th or later elements"
nothingmuch would anybody be willing to discuss Class::Events to death?
hlafarge gaal: how do i do that (unregistering) ?
Alias_ Arathorn: It works very well, and doesn't do unexpected things. What I wouldn't expect would be for it to do something to the 4th, 3rd, 2nd, etc...
Arathorn well, it's only because you're used to the perl5 semantics of it 14:25
and i have to admit that i've used 3..0 to return () in perl5 quite a bit
Arathorn shrugs
Alias_ Arathorn: True, but also because I'm used to it being a flip flop and not a "range" operator
Arathorn hasn't the bistable functionality been dropped for it for perl6?
(and does reverse (0..100000) happen lazily under perl5, out of interest?)
PerlJam Arathorn: I hope not 14:26
Arathorn i'm prolly imagining that, then
Alias_ foreach ( reverse 0 .. 100000 ) { ... } has been explicitly optimized since 5.8.something
Arathorn nods
Alias_ it's turned into $what_you_expect
I've never seen anyone use a flip flop properly in Perl 5 except me anyway :) 14:27
svnbot6 r3973, autrijus++ | * fire DESTROYALL in reverse order of BUILDALL.
Alias_ foreach my $line ( @perlcode ) {
Arathorn i've written in perl5 for years and I only noticed the flipflop behaviour of .. a few months ago 14:28
Alias_ next unless $line =~ /^=\w/ .. $line =~ /^=cut\b/;
...
}
PerlJam heh
Arathorn: Have you ever used the ... operator in perl5?
Alias_ Standard "do stuff while in POD" flip flop
PerlJam Arathorn: have you ever used the match-once pattern matcher? ?pat?
Arathorn didn't know there was a ... :) 14:29
PerlJam Arathorn: have you ever used reset() or dump()?
Arathorn recognise ?pat? from some perl4, though
and never used reset() or dump().
(not sure what that tells you...)
Alias_ I should run a PPI tinderbox scan to see if anyone uses ?pat?
PerlJam Arathorn: Perl is big enough that there are probably lots of things that you'll never use or even know exist unless you happen upon them in some stray bit of code.
Alias_ PerlJam: Or write a Perl parser ;) 14:30
The flip flop is kind of cool, but absolutely write-once code
I recomend against using it even when it's appropriate 14:31
PerlJam Arathorn: ... is just like .. mostly, except it modulates the behavior of the flip-flop version.
Alias_: what?!? you're clearly insane.
Alias_ PerlJam: I'd rather maintainable code than small code
PerlJam Alias_: .. *is* maintainable.
Alias_ Hell, _I_ have trouble reading flip flop code
1 .. $x is maintainable. next unless $line =~ /^=\w/ .. $line =~ /^=cut\b/;
... is less so
I mean when you actually use flip flops full on, not when you it in it's "range" thinkalike 14:32
Because there's invisible state 14:33
the flip flop holds state over the calls, and invisible state is nasty stuff
PerlJam how is the state invisible exactly? 14:34
broquaint Yeah, filehandles suck ...
Alias_ It isn't stored anywhere you can see
PerlJam you don't think the flip-flop syntax is distinctive enough?
Alias_ or manipulate, or pass around 14:35
broquaint And /.../g too ...
Alias_ broquaint... huh?
PerlJam Alias_: He's pointing out other instances of "invisible state"
Alias_ I don't see any invisible state there...
broquaint e.g /.../g also holds state invisibly.
Arathorn capture groups are fairly visible..
oh, the /g
point
Alias_ broquaint: Since when...
$_, $1, $2 14:36
...
broquaint Have you not done while(/.../g) {} before, Alias_?
Arathorn i think he's referring to the current position in $_
Alias_ yep, and you can always see the current value in $_
or the position magic, which I can't remember right now
Arathorn (which is accessible visibly through some random punctuation $variable, surely?)
broquaint pos() 14:37
Alias_ right, or that
Tell me how to find the current value of a flip flop
broquaint But it is implicitly maintained state nonetheless.
Alias_ except by implication
PerlJam And you evaluate the flip-flop :-)
broquaint I'm not arguing for flip-flop though. There's a way of accessing it's state but it's fairly magical. 14:38
svnbot6 r3974, autrijus++ | * add correct probing of hs-plugins so we don't trip over
r3974, autrijus++ | older versions of it.
r3975, autrijus++ | * refactor newSVval to handler Val->PerlSV casting.
r3975, autrijus++ | * beginning of unifying two GCs together.
r3976, iblech++ | modules/Class-Events -- Usual svn properties.
r3977, iblech++ | A handy utility added for screen(1) users --
r3977, iblech++ | examples/network/screen-nodestatus.p6 monitors the status of your hosts.
Alias_ regardless, I've abused flip flops before, and even I couldn't read my own code when I came back to it
I prefer to write Perl at one level below my skill level, so debugging is possible and less skilled people can maintain it 14:39
PerlJam Alias_: so ... because *you* abuse it, it shouldn't be part of the language?
Alias_ PerlJam: You think it shouldn't be part of the language?
PerlJam It's a useful tool that I've grown accustomed to you might say.
Alias_ It has it's place, but I still recommend not to abuse it, because it's very write-once 14:40
PerlJam It's like having a tack hammer. It's something you rarely need but when you need a tack hammer it's the right tool. Never mind that you can also use it to drive other things.
Alias_ Range-like uses are fine, but algorithmic ones are less so
PerlJam I will admit that the difference between .. and ... is to subtle for most people 14:41
Alias_ And it's often fairly easy to implement the same thing as you would abuse a flip flop for using 2 or 3 additional lines
I'd rather do that and retain readability
PerlJam (this is why most perl programmers who know about .. as flip-flop have no clue about ...)
sproingie i grew to hate the .. operator
its extent is apparently global 14:42
Alias_ as range or otherwise
sproingie as flipflop
PerlJam sproingie: really? I've been using it for nigh on 15 years now. I like it alot :-)
sproingie it's good in scripts, not programs, because it has global extent
Alias_ I've used it enough times to know that it's hard to debug easily...
And that alone is enough for me to avoid it
PerlJam boggles
debug? I've never had a problem with it. 14:43
Even when I was a perl newbie.
Alias_ as in, running through the debugger and understanding what's going on
sproingie it's quite nice for text munging scripts ... that operate on one file, and one only
because it doesn't reset
PerlJam sproingie: do you mean one filehandle? 14:44
Alias_ Doesn't it reset when it loses scope?
PerlJam (including ARGV for the magic <>)
sproingie Alias_: it apparently never loses scope
Alias_ hmm... that seems odd
sproingie i dont know the precise rules, but the behavior is mysterious
Alias_ right
I don't like mysterious
PerlJam sproingie: It's simple!
It's true as soon at the LHS is true and stays that way until the RHS becomes true. 14:45
the difference between .. and ... is when the RHS is first evaluated.
sproingie right, but i had thought when it went out of scope, it would reset
it does not. it has global extent
PerlJam what do you mean by "scope" in this instance?
sproingie put it in a function
integral err, that would break: while (<>) { //..// } since it would reset on every loop
Alias_ he means it never resets
integral: You never leave scope in loops
not till the end 14:46
which is why you 'my'ed variables aren't re-initialised each loop
errr.. not reallocated rather
integral oh, ok. I really hate that bit of perl's pads. 14:47
nothingmuch /me never sufferred from it
do you mean at the XS level?
Alias_ nothingmuch: From time to rare time you can encounter memory bloating of a type 14:48
svnbot6 r3978, iblech++ | Makefile.PL -- It's not "you are building perl5", but "building with embedded
r3978, iblech++ | perl5".
Alias_ because things aren't cleaned up exactly like you would initially expect
It's one of those very rare gotchas 14:49
I've hit it once in 7 years
nothingmuch isn't that perl's memory allocation behavior in general?
Alias_ what, bloat? :)
nothingmuch do you remember the scenario?
Alias_ not the particulars any more
I think it had something to do with a 100meg memory construct
Arathorn i got the impression that perl5 basically never gc'd :)
Alias_ And it built into another 100meg construct
nothingmuch hmm
Alias_ but I was careful to reduce the first one as I went
nothingmuch destructors were called, but data not deleted?
since scope was not exited memory was not reusable, so more was mallocced? 14:50
PerlJam nothingmuch: debugging someone's imagination isn't very productive :)
Alias_ and then on the next loop, the next 100meg thingy was built, but the previous end one wasn't cleared until I started reallocating to it
But it was a couple of years ago, so I'm guessing a little
nothingmuch PerlJam: i'm still working on the context switch 14:51
define reallocating and cleared
Aankhen`` What is the signature of `join`? 14:52
PerlJam still doesn't understand what people were expecting of flip-flop and not receiving.
nothingmuch reallocating is creating new perl values on memory space that was prevously mallocced?
PerlJam Or even why they were expecting flip-flop to be that way.
nothingmuch this discussion is about ..?
Alias_ nothingmuch: The 'my' variable didn't cease to exist at the end of the loop. Perl kept it to avoid having to create it again
nothingmuch err, C<..>?
PerlJam nothingmuch: no no, that was the previouis discussion that segued into scope and allocation
nothingmuch so, for example, a reference to it was still the old value? 14:53
Alias_ But again, I'm really guessing here. You'd need to search the newsgroups and mailing lists
nothingmuch i'll pass, i think
integral perl -le 'sub foo { scalar($_[0] .. $_[1]) } print foo(1, 0); print foo(0, 1); print foo(0, 0)' # hidden global state... "bad" 14:54
sproingie nice example 14:55
what gets me is that the documentation for .. doesn't seem to let on to this
or perhaps it does and i ignored it
integral perl -le 'sub f { sub{scalar($_[0] .. $_[1])} } sub foo { my $f = f; &$f } print foo(1, 0); print foo(0, 1); print foo(0, 0)' # very hidden global state 14:57
sproingie i wonder if it would reset if i evaled it
PerlJam you know ... maybe the docs don't explain it clear enough and I'm just so used to the behavior that it's not a problem for me to figure out what's going on.
sproingie integral: while you're testing it, could you try eval?
nothingmuch anybody got a comment on this: www.nntp.perl.org/group/perl.perl6....uage/21560 ? 14:58
integral the .. state is just stored in the OP for the .. (the flip or flop?), isn't it?
Alias_ PerlJam: Or you've just never hit the corner cases
nothingmuch it's really bugging me
integral PerlJam: it's not a problem at all when you don't have subroutines
PerlJam Alias_: um ... there *aren't* any corner cases as far as I'm concerned ;-)
integral PerlJam: ever used .. in a module?
PerlJam integral: many times!
sproingie huh
.. basically isn't reusable from what i've seen
Alias_ PerlJam: Then you obiously aren't looking hard enough :)
PerlJam sproingie: sure it is. 14:59
crap ... I'm missing a meeting.
we can argue later :)
sproingie carp("missing a meeting")
integral perl -le 'sub foo { my $x = 0; for(@_) { $x++ if $_..0 } return $x } print foo(0, 0, 0); print foo(0, 1, 0); print foo(0, 0, 0)' # hmm 15:00
fireartist Is anyone working on implementing "has $.attrib" in pugs ? 15:01
Aankhen`` Isn't that already working?
integral perl -le 'sub bar { scalar($_[0]..$_[1]) } sub foo { my $x = 0; for(@_) { $x++ if bar($_,0) } return $x } print foo(0, 0, 0); print foo(0, 1, 0); print foo(0, 0, 0)' # !
err, no, that's not surprising. err.
sproingie perl -le 'sub foo { eval { scalar($_[0] .. $_[1]) } } print foo(1, 0); print foo(0, 1); print foo(0, 0)' # really REALLY hidden global state... "bad" 15:02
fireartist (it's marked :todo<feature> in attribute.t) 15:03
sproingie however: perl -le 'sub foo { eval q[ scalar($_[0] .. $_[1]) ] } print foo(1, 0); print foo(0, 1); print foo(0, 0)' # no hidden global state here
er actually, doesnt seem to work at all :) 15:04
Aankhen`` ok 5 - .. autogenerated accessor works 15:05
ok 6 - .. autogenerated accessor works w/out parens
svnbot6 r3979, iblech++ | screen-nodestatus.p6 -- Added a link to a screenshot.
sproingie should be getting 1E0, right? .. seems to work if i use qq
Aankhen`` BBL. 15:06
fireartist thanks Aankhen`` - I'll see if there's enough implemented for my needs! 15:07
broquaint Aren't all operators reduceable now? 15:31
i.e my $str = ', ' [~] @arr; 15:36
clkao seen autrijus
jabbot clkao: autrijus was seen 11 hours 4 minutes 14 seconds ago
clkao i guess it's time to wake up
nothingmuch clkao: actually he committed changes up to around 2 hours ago, i think 15:38
Juerd broquaint: All infix and postcircumfix operators 15:39
nothingmuch but i don't think that invalidates your remark
Juerd broquaint: However, [op] itself is prefix, so ', ' [~] @arr is wrong
That should be either ', ' ~ [~] @arr or simply [~] ', ', @arr
broquaint Aha, thanks, Juerd. 15:40
iblech Aankhen``: (backlogging) HTTP::Headers is already ported, see ext/LWP-Simple/HTTP/Headers.pm 15:45
clkao yay, more perl5 fixes 15:46
ingy anyone here want to help me test a subethaedit connection?
nothingmuch ingy: sure
ingy nothingmuch: one sec
nothingmuch: try connecting to freepan.org 15:47
nothingmuch works
ingy type something? 15:48
nothingmuch see it?
broquaint Shouldn't this work?: pugs -e 'class foo {} foo.can("bar")' 16:04
Currently it complains: *** No compatible subroutine found: "&can" 16:05
Limbic_Region correct me if I am wrong, but while my @foo = $sub() { ... } is valid syntax, it just means that @combo is going to persist beyond the loop? 16:06
iblech broquaint: I'm not sure if .can works on Class objects, too (but, for consistency, it should). 16:07
broquaint: But IIRC .can isn't implemented yet.
Limbic_Region: Yep, should work.
Aankhen`` iblech >> I really should check for prior art... :-) Mind if I move LWP-Simple/* to libwww-perl/? 16:08
iblech Aankhen``: Go ahead :)
Aankhen`` Thanks.
Limbic_Region iblech - should, but IIRC isn't currently supported in Pugs
Limbic_Region has given up waiting for stuff and am just writing his cribbage solution as if it were 16:09
broquaint Ok, thanks, iblech.
iblech Limbic_Region: I think it's because of the my(). my is currently only supported on statement level, e.g. "my $a = 3;", it doesn't work yet in expressions (there's already a test for it, t/pugsbugs/parsing_my.t IIRC)
Limbic_Region no worries iblech 16:10
I plan on having a few people review my code for obvious errors and suggestions
and then check it in
hopefully someday it will run
iblech Limbic_Region: I'm sure it will, eventually :) 16:11
Limbic_Region well - I have p6ified it quite a bit with the help of revdiablo and dragonchild - but I still feel it has room for improvement 16:12
perlbot nopaste
err um 16:13
revdiablo Limbic_Region: sial.org/pbot/perl6 16:14
pasteling "Limbic~Region" at 129.33.119.12 pasted "Work in progress so far" (89 lines, 2.7K) at sial.org/pbot/10482
Limbic_Region about to start working on new_deck() and then going to look at incorporating gather/take 16:15
Aankhen`` I think I messed up something. 16:25
I `svn move`d all the files from LWP-Simple/ to libwww-perl/... then I tried to remove LWP-Simple/, but I get an error saying that it isn't a working copy directory...
gaal someone remind me what's the "usual svn properties" incantation for new files? 16:26
broquaint Where's all the OO code in pugs located?
clkao autrijus: still spins
and coredmp
pasteling "Limbic_Region" at 129.33.119.12 pasted "Please review for syntactic correctness and any suggestions for inclusion as an example" (82 lines, 2.3K) at sial.org/pbot/10485
clkao maybe because the coderef is gc'ed?
hm, or because there's no & 16:27
gaal is it just "eol-style native" ?
clkao autrijus: btw roundtrip fails now 16:28
iblech gaal: See docs/other/development_tips 16:29
gaal just found it, thanks :)
iblech Aankhen``: IIRC this is normal, after a directory mv you'll have to rm -rf the original dir by hand
Aankhen`` C:\DEVELO~1\pugs\ext>rm -rf LWP-Simple 16:30
rm: cannot remove directory `LWP-Simple/lib/HTTP': Operation not permitted (EPER
M)
rm: cannot remove directory `LWP-Simple/lib': Directory not empty (ENOTEMPTY)
rm: cannot remove directory `LWP-Simple': Directory not empty (ENOTEMPTY)
iblech Aankhen``: Huh? I don't know what rm you have, but mine does delete non-empty dirs (when executed with -rf) :) 16:33
Aankhen`` Hrm. 16:34
Strange.
Apparently it's being used by another process.
knewt i find that sort of thing on windows all the time. nothing running that should have possibly been anywhere near a folder other than explorer, but explorer still complains that the folder is in use and can't be removed :( 16:35
svnbot6 r3980, gaal++ | Unicode editing quickref
Aankhen`` Yea.
knewt i should get round to grabbing the program to tell me exactly /which/ bloody program it is. of course, knowing microsoft it'll be explorer itself keeping the folder open 16:36
gaal can someone with Haskell-fu try svn uppping and see why Pugs::Internals::openFile doesn't work?
iblech gaal++ # svn props :) 16:38
integral do I need a even newer hs-plugins for System.Eval?
gaal integral, i *think* it's the newest ones that are broken
but i may be wrong. 16:39
Aankhen`` Hmm.
gaal iblech, heh, thanks
integral oh, ok
Aankhen`` knewt >> You have AV software running?
gaal what the velvety antlers is wrong with my code?
integral hmm, it seems that enabling perl5 embedding for me (10.3) puts a '-optl' arg on the ghc command line which it chockes on 16:40
svnbot6 r3981, gaal++ | openFile updates (doesn't work yet, but shouldn't break anything)
knewt Aankhen``: ah, yeah. SAV. thing is, this only happens sometimes. i can be dealing with lots of different folders, and then randomly 1 or 2 will refuse to be deleted 16:41
Aankhen`` Same here.
I think I'll have to reboot for this to work. 16:42
And I can't do that for another 1 1/2 hours.
knewt have you tried killing explorer, just to see if that will help?
Aankhen`` No.
Ah, let it be. 16:43
I'll try later.
BTW, you wanna see the most useless "troubleshooting" article ever?
www.microsoft.com/resources/documen...s/3729.asp
knewt Aankhen``: explorer is great as a source for hate. the other day it decided to totally lock up one of the several instances of itself it had randomly decided to create. killed all of the explorer processes, and then ran a new instance, but all i got was an explorer window. no taskbar at the bottom. had another look at the task manager and found it had managed to spawn a new instance of itself in between me killing them all and starting a new one 16:45
Aankhen`` I will be so happy when we move abroad and I get a computer of my own on which I can run Linux. :-) 16:46
knewt well, this is my dads computer which i (ab)use for p2p use
Aankhen`` LOL.
This is really annoying though. 16:47
I can't even `svn up`.
knewt i've managed at least to convert him from IE/OE to Firefox/Thunderbird
Aankhen`` Good work. =)
knewt he hates having to use IE nowadays
Aankhen`` Heh.
knewt his two main reasons being tabs and popups
but lots of other little things as well which just add up to a much nicer experience 16:48
gaal btw, today i noticed Prim.hs was hampered by tabs somebody had left in it. should we perhaps put in emacs and vim magic lines to prevent that kind of thing? at least in haskell source. 16:54
the vim line is vim: set ts=4 et :
actually, we can s/vim/vi/g that 16:55
iblech FWIW, I use tabstop=8 shiftwidth=4 expandtab smarttab softtabstop=1 16:56
So I can use <Tab> and <Backspace> for indenting, but Vim indents only 4 chars
gaal okay -- the really critical bit is expandtab.
iblech Yep
gaal so i propose we add it to all .hs files
what's the equivalent emacs incantation?
pjcj perl5 uses: 17:00
* Local variables:
* c-indentation-style: bsd
* c-basic-offset: 4
* indent-tabs-mode: t
* End:
*
* ex: set ts=8 sts=4 sw=4 noet:
gaal in perl the code itself doesn't break as easily. before pugs i always used noet for my own projects, but it looks like it's just dangerous in haskell code. 17:01
wilx noet? 17:02
iblech noexpandtab, I thin
noexpandtab, I think
pjcj that's for perl's C code - just for reference, not copying
gaal yes. == leave a literal ^i in the file
er, yes, c isn't fragile that way either. 17:03
mauke that's why perl's C code sucks so much
it uses mixed spaces and tabs
gaal encouraging *just* expandtab sounds like a good idea to me
because it avoids getting into religious wars
mauke I use tabs only
pjcj mauke: so if we run it through expand perl5 will be maintainable? 17:04
gaal everyone can have their favorite tab width
mauke pjcj: huhu
PerlJam gaal: nah, perl should do like guido wants to do in python: all indentations must be spaces and come in multiples of 4
knewt the topic of tabs came up recently elsewhere. some tabs had managed to work there way into some python code on another project. and apparently python evaluates tabs as being "up to the next 8 character division"
gaal mauke, you do that in haskell?
mauke wait, let me check
Aankhen`` Nooooo!
SamB knewt: what did you expect?
Aankhen`` 8 character Tabs!
SamB you aren't supposed to use tabs in python anyway 17:05
pjcj s/in python anyway//
integral PerlJam: no, no, rather than waste space with 4 spaces, use a special "indent" character! :-P
knewt converting is theremore more interesting than it otherwise would be, because you can't just convert the tabs into a fixed number of spaces, you have to know where in the line the tab occurs
luqui decides he's going to try to learn Haskell again
does anybody here know it?
knewt pjcj: linux insists on tabs :)
PerlJam integral: Any good unicode candidates?
SamB pjcj: well yeah, but its harder to convince people of that!
gaal no, pjcj PLEASE let's not do the war thing.
integral PerlJam: \t !
mauke gaal: yes, my haskell settings are ts=3 sw=3 noet 17:06
integral &
PerlJam mauke++ a man after my own heart
luqui what is op2Cast?
wilx GNU style! 17:07
wilx runs for cover
Juerd 19:11 < integral> PerlJam: no, no, rather than waste space with 4 spaces, use a special "indent" character! :-P 17:08
That would work, if tab wasn't whitespace.
Odin-LAP wilx: While calling out 'fire in the hole'?
wilx While calling "GNU style" :)
svnbot6 r3982, ninereasons++ | add :todo<bug>s to relational operators
Odin-LAP Tab is the most spectacularly messed up character found in ASCII. And that's saying something.
wilx Anyway. Many people thing it is odd one while I really like it :) 17:09
Odin-LAP (It's messed up because of what people have done to it, not because of how it's specified.)
Juerd Odin-LAP: It should not have been specified as displaying as whitespace
PerlJam Juerd: I nominate ` as the indent char :-)
Juerd PerlJam: It's too small
gaal my, where's the UNDO button on this tab debate thing? 17:10
wilx Using \t is odd too. It breaks the intended look in any editor with different tab width.
Spaces everywhere!
Odin-LAP wilx: Huh. Frankly, I tend to more positive towards it with Linus's comment on the GNU coding standards, but that's just the oppressed-minority-whore in me, I think...
wilx :)
Odin-LAP Juerd: Hm. Maybe not...
SamB wilx: and it breaks the code in Haskell!
wilx Right! 17:11
gaal luqui, see Prim/Lifts.hs:52
pjcj Juerd: I set vim to display tabs as ōæ½xBB - so Perl6 messed that up.
SamB let them play with their silly unicode chars for a while 17:13
luqui how do I get the haskell list out of a VList?
gaal hl <- fromVal vl 17:14
Limbic_Region ok - now that more people are talking
sial.org/pbot/10485 - Please review for syntactic correctness and any suggestions for inclusion as a Pugs example
luqui mk
Limbic_Region warning - it won't run on Pugs now as several features are unimplemented (which is why I need help verifying syntax) 17:15
Juerd next if none( @combo>>.<val> ) == 5; 17:16
I'd personally write that the other way around
Limbic_Region which other way?
Juerd any( @combo>>.<val> ) == 5 or next;
Limbic_Region ok - but it equates to the same thing - so that's personal preference 17:17
I am more interested in things that are flat out wrong - or could be re-written better using a different p6 feature
Juerd Limbic_Region: Well, instead of the while, a real iterator, used with =, would be better 17:18
But I don't know how to write that
Limbic_Region I agree - I would like to use something with lazy gather/take
Juerd ?? [eq] @hand[3,4]>>.<suit> ?? 5 :: 4
Limbic_Region but also don't know the right way to do it
Juerd Just because nobody actually can remember the precedence of the ternary operator, use parens around the inner.
say join ' ', @combo.map{ $_<suit> }; 17:19
Just ~ it.
say ~@co...
gaal actually Juerd, i think there's an open bug right now there.
Limbic_Region that's a space
gaal (in ?? :: precedence)
Juerd Limbic_Region: Yes.
Limbic_Region so ~ doesn't work Juerd
Juerd Limbic_Region: ~@array joins on the exact same string.
A single space.
Limbic_Region oh 17:20
news to me
thanks
will change
but I also need to deref
how does that work?
Juerd Well, . is tighter
Limbic_Region ~@array>>.<suit>
?
Juerd Yes 17:21
revdiablo perhaps ~@combo.map{ $_<suit> }
Juerd And $_<suit> can be written as .<suit>
revdiablo ah, hyper ops
Limbic_Region say ~@combo>>.<suit>;
Juerd for %ordval.values { $score += [*] $_, $_ - 1 } 17:22
Limbic_Region that's how I am doing it now
Juerd That's just obfuscation
Limbic_Region if that works
Juerd Use $_ * ($_ - 1) instead
Limbic_Region well, it wasn't originally
there was more to it before
Juerd $score += [eq] @hand[0..3]>>.<suit> 17:23
?? [eq] @hand[3,4]>>.<suit> ?? 5 :: 4
Limbic_Region but I agree - with only 2 elements it makes sense to be explicity
Juerd () around both [eq] list
You do want "say $foo ?? 3 :: 5" to be parsed as "say ($foo ?? 3 :: 5)", don't you?
Limbic_Region ok - so what you are saying is add parens to make it clearer
Juerd [op] is a listop just as say.
No, to make it work. 17:24
Limbic_Region I'm not sure I agree with you
let me explain in english
Juerd See precedence table.
Limbic_Region you tell me if the code is right or wrong
mauke would $score += map { $_ * ($_ - 1) } %ordval.values; work?
Juerd mauke: No, comma is required after the closure, I think.
Limbic_Region score should be 4, 5, or 0 depending on a couple of tests
mauke ah, right :-)
Limbic_Region grrr - no plus symbol in this client
score should be incremented by 4, 5, or 0 depending 17:25
if the first test is false - then 0, if it is true then need second test, if second test is true then 5 if false 4
Juerd Limbic_Region: [op] foo ?? bar :: baz is parsed as [op] (foo ?? bar :: baz)
Limbic_Region: So here you need ([op] foo) ?? bar :: baz.
PerlJam Limbic_Region: Just use %2b everytime you want to type +
Juerd I'm really not interested in even thinking about what the code actually does 17:26
Limbic_Region %2b won't work
unless you know what I mean
Juerd The way you lay out the code makes it clear, but perl isn't quite so whitespace sensitive that it will dwym just because of that.
Limbic_Region ok - let me update again and see if what I did makes sense
mauke is this with CGI:IRC?
Limbic_Region Juerd (and everyone) - new version is at perlmonks.org/index.pl?node_id=461131 17:27
mauke - yes
mauke weird, it worked with elinks IIRC 17:28
Limbic_Region Juerd - do those parens now make it work correctly? 17:29
17:29 _SamB_ is now known as SamB
Arathorn stupid question: what do the rightangled brackets in a .<val> or .<suit> mean? 17:29
Juerd ?? ([eq] @hand[3,4]>>.<suit>) ?? 5 :: 4
Still could use extra parens :)
PerlJam Arathorn: It's a form of quoting. 17:30
Juerd # Count 15's
15s
my @vals = @hand.map:{ $_<val> };
@hand>>.<val>
Be consistent...
Limbic_Region oh yeah
sorry
wolverian I despise the ascii hyper ops.
Juerd if [+] @combo == 15
huh?
Limbic_Region if the sum of @combo == 15 17:31
Juerd if ([+] @combo) == 15 then
Limbic_Region k
Juerd list ops have very low precedence
print and say are listops.
[op] is a list op
Limbic_Region changed 17:32
Juerd $score += [*] %ordval{$start .. $start + $span}, $span; 17:33
needless obfu again
Limbic_Region I don't see that as obfu
Juerd return map -> $num {
map -> $suit {
Everywhere you use the .map method
Obviously, here, that would be ugly 17:34
But you can maintain left-to-right reading...
with ==>
Limbic_Region ok - take a step back
$score = [*] %ordval{$start .. $start $span}, $span;
what wouldn't be obfu to you?
svnbot6 r3983, ninereasons++ | perl6.vim: small improvement to pod matching
wolverian ninereasons++ # yay, someone works on it!
Juerd [op] with only 2 *literal* elements is better written as el op el
Limbic_Region that's not 2 statements 17:35
that's a hash slice
Juerd Ohhh
Limbic_Region each one of the returned values is going to be multiplied together
along with $span
Juerd You're right.
luqui okay, how do I get a list out of a VList without using monads
(I think that's the problem)
Limbic_Region Juerd - finally, WRT map consistency
Juerd Change of sigil use threw me off
(I still don't like that change, really)
Limbic_Region those two maps are the only two maps left
the rest were turned to hyper dereferencing 17:36
Juerd Oh, okay
I still like left-to-right reading better though
Limbic_Region ok - your preference
Juerd Because for works that way too
iblech luqui: That's not possible, IIRC
Limbic_Region as long as what I have isn't wrong
Juerd for list -> var { action } 17:37
list ==> map -> var { action }
iblech luqui: Ok, it is possible: case vlist of Val (VList hlist) -> ...do sth. with hlist
Juerd map -> var { action } list
iblech luqui: But hlist <- fromVal vlist is nicely polymorphic, I think
Juerd Limbic_Region: There are many levels of wrong in this universe
luqui yeah, but I think I'm in a context that doesn't allow monads
Limbic_Region yeah, I agree Juerd
Juerd Limbic_Region: And I do consider style one of them, when we're discussing non-production code :) 17:38
Limbic_Region people are more then welcome to change it to their personal preference once I check it in
but for now (while it is on PM anyway) - I just want it to be able to run
Juerd You use qw
That's unperl6ish
Just use <H D C S>
Or better, a global @SUITS 17:39
Limbic_Region thought so but wasn't 100% positive
Juerd H D C S is far removed from $suit
svnbot6 r3984, Aankhen++ | * moved ext/LWP-Simple/* to ext/libwww-perl/.
Juerd That's my main argument against right to left with specified variables
It isn't immediately obvious that they're suits 17:40
Limbic_Region I do appreciate your time - we will just have to agree to disagree with style/clearness
-> $suit and suit => $suit isn't obvious?
Juerd (In code, I want everything that CAN BE obvious to BE obvious, so that comments are needed less - I do think that for this H D C S, a comment is needed.)
Limbic_Region the key to the hash says what it is 17:41
Juerd Limbic_Region: As said, the <H D C S> is too far removed from the -> $suit to be visually linked.
compare:
map -> suit {
aoeuaoue
aoeuaoeu
aoeu {
aoeuaoeu
}
Limbic_Region hang on - real $work
Juerd } <H D C S>
Limbic_Region bbiab
Juerd versus:
<H D C S> ==> map -> suit { 17:42
aoeuaoeu
aoeu {
aoeuaoeu
}
}
where suit should have been $suit
luqui iblech, vCast did the trick
Juerd Declaration and definition shouldn't be spread apart when that isn't necessary
Limbic_Region: Now, to be honest, I'd have written this entire thing as a gather/take 17:43
Limbic_Region: return gather { for 1..13 -> $num { for <H D C S> -> $suit { take { ... } } }
RixiM is there a #parrot someplace? 17:44
svnbot6 r3985, ninereasons++ | hyper.t: todo failing hyper-dereferencing tests
iblech RixiM: Yep, on irc.perl.org 17:45
RixiM ah.
thanks.
Aankhen`` iblech >> In HTTP::Headers, what do you think of making @header_order, %header_order and %standard_case private members of the class? 17:49
luqui hmm, I wrote a new op2 entry in Prim.hs, but it says it can't find the operator 17:51
svnbot6 r3986, Aankhen++ | * ext/libwww-perl/lib/HTTP/Headers.pm: flatten arrays when creating @header_order.
iblech Aankhen``: Feel free to do everything you want with HTTP::Headers :) 17:52
luqui: You have to add it to the big table at the end of Prim.hs
Aankhen`` That's all very well, but I'd rather not do stupid things. :-)
luqui right... I remember now
tahnks
s/ah/ha
SamB is there any particular reason for showType to be exported from Pugs.Types? 17:54
gaal iblech: actually i was doing that earlier today and it didn't work, strangely. 17:56
svnbot6 r3987, Aankhen++ | * made @header_order, %header_order and %standard_case private members of HTTP::Headers.
gaal i know that sounds insane :) 17:57
luqui I'm getting a lookahead conflict 18:04
svnbot6 r3988, iblech++ | Add the svn props for libwww-perl.
r3988, iblech++ | Why doesn't svn keep the properties when moving dirs?
Aankhen`` Ah, sorry, iblech.
luqui When I say 1 ^.. 5, it dies, because it sees the ^ as a junctive xor
any idea how to fix?
integral SamB: remove the export, and see what the compiler complains about 18:07
SamB: if it doesn't complain, commit :-)
iblech Aankhen``: np
integral SamB: and of course if nothing inside or outside the module uses it, delete the code :-)
iblech luqui: Hm... you have to move "^.." further up in some list, I think. But I don't really know Parsec 18:08
SamB integral: well the Show instance for Type does. I was just thinking it might make sense to s/showType/show/ everywhere something referred to it.
integral SamB: hmm
SamB I don't think anything would be slower, though it is of course possible in theory that this would result in some typing ambiguity... 18:09
svnbot6 r3989, Aankhen++ | * re-applied changes clobbered by previous commit.
SamB anyway, I guess one thing at a time is good, so I just wrote a whole export list and didn't take anything out. 18:10
integral SamB++ 18:11
SamB now my question is why is it taking so long to hit my inbox from darcs send (via [email@hidden.address] 18:12
integral hmm? you're not a SVN committer? 18:13
SamB nope
SamB doesn't even have svk yet 18:14
SamB wishes apt would not insist on upgrading perl before installing svk 18:18
integral hmm, what's the difference between: my $a ::= foo(); and my $a = BEGIN { foo() } ?
luqui The :
Well, nevermind
The latter is like:
my $a; BEGIN { $a := foo() } 18:19
While the former is like:
my $val; BEGIN { $val = foo(); } my $a = $val;
integral hmm, is that the right way around? 18:20
SamB neato. I wish C had blocks as expressions...
luqui Yeah, I said latter first.
er
no
mauke SamB: gcc has them
integral jmm, err,
luqui yeah, it's the right way around if you take them respectively rather than antirespectively (switch former and latter)
integral *nod* 18:21
SamB masak: yes I know that!
integral the first though is the issue I tried to raise on the list though
About accessing variables that don't really exist at compile-time
SamB but I am reluctant to use such obscure GCC features, especially ones as syntactical as that one
luqui Hmm... yeah, that my is a bit disturbing
integral but, of course, that thread was warnocked. 18:22
luqui Well, it's a hard question, and one whose answer will come with the implementation (I think)
integral hmm, well autrijus chose the perl5 route, but it complicates the internals
so the whole point of asking was to find out if it needed to be complicated
luqui ahh 18:23
I suppose we'd like: my ::Foo ::= MyClass to work
because otherwise ::Foo doesn't have compile-time meaning
... I don't know
SamB how do you define functions in perl?
luqui what do you mean? 18:24
SamB, are you a perl5er?
wolverian hmm, I just converted this 52-line C routine to 19 lines of a lot clearer perl6, with less idiotic requirements on the user. yay.
SamB like in Haskell I would do foo x = x+1
integral sub foo($x) { return $x + 1 } 18:25
&foo ::= -> $x { $x + 1 } # ...
luqui You can leave out the return if you like
sub foo ($x) { $x + 1 }
SamB no
this is why my example was in Haskell
luqui ? 18:26
wolverian SamB: you'll have to be a bit more informative than that :) 18:27
integral thinks these are just like the haskell
iblech SamB: Perl-return only returns from the current sub, it does not make a monad or so...
SamB I'm lagged. The "no. this is why my example was in Haskell." were in response to '<luqui> SamB, are you a perl5er?' 18:28
svnbot6 r3990, luqui++ | Implemented ..^ ^.. and ^..^, but the latter two don't work because of &infix:<^>. Needs parsing fixes.
luqui ahh I see 18:29
integral what's ..^ and co ?
wolverian ouch, that's some serious lag.
theorbtwo wonders what ..^, etc, are.
Heh.
wolverian integral: exclusive ranges
luqui exclusive range
Forth they look cute......
integral ah
luqui 0..^5 <=> 0..4
vcv 1..^ eq 1..Inf?
oh nm
luqui uh, no
integral but .. are still ranges on enumerable types, not intervals on continuous ones?
wolverian well, 1...^ would be 1..Inf-1 ...
luqui integral, yeah 18:30
Juerd thinks ^.., ..^ and ^..^ are bloat.
luqui continuous intervals are tricky to implement
Juerd And unperlish.
luqui I think ^.. and ^..^ are bloat
integral well, they'd be sets not lists
luqui But I think ..^ is wonderful
SamB what is the difference between -CPugs and -CHaskell?
luqui And we should have ^.. and ^..^ by symmetry, maybe, then
SamB luqui: especially with finite-precision numbers!
Juerd luqui: Yes, if one, then all
luqui: Same goes for if ->, then <-> too :) 18:31
(<- has very little practical use, and is ambiguous with qw)
luqui hehe. Yeah, I like that operator too
SamB continuous intervals are un-almost any language I've ever seen-ish
luqui Mathematica?
integral hmm, well they go with sets fine if you allow infinite sets
Juerd infinite sets 18:32
oh my.
luqui are there languages that allow infinite sets?
luqui thinks that such languages would make you prove everything useful about them
integral perl hopefully, just an object which does Set
theorbtwo luqui: Yes, perl6.
Juerd Great in theory, but trying to figure out if $element is in $set is hardly possible.
integral hmm? Just define it using a predicate...
Juerd It can be done only if you know how the set was built.
theorbtwo $element ~~ $set 18:33
autrijus greetings lambdacamels.
luqui How about: { Rational $x : $x**2 < 2 }
wolverian Juerd: is it possible to get a UTF-8 locale on feather?
integral but is-a-member-of is just a method on an object and can do anything
Juerd wolverian: Have you tried?
luqui: That would work.
wolverian Juerd: no. I don't know anything about the locales system, beyond dpkg-reconfigure locales.
luqui needs to think of a more convoluted example
Juerd wolverian: locales wasn't installedh 18:34
SamB huh, this is odd
where are we pasting things?
integral the only thing you lose over finite sets is simple ways to iterate over them
theorbtwo Ho, autrijus, alpha and omega of lambda!
integral perlbot paste
perlbot Paste your code and #perl will be able to view it: sial.org/pbot/perl dragon.cbi.tamucc.edu:8080 erxz.com/pb
integral err, sial.org/pbot/perl6, SamB
Odin- Next step ... making Perl6 into one of those computer algebra thingamajigs? :p
luqui Odin-, clearly
:-)
theorbtwo It isn't already?
Seriously, ^?..^? seems like the better WTDI. 18:35
luqui Hi autrijus. How do I get ^.. to parse? (Right now it thinks it's infix:<^> then ..)
theorbtwo Er, being noncore seems like the better wtdi.
autrijus Ī±^..^Ī©
Odin- theorbtwo: Nah. I don't think it does integration of mathematical functions ... yet.
Juerd wolverian: Generating locales. This can take a few hours.
Odin- (And other such fun thingies.) 18:36
gaal ./pugs -e 'my $h = Pugs::Internals::openFile("AUTHORS", "r"); say =$h'
*** No compatible subroutine found: "&Pugs::Internals::openFile"
autrijus luqui: ah. that's because ^ is tighter
gaal why?
autrijus luqui: so the usual longest-string rule fails at that level
luqui how to fix it?
autrijus luqui: if you have tuit you can redo the lexical scanner to consider all strings
wolverian Juerd: thankyou.
Juerd longest-string should come before precedence
autrijus luqui: but a cheap fix is in Lexer.hs line 122
luqui agreed
luqui goes for the cheap fix 18:37
theorbtwo Odin-: If we have full reflection, defining inegration and derivitive-finding (how do you verb that, anyway) shouldn't be too hard.
autrijus luqui: add ^. to aheadSym and you're done
luqui has forgotten enough Haskell not to want to do a lexical scanner quite yet
integral derivation
autrijus luqui: that's fine... usually the prec is sane enough
but there are a handful of exceptions, ^.. being one
luqui: hey. I'd like some way to force GC run. 18:38
theorbtwo Isn't that the what you get when you derive, not when you take the derivitive?
vcv 1>..<5
autrijus what would you suggest it in syntax level?
theorbtwo Make it a method on a singleton pugs object?
autrijus Internals::perform_gc(); # ?
pasteling "SamB" at 64.0.112.225 pasted "wierdness" (20 lines, 306B) at sial.org/pbot/10492
theorbtwo smacks himself.
wolverian Internals::GC::collect?
autrijus gaal: let me check..
theorbtwo Bad orb. Multipragmatic.
luqui autrijus, in parrot? 18:39
autrijus luqui: no, in p6 level
luqui: one way or another we need a language level thing to force GC run
because there's no timely destruction anymore
luqui how come?
oh
We might just make it sweep
autrijus and sometimes you want to run finalisers
sweep() ?
wolverian there's timely destruction when you tag an object with a flag that says it needs that
autrijus that sounds like bad huffmanise
integral and namespace polluting... 18:40
autrijus right
luqui I would put it in Parrot::sweep, but maybe that's tying it down to the implementation too much
autrijus wolverian: but you cannot tag them recursively
wolverian autrijus: hmm. okay.
theorbtwo Internals::do_gc sounds good to me.
("Perform" when "do" will do seems like overkill.) 18:41
SamB autrijus: any idea why patches I darcs send to perl6-compiler might be taking a long time to get back to my inbox?
autrijus SamB: no idea 18:42
SamB: also, parrot codegen currently doesn't handle the various Syn constructs properly 18:44
SamB oh, that would explain that. 18:45
autrijus SamB: that will be fixed next week when I'm in Vienna
where I can work with two parrot hackers
Shillo Hullo, all! 18:46
autrijus greetings Shillo-san
Juerd Hello
SamB any chance of a human readable but not incorrect output format?
Shillo autrijus: Konbanwa! 18:47
autrijus SamB: not sure what you're after
SamB well, when I do: 18:48
./pugs -CPugs -e 'sub foo() { 1 }' -e 'my $a ::= foo();'
autrijus SamB: if you want a AST, dump, use interactive pugs 18:49
SamB I get a line that is, well, huge.
autrijus it has the :d and :D mode
18:49 knotty_ is now known as knotty
SamB its over 7000 characters wide 18:49
autrijus SamB: if you'd like tp prettyprint the -CPugs output, feel free to hack Pugs.Compile.Pugs
it's just Text.PrettyPrint 18:50
svnbot6 r3991, luqui++ | Fix ^.. and ^..^ .
autrijus luqui++
Juerd wolverian: locale generation done
autrijus SamB: can I make you a committer, and are you willing to use svn to commit?
SamB autrijus: you can 18:52
autrijus gaal: space after comma
SamB: ok. email?
SamB but I don't have svk installed yet 18:53
gaal huh? where?
autrijus gaal: Str,Str won't split
it's all isAlpha
integral you just need svn, SamB
autrijus I mean not isSpace
SamB dialup is slooooow
autrijus gaal: in the Prim prototype
gaal ahhh, i see it now. thanks!
autrijus np :) 18:54
gaal hmm, shall i hack it so that ',' is enough?
autrijus gaal: I don't think it's worth it ... that table wants to be shifted to parse prototypes in Prelude.pm 18:55
gaal: but if you really want to do so, feel free
SamB naesten at gmail dot com
gaal anyway, the next problem is that when i modified code (eg. builtins/io/io.t to use this openfile, it was *way slow* on an unoptimized pugs.
ninereasons iblech, and autrijus: t/builtins/map_constant_list.t is an eval test that fails. It does not fail if uneval'ed 18:56
gaal rather: to use the Prelude openfile which calls openFile
luqui eval_is looks broken
autrijus luqui: hey. how can I unbreak eval_is from a lang pov?
luqui: that is, how can I enter the lexical context of my caller? 18:57
%OUTER:: := %CALLER::;
probably doesn't wor
k
but what does?
luqui Your &is.goto looked good
iblech ninereasons: I think this particular bug has the same reason as the bug causing many unexpected suceedings lately
luqui but it's not an ordinary method
autrijus luqui: it doesn't fix the eval problem
ninereasons ok, iblech.
luqui what's the eval problem?
oh 18:58
autrijus luqui: say you have my_eval''
luqui I get it
autrijus good.
ninereasons iblech, where does that corner-case test belong? It doesn't seem to merit its own file (imho)
luqui Ruby has a WTDI, doesn't it?
autrijus luqui: I don't know about about ruby to answer this
luqui I think an option to eval might be in order
Give it a ref to a lexical pad or some such 18:59
autrijus mmmm.
integral like scheme's eval?
luqui sure (/me knows not scheme)
iblech ninereasons: map.t probably... (It was originally a pugsbugs test, which just got moved because it passed)
ninereasons iblech, yes, now that it succeeds. Ok, I'll move it then.
luqui A related question: does %CALLER:: give the immediate lexical variables, or the entire pad stack, of the caller 19:00
svnbot6 r3992, ninereasons++ | builtins/map_constant_list.t: hmm. failing test succeeds if unevaled
luqui because if it's the latter, you might do: eval 'foo', :pad(%CALLER::)
autrijus luqui: I think only immediate. 19:01
you need %CALLER::OUTER::
integral does %CALLER:: also hold details about what pragmas are in effect?
autrijus to get to outer of caller
luqui right
well, maybe
autrijus but you can argue that %CALLER:: contains a %OUTER:: entry.
much like how perl5 globs are handled
in any case it sounds like either a p6l or a cabal question :)
luqui Yeah. And you'd expect: my $x = 4; { foo() }
to give foo() 4 for %CALLER<$x> 19:02
autrijus will I? hm.
luqui As far as access, making it "deep" seems right. As far as keys, though..
Well what if they're saying $CALLER::_ and the immediate block isn't a topicalizer?
autrijus I don't know. :) currently in pugs 19:03
all pads are cumulative
so your %MY:: pad inherits all nonshadowed %OUTER:: syms
luqui I think that's correct
autrijus I'm doing it only because it seems sane
but it's unspecced as such
luqui Well, this reflection-like stuff doesn't need as much speccing as the rest of the language 19:04
Juerd Miscellaneous advice of the day: do NOT spray silicone spray in the direction of a fan.
luqui we're in the "possible" realm, out of the "easy" realm
autrijus sure, that's why I felt comfortable of implementing it without asking :)
Juerd Question of the day: how the heck can one remove silicone spray from a mirror?
autrijus ponders &call_with_current_continuation();
Juerd autrijus: $?CONTINUATION 19:05
autrijus Juerd: not exactly the same thing
Juerd Oh
autrijus and we have &?CALLER_CONTINUATION as escape-only thing too
luqui apparently $?CONTINUATION is a bit scary to work with
Juerd Can't it be made the same thing?
autrijus that should maybe called &?RETURN_CONTINUATION
Juerd: callcc is more generatl than $?CC.
integral are shift/reset exposed too?
autrijus integral: they are exposed as "sub {}" and "return" :) 19:06
integral ah.
autrijus er, no, "return" is hardly shift.
it's shift with a const ret.
but I don't know how to expose a full shift.
that sounds very dangerous :)
luqui I think your lexical stacking is good, and we should go with %MY{'%OUTER::'}
autrijus tho if we have that we can do eval is caller pad 19:07
sub my_eval ($str) { &Internals::shiftT({ eval $str }) }
will easily DWIM.
except we probably don't get to call it shiftT.
luqui what are you talking about? 19:08
(shiftT?)
autrijus luqui: so you're familiar with delimited continuations
luqui delimited?
autrijus in the form of a function call (enter dynamic scope)
luqui oh.. I think so
autrijus and a return() call (escape dynamic scope)
luqui okay
autrijus now return() is defined as 19:09
\x -> shiftT $ const $ return x
that is, "evaluate in outer dynamic scope"
"take the continuation as argument"
"ignore the continuation and just yield x" 19:10
luqui hmmm
autrijus so shiftT is like invoking a continuation
but it's delimited
svnbot6 r3993, ninereasons++ | move a map test from bug file to main file
autrijus so you can't escape all the way to program exit
luqui weird
autrijus you can just act in the caller's scope
luqui that's pretty cool
autrijus pretending you have already returned
and do things.
sure. shift/reset can be used to model all control flow structurs 19:11
as Ken Shan showed a while ago
so if you expose shiftT, then it can do strange things on the caller's continuation
such as suspending it and reactivate it as a callback -- instant coroutines 19:12
as well as other more interesting things.
luqui maybe this should be in Perl::Control::shift or Perl::Continuation::shift
autrijus *nod*
Control::call_cc
Control::shift
or is the Perl:: namespace the new Internals:: ?
luqui dunno
autrijus sure. we can improvise 19:13
iblech: so do you have an idea why :todo is suddenly being set?
integral hmm, well Perl is being used on CPAN at the moment
autrijus iblech: I think it might be the single-invocant fix I did a while back
luqui I think we get veto power over that module :-)
autrijus I'll try to track it down if you don't have already
mm I also have Inline::Perl 19:14
theorbtwo I'd prefer it not be named "shift", since then importing it overwrites the shift builtin.
luqui (there is a Perl:: namespace, for perl grammars and whatnot)
autrijus call_shift_cc etc
iblech autrijus: I think so, too. I did a little bit of investigating only -- is 1, 1, "..." works fine, the only things which are broken, I think, are dies_ok and friends. Not sure, though
autrijus: It seems that the test descriptions are used as todo reasons 19:18
autrijus: But I didn't investigate more
autrijus oh, there's a perl task 19:21
perl5 rather
the current Makefile.PL should be factored to enable EMBED_PERL5 for parrot and perl5 by default
if it can probe for it correctly
integral oh, btw my pugs isn't compiling with embedded perl5
autrijus if anyone wants to take a stab, I'd be grateful, otherwise I'll try to do it around release time
integral: what's the errmsg? 19:22
integral my perl link flags have a "-optl" in them that ghc doesn't grok
or, that's what I'm guessing, because it seems to be ignoring the -o and saying that "pugs" is a module to chase from
gaal autrijus, how to do that w/o a PARROT_PATH ?
autrijus gaal: look for parrot in PATH and query it... or just fail the probe and warn as we do now 19:24
luqui pianos &
autrijus integral: try quote the optl
in line 134 Makefile.PL
integral quote it with what?
autrijus try escaping it somehow and put the entire thing in double quote
or something. improvise
integral GHC has something like -Wl ? 19:25
Juerd So far, feather holds just fine with people aggressively running only bash :)
autrijus i.e. if your ccdlfags is "foo bar"
it currently is
integral oh, right
autrijus -optl foo bar
but it should be
-optl "foo bar"
but foo may contain double quote etc
so need a bit munging. commits welcome
integral I know what is is, there's nothing after my -optl
autrijus then the quoting will also fix that.
Juerd wonders if during the weekend, feather will be used more 19:26
integral ty, autrijus that fixed it 19:27
autrijus woot. commit? :) 19:28
(that will also likely fix it for some other people)
integral r3994
autrijus integral++ 19:29
cognominal_ juerd, I need to build a submission system for the JAPH constest, do you mind if use feather? 19:32
I will probably write it first as perl5 thing though
and may be I will use my fresh knowledge of mozile to create a sexy firefox-based interface. 19:33
svnbot6 r3994, bsmith++ | Quoting the argument to -optl.
autrijus the apw/fpw ja*h contest? 19:35
that sounds like definitely p6 related :)
cognominal_ yes
I am currently writing an article about ajax and mozile. So I could revamp the code to use it as gui. 19:36
autrijus I like mozile. I think it's neat.
cognominal_ but I see that one can access DBI from Perl6 know 19:37
SamB how hard would it be to make it possible to build pugs without actually statically linking more than a few modules?
integral SamB: pugs can run under ghci
cognominal_ do we have a database system on feather?
autrijus SamB: "make ghci" 19:38
svnbot6 r3995, ninereasons++ | pairs.t: add a pod section to an interesting test, and move some comments around to make them stand out.
SamB I also want to avoid invoking ld on everything, because it eats so much ram...
autrijus SamB: adding more ram will likely to help also 19:39
SamB but I do want to be able to run something like ./pugs_dynamic or ./pugs_ghci
cognominal_ I have a page about perl6 in linuxmag that has been published in linuxmag French, I submitted it four weeks ago. It seems so much out of date now.
SamB I don't want to buy another 128 MiB...
autrijus SamB: "make ghci" then type "main"
Juerd cognominal_: No, it has to somehow be good for perl 6 19:40
autrijus cognominal_: ooh. may I look at it? :)
SamB autrijus: what about command-line-args
cognominal_ I said no rules and no object yet :)
autrijus SamB: withArgs
Juerd cognominal_: And you're probably best off using sqlite
autrijus heh. so out of date. we have rules and objects and cpan
Juerd cognominal_: So the system is very portable
cognominal_ autrijus: I need to hook to find where is my scanner to hook it to my computer
autrijus cognominal_: you mean you did not write it in a text editor? :) 19:41
SamB what is withArgs?
autrijus withArgs :: [String] -> IO a -> IO a
cognominal_ autrijus: I thought you wanted the real published thing
SamB where is withArgs? 19:42
autrijus SamB: it's available in the ghci prompt
SamB: so you can do something like
ghci> withArgs ["foo.p6"] main
to get what you would get for "./pugs foo.p6"
"make ghci" will launch you to that prompt.
SamB oh, sorry, I had Main loaded without *
autrijus need the *
integral bother that patch doesn't fix it. I get an exciting "gcc: : No such file or directory" at the link phase 19:47
guess I need to use an if statement 19:48
autrijus nod
cognominal_ autrijus: stefp.dyndns.org/news6-0605.pod 19:49
autrijus merci
cognominal_ he, you use freebsd? 19:50
"Mozilla/5.0 (X11; U; FreeBSD i386; zh-TW; rv:1.7.6 :) 19:51
integral hmm, I guess it shouldn't be trying to link both src/perl5/perl5.o and src/Pugs/Run/Perl5_stub.o
SamB autrijus: are you going to apply that Pugs.Types export list patch?
autrijus SamB: hm, on p6c? had not seen it 19:52
cognominal_: yes
SamB: you are a committer now anyway... you can use svn/svk to commit that yourself :)
in generaly yo udon't need reviews before committing, as long as it doesn't break the build in a horrible way 19:53
(but your patch may be reverted too -- in that case discussion on #perl6 is useful)
cognominal_ I can't believe it, googlebot has already scanned my page!
do they spy our channel?
autrijus the channel is logged on web 19:54
SamB would I have to check out the repository again to use it with svk if I check it out now with svn?
integral well the logs are online
autrijus SamB: yes. but feel free to just check in with svn
integral mutters about his build
it now can't find _pugs_Apply, _pugs_MkSvRef and _pugs_PvToVal 19:55
autrijus integral: with or without embperl? 19:56
integral with
it may just be my build dir being broken...
yep, it was. committing fix 19:57
kolibrie is it currently possible to get humanly-readable time() 19:58
svnbot6 r3996, bsmith++ | Fixed -optl again. (if statement to only add if needed)
autrijus kolibrie: you mean like scalar localtime?
not sure
kolibrie or any type of localtime
array is fine 19:59
autrijus no. you can hack in localtime with calendarTimeToString
Prim.hs would be the place to do so
or, at least write a test
luqui localtime will probably be done a bit different this time
kolibrie are there specs on localtime, so I can write a correct test? 20:00
SamB autrijus: if only it were "just". it is not "just" on dialup...
luqui look at s29
haven't looked there in some time
chromatic suspects that Pugs doesn't currently call new() on objects.
kolibrie luqui: looking
luqui chromatic, is it supposed to?
chromatic When I write a constructor, I intend for *something* to use it! 20:01
luqui you mean BUILD
wolverian BUILD is the constructor.
wolverian feels like a bot
luqui haha
chromatic I want a singleton.
luqui wolverian, seen nothingmuch? :-p 20:02
jabbot luqui: nothingmuch was seen 4 hours 13 minutes 43 seconds ago
chromatic I don't want to allocate an object I'm never going to use.
Corion Shouldn't localtime() return a DateTime object?
luqui something like that
Corion (not that I like everything about DateTime)
integral ah, bug, use Data::Dumper--perl5; Data::Dumper.Dump("hello") throws a perl5 exception which exits the pugs REPL 20:03
autrijus chromatic: you mean you define new for your own class?
chromatic: which test in t/oo/ is that?
chromatic ext/Test-Builder/010*.t 20:04
ext/Test-Builder/020*.t 20:05
autrijus danke
kolibrie localtime() is listed under Time::Local in S29draft.pod, but no particulars
autrijus I'm fixing the misTODO bug first
turns out .goto is broken
chromatic Test::Builder creates and returns a singleton, while Test::Builder::Test is a facade factory.
luqui dunnos the facade pattern 20:06
kolibrie so does anyone know of a thread about localtime() ?
chromatic Just consider it a factory then. 20:07
When you call Test::Builder::Test.new() it returns a different kind of object based on the arguments.
luqui ahh 20:08
Juerd Hi, chromatic 20:10
autrijus: Does pugs have something like Data::Swap? 20:12
chromatic Hi, Juerd.
Juerd autrijus: That is: something that entirely swaps containers' contents
Keeping containers themselves, and references/names to them, untouched 20:13
autrijus Juerd: why do you want that?
and why can't you use use ($a, $b) := ($b, $a) ?
Juerd autrijus: To change the invocant. 20:14
autrijus: So that afterwards, it's an entirely different object
Corion Oooo - I get parsefails in the tests :)
Juerd So the REAL object can be stored in a cache for re-use
autrijus yeah. again, why not := ?
Juerd autrijus: because that does things with names only
autrijus Corion: all over the place?
Juerd It simply swaps the names, not the contents of the containers 20:15
Corion autrijus: No, just a few tests: t/operators/short_circuit.t, t/operators/assign.t
autrijus Juerd: how is it operationally different.
Corion: ok. may be related to luqui's change of ^..
or not. not sure
svnbot6 r3997, autrijus++ | * move GC run into end of runEvalIO. 20:16
r3997, autrijus++ | * fix .assuming and .goto.
Corion I think Data::Swap was written to (re)implement $a := $b :)
autrijus the TODO bug should be fixed
integral no, Data::Swap is different from $a := $b
Juerd autrijus: $foo.bar(), and I want bar() to change what $foo represents entirely. Another object, with another class, but still that $foo, and everything else pointing to that container.
Corion: := swap swaps names 20:17
Corion: = swap swaps primitive values
Corion: What I want is containers to be swapped without damaging the containers
This includes the value, but also any properties
SamB so basically you want become:
autrijus Juerd: I think pugs's := does that. and I can't still see a difference. 20:18
integral it's like updating in the STG-machine (and I use it to implement that in perl5)
Juerd If you talk about harbor containers, it would be swapping the contents entirely, and painting them both in the other's colour.
autrijus or rather, i thought that was what := is about.
Juerd autrijus: If it does that, it does the wrong thing. := only aliases/binds. I actually want something destructive.
autrijus := in pugs does _not_ switch names.
Juerd Oh
luqui if you switch names and containers, don't you end up doing nothing? 20:19
autrijus it repoints references into containers
Juerd autrijus: So ($self, $new_object) := ($new_object, $self) does what I want?
autrijus Juerd: I really think so, yes.
Juerd autrijus: That's not Perl 6 then...
autrijus Juerd: if you uncomment line 481
chromatic If you had a reference to $self, it'd be a reference to $new_object after that?
autrijus then it becomes name binding. 20:20
Juerd chromatic: Yes.
autrijus Juerd: but larry ruled explicitly against that
Juerd autrijus: It is meant for name binding
autrijus: Huh?
o...kay.
That's weird.
autrijus $x := @y[1];
SamB man, Pugs.AST.Internels even takes a long time to compile in GHCi
Juerd That'd have to *copy* then.
Either it makes $x point to @y[1]'s container 20:21
Or it copies its value...
integral ./pugs -e 'my $a = [123]; my $b = [456]; my $c = $a; say "$a[0] $b[0] $c[0]"; ($b, $a) := ($a, $b); say "$a[0] $b[0] $c[0]";' # $c doesn't change
Juerd But you say it copies the value with all the container's properties?
autrijus pugs> my (@y, $x); $x := @y[1]; $x = 3; @y.perl
'[undef, 3]'
Corion Hmm - either we get lots of unexpected successes, or :todo is broken: datenzoo.de/pugs/win2k.html 20:22
chromatic autrijus, isn't that binding an alternate name to an existing container?
integral perl -le 'use Data::Alias qw/swap/; my $a = [123]; my $b = [456]; my $c = $a; print "$$a[0] $$b[0] $$c[0]"; swap($a, $b); print "$$a[0] $$b[0] $$c[0]";' # $c changes
Juerd autrijus: Yes, that's binding, isn't it?
binding a name to a container.
integral Hence := currently doesn't do the same as Data::Alias::swap.
autrijus integral: how would Data::Swap do it?
I'd like to see code where I can tell the difference
integral there you go.
chromatic My TODO problems went away with 3997. 20:23
Juerd autrijus: Data::Swap uses a perl 5 specific trick, copying the raw scalar into the other's memory space.
autrijus: Such code I can't currently think of, except how DBIx::Simple uses it
chromatic I'd also like to see code where that's the best way to do something.
It feels somewhat face-stabby to me.
Juerd chromatic: DBIx::Simple
luqui I think pugs's current semantics are correct 20:24
Juerd It replaces the existing object with a DeadObject, so that next uses give you very useful debugging info
autrijus luqui: me too :)
integral I use it for lazy evaluation
autrijus Juerd: I see what you mean.
you want destructive type-violating pointer assignment.
Juerd Without needing checks for that in every single method, which would make the entire module much less efficient
autrijus: If I knew what that meant, I'd probably acknowledge :)
autrijus Juerd: i.e. you can use it to violate type constraints at runtime 20:25
chromatic It doesn't necessarily have to be type-violating. You could use a null object with a similar interface.
pasteling "integral" at 131.111.8.102 pasted "swap sample" (12 lines, 316B) at sial.org/pbot/10502
Juerd I have little notion of types, autrijus
integral in that one I have an interface where objects has a "to_ops" method
Juerd Damn, I have to go 20:26
I'll try and explain it better later
afk
autrijus Juerd: ok...
"This module allows you to swap the contents of two referenced variables, even if they have different types."
that's what I get from Data::Swap
swap() has a type of
a -> b -> Eval ()
integral the type bit isn't really critical, it's just nice. Since you can implement objects with either arrays or hashes
autrijus but a will become b
and b will become a
and that's patently weird :)
integral it's perl ;-) 20:27
autrijus I don't think perl let you do that.
perl is a strongly typed language.
integral well it does :-)
But it's *very* not good to use it on named arrays/hashes.
autrijus heh. Data::Swap::PP
how do you write that?
integral It's safe enough on anonymous ones
autrijus Data::Swap::PurePerl that is
if you can't, perl doesn't let you that.
integral hmm, tieing and indirecting?
autrijus then it's swapping Dynamic variables 20:28
which is fine
Dynamic -> Dynamic -> Eval ()
chromatic Perl doesn't let you do that unless you always do named lookup.
autrijus right.
chromatic Or if you have another similar type of indirection.
luqui though you can rebless 20:29
autrijus anyway, in parrot level you can get by using pointer assignment on PMCs
luqui: you can't rebless across types
chromatic Yeah, I'd rebless into a null object.
autrijus (here types means array/hash/code)
integral "null object"?
integral doesn't see what null has to do with swapping...
autrijus and in haskell level you can use unsafeCoerce
luqui i mean in perl 5, and no, you can't change a hash-based object to an array-based one
chromatic The ultimate in passive-aggressive.
autrijus but neither is supported in pugs -- I'm not even sure how to support that in perl 6 syntax level.
so a definite "not core" thing I'd argue. 20:30
luqui but also keep in mind that perl 6 lets you use references as if they weren't references
chromatic integral, it's what I might do if I were to do what Juerd needs to do in DBIx::Simple, based on me skimming the code.
autrijus luqui: that's syntax sugar.
luqui: esp now larry ruled deref is only one level and not infinite
integral chromatic: oh, ok.
luqui he did?
where?
integral swap's a bit more general though (see my code paste)
autrijus luqui: yes and it caused me great grief
luqui: www.mail-archive.com/perl6-language...19462.html 20:31
luqui: *sigh*
it's not yet in pugs
because it will mean yet another total rewrite of Eval/Monad logic.
chromatic Don't fret. There's more grief to come.
autrijus (the part that deals with casting, anyway) 20:32
chromatic: oh?
luqui hmm
I hated that at first, but then grew to see its merit
and now it's gone again
chromatic Unless you already believe that signature types check roles first, then inheritance.
luqui Hmm...............
autrijus chromatic: I believe that it checks .does.
and .does checks roles first, no?
luqui I wonder if we could just have two reference types
chromatic It ought to. Inherits is a more specific case of does. 20:33
autrijus chromatic: so no grief for this
luqui one that infinitely derefs and one that single derefs
autrijus luqui: ...principle of most surprise, eh?
chromatic Darn. How about grief from people who want to override new()?
luqui except that the one that infinitely derefs would only single deref, but transparently
autrijus chromatic: I'll fix. bbiab
chromatic How about lazy derefs?
luqui well not really
there are two kinds of references: concrete and abstract 20:34
abstract references just provide a transparent level of indirection when you really need it
concrete ones work like perl5ers expect "references" to
chromatic autrijus, is it something like checking if there's a &new in op1 "new" and calling that and potentially walking up SUPER if the method calls that and using the old behavior if not?
luqui like pointers
chromatic luqui, which one do you get when you take a reference? 20:35
luqui dunno
chromatic I like the idea except for that part.
luqui yeah. well we could deem one inferior, and give it a longer name 20:36
like makeref or somesuch
chromatic Maybe that's an argument for a bit of syntax that indicates how far you want to dereference a reference.
1, many
luqui but I think the decision is really up to the person making the reference, not dereferencing it
especially in the transparent case, you would use those for what juerd is wanting 20:37
chromatic I still think what Juerd wants is pretty face-stabby.
luqui give somebody a reference to an object when they're expecting a real object, so that you can change it underneath them
I would think so too, but I've done it, and it has been good to me
chromatic Yeah, now I think of proxies and pools. 20:38
luqui (in the ODE module, when you destroy something explicitly, it replaces it with a ODE::Dead object)
chromatic Where's Parrot ODE?
luqui Perl5 ODE
chromatic Yes. Where's Parrot ODE?
luqui huh?
oh 20:39
chromatic I want a port, darnit.
integral (lazy proxy objects)++
luqui man, ODE's interface is probably easy enough to do with nci alone
but I object-orientated it
chromatic Opaque pointers?
luqui well, we can pretend they're not opaque 20:40
if ODE changes the interface, we'll just have to change with them
chromatic I like opaque pointers. Don't make me define structs.
luqui er 20:41
I'm not sure what you're talking about anymore
chromatic Instead of making the user allocate data structures, you call a function in the library which returns the data structure for you and you use functions to get and set its members.
luqui oh, right 20:42
chromatic Sure, it's a struct underneath, but you don't have to fiddle with it.
svnbot6 r3998, Aankhen++ | * small formatting changes to HTTP::Headers.
luqui I think ODE actually makes it a C++ class underneath
but it's a C library interface, so you can't rightfully touch it 20:43
chromatic That's the best thing to do with C++ libraries.
Not coincidentally, it's the best thing to do with C libraries. 20:44
luqui heh
autrijus back. 20:50
svnbot6 r3999, iblech++ | examples/network/screen-nodestatus.p6 -- Added a usage().
r4000, iblech++ | Implement nothing().
r4001, iblech++ | * Implement evalfile()
r4001, iblech++ | * Added a test for evalfile (t/builtins/evalfile.t).
r4001, iblech++ | * Test for nothing() added, too (already in the previous commit, sorry)
r4001, iblech++ | (t/builtins/nothing.t).
autrijus chromatic: so class method and object method invocation are still mixed in p6?
Foo.blah and Foo.new.blah calls the same thing? 20:51
chromatic In what sense do you mean, autrijus?
autrijus in that sense
chromatic I believe so.
autrijus oh well.
then you want to fix the dispatch code in Eval.hs.
I'll attempt a fix.
chromatic As I understand it, you can specify more specific invocants in the method signatures.
autrijus so basically 20:52
evalExpType (Val (VType typ)) = return typ
because otherwise it will look at type of Foo 20:53
and dispatch to "Class"
because that's the class of ::Foo
chromatic Right.
See? More grief. I promised.
autrijus pugs> class Foo { method bar { 3 } }; Foo.bar 20:54
3
happy?
chromatic Yep, that fixes what I need. 20:55
Does it break new() though in the absence of a defined new()?
autrijus no. it will fallback to the builtin Class::new
chromatic Beautiful.
autrijus praises the maintainable hackability of the pugs codebase 20:56
chromatic Did you replace evalExpType (Val val) = evalValType val?
autrijus no, it turns out it's simpler
I can intercept it at type variable level.
basically assuming, for type dispatch purpose, that ::Foo is Foo 20:57
chromatic Yeah, I can see that.
autrijus not Class
chromatic Which file is that?
autrijus Eval.hs
evalExpType (Var (':':typ)) = return $ mkType typ
I added one line
that's all
it's not neccessarily correct 20:58
because that rule only applies on the invocant position
I'll refix
chromatic Ahh, very clever. 20:59
The charlist/string thing still confuses me sometimes.
wolverian I've missed walking outside a lot. slightly rainy is the best weather ever.
ninereasons iblech++, I tried the livecd yesterday. There's potential for something very interesting, there. But ...
wolverian (I also enjoy saying OT things in channels when I'm bored.)
ninereasons anyone who's tried it, when I tried to build my own, it complains that there is "kernel panic: not syncing, no init found" . What did I do wrong?
autrijus chromatic: properly fixed. r4003 21:00
svnbot6 r4002, autrijus++ | * Class methods like `Foo.bar` now dispatches to Foo::bar
r4002, autrijus++ | instead of to Class::bar. Requested by chromatic.
r4003, autrijus++ | * properly fix for class objects in invocant position
r4003, autrijus++ | to assume a type of itself; otherwise it should still
r4003, autrijus++ | be of Class type when used in MMD -- consider `foo(::Int, ::Int)`
r4003, autrijus++ | when foo is a normal sub.
autrijus chromatic: I believe that's correct.
i.e. `multi sub foo (Class, Class) ` will be invoked
not `multi sub foo (Int, Int)` 21:01
when you do `foo(::Int, ::Int);`
dbarth hi all
autrijus hi dbarth
chromatic: sounds sensible to you?
chromatic Yes, it does.
eric256 how do i tell vim that .p6 files should have perl highlighting? 21:02
Corion r3997 - datenzoo.de/pugs/win2k.html - 5620 ok, 273 failed, 738 todo, 314 skipped
chromatic I'm not 100% confident about the MMD, but it sounds sensible.
dbarth it seems I have the same pb as integral, trying to link with PUGS_EMBED="perl5" : my suse ld breaks with -Wl,-E : unknown option
autrijus cool.
dbarth: does the newest trunk fix it?
dbarth: I think theorbtwo did some Makefile.PL fix
dbarth i've done a svk pull a few minutes ago, but not sure about the 'trunk' :
autrijus ooh t\builtins\system\01-strings-with-spaces.t parsefail
dbarth: ah. trunk is merely newest version of svk repo 21:03
dbarth: did you rerun makefile.pl?
dbarth yes
what is the repo URL for trunk ?
autrijus Corion: wait, what is 01-strings-with-spaces.t ?
Corion: is it another of your local invention?
Corion autrijus: My (preliminary) test for system(@LIST)
autrijus dbarth: svn.openfoundry.org/pugs/
ah k.
Corion not yet working, thus not yet committed :)
autrijus so 273 failures isn't what it seems 21:04
Corion Ah - errr, true ...
autrijus since 228 of it is that test
we're more like 45 failures :)
chromatic Excellent, it's calling my new() now.
autrijus <- was briefly worried
chromatic: cool!
chromatic Failing on SUPER, but that's one step closer!
kolibrie I give up for now - I don't see any localtime() specs on any of the perl6 lists
autrijus kolibrie: that means you get to write the spec
kolibrie :)
autrijus kolibrie: and post to p6l and maybe patch s29 locally 21:05
kolibrie I'll think about it over the weekend
autrijus chromatic: so. what does SUPER mean?
chromatic: does it mean NEXT now?
or is there NEXT support separately? or not at all?
dbarth autrijus: I confirm I'm using trunk (did a 'svk cp svn.openfoundry.org/pugs pugs')
autrijus dbarth: ok... give me the first line when you type make 21:06
the ghc invocation line
chromatic SUPER and NEXT are separate.
Aankhen`` Does the "/x by default" behaviour of Perl 6 rules extend to comments within rules?
autrijus Aankhen``: I believe so. but I'm
not very sure
Aankhen`` OK. 21:07
autrijus chromatic: why not adopt ruby's super?
chromatic: it strikes me as very sensible
chromatic MMD.
autrijus point.
sigh.
dbarth I'm doing a make clean; + reconfigure + make cycle, just to be sure
chromatic You may want to call all or some of the several candidate methods.
autrijus yeah already grokked.
ok. let's implement SUPER. 21:08
what happens when you have no superclass?
i.e. you're Any
is the SUPER of Any Any?
chromatic I say it just stops.
autrijus as in not calling anything
chromatic You're at the top floor. That elevator's staying put. 21:09
Yeah, if there's no parent, there's nothing else to call.
Ahh, now I understand (':':typ). 21:10
It just needs to have at least one leading colon.
autrijus yup. 21:11
svnbot6 r4004, eric256++ | Fixed some spacing issues, removed some debuging output.
r4005, Aankhen++ | * HTTP::Headers -- fixed a few errors in porting.
r4005, Aankhen++ | * modified bits to use OO on builtins.
autrijus since I hear you can use :Int.
luqui Is using those cascaded :s really the best way to strip a prefix? 21:12
autrijus luqui: no.
luqui: you can use break and span etc
or just isPrefixOf
luqui I'm just surprised that foo ("bar" ++ x) doesn't work
on the lhs
autrijus you can do
foo arg | "bar" `isPrefixOf` arg, x <- drop 3 arg = ... 21:13
allowing ++ on lhs will amount to prolog.
so maybe you can write it in Curry, but not Haskell
luqui ... hmmm, I guess that's true
bar (x ++ y) = ... 21:14
autrijus right.
how very confusing.
chromatic No reversable guard clauses!
mauke but they could allow it with constant strings
dbarth is hs-plugins required to get an embeded perl5 ?
autrijus dbarth: nope
mauke: (x ++ "bar" ++ y) ?
mauke succ' (n-1) = n works too
autrijus mauke: it's too inconsistent.
and n+k patterns are considered by many as mistakes :) 21:15
iblech ninereasons: Hmm, I've just tried it -- purged livecd/ and svn up'd again, then ./mklivecd.pl, and finally qemu -cdrom cd.iso -- works as it should...
luqui sqrt (n*n) = n
hmm.. this is getting, er, tricky for a compiler
autrijus luqui: Curry and Oz can all do that
luqui: maybe Perl 6 should too ;) 21:16
(but not by default)
mauke I wanted to do something like that in perl5
autrijus mauke: AI::Prolog on cpan
mauke except without all the features
just stuff like sqrt($n) = 3;
luqui there's no ebuild for curry
eric256 should defined() work on $_ if not specificed so that given $x { when defined {} default {} } would work? 21:17
luqui yeah, I wrote Logic.pm which does that sort of stuff too
but the syntax frontend didn't end up being very pretty
chromatic &
luqui I'm going to be writing a logic module for parrot, because I need it for a game I'm writing
chromatic luqui, talk to Ovid about that 21:18
luqui ovid?
chromatic Curtis Poe
Limbic_Region curtis poe
Corion "Just" implement a WAM in Parrot
autrijus chromatic: what happens when you have multiple parent types? 21:19
luqui WAM?
autrijus chromatic: does SUPER:: check each of them in turn?
Limbic_Region luqui - see www.fcnp.com/448/restaurant.htm
err - not that
autrijus chromatic: does SUPER::SUPER:: check recursively two level?
Limbic_Region oops
luqui SUPER probably ought to die then
Corion Warren Abstract Machine - the usual way of implementing Prolog
Limbic_Region luqui - see use.perl.org/~Ovid/journal
chromatic No, I don't think SUPER::SUPER:: checks recursively.
autrijus luqui: p5's SUPER certainly does not die :)
Corion Pretty fast from what I know, at least for the commercial Prologs 21:20
Ovids WAM is fast. For Prolog in the 90s. :)
chromatic dev.perl.org/perl6/doc/design/apo/A...tml#___top
luqui oh the warren abstract machine or something 21:21
autrijus, but maybe perl6's should, so that you have to be more specific 21:22
perl 6 gives you a lot of ways to walk superclasses, and SUPER doesn't make sense for more than one
luqui is behind in the irc reading
chromatic SUPER is just syntactic sugar for $self.WALK[:super]::method( @args ) anyway. 21:23
Aankhen`` We broke 4,000 commits? Yay! :-D
Par-tay!
luqui wants forward-chaining logic too, so he might have to settle for a simpler algorithm
chromatic "Call method with $self as the invocant on the first parent found that has it defined."
luqui will research WAM (he read the beginning of it once)
chromatic really &
eric256 with pugs can i tell what type of object something is? ... like $x.isa("Whatever"); ? 21:24
Aankhen`` eric256 >> $x ~~ Whatever
eric256 i tried $x ~~ Whatever. and the above. and neither work
Aankhen`` Like $x ~~ Array
Corion luqui: I'm told that you need to read it two or three times, because while it implements everything as needed the explanation of why is lacking.
Aankhen`` Well, is Whatever a class that's being `use`d? 21:25
eric256 yes
Corion falls asleep
eric256 :q
opps.
Aankhen`` And you're sure the object is of that class?
eric256 yes.
Aankhen`` Hrm. 21:26
Could you try: say $object.ref
autrijus pugs> class Foo { method z { 3 } }; class Bar is Foo { method z { ./SUPER::z } } Bar.new.z
3
chromatic: oh, you're gone.
ninereasons iblech, belated thanks !
autrijus pugs> 1.isa(Int) 21:27
bool::true
svnbot6 r4006, Aankhen++ | * docs/other/porting_howto: added translation of "scalar @foo".
r4006, Aankhen++ | * translation of length("foo"): use "foo".chars instead of "foo".bytes
r4006, Aankhen++ | * added note about using m// with split().
r4006, Aankhen++ | * added section about arrays and hashes turning into references in scalar context.
r4007, autrijus++ | * SUPER:: implemented.
autrijus eric256: ahh. sorry. a new bug. fixing
Limbic_Region oh, crossed the 4000 mark 21:28
woo hoo
Aankhen`` Ah, that /x does extend to comments.
From S05: # now always introduces a comment.
luqui I think our bot should be able to execute pugs code
autrijus like rm -rf etc
luqui hmm 21:29
iblech If you are willing to host the server for it :)
luqui I guess we need a Safe.pm-like thing first
autrijus lambdabot has a @plugs for eval
but that's because you can't feed it monadic type :)
luqui wonders what it would take to keep things safe 21:30
iblech What happens if one says @plugs sum [1..]?
autrijus iblech: it timeouts
resource starvation are easier to counter 21:31
eric256 autrijus what bug did you find? 21:33
iblech Hm, if we had (for now) a simple --safe switch, which'd simple die() on things like system and open...
autrijus eric256: Foo.isa(Foo) was false 21:34
eric256 that list of things like gets long though.
autrijus iblech: we actually do
eric256 yea autrijus i guess that would count as a bug. ;)
autrijus iblech: just change runEvalIO to runEvalSTM
iblech: and all IO things will magically die.
I intended to implement atomic {} with that
but had not have tuits 21:35
Aankhen`` G'night.
autrijus iblech: so you can do a eval_safe
by copying "eval" op code
and change runEvalIO to runEvalSTM
iblech Hm, that's seems easy :)
autrijus all the debug statements etc are liftIO though so you need to watch out
but yeah, it's not hard
certainly much easier than Safe.
isn't effect typing great? :)
iblech Hm, I'll ponder this tomorrow then, probably :) 21:36
autrijus :)
iblech Gonna sleep... 21:37
autrijus nite
eric256: fixed as r4008 21:40
svnbot6 r4008, autrijus++ | * `isa` now handles class names are the invocant, so
r4008, autrijus++ | `Foo.isa(Foo)` is now always true. eric256++ for reporting this.
autrijus breakfast & 21:42
cognominal the sun never sets on perl6 realm :) 21:43
autrijus :)
parrot is now at r8185
we still have some way to go :) 21:44
luqui wonders if it is time to finally install windows 21:49
svnbot6 r4009, ninereasons++ | tidy up a little
luqui cringes
autrijus luqui: why?
luqui my composition programs are windows only 21:50
and I have an urge to play counterstrike
autrijus oh. have fun then
you may find Flyakite useful.
luqui yummy 21:51
luqui goes off to do the unspeakable
chromatic Hmmm. 21:57
autrijus chromatic: SUPER:: is in. what else you need? 21:58
chromatic I'm running 010*t right now to see. 21:59
autrijus seen on #ruby-lang:
jabbot autrijus: I havn't seen on , autrijus
autrijus 05:56 < Aredridel> Hehe. lucsky: it's matz' least surprise, not yours ;-)
chromatic Ah, there was an error in that test. 22:01
There was also an error in Test::Builder::TestPlan. 22:03
If I do $Test = Test::Builder.new(), what should $Test.ref return? 22:08
autrijus ::Test::Builder surely? 22:11
svnbot6 r4010, naesten++ | Add export list to Pugs.Types
r4011, autrijus++ | * Patch from Samuel Bronson to add export list to Types
r4011, autrijus++ | ...but he already committed it. Oh well. add Samuel to AUTHORS.
chromatic I would have thought so too, but it's returning Class. 22:12
autrijus because you told it to. 22:13
ah.
I see.
you have rebound ::Class.
clever, but not yet supported.
chromatic In the BUILD signature, you mean? 22:14
autrijus if you had said
method new ( Test::Builder $Class: *@args )
return $:singleton //= $Class.SUPER::new( @args );
then it would work.
and I think it's valid anyway
chromatic Good question on that. I stole that code almost directly from Damian.
autrijus I think $class
is better. 22:15
makes more sense to me anyway.
otherwise it's very confusing
because there's a builtin ::Class too.
chromatic Me too. Let me see if he explained his reasoning.
autrijus anyway I'm not going to be able to implement rebindable class space right now :)
chromatic No, that's a biggie.
autrijus if you insist, write a separate test under t/oo/.
chromatic I imagine you probably want separate tests for some of these anyway, including the wrong order of DESTROYALL? :) 22:16
autrijus the wrong order is already fixed.
:)
I slapped a reverse on your code.
chromatic Yes, I saw.
autrijus but sure.
oh also, a GC run is performed during each runEvalIO now
chromatic I tried to refactor out the duplication, but it's hard to reverse an Eval [Val].
autrijus so in interactive shell or in ghci it's very convenient 22:17
because DESTROYALL is always called between evals
chromatic: huh? you pass in another ([a]->[a])
and in buildall's case pass in "id"
and in destroyall's case parameterise it with "reverse" 22:18
chromatic Okay, it was hard for me.
autrijus and call that argument "f" :)
that's fine :)
chromatic I'm going to see what low-hanging fruit this picks in the T::B suite and I'll check in my modifications.
autrijus refactored done 22:19
op1 "BUILDALL" = op1WalkAll id "BUILD"
op1 "DESTROYALL" = op1WalkAll reverse "DESTROY"
chromatic That's what I had in mind for the results. 22:20
eric256 should 22:21
when $_.isa(Weapon) { } work? when $_.ref = Ref Weapon, and $_.say = <obj:Weapon>
svnbot6 r4012, autrijus++ | * refactor BUILDALL and DESTROYALL into op1WalkAll
autrijus you can use .isa(Weapon) :) 22:22
but yes
oh wait
not really
fixing 22:23
eric256 lol. i was just about to commit my wizard.p6 displaying this problem....i should right tests ;)
is there a good place in oo for these? 22:24
autrijus it's not a oo bug
it's a when bug
it was parsed as
if ($_ ~~ $_.isa(Weapon)) { ... }
eric256 that won't work 22:25
;
autrijus right. and fixed.
r4013.
eric256 thanks
autrijus np :)
eric256 should i go ahead and right a test to make sure it doesn't break in the future?
autrijus btw it's really clearer to write
when .isa(Weapon) { ... }
sure, go ahead
eric256 i tried that
autrijus yeah, now it's fixed, you can use that :)
eric256 and $_.isa and Weapon and $_ ~~ Weapon. ;)
autrijus sorry ;) 22:26
oh right. when Weapon
that should work, no?
eric256 np. i just thought i was crazy
i thought it should
autrijus fixing
oh wait 22:27
it already works, lol
svnbot6 r4013, autrijus++ | * `when .isa(Foo)` is no longer parsed as `when $_ ~~ .isa(Foo)` now.
autrijus but you need to write 22:28
when (Weapon) { ... }
currently
there's a space globbing bug somewhere.
not too worried, though. I'll really find food now :) 22:29
bbl &
eric256++ # first Perl 6 OO guinea^Wpioneer
chromatic Sure, just because his code *works* :P
autrijus :) 22:30
eric256 lol
how do i convince my feather acount that it knows where the Test module is?
autrijus set PERL6LIBS env var? 22:32
er, PERL6LIB
chromatic Hm, do roles compose in methods yet?
I *think* they do, but I'm seeing weird failures.
autrijus pugs does not support roles yet ;) 22:33
roles are just classes, and "does" is just multiple inheritance.
chromatic That explains a lot of these failures then.
autrijus I'm going to pick up MetaModel works 22:34
and inject them to haskell lnad
but perl5embed took away that batch of tuits
so if you'd like to help, audit MetaModel docs
to make sure it agrees with design
chromatic Yep, it must not be injecting the BUILD submethod.
autrijus because that's ultimately what I'm going to base on
esp. ext/Perl-MetaModel/perl6* 22:35
ext/Perl-MetaModel/docs/perl6* even.
really &
svnbot6 r4014, eric256++ | Testings some tests for given and OO 22:40
r4015, eric256++ | Second to last test is causeing Heap exhausted error!
eric256 could someone review \t\statments\given.t second to last test is dieing and i'm not sure why.
Juerd chromatic: What does "face-stabby" mean? 22:41
chromatic It means it makes me want to stab myself in the face with a fork instead. :)
Juerd And why is that?
chromatic It feels dirty to me. 22:43
Juerd It is dirty.
However, it is a useful technique that in this case adds functionality
Although I've been wondering lately if reblessing could suffice
chromatic I don't like reblessing much, but in this case it seems appropriate. 22:44
Juerd When you're working with connections, something can invalidate an object
It's then very useful to have the reason for that in the error message when the object is used 22:45
Wouldn't it be terribly useful if you knew when printing to a closed filehandle, not only that it was closed, but also WHEN it was closed, if that was because the virtual filesystem ceased to exist? 22:46
$db->disconnect makes all statement handles useless
However, they're not yet officially finished, on your side
chromatic Yeah, it's nice to store that information somewhere.
Juerd Now, if you explicitly disconnect, you probably know what's going on
But this one's subtle to newbies: my $result = DBIx::Simple.new.query('...'); 22:47
I actually implemented it in DBIx::Simple after writing $r = DBix::Simple->new(...)->query(...) myself. 22:48
It was in a rather huge script, and the $r was initialized in the beginning, and then later on used often.
But even the first failed, and before I had better error reporting, I couldn't figure out why my dbh had gone. When I found it, it was all too obvious, of course. 22:49
chromatic Yeah, the Portland Pattern Repository has a good page about the null object pattern for that reason. 22:50
Juerd brb
eric256 given $t { when .isa(TestIt) { $passed = 1;}} causes some sort of infinite loop to happen. any ideas? 22:51
chromatic I'm seeing infinite loops in SUPER calls too, but I don't know if that's related. 22:55
eric256 that would make sincen. i'm sure isa must use SUPER in some form
chromatic Revert to 4006 and see if that helps maybe. 22:56
eric256 hmmm.. $_.isa() alone works fine. when .isa() failes though
chromatic Could be something wrong in findSuperSub in Pugs.Eval. 22:57
eric256 seems wierd for it to only occur in the When block 22:58
Juerd roe 22:59
ehh
eric256 ?
Juerd s/roe/re/
eric256 error with smartmatch? ~~ ? 23:00
Juerd .method() is special cased for smartmatch
But in this case that shouldn't matter, because $_ and the LHS of the implicit ~~ are both the same $t 23:01
chromatic Hm, in a submethod BUILD, a signature member of $expect = 0 doens't work, while $:expect = 0 does. 23:07
Curious.
s/ens/esn/
SamB autrijus: so that darcs thing doesn't go both ways? 23:10
kelan the darcs repo is read-only 23:15
although 'darcs send' defaults to sending to perl6-compilers, where i'm sure someone would see the patch and maybe apply it 23:16
SamB kelan: yeah, but it would then conflict with the sent patch... 23:17
Juerd Every time I read things like 'bidirectional' and 'both ways', or the opposites, 'unidirectional', 'one way', and 'read only', I think about how much I want <->.
And that's often.
kelan what would conflict? 23:18
SamB kelan: the darcs patch resulting from the svn commit would conflict with the darcs patch that was sent 23:19
eric256 debates wether to use any([//]<<<<@x) or @x.map:{$_[0]}.any ..one is pretty, one is readable. ;)
SamB I don't think either of those is terribly pretty 23:20
but @x.map:{$_[0]}.any looks like it might be easier to decipher
kelan i meant maybe autrijus would use darcs to apply it, too
eric256 you have no flair for art! ;)
kelan not sure
Juerd @x.squish.any :) 23:22
Or whatever the method name for recursive flattening is.
eric256 except i just want the first element of each sublist. yea i know i'm picky 23:23
Juerd eric256: Then I don't understand how the first works.
As it has no index.
SamB eric256: well, I don't know how to read folds or maps or anything like that in Perl 6 yet...
Juerd Shouldn't that be @x>>.[0].any then? 23:24
Which I expect to be writeable as @x>>[0].any too
eric256 [//] will pick off the first defined value of the list. ;)
Juerd That's obfuscation.
And will hurt when you change your mind about whether the first has to be defined. 23:25
It makes the code much less generic.
eric256 and who says obfuscation isn't pretty. ;) like i said one is pretty one is readable. don't worry i went with teh more readable of the two.. actualy i made it two steps....my @choices = @x.map:{$_[0]}; until ($choice eq any(@choices) { 23:26
Juerd I don't agree about the other one being particularly readable. I think a hyperop version of that, so @x>>[0].any is much prettier and morre readable.
Why is your array called @x? 23:27
And if only the first element matters, why are you using an array, not a hash?
eric256 its not. it is called @data. which is clear enough at the time
Juerd "data"
Right.
SamB Juerd: why do we call all our lists xs in haskell?
mauke because we treat them as generic lists without looking at their content 23:28
Juerd I still need to encounter an array that doesn't hold data, data for sake of simplicity being able to be 0 elements.
If you use only the first element, it's no longer a generic list.
mauke (\@function)->("hello"); 23:29
(perl5 code)
Juerd not a CODE reference ...
SamB mauke: for a half second that looked like haskell
mauke Juerd: just bless \@function into a package with an overloaded &{} operator 23:30
eric256 hehe. you can look at the commit of wizard.p6 i just made. in short it is a prompt sub that takes an array of data, the first element can be a key or undef, it then takes that and makes the first element an integer if it is un specified. so a list of weapons can be given and it assigns them 1..5 or some of them can have designated shortcuts.... like f for flee! ;) it is definitly a work in progress and suggestions are welcome
kelan SamB: lists are called 'xs' because its the plural of 'x'
eric256 okay so that wasn't soo short.
Juerd eric256: Names.
SamB kelan: I know that!
kelan oh 23:31
Juerd eric256: My primary suggestion is that you use names. Not meaningless indices that make sense only to you, only this month.
kelan nevermind, i didnt read far enough back:)
SamB kelan: ;-)
Juerd If the first element is a key, then call it key.
eric256 read my code. you will see that one array is probably the worse named. and it is named as such because it is good enough for the few line sub it is in. 23:32
Juerd And @data for this sub should probably be called @options
eric256 better names are welcome... @options is certainly a better one...i thought of prompts at first but didn't like it.
in hindsight either are better than @data.
Juerd Everything is better than @data.
$data sometimes makes sense 23:33
eric256 lol. i noticed you have no love for @data
Juerd Of course not.
svnbot6 r4016, eric256++ | Some updates and a bug work around
Juerd It's like naming your sub "function"
function(@data)
wolverian open "file";
Juerd Now what the heck does that do, and what is the data it uses for that?
It's generally better to use foo(@bar) instead 23:34
pjcj $datum </pedant>
mauke well, it's better than data(@function);
Juerd pjcj: Then it's a given, thus input, or a date.
mauke: Yes. That's why we have sigils for this 23:35
mauke meaningless names > misleading names
Juerd Certainly.
kelan Juerd: wouldn't it have to be `any(@x>>[0])`?
sorry to back up so much
Juerd Thanks for not using >>. Although mathematically appropriate, it'd be very confusing and make me think about hypering the two sides.
kelan: Why specifically that?
As far as I know, hypers don't change precedence. I may be wrong, 23:36
s/,$/./
SamB what is >> anyway?
Juerd SamB: Hyper operator 23:37
SamB: Also known as ōæ½xBB.
SamB the singular of data is bit
not datum
Juerd SamB: No, it's datum.
kelan `@x>>[0]` is just a list of elements, isnt it? do lists have methods?
Juerd Although data is used singular.
kelan: Lists, when used with a dot operator, turn into arrayreferences 23:38
svnbot6 r4017, eric256++ | wizard.p6: renamed @data to @prompts, to satisfy Juerd++ ;)
kelan ah
SamB Data has morphed into an aggregate noun in programerly usage...
Juerd kelan: This is why you can use (1,2,3)[1] to get 2.
SamB: In English.
SamB and I can't spell
Juerd eric256: multi sub infix:<<.?.>> ($low,$high) {int( rand($high - $low) + $low ) + 1; }; 23:39
SamB Juerd: In standard english maybe.
Juerd eric256: That really needs a comment explaining why, and probably what too.
SamB if there is such a thing as standard english
wolverian isn't that the same as any($low, $high).pick? 23:40
Juerd eric256: This is efficient, but a comment in the form of [$low .. $high].pick
wolverian: No, s/,/../
wolverian er, yes.
thanks for spotting that :)
eric256 any($low..$high).pick but without building the whole list in memory
Juerd eric256: This is efficient, but a comment in the form of [$low .. $high].pick would be very useful.
wolverian Juerd: is .pick a method on arrays too?
Juerd eric256: Do you ever maintain code? :)
wolverian: Yes 23:41
wolverian oh. then what you showed is probably nicer.
Juerd wolverian: Because it'll be used much, and skipping the any means a huge efficiency gain :)
wolverian yes. :)
eric256 Juerd: only my own 23:42
Juerd Does .chomp mutate?
It shouldn.t
kelan does the dot operator infer scalar context, then?
which is what turns the list to an arrayref? 23:43
Juerd eric256: @prompts is a bad name. They're options, not prompts
eric256: The same goes for @choices, which is a name used for the same thing
eric256: But more importantly: you have two names for the same thing in different places. Use one name to improve legability... 23:44
kelan: Sort of.
kelan legibililty
Juerd kelan: It doesn't hurt to think of it that way
Eh, yes.
kelan well tell me what it really does:) i like knowing the details
Juerd kelan: It's what makes the list not a list and my previous statement rather incorrect :) 23:45
kelan: You're right.
eric256 two names for one thing?
which one thing?
Juerd eric256: @choices and @prompts
When you call prompt, you use @choices, and then in prompt, you alias it to @prompts
That's awful :)
eric256 @choices are what can be typed in. @prompts are the actual prompts include they key, they prompt, and the return object 23:46
Juerd And an option is a choice only when it's chosen </nitpick>
eric256: They are the same thing!
multi sub prompt ($prompt, @prompts is copy) {
...
}
$choice = prompt("Your choice?",@choices); 23:47
eric256 well feel free to rearrange it if you like
Juerd (Okay, not alias, but copy)
Why would a choice, when copied, become a prompt?
eric256 because it takes the choices they could chooce and gives them to the user in a prompt. ;)
the prompt takes a set of choices and returns them.. perhaps it should be @choices instead of @prompts. dunno.. you seem to care alot more than me though so feel free to fix it up, it would be interesting to see what you do with the same code to make it more readable 23:48
gtg now though. later
Juerd Oh my 23:58
svnbot6 r4018, chromatic++ | Untodo passing tests. 23:59
r4018, chromatic++ | Use $Class as invocant in Test::Builder, not ::Class.
r4018, chromatic++ | Use attribute as potential default variable container in T::B::TP::BUILD().
r4018, chromatic++ | Use SUPER() instead of WALK:: in T::B::TP.
Juerd That script is awful :|
Inconsistent style, code that never runs, meaningless variable names

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!