Feather users: perlcabal.org/~autrijus/ | pugscode.org/ <Overview Journal Logs> | pugs.kwiki.org
Set by autrijus on 26 May 2005.
luqui Darren_Duncan, you can do that in Perl 5. And yes, you can do that in perl 6. 00:00
Darren_Duncan I guess I never really tried to do that in either language 00:01
it smacks of using something before its declared
which seems to be what 'use strict' should prevent but doesn't 00:02
Jonathan_ autrijus: Any chance you can make Pugs look for Parrot in the current working directory, not just in the path? If you can, then I should be able to get this Pugs binary distribution thing working...
Darren_Duncan doesn't the path include the cwd? 00:03
luqui I use autovivi all the time
Jonathan_ Darren_Duncan: It appears not.
obra 84
Darren_Duncan odd, I saw it there before 00:04
obra mischan
Jonathan_ Darren_Duncan: Have a parrot.exe anywhere in the path, it works out fine. Drop one in the CWD instead, I get errors.
autrijus obra: yes, I can test on whiteg's laptop
Darren_Duncan maybe there are multiple paths
autrijus Jonathan_: problem is that the "current directory" is ~/src/pge/ 00:05
Darren_Duncan when I 'use' a module that doesn't exist, the error message includes the path, which ends with '.'
autrijus (For PGE)
clkao autrijus: fix av return before release! :)
autrijus Jonathan_: and it has to be that way
obra autrijus: cool
autrijus clkao: p5_apply _always_ return a av?
clkao: even when GIMME_SCALAR?
clkao autrijus: ya. when multiplevalues are returned. 00:06
should i make it scalar directly?
autrijus clkao: nah... that is fine
Jonathan_: so I don't know... place parrot.exe inside ~/src/pge
should fix it
clkao right, i thought it'd be handled up there. since it's already generated it doesn't matter
Jonathan_ autrijus: I did, and
Darren_Duncan as an example, *** Can't locate Fafdsfd.pm in @INC (@INC contains: /System/Library/Perl6/darwin-thread-multi-2level /System/Library/Perl6 /Library/Perl6/darwin-thread-multi-2level /Library/Perl6 .).
Jonathan_ pugs> /\w+/;
*** Cannot parse PGE: \w+
*** Error: Cannot find the parrot executable in PATH
undef
pugs> system('src\pge\parrot.exe');
parrot -[abcCEfgGhjprStvVwy.] [-d [FLAGS]] [-O [level]] [-o FILE] <file>
undef
Darren_Duncan the CWD is at the end of the list 00:07
autrijus oh well...
clkao autrijus: btw i'm diving into monad this week i think
autrijus clkao: cool. I think you'll like what you'll find.
clkao is there a taichi glyph ?
can we overload that when defining monad?
Darren_Duncan back in a bit
autrijus ā˜Æ 00:08
clkao orz 00:09
autrijus but monad is more like ā˜®
the identity supported by an endofunctor and natural transofmration 00:10
to form a triad
Jonathan_: hm.
clkao not dual before that? 00:11
Jonathan_ autrijus: Gotta run, I'll look at it again soon... 00:15
autrijus jonathan: I attempted a fix. let me know if it works when you backlogeth 00:17
clkao autrijus: what happens if p5 land is returning a raw hash not hashref? 00:19
autrijus clkao: how do you do that. 00:20
(without a lvalue sub)
clkao ? just return %foo
as multiple value
autrijus I thought that explodes %foo?
(into k,v,k,v)
clkao yes. so can this be somehow easily reconstructed to p6 hash?
so you can just add such wrapper around those sick function 00:21
autrijus it's not what we want.
in p6 when you say return %foo
er
in p5 when you say return %foo
jql no. it explodes into :k<v>, :k<v>
autrijus it means return *%foo in p6
jql I thought we went over this last night. :)
clkao ok
autrijus jql: we're talking about p5 man.
luqui depends on context in p6
jql ahh. p5. true
clkao alright. it's fine then. 00:22
luqui return %hash when want(Scalar); # returns \%hash
clkao next. getopt.
autrijus luqui: p5!
luqui okay
jql sees the confusion
clkao sorry for brining in perl6 here. but it's a necesary evil
autrijus: \$foo in perl6.
should became a tied evil in p5
luqui I just assumed you were discussing perl 5 for some analogy to perl 6
but I guess you guys are doing the p5 bindings for haskell?
autrijus luqui: ah. no. clkao is writing SVK in perl6 00:23
reusing existing cpan modules
clkao luqui: we are trying to make pugs looks lovely.
perl5 isn't lovely without pcan
cpan
autrijus CPAN is the language, Perl5 just the syntax
(samtregar++)
clkao bwaha
luqui CPAN is the vocabulary
clkao autrijus: so, scalarref?
autrijus clkao: a Ref in p6 when passed in p5 land is just another wrapped thing 00:24
clkao yes. but it needs to be flag
flagged
autrijus clkao: but we need to defined ${}
clkao so i can tie it
autrijus for it
luqui btw, what does svk buy me if I'm always online
autrijus via overload
clkao ${} ?
autrijus or some such.
clkao luqui: faster
autrijus luqui: no .svn/
luqui: incremental merges.
luqui doesn't understand that last one 00:25
autrijus luqui: say you have a branch and trunk
clkao you don't have to, justjump on the wagon
autrijus you can say "merge new changes from trunk to branch"
and then some 10 days later
luqui clkao, that's what I did :-)
autrijus "merge new changes from trunk to branch"
clkao autrijus: what overload? that should be just tied no?
luqui ooh
autrijus and svk will just pick up the changes in the last 10 days
can't do that with svn.
luqui you can, but you have to be explicit
autrijus that is the same to say you can't. 00:26
luqui heh ok
autrijus clkao: you can approach it with tie or with ov
clkao ya being explicit = too much invasion from the tool to real work!
i would avoid overload
it's bringing lots of problem
luqui autrijus, where should I look to implement BUILD ($.foo) ? 00:27
clkao plus if someday people are crazy enough to support io::all...
luqui we're doing that in p6...
autrijus luqui: what does that do again?
luqui it automatically initializes $.foo
autrijus clkao: alright.. so if I pass in a [] from p6 you'll TIEARRAY
and a \$ you'll TIESCALAR
etc
clkao *nod* 00:28
autrijus clkao: that's not for this release ;)
clkao wait
autrijus I'll look at handling av returns
clkao sure av.
but
pugs_MkValRef
do the same isa thing and bless them into pugs:scalar and pugs:array 00:29
i'll do the rest
maybe this can get finished tomorrow.
autrijus luqui: it automatically initialized in what way?
clkao when's the release?
autrijus clkao: in 12 hours or so
luqui new Bar(42); # $.foo becomes 42
I mean Bar.new(42), of course
autrijus luqui: oh. wow.
sick. 00:30
clkao sucks. i am busy in the morning i think. but it should be trivial right?
autrijus: can i use ref() rather than doing multipl &isa call to pugs?
autrijus clkao: sure, you can just call &ref
and stringify the rv
clkao good.
want to make a stub and i continue when i wake up?
if getopt actually works i can do real first step svk in p6 :)) 00:31
autrijus so in pugs_MkValRef
you have
sv_setref_pv(sv, "pugs", val);
and you don't want that
instead you want to call &ref
first
and analyse the rv
and bless accordingly
I don't see why it needs anything at hs side
it's all C
clkao it still has to be blessed as the entrace back to haskell 00:32
ya, it's a call. how to stringy return value from &ref
autrijus "bless accordingly"
clkao no, blesseed and then wrapped in closure.
autrijus eh.
you just clal pugsapply
then SvPV_nolen
piece of cake
clkao otherwise you can't get back
ok. good.
autrijus yeah yeah
clkao i'll do it tomorrow. 00:33
autrijus clkao++
clkao \
now zz &
make av return work!
autrijus (sigh) sure
clkao and maybe write some tests for ref accessing
(or find people to write tests :) 00:34
autrijus luqui: I'm not sure how to attack this.
luqui: I guess you can hack Bind.hs
luqui shouldn't it go right where the parameters are bound from the signature 00:35
autrijus but is it just BUILD?
method blah ($.y) {}
luqui I think it's everywhere
autrijus it will automagically become set-accessor to $.y?
instead of just lexically set $.y ?
and how does it play with the "param is ro by default"?
luqui right, because $.y isn't a valid parameter name anyway
autrijus surely you can always write to $.y?
luqui yeah, that has to be an exception 00:36
autrijus ($:y) too?
yay exceptions
luqui I think ($.y) is just short for ($dot_y) { $.y = $dot_y }
maybe with a : before the =, not sure.
autrijus except it happens at binding time
luqui oh, right
autrijus method foo ($.y, ?$z=$.y) {}
having it to be := will be terribly surprisin. 00:37
luqui I expect $:y works too
autrijus because you end up with random aliasing.
foo(@x[1])
$obj.foo(@x[1])
svnbot6 r4181, autrijus++ | * VType to p5 land becomes a String, so clkao can typecase from C.
r4182, autrijus++ | * try finding parrot in cwd, for Jonathan.
autrijus $obj.y = 4; # touches @x[1]
oops
luqui good point 00:38
autrijus so line 1167 Eval
is where you want to attack
analyse "name" for $. and $: forms
and turn it into assignment somehow.
I need to run -- $work and $food
good luck!
luqui okay thanks 00:39
meppl gute nacht 00:43
svnbot6 r4183, autrijus++ | * delay the ifdef call in Parrot.hsc
Darren_Duncan autrijus, I want to revisit that test suite thing where uppercase file names have special meanings ... 00:47
if it wasn't already done, that hack should be removed where files with uppercase chars are not executed
I want to name my tests using uc chars and have them execute 00:48
I think a better way to say 'don't run this' in the short term is to put the non-running files in a different dir name, such as 'todo_t' rather than 't'
its much more self-documenting too
luqui why do you want uppercase chars? 00:49
Darren_Duncan it reads better
luqui than lowercase with underscores?
Darren_Duncan for example, one of my desired test file names is 'LKT_00_Compile.t'
luqui oh I get it 00:50
Darren_Duncan the hack requires it to be 'lkt_00_compile.t' which is harder to read
luqui yeah I agree with you
Darren_Duncan besides, module names are mixed case
luqui maybe non-running tests should just be named .T
Darren_Duncan that's fine too
jql works like \s and \S in rules 00:51
I like
Darren_Duncan I just want the hack gone that prevents me from using uc chars
since the next release is 12 hours away, is it reasonable to get this change in for that? I hope so 00:52
luqui the next release is 12 hours away? 00:53
Darren_Duncan autrijus: clkao: in 12 hours or so 00:54
twas said about 30 mins ago
IshaCocoa_ Anyone here built Parrot on win32? I'm getting a "Expanded command line '...' too long error whenit tries to "lib -out:blib/lib/libparrot_s.lib". Workarounds? 00:56
Darren_Duncan no win here
I believe the problem is this portion of the Makefiles: @ARGV = grep !/[A-Z]/, map glob, @ARGV; 01:03
right now I'm trying to figure out where that is added to them 01:04
actually, we can't use T for the alternative, unless the whole test suite is turned off 01:06
because some platforms that Pugs has to run on don't have case insensitive file systems
so we can't have both a 't' and a 'T' in parallel on them
luqui right 01:08
.tx?
Darren_Duncan sure
luqui nobody likes textans
texans
Darren_Duncan it just has to be different characters 01:09
okay, I never found it 01:36
where is this being put into the Makefiles, so we know what to edit so this doesn't happen: @ARGV = grep !/[A-Z]/, map glob, @ARGV; 01:37
?
mugwump Darren_Duncan: didn't you say you were going to rewrite ExtUtils::MakeMaker? :) 01:50
svnbot6 r4184, autrijus++ | * handle multiple return values. It segfaults still --
r4184, autrijus++ | clkao, please take a look.
Darren_Duncan I thought about it and then changed my mind 01:51
also it sounds like someone else was going to do it
mugwump that statement has conflicting tenses 01:52
Darren_Duncan wasn't there an announcement recently where someone else said they would do it? 01:53
mugwump anyway, makefiles are generated there. I think ingy played with them last, but first he was possessed by a demon. We later revived him and now he's OK
Darren_Duncan I tried doing a number of string searches and couldn't find what I wanted in either the Pugs or Perl distro 01:54
I figured the ignore-uc thing was added explicitly for Pugs as a temp measure 01:56
normal CPAN makefiles certainly don't act that way
I just did a search for "!/[A-Z]/" and it only appears in the Makefiles themselves, not the makefile generators
still, I'm hoping that someone who wrote that functionality can speak up 01:57
seen ingy
jabbot Darren_Duncan: ingy was seen 1 days 9 hours 5 minutes 1 seconds ago
Darren_Duncan mugwump, re makemaker rewrite, more to the point it seemed like since I last discussed it someone already had did what needed doing, a month or 2 ago 02:00
ingy probably 02:01
so I thought it was done
eric256 quite in here 02:50
err quiet
Darren_Duncan tis 02:56
luqui AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA' 03:01
mugwump what's all this racket about!?
luqui contrast? 03:02
Darren_Duncan the A key is stuck down
luqui no, I was screaming
Darren_Duncan I'm writing a quick note to p6c re the test file name issues so it gets more attention 03:03
luqui That or you just have to be here when somebody who has any idea what's going on is here
Darren_Duncan yes 03:04
I'll try both
mugwump Darren_Duncan: just hack a s/// into lib/Perl6/MakeMaker.pm, I think 03:10
Darren_Duncan could do, but isn't there a more elegant solution? 03:11
mugwump NO
not unless you want to rewrite ExtUtils::MakeMaker ;)
Darren_Duncan I get the impression that Perl6::MakeMaker isn't being used anymore, but rather inc::Module::Install is, so the latter should be hacked, correct? 03:15
mugwump Somewhere like inc/Module/Install/Makefile.pm (sub fix_up_makefile) perhaps 03:19
Darren_Duncan that's what I thought
mugwump That might be copied from the Module::Install distribution, but in either case a commit there will communicate more clearly than a p6c post methinks :) 03:20
Darren_Duncan I'm deferring the p6c post 03:21
it may never go up
commit instead
how does this line look to you ... 03:27
$makefile =~ s|\@ARGV = grep !/\[A-Z\]/, map glob, \@ARGV;|\@ARGV = map glob, \@ARGV;|;
IshaCocoa Does (GNU) make support inline files?
Darren_Duncan that's in fix_up_makefile() after all the other substitutions
mugwump use \Q ... \E please :)
oh, wait, no that wouldn't work either would it
Darren_Duncan what does that mean?
mugwump would interpolate
Darren_Duncan I thought my backslashing would prevent interpolation 03:28
all characters in regexp should be treated literally
mugwump \Q disables pattern metacharacters, but not interpolation
extremely handy with untrusted search strings 03:29
or when you've got a lot of brackets etc in your pattern
that change looks sufficiently harmless to me, commit and I'll smoke it for you 03:30
Darren_Duncan okay
mugwump jabbot: seen svnbot6 ? 03:31
jabbot mugwump: svnbot6 was seen 1 hours 40 minutes 25 seconds ago
mugwump smokes r4184 as a control 03:32
jql hmm
Darren_Duncan committed 4185 03:34
jql I think literal syntax is <'@ARGV = grep !/[A-Z]/ ...'> 03:35
mugwump that's perl6 rules, though, innit? :)
jql yes
mugwump This is a Perl5 module
jql I'm trying to remember p6
mugwump good
Darren_Duncan the literal text is: @ARGV = grep !/[A-Z]/, map glob, @ARGV;
I removed the grep 03:36
mugwump you could use \Q${\( '@ARGV = grep !/[A-Z]/, map glob, @ARGV;' )}\E :-)
but I think at that point it starts to defeat the purpose ;)
jql heh 03:37
@ doesn't interpolate, at least
you just need to escape [
mugwump it doesn't?
mugwump tries
luqui what would it interpolate as?
Darren_Duncan while I could find nothing else in Pugs containing just '!/[A-Z]/', I thought a s/// with the longer string was more future proofing
svnbot6 r4185, Darren_Duncan++ | updated inc/Module/Install/Makefile.pm so that generated Makefiles do not ignore t/*.t scripts having uppercased chars in file names -- if you want your scripts to be ignored during 'make test', place them in a different dir than t/, such as tX/
jql well, that was a guess
jql didn't try it 03:38
ok, you can laugh at me
luqui hahahahaha 03:39
mugwump perl -le 'my @a = qw(1 2 3); my $qr = qr/@a/; print $qr'
(?-xism:1 2 3)
luqui really
that's unexpected in an expected sort of way
jql s/@// doesn't interpolate @, s/@@// doesn't interpolate. s/@foo// interpolated
damn perl
mugwump notes that the perl5 embedding is causing a compile failure on 4184, and switches it off 03:40
jql even though @@ is a perfectly legal variable
Darren_Duncan is it now?
jql I know and love @@
perl -le 'push @@, 1; print @@'
%% is also good 03:41
I love me some @@
luqui it doesn't mean anything, right? 03:42
'
jql define 'mean'?
luqui it's not magical
jql no, not magical
it's legal because $@ is legal
luqui right
mugwump but what about %%? 03:43
luqui that's legal too, I believe
jql %% is legal because $% is legal
Darren_Duncan how about @$ or %$
jql && is probably not legal, for obvious reasons
those are all fine
luqui Darren_Duncan, those are dereferencers
mugwump &{'&'} :)
jql well, of course. :)
Darren_Duncan by themselves?
luqui oh my god, it looks like @$ is fine by itself 03:44
jql sure it's fine
luqui just don't put \w after it
Darren_Duncan I always use braces for derefs for readability
eg, @{$foo}
mugwump yuck :)
luqui yeah, I always avoid braces when I can
but, tmtowtdi, teho, etc. 03:45
mugwump I use them when ambiguous
Darren_Duncan of course, I'm often doing this: @{$foo->{'bar'}}
jql perl -le 'print %++%%
jql hides
luqui in which case you have no other choice
Darren_Duncan you can't take the braces out then
luqui well, yeah you do I guess
@{$$foo{bar}}
but, uh, no
mugwump I use $foo->bar and let the accessor and context return a list :) 03:46
luqui and people say perl is linenoise
Darren_Duncan the result of $foo->bar is an array ref, which is a scalar 03:47
jql @{ my $bar = $foo->{bar}; my $baz = $bar->{baz}; $baz->{blurfl} }
Darren_Duncan I'm talking p5 here
jql takes indirection to the Nth
luqui jql, I'm not even sure that's legal
jql it's legal
mugwump I like my array accessors to return lists in list context.
luqui no, it is
jql the {} is a do{} block which has its return-value array-derefed 03:48
except for a bareword, which is special-cased
ala. hash
Darren_Duncan question ... 03:49
do you know what the insanta/outsanta files are for?
mugwump elves? 03:50
jql sleds?
Darren_Duncan they seem related to Pugs' compilatoin
eg, insanta.988.59
but I don't know why they stick around
jql knows not the ways of GHC 03:51
Darren_Duncan it seems my file change didn't work 03:53
trying again
mugwump, hold off the smoke for the moment 03:54
mugwump puts down the crackpipe
Darren_Duncan I mean my version; continue with the control one
mugwump yup 03:55
Darren_Duncan: there seem to be lots of test scripts that make foo.$$.NNNN temps.. 04:14
svnbot6 r4186, mugwump++ | force_todo / un-TODO various tests for release - round 1
Darren_Duncan I'm continuing to isolate the problem 04:15
I think the Makefile is post processed after Build is done with it, and that's where the A-Z etc is added, hence my change had no effect
mugwump ok, well why don't you commit something that works for you first ;)( 04:17
Darren_Duncan working on it 04:18
meanwhile, my change didn't break anything 04:20
twas a no-op
I have confirmed a change following when fix_up_makefile() is run 04:32
luqui mugwump, by the way, I warnocked the message that you mentioned got warnocked because I didn't understand it 04:33
There aren't too many people on p6l who grok Haskell, and I couldn't understand the message by ignoring the haskell
mugwump fair enough. I barely understood it myself ;)
Darren_Duncan found the problem 04:35
inc/Module/Install/Pugs.pm is what's adding the offending line in the first place 04:36
I failed to find it before because my grepping didn't account for the target having the slashes escaped
will commit momentarily
svnbot6 r4187, mugwump++ | :todo / fix various tests for release - round 2 04:40
Darren_Duncan committed 04:44
mugwump, smoke away 04:46
svnbot6 r4188, Darren_Duncan++ | undid r4185's change to Makefile.pm and redid the fix properly, by altering inc/Module/Install/Pugs.pm to not add the offending section in the first place 04:47
mugwump I'm still smoking 4184 :)
Darren_Duncan no prob
mugwump or, rather, I'm marking the failures as known
Darren_Duncan meanwhile, I'm doing a clean smoke on my machine
quickie ... 04:50
is there a svn command to rename a file, or do I have to delete + add?
nevermind
found it
nothingmuch good morning 04:52
Darren_Duncan time to rise and shine
luqui feels like watching a movie 04:57
any suggestions?
mugwump The Triplets of Belleville
luqui yeah, I loved that one
mugwump Shaolin Soccer
luqui oh right 04:58
I saw kung fu hussle
okay, thanks
mugwump if you've seen that, ... ah, beat me to it :)
what about Snatch, or Two Hands 04:59
nothingmuch Battleship Potemkin 05:02
Darren_Duncan so now the LKT test file names are back to what they should be 05:07
svnbot6 r4189, Darren_Duncan++ | LKT test file rename stage 1 of 2
r4190, Darren_Duncan++ | LKT test file rename stage 2 of 2
mugwump Darren_Duncan: sadly I've run out of time to smoke the new version - those tests took 47m to run on my system! Must be due to the NFS directory :) 05:17
svnbot6 r4191, mugwump++ | :todo / fix various tests for release - round 3
Darren_Duncan mugwump 05:18
that's fine, I'm doing a full run on my system 05:19
realclean, make, test
mugwump not only that, but the smoking I did didn't even cover testing the modules ... but hopefully it will free up autrijus' time a bit :) 05:20
Darren_Duncan yep
now, you only updated the test files, right, no haskell files? 05:21
mugwump that's right 05:22
Darren_Duncan okay, I'm on 4191 and beginning make test 05:27
06:00 joepurl is now known as joepurl28730292
svnbot6 r4192, mugwump++ | Document Set.pm unicode operators as requested on p6l 06:00
06:08 joepurl28730292 is now known as joepurl
svnbot6 r4193, autrijus++ | * allow calling Perl 5 methods with named arguments 06:17
r4193, autrijus++ | (which are flattened into two positional arguments each).
mugwump sidles off home 06:24
svnbot6 r4194, mugwump++ | Add some more unicode operators for the hell of it ;)
jql good one 06:30
Darren_Duncan finished 'make test' for 4190 ... 06:32
Failed 1/319 test scripts, 99.69% okay. 498/6489 subtests failed, 92.33% okay.
autrijus lkt?
Darren_Duncan failed test: t/rules/Disabled/rules.t 683 498 72.91%
autrijus oh, all tests
the "Disabled" test is not supposed to run anyway
Darren_Duncan I didn't do anything with LKT since yesteday except rename the test scripts
yes 06:33
so Disabled should be moved out of t/
autrijus I thought our harness already ignored it. hm
Darren_Duncan it did, but I fixed that
autrijus heh, "fixed".
Darren_Duncan now the harness will run test files no matter what their name is 06:34
luqui \did someone try to implement that "describe" program that was mentioned on the list?
autrijus Darren_Duncan: please svn mv the disabled tests somewhere else then.
t_disabled
or something. improvise.
Darren_Duncan will do
autrijus darren++
<- in $work... doing a Class::DBI based perl6 script 06:35
Darren_Duncan and thats a much more elegant solution than the old hack
autrijus Darren_Duncan: sure, I agree
Darren_Duncan funny thing is, we discussed this before, but such a remedy didn't come up
luqui autrijus, oh, I just saw the six in @autrijus[-2] 06:36
lucky you, wish I could use p6 in my job
Darren_Duncan autrijus, r4185,4188 contained my fix ... it even says in the checkin comment that a good alternate way to disable is to move out of t/
so your suggestion is in line with that
autrijus luqui: the six?
Darren_Duncan: cool, please proceed then
luqui Class::DBI based perl >>6<< script 06:37
autrijus ahh.
luqui hmm.. we don't have a hyper 6 operator yet...
so you don't know if anybody tried to write "describe" ?
autrijus no. 06:38
luqui okay. time to re-invent if someone did, then :-)
Darren_Duncan autrijus, there are a lot of other files in Disabled, and they did not fail 06:39
autrijus Darren_Duncan: do you have parrot in path? 06:40
Darren_Duncan no
autrijus then those tests didn't run.
Darren_Duncan probably the reason
will move entire folder 06:41
autrijus nod.
06:45 castaway_ is now known as castaway
Darren_Duncan autrijus, is it fine to 'svn mv' a whole folder, or should I do each file inside separately? 06:56
autrijus Darren_Duncan: the whole folder.
(that will preserve history)
Darren_Duncan ok
I wasn't sure, cause when I did a trial run the info spat out just said "A + folder" but didn't mention each file, whereas it did mention each file for the "D" part 06:58
actually it just said "A folder" 06:59
autrijus the "+" is right. 07:00
ah.
don't worry. it's fine
Darren_Duncan committed 07:06
autrijus cool. bbl & 07:07
svnbot6 r4195, Darren_Duncan++ | moved tests from t/rules/Disabled/* to t_disabled/rules/*; removed old Disabled 07:16
Darren_Duncan fyi, assuming that 6.2.6 will be going up in a few hours, I'll defer any further LKT work until afterwards ... its midnight here
fyi, all of my commits today are related -- they are all specifically about letting test files have uc chars in their names, allowing those to run, and making sure files that should be disabled stay that way 07:18
I didn't do anything else 07:19
oh yeah, and actually renaming some tests to uppercase
is there a dedicated ChangeLog updater, or does anyone do that? 07:36
castaway you can volunteer :) 07:37
(generally someone does, or gets volunteered by autrijus) 07:38
Darren_Duncan in any event its, obviously, out of date right now 07:44
castaway quite likely 07:47
someone usually compiles it from the journal entries iirc
Darren_Duncan good night 07:54
svnbot6 r4196, Darren_Duncan++ | some MANIFEST.SKIP updates
masak castaway: I think "gets volunteered" is a quite blatant contradiction in terms 08:25
arcady say, does perl6 allow underscores in numeric literals? 08:27
because apocalypse 2 says it does, and pugs doesn't allow it 08:28
scook0 arcady: probably just unimplemented--care to write some tests? 08:31
arcady I kind of don't have any free time at the moment 08:34
castaway grins
I think its correct tho masak :)
arcady eval("42_000") should be 42000
castaway ?eval 42_000 08:35
evalbot6 Error: unexpected "_" expecting digit, fraction, exponent, term postfix, operator, postfix conditional, postfix loop, postfix iteration, ";" or end of input
castaway pats evalbot on the head
scook0 I've just written up that one, any other useful test cases? 08:36
arcady with operators
castaway umm.. 42_000_000 and 42_000.31 ?
right
arcady and 42.0_32 08:37
scook0 castaway: good thinking
autark-jp and 4_2000
arcady and possibly 42__0 not being valid
scook0 arcady: 42__0 is valid P5 08:38
arcady oh, is it? 08:39
didn't know
castaway umm, whats 42.0_32 ?
arcady underscores in a number are ignored
btw, this goes for all kind of numbers 08:40
float as well as int
and octal and hex too
castaway oh, sounds like an easy implementation then
autark-jp throw away all '_' except at the beginning or something maybe.
arcady yes 08:41
castaway right
svnbot6 r4197, bsmith++ | Added an export list.
r4198, bsmith++ | Moved some parser types to Pugs.Parser.Types.
autark-jp anyone on good terms with the haskell type system btw? 08:42
scook0 autark-jp: depends, what's your issue?
autark-jp scook0: in Pugs.Lexer.hs, interpolatingStringLiteral 08:43
scook0: it takes in something called 'RuleParser x'
scook0 autark-jp: aye, do you understand what that is supposed to mean? 08:44
integral autark-jp: A (RuleParser x) is a just a parser than returns some unspecified type
autark-jp scook0: it says (in comments) that it is a delimiter (for strings) 08:45
integral autark-jp: notice that where endrule is used lower down we don't use it's value
autark-jp integral: ok, but I want to use it's value ;)
integral: and then the type system starts yelling at me :/
integral hmm, you'll probably need to change it to RuleParser Exp then
what type of value do you expect it to have? an expression? 08:46
scook0 the type system should be telling you the type it's expecting
autark-jp scook0: it says something about rigid variable x can't be match to 'char' or 'string' (depeneding on whether delim is a single char or mulit-char)
ie. <<foo>> vs. <foo> 08:47
scook0 hmm
autark-jp wants q(foo(bar)baz) to be legal.
scook0 so, people are already giving it both chars and strings?
or is that what you're trying to do? 08:48
integral this is harder. You'll have to change where it's called so that they always use strings
castaway hmm, that got discussed a while back , but I cant remember what the outcome was
autark-jp it is being given both both string (symbol?) and char it seems.
integral: hm, I should maybe try that.
clkao bad autrijus can't write c 08:50
castaway who wants to write C :) 08:51
luqui autark-jp, I think we've said that the delimiters nest unless they're backwhacked 08:57
svnbot6 r4199, clkao++ | Fix perl5_apply returning multiple values and the use in mkval generating closure.
autark-jp luqui: sounds reasonable. 08:58
08:58 theorbtw1 is now known as theorbtwo
svnbot6 r4200, scook0++ | Tests for numeric literals with underscores 09:08
fireartist warnocked :) Could someone take a look and set me straight?... 09:10
www.mail-archive.com/perl6-language...20241.html
(question about class construction)
luqui lookin 09:13
g
Oh, I read that 09:14
I'm not really sure what you were asking
But in any case, if you write your own new, you have to call BUILD yourself
fireartist oh, ok, I'll try again now,
hmm, didn't realise I'd have to call build myself
luqui the recursive BUILD call is something that is done by the default new 09:15
we generally don't write our own news (no pun intended) in perl 6
fireartist is build not called by BUILDALL? Do I have to call that too?
luqui maybe that's the one you call 09:16
fireartist the reason I defined my own 'new', is the object's usually constructed with only one arg
I wanted a shortcut, rather than having to pass it my name everytime
luqui oops 09:17
I'm wrong
fireartist (by name everytime)
luqui apparently bless calls BUILDALL for you
fireartist the email was trying to clarify if that's what is supposed to happen, because I was seeing some strange behavour with pugs... 09:18
luqui what were you seeing 09:19
fireartist just a minute, I'm opening my editor to make sure I get this right 09:20
luqui Couldn't you just write:
submethod BUILD (Str $.date, +$.time, +$.offset) 09:21
fireartist my explicit new method was being run, when I was calling 'new' with a named pair of args
luqui oh, you want your multi to "append" behavior to the default new?
hmm 09:22
fireartist and in my build, the arg was undef
luqui writes up a little test case to see what you're talking about
fireartist I wanted the option of either a single arg, or all named args 09:23
www.fireartist.com/foo.pl shows the behaviour
it prints 'ok', showing my new is called, rather than the new inherited from Class 09:25
it also prints date: ''
meaning the arg in build is undef
luqui hmm, I think it should call your new.
but the fact that date is undef looks wrong 09:26
fireartist I thought Foo.new(date => 'blah') wouldn't match my 'new' signature
luqui that depends on whether the default new is a multi 09:27
well, and whether that makes any sense with our multi system
a couple of us (including me) think that the multi system needs a rework
fireartist I'm not sure whether the design I'm going for is perl6'ish
luqui what I find perplexing is the fact that your "return $class.new(...)" doesn't fall into infinite recursion 09:28
fireartist if I change it to Foo.new, it does! 09:29
Juerd For some reason, pugs locks itself
luqui But $class should be the Foo object
Juerd So I had to kill evalbot6 in order to be able to upgrade pugs
luqui I mean Foo class object
fireartist If I add 'say $class.perl' in 'new', it reads \{obj:Class}
Juerd autrijus: Any idea how to avoid this?
luqui right, not very informative
Juerd 1;0 juerd@feather:~/auto/pugs$ cp pugs ../bin 09:30
cp: cannot create regular file `../bin/pugs': Text file busy
fireartist Should it not be \{obj:Foo} ?
luqui something like that
clkao how's mod_pugs?
luqui If you just "say Foo.perl"
you get the same thing
okay, this is definitely not what is supposed to be happening 09:31
I don't grok the object internals though
fireartist outside of the class, after I've constructed $foo, is reads \{obj:Foo}
luqui maybe you should post the example and behavior to p6c
Not for me 09:32
fireartist I'll do that thanks, I'll try to be more clear what I'm asking though!
luqui $foo.perl does, but not Foo.perl.
as far as how you would do it in Perl 6, though
I think you're best off putting all the logic into BUILD
and avoid writing your own new altogether
vcv how do i create a list of 26 items (the letters of the alphabet)? i feel stupid, but my brain isnt working
luqui 'a'..'z' ? 09:33
vcv doesnt seem to be working.
luqui works for me
what codE?
fireartist thanks for your help luqui, I'll start on that email now.
luqui thanks for the email ;-)
vcv pugs> @chars ōæ½xBB~ōæ½xAB '+' 09:34
'a b c d e f g h i j k l m n o p q r s t u v w x y z+'
that cant be right
@chars >>~<< '+' works
luqui huh?
it looks like its working to me 09:35
arrays join on spaces when interpolated (just like in p5)
scook0 vcv: maybe your pugs is hosed
luqui the example that he just showed looks like it's working fine 09:36
vcv ok. why would @chars ōæ½xBB~ōæ½xAB '+' and @chars >>~<< '+' return different results
luqui because one is stringifying @chars and then appending '+'
the other is appending '+' to each element of @chars, and returning that list
the same reason @chars + 1 and @chars >>+<< 1 are different
vcv i thought >><< was ascii version of ōæ½xBBōæ½xAB
scook0 vcv: Ā»~Ā« worksforme 09:37
vcv i just cant understand why ōæ½xBB~ōæ½xAB and >>~<< are working differently
luqui so, "@chars >>+<< 1" means "map { $_ + 1 } @chars" 09:38
scook0 vcv: I'm not seeing any difference when I try it
luqui whereas "@chars + 1" means [email@hidden.address] + 1"
vcv, are you a perl 5 programmer?
vcv i understand what it does luqui
im tellin you ōæ½xBB~ōæ½xAB and >>~<< are doing different things
scook0 vcv: maybe you need to update & rebuild
luqui than what they're supposed to, or than each other
vcv just did
than each other
luqui they're supposed to
oh! 09:39
I must be unicode-stupid
You're probably typing the unicode version of >>~<< and I'm just seeing ~
okay
vcv oh! haha
i bet i looked like a real idiot ;)
luqui but now it is I hahaha!
:-)
are you doing this on the pugs command line? 09:40
vcv yes
luqui have you tried writing it in a script?
vcv ill try that
luqui the pugs command line might be unicode stupid too
arcady it works fine for me though
luqui terminal settings? all that unicode bullshit that we knew we'd run across once we accepted the latin-1 variants 09:41
scook0 hmm, it must be the command line
theorbtwo ...or your shell might be unicode-stupid.
Mine was, until I started tweaking things.
arcady ?eval ("a", "b", "c") Ā»~Ā« (1, 2, 3)
scook0 Juerd had to kill evalbot for a bit
arcady oh whoops 09:42
it's kinda early
didn't notice that
castaway hmm? it was there just now..
arcady yeah, I think it's the unicode issue
castaway oh, missed that, so he did
vcv >pugs test2.p6
a b c d e f g h i j k l m n o p q r s t u v w x y z,
a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,t,u,v,w,x,y,z,
code is: my @chars = 'a'..'z'; 09:43
say @chars ōæ½xBB~ōæ½xAB ',';
say @chars >>~<< ',';
scook0 vcv: your terminal seems to be throwing away Ā«Ā»
arcady yeah, that really looks like it
vcv but im running it from a file darn it! :/
arcady because you get @chars ~ ',' for the first one
luqui vcv, it's got to be unicode, because it works for me 09:44
scook0 vcv: yeah, the file seems to be working correctly
vcv no that cant be it
watch..
scook0 sorry, no, that's wrong
lumi vcv: Are you sure you're doing it in unicode and not latin1?
vcv abcdefghijklmnopqrstuvwxyz
a b c d e f g h i j k l m n o p q r s t u v w x y z,
a,b,c,d,e,f,g,h,i,j,k,l,m,n,o,p,q,r,s,t,u,v,w,x,y,z,
say @chars;
say @chars ōæ½xBB~ōæ½xAB ',';
say @chars >>~<< ',';
luqui that makes sense too
vcv errrr
theorbtwo vcv: I think you are outputting latin1 into the channel instead of utf8.
vcv whoops made a mistake 09:45
theorbtwo Could you give us a euro sign, please?
vcv alright. thanks. stupid latin1
hm.. yeah
theorbtwo (A good test because the euro sign is not in latin-1, since it didn't exist at the time.)
vcv im a unicode newb. 09:47
well i use alt+0171 and alt+0187 to get the characters
if that helps
luqui you're on windows
?
vcv yes 09:48
luqui doesn't know anything about windows unicode
theorbtwo knows very little about it.
What IRC client are you using?
Somewhere, there should be a setting for what charset it expects data from the server to be in.
vcv mirc because there is no good client for windows. yet. :) 09:49
luqui xchat on cygwin ;-)
vcv ha!
lumi irssi
vcv irssi is nice, but some reason i cant stand it when im on windows.
osfameron xchat on windows it pretty good
luqui oh, there's an xchat for windows 09:50
well, screw cygwin then
vcv Yeah. used it and didnt like it
i like to customize my stuff a lot and mircs scripting is easy, so
luqui afaik xchat is scriptable with perl
never used it though
arcady is there no irc client written in perl6 yet? 09:51
luqui hahaha
no, I don't think so
kungfuftr you can, Net::IRC runs on pugs
arcady yeah, someone better write one
luqui does curses or gtk or ptk or any of those? 09:52
vcv i use this when im bored: weakmind.org/zephyr/zephyr.png
theorbtwo Looks like mIRC has no utf8 support. 09:53
vcv figures
though it doesnt look like its mirc, since the file does it too 09:54
IshaCocoa Anyone know there way around the Parrot makefile on Windows? 09:55
vcv doubt there is
whats wrong with the makefile?
luqui heh, I was trying to see if I had a combining key and accidentally pushed ctrl-alt-backspace :-p
theorbtwo Swift there, luqui.
scook0 luqui: fantastic
Aankhen`` theorbtwo >> That is correct. 09:56
There are many petitions and requests for Unicode in mIRC.
The author has said he will add it... eventually.
theorbtwo xmodmap -pk|grep -i multi will tell you if you have one.
vcv No he wont. he's too busy sailing on his yacht
luqui is unskilled in xmodmapfu
Aankhen`` vcv >> I'm just relaying what he said. :-P
vcv Yeah. he's said a lot of things :p 09:57
luqui theorbtwo, doesn't seem I do
theorbtwo I use xmodmap -e 'keycode 94 = Multi_key', which makes the 102nd key on my european keyboard in en_US layout be the compose key.
luqui how do I give myself one
ok
vcv then i realize mirc sucks and is going no where, so i starting writing my own client
luqui what's the keydump utility so I can find the scancode of a key
theorbtwo Do xev, and hit the key until you figure it out. 09:58
It's not exactly nice, but it works.
luqui yep
?woo? unicode 09:59
and that goes into .xmodmaprc? 10:00
vim thinks so, so I'll trust it
Aankhen`` vcv >> Are you seriously writing a new IRC client? 10:01
luqui that may be a bit hubristic 10:02
got to make sure you're lazy and impatient about it, otherwise it will go nowhere
(I've found that the three virtues only work well when you have all three together in one room)
theorbtwo luqui: Haven't quite figured out where it goes, sorry. 10:04
luqui: Since you probably write better then me, put that info in the unicode minihowto thingy? 10:05
luqui where's that?
theorbtwo docs/quickref/unicode 10:07
clkao summon autrijus
VType typ -> vstrToSV (showType typ)
luqui oh, it seems that there's already a url there that does it 10:08
clkao this doesn't seem to catch what ref() returns
so it's doing infinite recusion on mkValRef
luqui nope, it doesn't
it's lower level than what we need
adding it
nothingmuch yum yum dum dee dum 10:11
theorbtwo wonders if nothingmuch is going to break our bones to make his bread.
nothingmuch instead eats a jewish chicken 10:12
eek, what horrible english.
Aankhen`` Racist!
What's wrong with us normal people, huh?
We're not good enough for you?
Aankhen`` pouts.
nothingmuch well, i was paraphrasing
theorbtwo s/jewish/kosher/
nothingmuch nay nay
Aankhen`` Hang on a sec while I look up 'kosher' in the dictionary. 10:13
nothingmuch jewish as in jewish mother (or father, in my dad's case)
Aankhen`` Hrm.
Aankhen`` goes back to calling nothingmuch a racist.
castaway the chicken had jewish parents?
nothingmuch www.amazon.com/exec/obidos/tg/sim-e...45-8113544 10:14
no, it was just behaving jewishly
nothingmuch goes to help something compile
theorbtwo No, just a jewish father. (Which is nice, and all, but neither neccessary nor sufficent, at least according to some people.)
Of course, I don't think chickens can have souls, much less Jewish ones, so it's really a pretty moot point. 10:15
Aankhen`` koshers theorbtwo. 10:16
luqui oh, now you're being politically correct for sure theorbtwo
Aankhen`` Why can't chickens have souls?
scook0 nominates this for "most OT discussion EVER" :-)
luqui this will lose
The jewish faith and perl 6 have tons in common 10:17
theorbtwo waits for luqui to give an example.
svnbot6 r4201, luqui++ | Add unicode tips.
luqui exactly what is left as an exercise for the reader
Aankhen`` waits too.
Cop out!
Aankhen`` glares at luqui.
theorbtwo Total cop-out.
luqui I'm american
roll a dice 10:18
Aankhen`` Ah, that explains it. ;-)
luqui if it doesn't come up six, then I was being sarcastic
theorbtwo "They are both not-so-secretly run by the Cabalists."
"They both require Unicode." 10:19
luqui huh?
oh
haha
there we go
and we were just talking about unicode
see, not OT
pjcj Old Testament? 10:20
theorbtwo The word Cabal comes from Kabbalah, a Jewish tradition of hidden knowladge.
luqui Apocalypses
or Perl 5
can't decide
luqui steps back to hide in his ignorance
and watch a movie
luqui &
castaway declares lunch 10:21
theorbtwo Blast, scared away the Great Summorisor.
castaway ? 10:22
theorbtwo luqui is Luke Palmer, who... oh, nevermind. 10:23
theorbtwo was thinking of somebody else.
theorbtwo is very good at confusing people.
castaway as long as you dont confuse me with someone :) 10:24
theorbtwo You're Jill, right.
castaway close
theorbtwo Jane?
castaway nope
theorbtwo Ann? 10:25
theorbtwo kisses his Jess.
castaway :)
castaway munches a salad sandwich
soup time!
theorbtwo wonders what the best way of making a haskell-level Param into a perl6-accessable scalar is. 10:27
scook0 theorbtwo: trying to expose params as perl6 values? 10:28
castaway hmm, dont bits already do that.. like $?PUGS_VERSION ?
nothingmuch wants tea 10:29
castaway passes the thermos
nothingmuch so anyway, in the stinky cheese man there's a giant, which says that he'll make bread out of several characters 10:31
and there is also an annoying chicken
which says that no one helped her make the bread or something
and then the giant eats the chicken with the chicken's bread, IIRC 10:32
Aankhen`` pokes nothingmuch and runs around laughing maniacally. 10:36
nothingmuch tee hee
anyway, i meant to get me some tea
theorbtwo Oh.
I was thinking of Jack and the Beanstalk, and didn't get the Jewish Chicken reference at all. 10:37
nothingmuch lemon tea with honey 10:40
nothingmuch feels sick
clkao my $rw = 1; warn \$rw; why is it still 1 ? 10:42
nothingmuch clkao: i think autodereferencing goodness, but I'm not up to date on that stuff 10:43
theorbtwo s/goodness/badness/ 10:44
nothingmuch i think autodereferencing could have it's goodness, if it were not confusingly special cased 10:45
the problem is that sane semantics for autodereferencing, are.... well... tough.
i don't like having to explicitly dereference with @{ } and so on in p5
but at least if I do it slowly and surely, balancing everything beforehand, things turn out OK
theorbtwo I don't like having to explicitly dereference all the time either. 10:46
But it means that I know what's going to happen in any given situation, and can treat things generically without too much worries. 10:47
nothingmuch exactly
theorbtwo I can have a tree class, into which you can put references or nonreferences, without having to do anything special.
I'd like a simple rule that says when something gets autodereferenced and when it does not.
...and I'd like the answer to be "does not" most of the time. 10:48
nothingmuch i think this is like lazyness and autothreading 10:49
the only place where it should be controlled is the absolute edge
either at the lowest level, or the highest level
theorbtwo Oh! Lathos is Simon Cozens!
Now I get his references to being in the Cabal. 10:50
nothingmuch Oh!
He Is Indeed!
I on the other hand googled yesterday, and didn't have to realize it in a boom 10:51
BUAHAHAHA
theorbtwo Oh.
I looked in AUTHORS, but he wasn't there.
nothingmuch ah
theorbtwo I didn't google, after assuming that "lathos" didn't sound very googleable -- there were probably too many.
autark-jp gah! any way to speed up 'make test'? splice.t just took 1m15s :/ 10:52
castaway get a faster machine
,)
autark-jp ;)
nothingmuch autark-jp: use feather
autark-jp feather?
nothingmuch Juerd's machine
autark-jp ah, and suddenly the topc makes sense :) 10:55
autark-jp <- slow
clkao how do i call assignment with a sub? &infix:<=> ? 10:58
nnunley theorbtwo: You probably got a few too many references to Chrono Trigger.
theorbtwo That should do it, clkao. 10:59
clkao it doesn't work. *** Undeclared variable: "&infix:=" 11:00
theorbtwo The emphasis may be on /should/.
clkao ok. if i want to write a failing test where does it belong? 11:01
to
scook0 currently '=' only exists as a Syn, not a regular function... 11:03
clkao eek. will that change?
scook0 clkao: I don't know what the specced behaviour is, that's just the current implementation
t/operators/assign.t ?
theorbtwo scook0: I don't know of anything specced on = that would make it different from any other op. 11:04
(And every op should be a function with a strange name.)
scook0 the only difficulty is evaluating the LHS in LHS context, but I suppose that should be feasible somehow 11:05
theorbtwo Hm, good point. 11:07
It might need to be a macro (and we don't have those yet... something which I should work more on fixing). 11:08
kungfuftr perl6 is moving way too quickly for me to keep track... =0(
castaway ditto, kungfuftr 11:11
theorbtwo macros have been semispeced for a while.
They just haven't been implemented.
nothingmuch hah! in our product we have a method named dearHierarchyGiveMePleaseHeightOfMember 11:13
theorbtwo laughs aloud at nothingmuch's product. 11:14
castaway .) 11:15
scook0 reminds me of INTERCAL...
nothingmuch i'm 90% sure I Know who wrote it too =)
theorbtwo You don't have a svn blame (or equiv) to tell you? 11:16
Mornin, IshaCocoa. 11:24
nothingmuch theorbtwo: p4 can do that in an acward way 11:25
ackward
i have to trace back through branches manually
theorbtwo Oh, perforce, not perl 4.
Ugh.
castaway wondered there too
nothingmuch and I have to convert changelist numbers to authors
theorbtwo You could write a script to do it... but it probably wouldn't be worth the effort.
nothingmuch and just to find out who has a nice sense of humor, it's too much work =) 11:26
clkao the accessing perl6 data as reference requires the assigment function to work. 11:27
failed test committted.
s/the/to
autrijus greetings.
clkao: I want auto TIEHASH
clkao autrijus: i'm having it. 11:28
autrijus: doesn't work because i can't assign
i'll finish the tie stuff later
IshaCocoa_ Howdy theorbtwo.
autrijus clkao: what can't assign?
clkao if you can fix the assignment, and ($rw = 1; ref(\$rw) eq 'Scalar') i'll be hapy 11:29
autrijus: see my last comit
autrijus ref(\$rw) is Int
always is
clkao it returns '1' :)
autrijus pugs> my $rw = 1; ref(\$rw)
::Int
?eval my $rw = 1; ref(\$rw)
clkao not right 11:30
castaway evalbot is gone walkies
svnbot6 r4202, clkao++ | Failed test for assignment operator.
castaway (Juerd said recompiling pugs confuses it or something)
clkao it's same as ref($rw) wtf?
castaway why not, if $rw is an int?
clkao so reference is the same? 11:31
because we have "is rw" so we don't use reference to scalar anymore?
castaway $rw = 'foo'; ref($rw); -> still an int?
oops ref(\$rw)
autrijus pugs> my $rw = 'foo'; ref(\$rw) 11:32
::Str
clkao and try print ref(\$rw)
theorbtwo say ref 1; say ref \1; say ref \\1;
autrijus pugs> my @x; (ref(@x), ref(\@x))
(::Array, ::Array)
theorbtwo At least one of those is wrong. I'd argue that two of them are.
autrijus pugs> ref \(\1)
::Int
theorbtwo Autoderef is ucky.
clkao autrijus: the value of that is wrong
autrijus clkao: the value of what?
castaway ok, I'm confused too now 11:33
autrijus theorbtwo: I don't disagree.
clkao my $rw = 1
print \$rw
this is what got passed into p5
and fscked
autrijus ok.
I agree it's a bug.
I further agree that infinite dereferece, although at one time mandated by larry, is now out of fashion 11:34
so pugs will fix
but probably not today
11:34 ChanServ sets mode: +o autrijus
theorbtwo On a completely different note... 11:34
clkao eek, fix that before i come back from lunch
autrijus clkao: not possible...
clkao otherwuise perl5 accessing p6 ref won't work
it invovles p6l?
11:35 autrijus sets mode: -o autrijus
autrijus clkao: no, it involves rewriting 30% of AST.Internals. 11:35
theorbtwo I'd like to have programatic access to the body of expressions. Is doing instance Value Exp sane?
autrijus theorbtwo: go ahead. see DirStream code
use opaque and fromObj 11:36
castaway ?eval $?PUGS_VERSION
evalbot6 \'Perl6 User\'s Golfing System, version 6.2.5, May 24, 2005 (r4201)'
theorbtwo Cool!
autrijus ok, dinner now. bbiab &
theorbtwo Allo, bro. 11:43
broquaint Hey, theorbtwo. 11:47
What's the latest?
theorbtwo I'm trying to make expressions introspectable. 11:48
broquaint Huzzah! 11:49
theorbtwo LISP with syntax, baby.
broquaint More introspection would be very handy at the moment.
LISP has syntax? ;)
theorbtwo No -- perl6 does.
broquaint Hehe. 11:50
theorbtwo wonders what the difference between vCast and doCast is.
Odin-LAP Which is rapidly becoming the single difference between the two...
theorbtwo (They have the same signature.)
broquaint vCast is a lazy cast?
theorbtwo Ah. 11:51
Wait, no.
broquaint That was a guess.
theorbtwo Those would have different signatures -- the nonlazy version would have a !.
nothingmuch blah! 11:52
Class::Events takes 20 seconds to parse on feather
and then generates an error
theorbtwo Oh. Everything in value has default instances anyway. 11:53
So I only need to implement vCast and castV and everything else hopefully just works.
broquaint tries to track down a recursive infinite loop in some mjd code ... 11:56
broquaint 's brain teh splode 11:57
Awesome. Tron. 12:02
12:18 sekimura is now known as sekimuraway
luqui just finished watching shaolin soccer 12:23
very much fun :-)
thanks mugwump 12:24
broquaint ?eval my $s = -> { say @_ }; $s.(q[b0rk]) 12:35
evalbot6 Error: Undeclared variable: "@_"
broquaint That's not right, is it? 12:36
Yeah, that used to work. Hrm ... 12:37
theorbtwo Looks wrong to me, but I'm not sure. 12:38
autark-jp isn't -> {} saying, I take no variables? 12:39
but $_ get set.
broquaint ?eval sub f(Code $a) { $a } my $s = f { say @_ }; $s.(1) 12:40
evalbot6 Error: Undeclared variable: "@_"
broquaint Using sub {} works, but -> {} and a bare {} doesn't work. It seems odd. It also means a program that was working, does so no longer.
castaway oops 12:42
theorbtwo ?eval [(sub {}).arity, (-> {}).arity, ({}).arity] 12:43
evalbot6 [1, 0, 0]
theorbtwo Yup.
castaway whats that?
broquaint Shouldn't they all behave the same?
scook0 broquaint: they used to, but it was explicitly changed recently 12:44
theorbtwo I seem to recall Autrijus mentioning the other day that they were clarified not to.
autark-jp so if you define f { say @_ }, f(1,2) will work, but not f(1)?
broquaint Could someone point me to where this was clarified?
scook0 I don't think it's on p6l...checking the irc logs... 12:46
Juerd re evalbot6: it can run on feather if it doesn't use the global pugs
Because for some reason, the pugs executable gets hard-locked when in use
And then the automatic upgrades can no longer take place 12:47
I hope autrijus has a solution for this, and that the locking isn't necessary
clkao autrijus: aarrg. :/
scook0 broquaint: start looking from colabti.de/irclogger/irclogger_log/...=891#l1346 12:48
broquaint Thanks, scook0 :)
theorbtwo r4160 | autrijus | 2005-05-30 11:04:07 +0200 (Mon, 30 May 2005) | 2 lines 12:49
* according to luqui, bare {} and pointy -> both have no
default argument at all, unlikes sub/method/coro that has *@_.
Whoops, sorry. 12:50
Too slow.
scook0 just checked S06
broquaint Thanks, theorbtwo, that makes sense quicker than reviewing IRC logs ;) 12:51
scook0 it implies that only sig-less (p5-style) subs get @_, but it isn't 100% clear
castaway hmm
broquaint That should probably be noted in autrijus' next use.perl update. 12:54
castaway wonders where that is declared.. 12:55
I cant find it anywhere in S06 or 04
scook0 castaway: the @_ thing? 12:56
broquaint It makes porting functional perl5 code a little verbose.
castaway yes
scook0 in S06 under 'perl5ish subroutine declarations'
but it doesn't explicitly say that @_ doesn't show up for {} & ->{} 12:57
castaway well under the stuff about pointy subs it says
The arrow operator -> is almost a synonym for the anonymous sub keyword, except that the parameter list of a pointy sub does not require parentheses, it does not require a preceding comma when included in a list, and a pointy sub may not be given traits.
broquaint Seems odd you have to declare the arguments for what is meant to be huffmanized :/
castaway (hmm, those are to anon subs not real subs.. ) 12:58
scook0 I think the 'almost a synonym for anonymous sub' thing is misleading
castaway the perl5ish bit doesnt mention anon subs not having @_ at all
scook0 In my mind, ->{} is closer to {} than sub {} 12:59
castaway ah but what about Larrys? :) 13:00
scook0 castaway: perhaps we need a p6l ruling
broquaint @_ should always be the default arg stack. *Having* to declare it every time seems oddish. 13:01
Oh well, another quirk to learn, or unlearn, as it were.
castaway sounds crazy to me 13:02
scook0 perhaps we need @?ARGS or something
then @_ can default to an alias to that
castaway surely thats just a name for the same thing?
scook0 (I'm trying to crystalise my thoughts at the moment...)
anyway, I think there's a difference between @_ as implicit argument, and @_ as an extra array that gets all arguments 13:03
we also need to figure out whether ->{} and {} take 0 args or unlimited args 13:04
broquaint I don't see why @_/@?ARGS shouldn't be the default for all codesque things that don't have an signature, no signature shouldn't imply no args/$_. 13:05
Pointy and bare subs should have 'unlimited' args.
scook0 broquaint: that's probably best
broquaint It just means that if the args aren't used then the compiler can just optimize them away.
Or something. 13:06
scook0 I think @_ should be an alias to (or copy of) all the arguments, rather than being an argument itself
broquaint For sure.
castaway makes sense to me.. isnt it now? 13:07
scook0 currently, sub {} effectively expands to sub (*@_) {}
which works, but is less general
(IIRC)
broquaint Where as -> {} and {} now have an implicit arity of 0.
&code.arity = Inf ;) 13:08
scook0 so sub {}, ->{} and {} should all get an implicit slurpy param that doesn't have a name 13:11
and you access the arguments through @_
(which isn't a param, but an array of all the arguments)
broquaint a la Perl5 :) 13:12
svnbot6 r4203, autrijus++ | * Kwid::Event::Parser unbroken and test fixed.
arcady whoa, the unary + makes the multiple dispatch confused 13:13
scook0 I must remember to take this proposal further...
arcady or am I just confusing myself
ah, so I am 13:14
broquaint A perl6-lang post perhaps, scook0?
arcady but I'd think that +$a is more like an Int than an Str 13:15
(though it's really a Num)
specifically, if I have a multi sub foo(Int) and a multi sub foo(Str) 13:19
foo(2) call the Int one
and foo(+2) calls the Str one
svnbot6 r4204, autrijus++ | * small random typo fix to hw2005 talk 13:23
r4205, autrijus++ | * &infix<=> can now be used as the asignment operator.
Limbic_Region having read the recent commits - are we preparing for a release today?
castaway AFAIK 13:28
scook0 broquaint: I'm writing up a p6l post, but I'm not getting very far--I'm just about to go to bed 13:29
svnbot6 r4206, autrijus++ | * unTODO and remove unneccessary unshift @*INC
r4207, autrijus++ | * guarantee access to blib6/lib in `make smoke`.
r4208, autrijus++ | * unTODO and fix bad :todo syntax.
r4209, autrijus++ | * declaring `sub Qulified::name` inside a package no longer
r4209, autrijus++ | prepends the package name to the sub name.
castaway karma autrijus
jabbot castaway: autrijus has karma of 511
castaway (hmm, that should read Inf :) 13:30
broquaint I could take a stab if you like, scook0.
scook0 broquaint: feel free--I probably won't get another chance in the next 24h :( 13:31
broquaint Alrighty.
scook0 well, goodbye all
castaway when do you start the job, bro? 13:32
broquaint June 13th.
castaway nifty
castaway hopes they allow PM/IRC access :) 13:33
broquaint So two weeks of intense relaxtion :)
castaway *g*
broquaint I should think so. They seem a like a very sensible outfit.
castaway yay
hmm, their list claims to still be looking for perl devs 13:34
mm, "other european languages" :) 13:36
broquaint Is Haskell a European language? 13:38
castaway Umm.. probably? 13:39
castaway was thinking of german :)
Limbic_Region isn't sure but he thinks efforts have been taken to prevent users from using anything other than IE 14:05
Limbic_Region had to go through strange steps to get FF working this morning
Khisanth with pugs?
Limbic_Region hopes it is all just a bad dream and tomorrow things will work fine
huh?
I was referring to sensible outfits allowing access to certain locations 14:06
and commenting on the fact that I am not sure I work for a sensible outfit
one could argue that "The U.S. Government" by definition could never be considered a sensible outfit
but hey - who am I to judge
Khisanth hmm slightly different imo :) 14:07
anyone who pays taxes should be able to access those! 14:08
what the hell are they doing that would require IE specific features anyway?
theorbtwo Not quite sensible -- the hemline on the skirt is /all/ wrong.
Khisanth (otoh a FF specific site is equally bad)
Limbic_Region completely wrong perspective Khisanth
it is about control
not features
theorbtwo Khisanth: Not IE specific features -- only allowing the computers to run IE. 14:09
Limbic_Region they can lock you down to a specific app with a specific set of plugins with a specific proxy with a specific set of configurations
Khisanth ah see THAT could be useful
for a politician
a press diverting tool ;) 14:10
Limbic_Region given the number of people and sub organizations and geographic dispersity - it makes sense to have security controlled at a cetral location 14:11
you can't trust that people will do the right thing (either intentionally or accidently)
anyway - back to being on topic 14:12
so there is a release scheduled sometime in the near future?
6.2.6?
theorbtwo I think Autrijus wanted to release it a bit ago, but that didn't happen somehow. 14:13
Limbic_Region I was just a bit confused - normally the topic changes 14:14
when I looked at the commit history I got excited
castaway afaik, yes there is 14:15
(didnt I answer that just now?)
Limbic_Region a while back
I have ulterio (sp?) motives castaway 14:16
castaway oh dear :)
Limbic_Region I need to go fight with the INS in a couple hours but was thinking about pinging a few people who have said they are waiting for a stable release of Pugs that has features x, y, and/or z
I think I convinced blokhead to join the madness 14:17
castaway we need more cooks? 14:19
svnbot6 r4210, autrijus++ | * MMD tiebreaking: remove sub type and globalness from the criteria
r4211, fireartist++ | * updated porting_howto with dotted-patenthesis requirement on method calls
Khisanth .()? 14:21
fireartist <whitespace>.()
S12
Limbic_Region castaway - yes, I think so
svnbot6 r4212, duff++ | Added whitespace sensitivity of subs/arrays/hashes to the porting guide 14:24
Arathorn cooks are a state of flux, anyway, surely :) 14:25
clkao my @a = (1,2);my $ar = \@a;warn $ar.elems; 14:38
is this right?
svnbot6 r4213, duff++ | Slight reformat of docs/other/porting_howto
clkao we i need to (@$ar).elems?
(it works in pugs, i am just unsure about p6l)
Limbic_Region why wouldn't it work? 14:39
that isn't meant to be flippant btw
Arathorn (very) vaguely remembers something about autodeferencing
uh, autodereferencing even
Limbic_Region you would expect $ar[1] = 42; # to work right?
the . is -> afterall 14:40
and so $ar.elems is dereferning
Limbic_Region can't speel
clkao ok, so @ar[1] and $ar[1] will both work. any of them is recommended form? 14:43
Limbic_Region hang on as that is a bit confusing
at least to me
autrijus clkao: depends on what you expect it to behave in slurpy context.
say @ar; # each elements
theorbtwo clkao: They both do something, but they do different things.
autrijus say $ar; # a single ref
Limbic_Region thanks theorbtwo and autrijus 14:44
autrijus clkao: think of sigils as a context hinter.
Limbic_Region was hung up on the "both work" part
autrijus they probably only have that much use now.
theorbtwo Eh? @a[1] works on the array in @a. $a[1] works on the array reference in $a. 14:45
Different things.
clkao autrijus: ya, but the problem is single ref odesn't work right?
autrijus clkao: single ref doesn't work right how? 14:47
clkao my @a = (1,2);my $ar = \@a;warn $ar; 14:49
i thought $ar should just give me a reference so i can pass it as a single value to perl5
castaway Im guessing it does, but warn is dereffing it?
warn $a.[0] also works there 14:50
clkao hmmm. ok.. i will try
autrijus clkao: $ar is a reference.
you don't need the \, even. 14:51
castaway (annoying tho, how do you now tell if you have a reference?)
autrijus castaway: I think the idea is you don't :) 14:52
castaway Ugg
why not? how does one debug there?
autrijus well you can do
?eval (1,2).perl
evalbot6 '(1, 2)'
autrijus ?eval my $x = (1,2); $x.perl 14:53
evalbot6 '[1, 2]'
castaway having @a and \@a printed looking the same sounds like it will just lead to confusion
PerlJam Who was it that was talking about transparent vs. opaque references on p6l?
This is exactly why.
autrijus luqui was it.
castaway well thats something at least 14:54
clkao oh ok.
castaway hmm, and is there a code-ish way of telling whether $ref is a hashref or an arrayref?
clkao ref($ref) ? 14:55
autrijus ?eval my $x=(1,2); $x.ref
evalbot6 ::Array
autrijus ?eval my $x={1=>2}; $x.ref
evalbot6 ::Hash
castaway did you just say that gave Str etc?
Ok
clkao autrijus: so if i use pugs_apply for @a, and i want to apply &infix:<=>, @a[1], 123
how do i do that? 14:56
autrijus clkao: simple
cv = pugs_eval("{ @^x[1] = $^y }")
Limbic_Region PerlJam - the opaque vs transparent references has implication on the number of levels dereferences (1 vs Inf) 14:57
autrijus pugs_apply(cv, @a, 123)
clkao what what what?
Limbic_Region basically, it sounds like Luke sees both sides of the argument and wants both
autrijus clkao: use the ā˜Æ!
Limbic_Region but isn't sure how the syntax should work
clkao i don't see pugsS_eval is exported?
autrijus clkao: it's exported. 14:58
clkao ok.
i'll try tonight
autrijus extern Val *pugs_Eval ( char *code );
in pugsembed.h
so just use ā˜Æ to implement them
clkao ok ok
but what was the \$ thing.
it was passed in as the int value 14:59
kolibrie well, I posted my time formatting proposal to p6l -- hopefully no flames erupt
clkao my $rw = 1, p5_code(\$rw)
PerlJam Limbic_Region: it's too bad we can't say something like $ref.*{$foo} to mean infinite deref, while $ref{$foo} would be only one
clkao what's in p5_code is 1, not a pugs blessed thing 15:00
autrijus clkao: you need to declare the prototype is "is rw" I think
Limbic_Region PerlJam - I don't fully grasp the implications
If you feel you do and have a good idea for solving, please post to Luke's thread
autrijus clkao: I can do that for all methods
clkao: in the p5 side thatis
?eval role Class::Events::Publisher {};
evalbot6 undef
Limbic_Region I am sure that there are at least 1 pending decision by @larry that is going to mean a significant overhaul of pugs code
PerlJam Limbic_Region: nah, I'm straddling the fence right now and don't see what I would consider a good solution
Limbic_Region as autrijus has said in the past - thanks for having lots of tests 15:01
clkao autrijus: ok. let me know when it's done
autrijus clkao: so just method invocation for now?
jhorwitz autrijus: fully qualified subs have broken mod_pugs -- so now i'd like to extract the package & subname using breakOnGlue in Compile/Parrot.hs, but of course i'm running into monad issues, which still perplex me...
clkao autrijus: wait, if declareed as rw it's still not a refernece that p5 sees? because i need to tie it
autrijus clkao: if declared as rw, ref won't deref. 15:02
clkao: which is what you want I persume
clkao: i.e. $x = 1; perl5_sub($x)
is _always_ going to pass $x as ref
only perl5_sub(1) gets 1
as int
clkao what? gets 1 then it's not ref?
autrijus perl5_sub(1) # gets 1 15:03
$x = 1; perl5_sub($x) # gets VRef
of course 1 is not a ref
clkao haha.
autrijus I hope, at least
clkao ahah, i misread.
ok, but does mkvalref already seess that ref? 15:04
autrijus clkao: yeah
clkao good then.
autrijus jhorwitz: hrm, sorry. :-/ what monad issue?
Juerd autrijus: Do you know why pugs locks itself?
jhorwitz autrijus: sorry, got distracted...
autrijus Juerd: hm?
let (Just (package, name')) = breakOnGlue "::" name
that should do
Juerd autrijus: The automated upgrades broke, because evalbot kept ~juerd/auto/bin/pugs locked 15:05
jhorwitz autrijus: i've tried that, except for the "let". will try that now.
Juerd autrijus: And it seems every running pugs locks itself
autrijus "locked" ?
Juerd 11:36 < Juerd> cp: cannot create regular file `../bin/pugs': Text file busy
autrijus as in flocked?
Juerd no, a flock can be ignored...
autrijus try "cp -f"
Juerd I have no idea what kind of lock this was.
Never saw this error message before, and I wonder why pugs is a "text file" 15:06
I'll try
autrijus clkao: meanwhile focus on hash and array first
clkao: I'll get the scalar ref thing done
Juerd autrijus: That work 15:07
autrijus: That works
jhorwitz autrijus: that worked. i'm still not completely grokking this though -- why did the let work where "<-" didn't?
autrijus jhorwitz: because <- takes, as rhs, a compuatation 15:08
of type (Eval a)
and the lhs is of type (a)
but breakOnGlue has type
[a] -> [a] -> Maybe ([a], [a])
clkao autrijus: yes, i've had something. i ugess i can finish them tonight if the cv pugs_eval thing works
autrijus so no Eval is involved
clkao: I think it'd work
jhorwitz autrijus: getting it, thanks
autrijus :)
clkao: nice... I got SB and Class::DBI working with pugs today 15:09
if passing hashref can work I wouldn't need the helper modules anymore
clkao autrijus: oh ?? but export doesn't work yet. 15:10
autrijus: see the readme items.
autrijus clkao: both modules are OO
clkao: so export is irrelevant
(but yeah)
clkao alright
i've got plans for export.
need some help from you
autrijus cool. I'd love to hear about them and what I can do 15:11
but I need to take a nap now
stayed awake for far too long again
bbl :) &
clkao autrijus: hear about them = i wrote them in perl5/README 15:20
gaal how do i get runEvalIO env & opEval to not dump the AST? 16:15
and does the eval name argument have magic significance anywhere? 16:17
ninereasons msg evalbot6 eval 1+1 18:05
oops
Arathorn :o
gaal where's the place to put a util i use in building pugs, but is itself in haskell? till now things like this were put in util, but they were all perl5. 18:11
yay! namespaces were fixed! who did that? 18:13
okay, i have a working prelude on my box. it's still a little rought though. 18:15
seen autrijus
jabbot gaal: autrijus was seen 3 hours 3 minutes 57 seconds ago
integral is there a spec for pack in perl6 yet? 18:21
gaal i don't think so, integral.
integral pity :-)
gaal inline c code and use a struct? *duck*
integral inline perl5 ;-) 18:22
nothingmuch fucking beurocrats! i have so far found 10 pages telling me when to replace my "Travel Document in lieu of National Passport" but I still don't have the slightest bit of a clue what the fuck it is! 18:40
GAH! 18:41
gaal what you need, my friend, is an agent.
nothingmuch i have an agent, i just forgot to call her
today, that is
so i said "oh, i'll google instead"
Chewie[] Do Parrot's string functions support multi-byte characters? 18:42
gaal call your agent. it's like calling your mother, only it's more likely to get you a trip abroad.
Chewie[] If so, is there a way to take bytes out of a string individually? I'm trying to implement base64. 18:43
nothingmuch Chewie[]: i think so, since parrot has unicode, but we don't know much about it as a rule of thumb
#parrot on irc.perl.org is a better place to ask
Chewie[] ah.
nothingmuch afaik leo posted some string changes about 2-3 months ago
regarding the way strings are stored inline, i think 18:44
but i'm not sure
string register OTOH have
are supposed to have full unicode support
Chewie[] Hmm. 18:45
gaal anyone with Makefile-fu around? Any reason why Config.hs isn't unshifted to @srcfiles? 18:50
how do i locally turn off debug prints, the ones trifggered by Pugs.Eval.evaluate ? 18:57
.... or is that a leftover debug message someone forgot to remove? 18:58
eric256 what is with all the temp files created? why arn't they cleaned up with some sort of batch rm temp* after testing is done? 18:59
gaal are you on windows? 19:00
eric256 yea
gaal there's an unhappy condition there. it goes like this:
they are unlinked
but handles aren't closed at the right time yet
and on windows you can't delete an open file
so... junk. :( 19:01
eric256 why don't the tests explicitly close the handles? or is that the borked part you mean?
gaal however! i'm just about to check in standard pipe open, so you're welcome to refactor some of the tests not to use temp files at all :-)
eric256 Pipe.open? that was part of the namespace thing you where waiting for right? 19:02
gaal they do close on the p6 level, but apparently there's a problem deeper down. Ask Corion next time he's around.
yes. for now you need to call it Pipe::pipeopen (note the ::)
eric256 Pipe::pipeopen seems a bit redundant. why not juse Pipe::open? 19:03
gaal because namespaces are still a little bit flaky :)
it will be Pipe.open
clkao autrijus: bah, pugs_Eval isn't working properly. i think it evals twice or something weird. i'm going to do some svk refactoring tonight instead 19:08
gaal r4216. if the debug prints bother you, don't svn up, or better yet, make a patch :-) 19:09
./pugs -e 'File::openfile(<AUTHORS>).slurp.print' 19:11
./pugs -e 'Pipe::openpipe("pugs -v", :r).slurp.say' 19:16
hmm, the :r is required, i think it's a bug.
when you say sub foo (+$r is Bool), but call &foo w/o :r, $r 19:17
ESCdd
my bad, i should 'is copy'. 19:18
(fixed as 4217) 19:29
20:02 khisanth_ is now known as Khisanth
gaal is eval(Str $code, +$lang = "perl") the correct signature? 20:13
and: if it is, will this now break? is(eval 'some_code', "expected_result", "look ma, no parens") ? 20:16
Aankhen`` G'night. 20:32
Corion Shouldn't @foo[-1] work? It's a parsefail for me ... 20:35
gaal Hey Corion! :)
Corion Hi gaal
gaal I'd imagine it should.
Corion Then my error is elsewhere I guess :)
gaal say, does concurrent smoke work for you? i want it back :-) 20:36
Corion gaal: No, doesn't work for me, but I didn't try it
It says "Concurrent smoking disabled" or something like that
gaal ah :) does homegrown perl on w32 have fork? 20:37
it says that because i put in that message.
Corion gaal: It has fork emulation. Fork emulation only works in very specific cases.
eric256 wonders if concurrent smoking is anything like chain smoking 20:38
gaal hmm. where's the specific specified? :)
Corion eric256: Faster.
gaal: In the minds of the people programming the fork emulation.
And incidentially, the fork emulation only does what these programmers wanted/needed
gaal that is unfortunate. well, i'll try it, i guess.
Corion (sockets and filehandles don't really survive fork)
(pipes are also a bad idea) 20:39
gaal pipes? really!? beh 20:40
Corion I think that's what made me decide that parallel smoking won't be. But you're free to try. I'd recommend a pipe-free approach though. 20:41
gaal hmm, this debug message is annoying. i think i'll remove it.
Corion I think that open(F, "|-") mostly works.
gaal no, it forks now.
i'm sure open |- works, because Test::Harness depends on it on all platforms. 20:42
btw, the channel is curiously quiet today. 20:43
Corion Not my fault I hope 20:44
gaal well, you could try making more noise.
but it's funny considering autrijus said he was planning a release this local_morning 20:45
eric256 why does smoke testing require so many modules? 20:47
what ever happened to the KISS theory? ;)
gaal eric256, 'make test' is the kissy alternative
mj41 hello, 20:48
my $fh = open 'mjserver-dyn.p6'; my $code = slurp $fh; print $code;
Has slurp problem with long (>550 B) files?
Corion It shouldn't
eric256 yea. since i can't get the modules needed for smoke running i guess i'll settle for that
gaal seeing as "make smoke" outputs nice HTML, not having it use external modules would require inlining a templating tool, thus violating KISS badly 20:49
what can't you get?
Test::TAP::Model requires a new Test::More. it's s a known bug that it doesn't say so. 20:50
eric256 Test::TAP was the problem. 20:51
in my world KISS normaly refers to the end users experience not the programmers. ;) in this case endusers just happen to be programmers.
thanks that seems to have fixed the problem 20:53
mugwump morning
gaal hmm, nothingmuch about?
we really should update the module's META.yml and fix that already.
eric256 couldn't Test::TAP::Model be updated to show that as a requirment? or is that what you meant gaal? 20:54
mj41 perlbot nopaste 20:55
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
gaal yes, it is.
Corion Oy. Is the build broken? Or do I need a clean checkout ? 20:56
pasteling "mj41" at 147.229.221.107 pasted "Should anybody verify slurp bug (for file >550B) ?" (18 lines, 300B) at sial.org/pbot/10624
Corion I get lots of loading errors for (for example) ext\CGI\t\params.t 20:57
gaal paste, Corion? those coudl be just the debug messages i was looking to eliminate 20:58
pasteling "Corion" at 84.176.33.233 pasted "Debug message?" (4 lines, 96B) at sial.org/pbot/10625
Corion Yeah - it could be a debug message
Yeah - it's a debug message. That'll screw with some of my tests ;) 20:59
gaal yes, it's that. gimmie a minute, it's on its way out. 21:00
do you grok ControlEnv? 21:07
Corion Who, me? 21:08
I don't grok much Haskell
gaal np, i'll figure it out. 21:09
Corion BTW, I imagine reading earlier that autrijus wanted glob() in Pugs ... That'll get ugly again, because Win32 usually has spaces in filenames/directories, and Unix folks never consider that. 21:10
gaal well, at least now you can implement it in perl.
Corion (other than that, globbing is a problem of the shell - it should come as a module instead of a builtin)
gaal (see Prelude.pm -- it's loaded automatically now.) 21:11
Corion Ah
gaal www.livejournal.com/users/gaal/164263.html
Corion gaal: Heh - maybe I should have given you %ENV<PUGSOPT> for that. %ENV<PUGSOPT> = '-MPrelude' should do the same, but PUGSOPT doesn't exist yet :) 21:13
gaal i can't imagine why anyone wouldn't want a way to open files :) (much of prim is going there too) 21:14
Corion Yeah - it's cool to have the things that are easy in Perl6 directly available as Perl6 code :) 21:15
gaal r4219 21:16
i hope i didn't turn on debugging globally, but i'm sure i'll hear about it if i did. :-p 21:17
eric256 shouldn't that be class Pipe { sub openpipe } so that stuff can happily go after it?
gaal yes, but i think that still doesn't work.
see t/pugsbugs/namespace.t 21:18
eric256 stubs of what needs implemented would be good too... like an S29 outline in there....maybe S29 could be pod in that document. dunno
gaal (which need more tests!)
indeed! go for it, eric256.
also, there's a big job waiting to be done to p6ify eval_* 21:19
it's big because of t/.
sysinternals.com++ # tools that make w32 nicer to use 21:21
Corion Does ext\Perl-MetaModel\t\00-compile.t Just Run very long or does it hang? 21:40
Pugs has already taken 6 minutes running it I believe ...
gaal i'd forgotten to start a smoke... checking. 21:42
are you sure you're optimized? 21:43
eric256 has been running a smoke for a while now...
almost an hour in fact...that is insane
Corion gaal: "nmake && perl run_smoke.pl" 21:44
gaal if that's from clean, it should be optimized.
Corion I guess there is an infinite loop, as pugs.exe has amassed 10 minutes of CPU now. Killing it.
gaal but you may have a leftover unopt ast?
Corion gaal: I do a "make clean" before that always.
ext\Perl-MetaModel\t\10_Perl_Meta_MetaClass_test.t crashes after a few seconds. That's better than running forever :) 21:45
geoff__ Is there a doc or url that lists all of the packages a pugs user should have installed to enable everything that pugs can do? I must be missing something obvious, because that info seems very distributed. 21:47
eric256 you shouldn't need any modules installed for pugs to enable all the core stuff.
gaal you need parrot for p6 rules... 21:48
that's pretty core imo
Corion geoff__: Parrot and Perl5
geoff__ eric256, the kind of stuff gaal is mentioning.
gaal geoff, that's a good document to have. start one?
Corion The rest is included with Pugs from what I know. And GHC of course...
geoff__ gaal, I will as soon as I have pugs running the first time. :-)
Corion (and Parrot isn't as crucial yet :) )
gaal well, i'd say test::tap::htmlmatrix and dependencies (many!) for smokes, but that isn't required for pugs itself. 21:49
geoff__ My eventual goal is to port some of my SDL / OpenGL stuff to PUGS, so any stuff I might need to do that linkup would be cool to know about as well. Inline for P5, at least, yes? 21:50
gaal Corion, I'm worried a bit about w32, because fancy ghc stuff doesn't work there. have you looked at native ghc there? or is that as crazy as it sounds to me? :)
Corion ext\Set\t\overload.t is taking 2 minutes already ... Weird ...
gaal: I haven't compiled GHC from source there, but I'm considering trying to build GHC for MSVC7 there so that it uses cl.exe instead of gcc 21:51
21:52 Obadiah is now known as mml
gaal can anyone with a fast machine back to r4215 and see if the problems appear there too? that's before the Prelude patch 21:52
Corion I'm going to bed now, sorry :) 21:53
gaal night, Corion!
won't hurt if i did that too actually, $work tomorrow... 21:54
eric256 needs a faster machine to run smoke tests like this on....or concurrent testing
827 todo tests.. thats huge even out of 5763 tests. 21:56
gaal that's TDD for ya.
better 827 todo tests than 827 missing tests. 21:57
22:00 geoff__ is now known as geoffb
mrborisguy i've spent all day pouring over the subroutine portions of the apocolypse and synopsis, and i think this is right, but i'd love for somebody to comment on these: 22:04
multi sub split ( Str $delimiter , Str $input = $CALLER::_ ) returns List
multi sub split ( Rule $delimiter , Str $input = $CALLER::_ ) returns List
multi sub split ( Str $input : Str $delimiter ) returns List
multi sub split ( Str $input : Rule $delimiter ) returns List
&split<> := &split<Str>.assuming:delimiter(' ');
(for S29) 22:05
gaal i think you want Str ?$input = $CALLER::_
("?" cause it's optional)
mrborisguy oh yeah... yep, you're right.
gaal zzzZZzz & 22:06
clkao summon autrijus 22:08
eric256 does that actualy do anything? or do they setup there client to alert them when it sees that? 22:11
mrborisguy i was wondering the same thing 22:13
it reminds me of the movie "The Anchorman"... "News team, ASSEMBLE!!!!" 22:14
(and other vague references in my head)
eric256 summon eric256
yea i don't think it acutaly does anything
mrborisguy do anything?
eric256 that is a BAD movie 22:15
mrborisguy summon eric256
clkao
perlbot: status? 22:16
evalbot6: uc('SUMMON')
evalbot6: ?uc('SUMMON')
?eval uc('summon') 22:17
evalbot6 'SUMMON'
clkao evalbot6: $summon = {((uc(shift)~' ') x 4) ~ 'COME TO ME'}.('autrijus') 22:18
evalbot6: $summon = {((uc(shift)~' ') x 4) ~ 'COME TO ME'}; $summon.('autrijus')
?eval $summon = {((uc(shift)~' ') x 4) ~ 'COME TO ME'}; $summon.('autrijus')
evalbot6 Error: unexpected ( expecting expression
mrborisguy can somebody give me commit rights to perl6 so i can upload those split prototypes? 22:20
ninereasons ?eval my &summon = {((uc($^name)~' ') x 4) ~ 'COME TO clkao'}; summon('autrijus') 22:22
evalbot6 Error: unexpected ( expecting expression
ninereasons hm
clkao ?eval $x = { ( ( uc ( $_[0] ) ~ " ") x 4) ~ "COME TO ME"}; $x.('autrijus')
evalbot6 Error: Undeclared variable: "$x"
clkao ?eval my $x = { ( ( uc ( $_[0] ) ~ " ") x 4) ~ "COME TO ME"}; $x.('autrijus')
evalbot6 'AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS COME TO ME'
clkao ?eval my $x = { ( ( uc ( $_[0] ) ~ " ") x 10) ~ "COME TO ME"}; $x.('autrijus')
evalbot6 'AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS COME TO ME'
clkao ?eval my $summon = { ( ( uc ( $_[0] ) ~ " ") x 10) ~ "COME TO ME"}; 22:23
evalbot6 \sub {...}
clkao ?eval $summon.('autrijus')
evalbot6 Error: Undeclared variable: "$summon"
clkao hate
?eval $::summon = { ( ( uc ( $_[0] ) ~ " ") x 10) ~ "COME TO ME"}; 22:24
evalbot6 Error: Undeclared variable: "$summon"
ninereasons my $summon = {((uc($^name)~' ') x 4) ~ 'COME TO clkao'}; $summon.('autrijus')
?eval my $summon = {((uc($^name)~' ') x 4) ~ 'COME TO clkao'}; $summon.('autrijus')
evalbot6 Error: unexpected ( expecting expression
ninereasons ??
mrborisguy does anybody here have such authority to allow me to commit? 22:26
clkao you need to get a hold of one of: autrijus, bestian, hcchien, iblech, jesse, nothingmuch, Stevan, theorbtwo 22:27
mrborisguy clkao: thanks 22:28
clkao damnit, autrijus keeps commiting but not answering to summon 22:29
obra clkao: hi. I'm here
what's up? 22:30
clkao ?eval my $x = { ( ( uc ( $_[0] ) ~ " ") x (70/$_[0].chars)) ~ "COME TO ME"}; $x.('autrijus')
evalbot6 'AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS COME TO ME'
clkao ?eval my $x = { ( ( uc ( $_[0] ) ~ " ") x (70/$_[0].chars)) ~ "COME TO ME"}; $x.('perl6') 22:31
evalbot6 'PERL6 PERL6 PERL6 PERL6 PERL6 PERL6 PERL6 PERL6 PERL6 PERL6 PERL6 PERL6 PERL6 PERL6 COME TO ME'
mrborisguy clkao: 70/($_[0].chars + 1) because you add the space? 22:34
clkao right
?eval my $x = { ( ( uc ( $_[0] ) ~ " ") x (70/(1+$_[0].chars))) ~ "COME TO ME"}; $x.('chicks') 22:35
evalbot6 'CHICKS CHICKS CHICKS CHICKS CHICKS CHICKS CHICKS CHICKS CHICKS CHICKS COME TO ME'
clkao too many nails
obra mrborisguy: hang on
mrborisguy: has someone reviewed your patches?
mrborisguy i haven't patched anything 22:36
theorbtwo Somebody was looking for me?
mrborisguy and i just posted my split prototypes here
theorbtwo Ah, I see.
mrborisguy (or, i mean, i haven't submitted any patches)
theorbtwo mrborisguy: Email address?
mrborisguy mrborisguy at gmail dot com
theorbtwo mrborisguy: Invitation is on it's way. 22:41
mrborisguy theorbtwo: thanks 22:42
eric256 waits for someone to mod evalbot so you can bind commands to perl snippets ;)
where is the svn bot? 22:43
mugwump it's a bit slow, eric256 22:47
jabbot: seen svnbot6?
jabbot mugwump: svnbot6 was seen 8 hours 8 minutes 54 seconds ago
ninereasons perlbot metacommit?
perlbot ?eval my @meta= <autrijus bestian hcchien iblech jesse nothingmuch Stevan theorbtwo>; sub commit { return 'summon'~ ~@_ }; commit @meta;
autrijus_ g'morning
22:47 autrijus_ is now known as autrijus
clkao autrijus! 22:48
obra HAH
ninereasons hmm. experiment didn't work.
autrijus gaal++ # Prelude
clkao ?eval my $x = { ( ( uc ( $_[0] ) ~ " ") x (70/(1+$_[0].chars))) ~ "COME TO ME"}; $x.('autrijus')
theorbtwo Allo, autrijus!
autrijus yo
autrijus just woke up and relenging
clkao autrijus: quick, ifx pugs_Eval
mugwump <evalbot6> 'AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS COME TO ME' 22:49
autrijus clkao: how is it broken?
clkao autrijus: it's evaling twice
autrijus clkao: test case?
clkao: can you export it to p5 space?
like bind it to xsinit
as eg. pugs::_eval("...")
clkao hang on 22:50
i did
gaal wakes up, checks flist, finds this: code.google.com/summerofcode.html
evalbot6 'summonautrijus bestian hcchien iblech jesse nothingmuch Stevan theorbtwo'
'AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS AUTRIJUS COME TO ME'
clkao autrijus: 22:51
eval ({ $^x[0] })
VError "Insufficient number of required parameters: 0 actual, 1 expected" [(MkPos "t/unspecced/p5/method.t" 126
5 126 26)]
that's what gets into pugs_Eval
it works in raw pugs
gaal what does 'is primitive' signify?
autrijus gaal: it signifies that it should not be compiled into Parrot (yet) :) 22:52
other than that it means its subtype is SubPrim
so doesn't add to caller stack
and you can't use return() there
ninereasons perlbot metacommit?
perlbot ?eval my @meta= <autrijus bestian hcchien iblech jesse nothingmuch Stevan theorbtwo>; sub commit { return 'PERLBOT can speak to EVALBOT6: summon '~ ~@_ }; commit @meta;
evalbot6 'PERLBOT can speak to EVALBOT6: summon autrijus bestian hcchien iblech jesse nothingmuch Stevan theorbtwo'
theorbtwo ?eval '?eval "Hi!"' 22:53
evalbot6 '?eval "Hi!"'
gaal ah! great, i was wondering how to do stacktraces and things like that without messing up the stack.
autrijus :)
gaal i want to add java-like response to SIGQUIT by default
clkao autrijus: so, any idea? can you make sure the pugs_eval thing work in haskell land? cause it's not used elsewhere 22:54
gaal now we can add caller() that returns a real object
autrijus clkao: wait 22:56
clkao: 22:57
$ ./pugs -e "say eval_perl5('pugs::guts::eval(q[{$^x[0]}])')"
CODE(0x8577910)
worksforme.
please check r4227. 22:58
clkao orz
autrijus $ ./pugs -e "say eval_perl5('pugs::guts::eval(q[{$^x[0]}])->([9])')" 23:01
9
seems it's not my problem
gaal: er. 23:02
pugs::guts:: is a perl5 side thing
gaal oh. oops. :) merging back.
autrijus not sure if we want to be consistent with that... seems to be more confusing 23:03
clkao weird
autrijus please reverse :)
Limbic_Region bit OT, but if anyone is a student and wants to earn $4500 USD this summer - see code.google.com/summerofcode.html
clkao Could not find module `Pugs.Prelude':
gaal clkao, perl Makefile.PL and make again
autrijus, i should do 'svn merge -r4226:4227', right? not 100% used to this yet 23:05
autrijus gaal: if your change was 4227, do -r4227:4226
gaal ah :)
autrijus the bigger number comes first
gaal thanks.
autrijus :) 23:07
breakfast, bbiab &
clkao my bug. hate 23:17
it's when applying that eval result
now it worked! ship it!
autrijus: make things rw! 23:22
i'm almost done
where's svnbot? 23:24
gaal: what's these pugs:guts open stuff for?
how to make subs with placeholder varaible rw ? 23:27
can't find in s06
gaal clkao, my mistake, it isn't p5-related 23:28
clkao ok..
gaal it's the glue functions between haskell openFile and perl6 open 23:29
stevan seen nothingmuch 23:30
jabbot stevan: nothingmuch was seen 4 hours 45 minutes 50 seconds ago
stevan hmm
hey all :)
eric256 hello
clkao win42
gaal hey stevan :)
stevan hey gaal, how goes it?
mugwump do we have units in Pugs yet? 23:31
gaal very well! see src/perl6/Prelude.pm
stevan units?
mugwump eg, 1`m
gaal of all the p6 features, /me is scared of that the most
eric256 what?
stevan gaal: this is very very cool 23:32
autrijus where is units specced?
stevan this gets loaded into Pugs?
gaal stevan: yes
autrijus, i don't thiung they were finalized, but there was a debate on p6-l 23:33
autrijus stevan: got time to catch up changelog a bit?
stevan so we can start writing Perl 6 in Perl 6 now ?
clkao autrijus: rw func! arrayref access is complte, pending the rw function in assignment only
mugwump autrijus: remember that discussion about a superposition of all possible unit types for numbers?
autrijus clkao: k k. do hashref too
stevan autrijus: I only have about 20 minutes right now, but I can find some time later
gaal stevan: exactly :)
autrijus mugwump: yes?
stevan: thanks!
clkao autrijus: can't do hashref tonight.... drank champagne
gaal however, limitations apply.
stevan when is the release scheduled for?
autrijus heh
mugwump well, this is what I was referring to, so I thought you saw a post from one(@Larry) about it..
theorbtwo Units is unspecced, AFAIK. I'd like to see the assumption that they are an external module.
gaal 1. can'y use "return()" 23:34
clkao champagne++
autrijus stevan: any time in the next 14 or so hours
theorbtwo: <aol />
stevan autrijus: sounds good to me, I will update later tonight (about lunchtime for you)
clkao quick, let's make arrayref access work in this release
gaal 2. can't use more files, it's all inlined from Prelude.pm (but i can fix that)
autrijus stevan: cool
clkao next one we fix scalar crap (30% in ast?) and hashref
autrijus clkao: just go sleep :)
I need hashref so i'll make it happen. 23:35
clkao no, i want tosee it owrks
stevan gaal: why no return?
eric256 for units it would be easy to make postfix operators that converted a number into a Unit object. then you don't worry about it at all internaly..
clkao autrijus: make it rw, and i'll make hashref tomorrow
autrijus: see the current STORE code
theorbtwo I think it should be an infix operator ` on Any and Type.
clkao in perl5.c
gaal autrijus, can you take a look at Run.hs:157 and see if i turned off debugging correctly? i don't understand why this is local :-) 23:36
autrijus $$self?
ah. I see.
eric256 infix / postifx... doesn't matter much. just the fact that we will have such power with creating operators means somehting like units doesn't need to be part of the actual language
autrijus gaal: that's correct
clkao autrijus: also have you seen my plan for exportation in perl5/README? 23:37
eric256 as soon as coerce:<as> or infix:<+> dispatch correctly for classes that will be easy...hint hint. ;) /me would love to see operator overloading for classes work...
mugwump So, if I'm returning a number of days, then you'd expect that to be a Num and a Num Gregorian::Day
autrijus clkao: yeah.
clkao anyawy, i shall go sleep and expect everything to work when i get up, right, autrjus?
autrijus clkao: right.
clkao ok. good night 23:38
theorbtwo (Though I'm not sure how you spell type... possibly on Any and Unit, and then export functions with the name of units.
gaal stevan: colabti.de/irclogger/irclogger_log/...1302#l1954
autrijus nite!
gaal autrijus, but why is it correct? :) am i passing a copy of the env and modifying it on the fly?
autrijus gaal: you're passing a modifier function. 23:39
or rather
gaal also: woudl it be wiser to pull out the modification from initPrelude and put it in the caller, so that someone else could theoretically call initPrelude themselves?
autrijus you're passing in the env that was modified
gaal (with different debugging options)
autrijus sure, if you want that (although I don't quite see why), move the {envDebug} thing up to line 151 23:40
mugwump I want it to be a conjunctive type. One a plain num and one a higher order type. Num&::(Num of Date::Gregorian::Day) or something
Actually maybe auto-casting is a better way for that to work.
gaal back to my previous question, what causes the change to the env to be temporary? how come if beforehand envDebug was Just True, it would revert to that afterwards? 23:41
mugwump or maybe just being Num of Something matches the plain type Num
autrijus gaal: updates are never destructive. 23:42
env is not a IORef.
you can't change anything really. 23:43
so this
env{ envDebug = Nothing }
merely creates a clone of env
changes one field
lazily reusing all other fiends
fields, even, freudian slip
gaal ah, like OS environment, cute.
autrijus something like that, yeah.
bbiab again... 23:44

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!