Upgrade to YAML::Syck 0.85 before smoking | moritz.faui2k3.org/irclog/ | pugscode.org | sial.org/pbot/perl6 | ?eval [~] <m oo se> | We do Haskell, too | > reverse (show (scanl (*) 1 [1..] !! 4)) | "Perl 6 Today" video from YAPC::Asia: xrl.us/v6op
Set by agentzh on 1 May 2007.
00:09 edenc joined 00:18 ludan joined 00:29 REPLeffect joined 00:30 mj41_ joined 00:34 cmarcelo joined
cmarcelo moose! 00:34
lambdabot cmarcelo: You have 1 new message. '/msg lambdabot @messages' to read it.
cmarcelo @moosages
lambdabot fglock said 2d 4h 58m 16s ago: - pugs can use this data to emit either perl6 or haskell - sial.org/pbot/24669
00:35 Aankhen`` joined
pasteling "rhr" at 65.94.38.10 pasted "pugs build failure" (17 lines, 857B) at sial.org/pbot/24715 00:50
rhr "Setup: cannot satisfy dependency network-any" what's that?
nevermind, found it 00:53
svnbot6 r16191 | Darren_Duncan++ | ext/QDRDBMS/ : added methods to AST.pm and PhysType.pm
00:59 justatheory joined 01:00 [particle] joined
cmarcelo Last Thursday I gave audreyt++ Perl 6 Today talk in my University (took 2.5x time to cover 90% :-P). There was a friend of mine there who worked with LLVM and I told about mp6 and the possibility of making mp6 work over LLVM. Is there another place where LLVM could be useful? Maybe making PIL -> 'llvm high level language' conversion? Ideas? 01:18
(or should I convince him to 'go parrot' :o) 01:19
avar It would probably be best to convince him to keep going away from parrot:) 01:22
cmarcelo: Have you looked at the emitter part of the kp6/mp6 source?
anyway, it looks pretty easy to write a backend 01:24
I've been thinking of how neat it would be to write an emitter for common lisp. You could run it on sbcl, perl6 in machine code:) 01:25
dduncan terminology opinion question: if you were to take an arbitrary expression and derive a simpler expression that produces the same result, eg, changing the expression 5+2 into the expression 7, is that better called refactoring, or folding, or something else? 01:27
avar constant folding
dduncan what about expressions that contain variables 01:28
avar constant folding is genarally used for things you can fold at compile time regardless of whether the values are actually constants
cmarcelo avar: I'm looking, seems doable, maybe it's time to spread a commitbit :)
dduncan eg, ($a / 7) + ($b / 7) --> ($a + $b) / 7 mayhaps
avar perl5 for instance folds sub a () { 1 } sub b () { 2 } a + b to 3 01:29
someone give the man commit!
avar doesn't know how
dduncan I think I can do it ... have before ...
looking ...
avar dduncan: oh that's not constant folding, that's like turning div operations into multilication operations 01:30
I don't think there's a specific term for that. That would just be generally re-arranging the AST to get more speed out of it 01:31
dduncan I'll specifically tell you what I want to know about ... look at the pod for ext/QDRDBMS/lib/QDRDBMS/AST.pm
in a DESCRIPTION sub-section, I make references to rearranging an AST to something logically equivalent, but simpler
in those docs, I called it refactoring 01:32
cmarcelo invite.pugscode.org isn't up :(
dduncan but I'm looking at a slideshow now about Perl 5.10 features, and it calls something similar as folding
the refactoring in question is a compile-time action effectively
as in the general case it only uses info known from looking at the AST itself, and known language built-ins, but treats user-defined stuff as free variables 01:33
note that in the language the AST is for, system-defined and user-defined stuff have a separate namespace, and one can't override the other 01:34
avar cmarcelo: damn, you can still do a patch I guess..
patches--
dduncan so, eg, you don't have to worry about users redefining sys.add or sys.divide etc 01:35
anyway, about commit bits, looking ...
cmarcelo svn.pugscode.org:9999
dduncan who needs one anyway? post your email address here, which is how they are handed out
cmarcelo avar: I do have one, but my friend still doesn't.. 01:36
dduncan what's your friend's email?
01:36 REPLeffect joined
cmarcelo 9999 doesn't work either... 01:38
dduncan but anyway, as soon as I recall what site to go to, unless it is one of the above, I do have Pugs admin privs
for a long time 01:39
dduncan off to dinner
cmarcelo dduncan: I'll try login invite via feather to see if it works.. 01:40
dduncan okay 01:42
rhr dduncan: I think that's called strength reduction 01:48
01:49 stevan__ joined
dduncan is back 02:11
rhr, "strength reduction"? ... interesting, haven't heard that term before... will look it up 02:12
if you're referring to the fold/refactor question
02:18 stevan__ joined 02:19 staunch joined
rhr dduncan: not sure that's exactly correct, it's probably closer to what you mean than constant folding 02:25
e.g. replacing multiplies with shifts and adds is strength reduction 02:26
dduncan that's just an example ... what I want is more generic
rhr "optimization" :)
dduncan generically, its just the process of reforming an AST into another that means the same thing but is spelled differently 02:27
optimization is one primary reason to do it
another reason is canonicalization ... if an AST is canonicalized, then you can take 2 such ASTs and then structural equivalence can mean value equivalence 02:28
otherwise you can have different structures with the same values
mind you, having the same structure isn't essential for use of the AST in general, and so the core doesn't implement refactoring itself, but I wanted to have a name for if someone made a third party module to do so 02:29
optimize is good for a common case
rhr woohoo, the big machine at work does a make smoke in 6 mins. Smokin! 02:31
dduncan example of different being the same is "a+b" vs "b+a", assuming + is commutative
02:32 Kattana joined
dduncan on a separate matter, I'm wondering about the feasability of, right after 5.10 comes out, making a new CPAN module have Perl 5.10 as a minimal dependency, rather than 5.8.1 ... eg, are people likely to be migrating to 5.10 quickly? 02:34
as I recall, people using some Windows packages can't use 5.8.1 anyway as they get 5.8.0, and would next get 5.10.0
so for them it would be no change 02:35
s/Windows packages/Perl binaries for Windows/
avar people will generally upgrade along with their distro 02:36
dduncan so then, when the OS bundles 5.10, that's when most people will get it? 02:37
avar yup
very few build their own perl
dduncan in that case, I hope that 5.10 comes out early enough that October's public Mac OS X 10.5 release bundles it 02:38
so far, only Mac OS X 10.x.0 versions update the system perl
though 10.3.0 used 5.8.1rc3, so I suppose they may not wait for the final 5.10.0 02:39
thanks for that note
avar bundling a maint rc is way different than bundling blead 02:40
dduncan what about 10.0.0rcN ? 02:41
avar in that case they probably manually went over the bugfixing patches (I wouldhope)
mmmmayybe
dduncan afaik, the 5.8.6 bundled with OS X 10.4 includes some extra patches
or I certainly recall that an OS X security update patched the system perl 02:42
avar why do you care about osx perl in particular?
dduncan but it didn't increment the number
that's the system I use, and so do a lot of other people
mind you, I build my own Perls anyway, so it doesn't affect me so much
personally
so in that case, all I really care about is that most of the people who would use my module, whatever situation they are in, they could easily have 5.10 02:43
whatever their OS
speaking a little more broadly, I have separate 5.x and 6.x versions of my module anyway ... so since the 6.x one is meant to be the future one, I wonder if I should make the 5.x one support as many 5.x Perl versions as possible, or also limit it to an extent by requiring 5.10 02:45
my current thought is to stick with 5.8.1 for now, and change to 5.10 for requirements in 1 year
anyway, this isn't very important 02:46
rhr debian stable will probably not get 5.10 until the next release, likely >1yr away 02:47
dduncan okay 02:48
mind you, I don't actually need 5.10 for anything, that I'm aware of, but I know that 5.8.1 is a minimum since good Unicode support is important 02:49
but some of 5.10's features look nice, and could considerably affect my module's API if employed 02:50
but my module will be pre-1.0.0 for a couple years probably, so I can change the API before then
eg, if Field Hashes let me use arbitrary objects as hash keys, I might use that as part of my API rather than analagous input being arrays of binary arrays 02:51
otoh, Field Hashes may not actually do what I'm thinking of
Khisanth pseudo hashes done right? 02:56
02:59 elmex_ joined 03:07 audreyt_ joined
svnbot6 r16192 | rhr++ | Fix to t/xx-uncategorized/io_finalize_part(1|2).t 03:13
r16192 | rhr++ | Part 2 can fail when tests are run concurrently. This ensures part 2
r16192 | rhr++ | waits until part 1 is done before trying to read the file.
03:15 audreyt joined 03:31 mithraic joined 03:32 yumiko joined 04:10 dduncan joined 04:20 tokuhirom__ joined 04:22 nekokak_ joined 04:25 stevan_ joined
svnbot6 r16193 | rhr++ | Cleaner fix for t/xx-uncategorized/io_finalize_part2.t 04:25
04:54 SubStack joined 05:18 SubStack joined 05:25 jisom joined 06:09 dduncan left 06:18 dolmans joined 06:33 Aankh|Clone joined 06:50 Aankh|Clone is now known as Aankhen`` 06:58 the_dormant joined 07:05 Aankh|Clone joined, rindolf joined 07:10 agentzh joined
agentzh fglock: some random thoughts on PCR: 07:15
lambdabot agentzh: You have 1 new message. '/msg lambdabot @messages' to read it.
agentzh @messages 07:16
lambdabot TimToady said 1d 14h 3m 10s ago: the other problem I noticed is that the build_info line is excessively long, with embedded \n sequences that could probably be turned into <br> at some point
agentzh TimToady: i've noticed that too. it seems to be a problem on the YAML::Syck side (0.85). not sure. 07:17
fglock: i see a lot of code duplication in PCR's test suite. and i've been pondering the idea of refactoring some of them using ingy++'s Test::Base. 07:18
because PCR is already using Module::Install, it will be straightforward to bundle T::B into inc/ without adding an extra module dependency. 07:19
furthermore, Test::Base offers ---TODO, ---SKIP, ---ONLY, and ---LAST, which are very handy while debugging. 07:20
ingy :) 07:21
agentzh hi, ingy!
ingy hola agentzh
agentzh :) 07:22
i hope you are able to hack now.
ingy is Test::Bast getting ported to p6
I am
agentzh you're porting T::B to p6?
ingy no, are you?
agentzh not yet :) 07:23
ingy is hacking on YAML2.pm
agentzh wow
i'm talking about using T::B in PCR which is p5 module for p6 regexes.
ingy with libyaml support as first thing
agentzh so it will be as fast as libsyck?
ingy ok cool
i don't know
maybe faster 07:24
maybe slower
agentzh thinks libyaml is in C.
ingy yes, it is the best yaml parser emitter currently
agentzh so i'd expect it to be faster. :) 07:25
ingy well it' more correct
it was written from the spec
agentzh that's cool.
07:25 iblechbot joined
ingy for python, then ruby got a port 07:25
agentzh dislikes YAML incompatibility. 07:26
ingy nod
rindolf Hi ingy, agentzh
ingy the YAML trio just last night decided to start having regular voice meetings to fix the nits in the yaml world 07:27
agentzh hi, rindolf.
ingy hi rindolf
rindolf What's up?
ingy hackin
rindolf ingy: oh, that's good.
Ingy, I wrote a Flickr uploader, using YAML.
ingy :)
agentzh ingy: one complaint on Test::Base: 07:29
07:29 BooK_ joined
agentzh positioning of failing tests is not trivial. 07:29
because there is no line number in the TAP output. 07:30
currently i'm using === TEST 1, === TEST 2, to identify the tests.
of course, i'm also using a perl script which can index the tests in *.t automatically for me 07:31
rindolf And the good thing about my Flickr uploader is that I can just delete the few lines of the files I've already uploaded in case I'm interrupting it.
ingy agentzh: commits welcome :) 07:32
07:33 revdiablo joined
agentzh ingy: cool! 07:33
ingy: where's the commit bit?
;)
ingy send me an htpasswd entry 07:34
agentzh k, a sec
ingy svn.kwiki.org/kwiki/trunk/src/ingy/Test-Base/ 07:35
lambdabot Title: Revision 246: /trunk/src/ingy/Test-Base
agentzh agentz:fDIUPjY7ncjrs
ingy I moved most of my modules to just ship with Kwiki ;) 07:36
agentzh Kwiki++
ingy: why put META.yml under version control? isn't it automatically generated by Makefile.PL? 07:39
ingy agentzh: I'm moving away from that 07:40
is "agentz" a typo?
agentzh no
agentz == agent zhang
because agentz is not available here, i chose agentzh. 07:41
miyagawa ingy: tried ShipIt?
ingy miyagawa: no, what is it 07:42
agentzh: please test the commit bit 07:43
miyagawa it's an app to ship the software, like CPAN module
agentzh ingy: okay!
miyagawa now I can't live without it
ingy huh
i'll check it out
miyagawa ingy: it's on CPAN or code.sixapart.com/svn/shipit/trunk/ (the svn head supports M::I) 07:44
lambdabot Title: Revision 63: /trunk
agentzh ingy: 403
07:44 dmq joined
agentzh ingy: is the command correct? htpasswd -nb agentz password 07:45
ingy I use htpasswd -bc /dev/tty agentz password 07:46
agentzh ingy: try this again? "agentz:a5pwh4S5zJ27w" 07:47
ingy agentzh: ok, but I made a config mistake 07:48
agentzh committed :)
ingy it worked?
ok
good
agentzh yup
ingy trac.kwiki.org/trac/timeline 07:49
lambdabot Title: Timeline - Kwiki 2.0 - Trac
ingy :D
I gave you write access to src/ingy/ 07:50
agentzh cool, thanks :)
ingy miyagawa: tell brad to use YAML :P 07:51
miyagawa he hates all kinds of serialization :) 07:52
ingy so he invents his own 07:53
silly
rindolf Hi miyagawa 07:58
07:59 nekokak joined
miyagawa ingy: the default is good (re: shipit) 08:00
08:17 VanilleBert joined 08:18 penk joined
svnbot6 r16194 | agentz++ | util/testgraph.pl - patched smoke.html to expand the \n\t stuff requested by TimToady++ 08:32
agentzh TimToady: fixed.
08:33 the_dormant joined
agentzh runs a smoke to test it. 08:36
FYI, if you has got a smoke.yml, you can run the following command to generate smoke.html from it without wasting off your last smoke: 08:38
util/testgraph.pl smoke.yml smoke.html
*have
rindolf agentzh: are you Chinese?
agentzh rindolf: yes
rindolf: where are you from? 08:39
rindolf agentzh: I'm from Israel.
.il
agentzh: where do you live?
agentzh Jiangsu province, China. 08:40
rindolf agentzh: OK.
agentzh (in a small town)
hehe
rindolf Mainland China?
agentzh yup
rindolf agentzh: nice.
agentzh: OK.
08:42 pack|pizza joined
agentzh Hmm, it would be useful to convert Pod6 to Pod5. 08:42
and vice versa 08:44
fglock: also, it would be nice to add smartlinks to PCR's test suite. 08:49
linking to S05.
i'd like to compare the performance of PCR and P::RD :) 08:50
08:53 bernhard joined
rindolf agentzh: so do you have problems accessing the wikipedia? 08:55
agentzh rindolf: yes
rindolf agentzh: too bad. :-( 08:56
I can no longer imagine my life without the WP.
agentzh rindolf: well, i have very cool proxies :)
rindolf agentzh: yes, so I was told.
agentzh WP has some political entries which our government dislikes. 08:57
rindolf agentzh: sports.yahoo.com/olympics/news?slug...-chinglish - Chinglish. 08:58
lambdabot Title: The word is out in China: Speak better English - Olympics - Yahoo! Sports
rindolf agentzh: maybe.
08:59 REPLeffect joined
agentzh "* The .yml and .html files don't seem to correspond. Aborting." 09:21
sigh.
okay, finally fixed. 09:32
09:34 Aankhen`` joined
svnbot6 r16195 | agentz++ | updated smokeserv-client.pl accordingly 09:38
09:41 VanilleBert joined 09:42 ofer joined 09:48 kanru joined
agentzh iblech++ seems to be very busy lately. 09:50
problems on the server-side of the smokeserver cannot be fixed while those on client-side can. :)
agentzh goes to run another smoke. 09:51
yay for smoke cycle of 6.8 min! :D 09:53
09:54 elmex joined 10:05 VanilleBert left 10:07 kanru joined 10:11 dolmans joined 10:15 ludan joined 10:20 ajs_ joined 10:22 b_jonas joined
svnbot6 r16196 | agentz++ | made 01-smartlinks.t more portable 10:22
r16197 | agentz++ | smartlinks.pl - now we require Perl6::Perldoc 0.0.5 10:31
10:32 larsen_ joined
ingy agentzh: ping 10:32
agentzh: do you know of some sane xs modules that i can study? 10:38
10:57 mj41 joined
rindolf ingy: maybe Math::GMP? 10:57
agentzh ingy: ? 10:59
agentzh believes ingy is the author of Inline::C which does a lot of XS magic. 11:01
svnbot6 r16198 | agentz++ | smartlinks.pl - made File::Slurp optional 11:14
11:18 riffraff joined
riffraff hi everyone 11:19
11:19 mithraic joined
agentzh hi, riffraff 11:19
riffraff dummy question: am I right in thinking that substr($str,$n,1,'') should delete the n-th character from $str? 11:20
agentzh as long as you're using 1 as the base.
index base, i mean 11:21
riffraff wow
11:21 TimToady joined
riffraff why not 0? 11:21
agentzh riffraff: sorry, i may be wrong. why not try that out in your perl/pugs? 11:22
riffraff because I think it is broken in pugs, but I wanted to check :)
given $str='ciao' and $n=0
it returns 'c' as expected, but $str is unmodified 11:23
agentzh i think it shouldn't modify $str.
riffraff but it does in perl5, doesn't it?
agentzh indeed. 11:24
riffraff ok 11:25
agentzh $ perl -e '$a='abc';substr($a,3,1,"");print "$a\n"'
abc
p5 doesn't delete the nth character it seems.
riffraff then the question becomes: how do I remove char $n from $str? I'd use s/<at($n).// buit it won't work with current Rule implementation
this is strange, I was basing my expectation on www.perlmeme.org/howtos/perlfunc/substr.html 11:26
lambdabot Title: Using the Perl substr() function
agentzh sorry, i was assuming your n is the length of the string.
11:26 TimToady joined
riffraff agentzh, wait, try 2 instead of 3 11:26
agentzh right. 11:27
riffraff eh exactly
agentzh and it removes the n+1th character.
riffraff well I think strings are 0-indexed so for me it removes exactly the nth char :)
polettix the whole point of the fourth parameter is the substitution of the substr in the original string, IIRC 11:28
riffraff yep, I'd prefer an interface where substr does'nt do side effect and returns two values $modifed, $replacement
agentzh riffraff: removing the nth char is easy with regexes.
riffraff agentzh, eh, try in pugs :)
agentzh (if not using substr)
polettix riffraff: I actually like the current interface. You can choose to use the three-args one, which has no side effect. 11:29
b_jonas I like the current interface as well
riffraff I think it depends in the fact tha yoiu're accustomed to that while I have an habit with different ones 11:30
agentzh s/(.**{n-1,n-1})./$1/
not sure if it's the correct syntax.
riffraff agentzh, I expect you mean $n in the first part.. won't work, can't interpolate in regex in current pugs
polettix I have to test, but I think that something like this my $extracted = substr(my $residual = $orig, $n, $howmany, ''); should work as well
agentzh riffraff: oh, pity 11:31
riffraff polettix, (my $tmp = $orig).substr($n,1,$rep) works
polettix riffraff: indeed, it probably boils down to what one is used to :)
riffraff well, it would if pugs did side effect :)
b_jonas I think it's good because if you're used to side-effectless programming, you can use the three-arg form 11:32
while if you want to modify the string in-place without copying, you can use the four-arg form
riffraff b_jonas, but it doesn't provide the same feature
b_jonas that's true 11:33
riffraff I mean, once I know what is the meaning of substr I'm fine with it, I don't complain
polettix riffraff: (my $tmp = $orig).substr($n,1,$rep) sounds very nice imho
agentzh riffraff: maybe hack that into pugs?
riffraff polettix, yep. $orig.substr($n,1,$rep) and $orig.=($n,1,$rep) seem better to me :)
but again, just habit
ingy agentzh: Inline::C was 7 years ago man 11:34
agentzh ingy: okay
ingy :)
polettix riffraff: I don't get the second
11:35 Belaf_ joined
riffraff if substr was side effect free, that would be the side-effect version 11:36
polettix $orig.=($n,1,$rep) isn't meant to append something to $orig? 11:37
ah no, the .
riffraff ah yea there is a substr missing 11:38
11:38 jrockway joined
riffraff $orig.=substr($n,1,$rep) 11:38
polettix riffraff: so what should $orig.substr($n,1,$rep) do?
riffraff I love the .= thingy :)
return a copy and not modify $orig
11:39 evalj joined
polettix riffraff: so why the $rep? 11:39
b_jonas it's the replacement
riffraff well a _modified_ copy :) 11:40
polettix yes, but if the orig is not modified, why the rep?
riffraff because you modify the copy
i.e.: 'hello'.substr(0,1,'f') #-> returns "fello"
polettix riffraff: sorry, I don't fully get the semantic. The $rep in perl 5 is supposed to modify the part you're extracting, so in your non-modifing version it should simply return $rep 11:41
uh, now I see
so it's more a subst than a substr :)
riffraff yep
polettix I don't like it, it completely changes the semantic 11:43
I mean, 'hello'.substr(0,1) # returns 'h', right?
riffraff maybe
:)
ah 11:45
substr($s,$n,1)= 'foo' works as expected 11:46
nice
polettix, yep, returns the fragment
that I'd return as a second value
but nevermind I don't plan to change the perl6 spec ;)
polettix riffraff: as long as what comes out is consistent and easy to remember, without surprises I mean, it's good IMHO 11:47
11:48 penk joined
svnbot6 r16199 | agentz++ | [util/smartlinks.pl] 11:59
r16199 | agentz++ | * refactored the duplicated code into sub gen_preamble
r16199 | agentz++ | [util/t/01-smartlinks.t]
r16199 | agentz++ | * trimed trailing spaces before diff'ing
12:12 iblechbot joined
agentzh iblech: ping 12:13
iblechbot: tell iblech to update the stuff on smoke server.
weird. there's still no interface for specifying external css file in the latest Perl6::Perldoc::To::Xhtml :( 12:15
hmm, S26.pod6 is not in UTF-8, it seems. 12:21
svnbot6 r16200 | agentz++ | [smartlinks.pl] 12:23
r16200 | agentz++ | * used Perl6::Perldoc::To::Xhtml's full-doc feature
r16200 | agentz++ | * now we don't set UTF-8 for S26 since it's in Latin-1
12:25 mithraic joined
svnbot6 r16201 | agentz++ | smartlinks.pl - fixed various typos in the POD 12:35
12:35 stevan_ joined
svnbot6 r16202 | agentz++ | smartlinks.pl - added note that it's still under active development 12:38
12:44 buetow joined 12:50 prism joined 12:55 prism joined 12:59 chris2 joined 13:29 xinming joined 13:32 fglock joined
xinming -rd syn 13:36
fglock agentzh: re PCR tests - I prefer to keep the tests in plain perl5, because I frequently need to insert debugging code
xinming sorry wrong channel.
13:36 buetow joined
fglock re PCR performance - it depends on which backend you are testing; there is a plain-perl5regex backend that is very fast, and the original lrep backend which is very slow 13:37
I've been replacing the slow parts one at a time 13:39
agentzh fglock: it's still very easy to insert debugging logic in the context of T::B. 13:45
(and i believe it can be *easier* :))
i'm not suggesting converting all the existing tests to T::B.
i just think it's good to organize some new tests of regular forms via T::B 13:46
fglock: is it okay to add some new tests in T::B format to serve as some examples? 13:49
fglock: i think you will like it :)
fglock agentzh: sure - please note that the actual specification-compliance tests are run with v6.pm 13:50
from pugs/t/
agentzh fglock: yes, i know. 13:51
fglock: but i'm more interested in PCR at this moment :)
PCR is much more practical for real-world apps. 13:52
fglock I'm doing some refactorings right now, in order to run Perl6-STD as the v6.pm grammar
agentzh fglock: i'm willing to offer tests to PCR if you don't mind :) 13:53
i won't touch the old ones.
i'll only add new ones.
fglock agentzh: sure, please do :) 13:54
agentzh fglock: thanks :)
fglock I'm implementing non-greedy quantifiers in ratchet mode, using 'forward checking' 13:56
re tests - there are some corner cases, in which I'm not sure what is the correct answer 14:00
agentzh *nod* 14:01
i'll look at PGE's test suite to see if i can import them into PCR's format.
as well as pugs/t 14:02
fglock agentzh: it is already ported to pugs/t/, isn't it?
agentzh fglock: ah 14:03
cool
agentzh has a lot of stuff to read. 14:04
now that i'm able to use pod6, i'd like to use p6 regexes *today* :) 14:05
fglock agentzh: take a look in www.perl.org.br/bin/viewfile/Artigo...ure.pl.txt 14:14
lambdabot tinyurl.com/ywhau2
agentzh looking 14:15
ah...a game using v6.pm? 14:16
hehe
fglock notice the '#!/usr/bin/perl' :)
agentzh yes, i've noticed that :)
it looks very nice :) 14:20
14:24 Psyche^ joined
moritz rhr: if you haven't fixed you build failure, install libghc6-network-dev 14:30
14:30 ozo joined 14:32 kanru joined 14:33 fglock joined 14:37 mithraic joined
Yaakov buu 14:41
14:48 Psyche^ joined, Psyche^ is now known as Patterner 15:15 intelin joined, intelin left 15:17 riffraff joined
riffraff hi 15:17
I notice that some tests marked as TODO pass 15:20
shall I remove the todo<feature> thing and make them real?
15:21 f0rth joined 15:24 torz joined 15:30 ruoso joined 15:33 avar joined 15:38 justatheory joined 16:04 ruoso joined
masak riffraff: sounds like a good idea to me -- unless it's somehow evident from the tests that the feature isn't really implemented and that the tests pass anyway 16:28
in the latter case, I think I would actually advise removing the tests, or at least think hard about why they're there 16:29
16:30 drupek12 joined, rindolf joined
riffraff good point 16:32
point is evidence is not so evident
:)
maybe I should write another test for the same feature, check that it pass, then remove the todo one and commit
agentzh oh, my god... 16:38
pugs installs p6 modules into my p5's lib and destroys everything. 16:39
integral The install is untested and no one uses it... 16:40
agentzh i've just used it.
i think "make install" worked fine on Win32 in the last year.
after trying this command on my linux tonight...
heart-burn... 16:41
agentzh looks for a way to cancel the damage.
riffraff ok, I really believe that t/builtins/want.t's first 5 test really work 16:43
probably because now caller is implemented correctly
or, well, better if not correctly
16:43 kattana_ joined 16:44 baest joined 16:45 ruoso joined, torz joined, chris2 joined, stevan_ joined, jrockway joined, b_jonas joined, tokuhirom__ joined, Kattana joined, cognominal joined, yves_ joined, JarJarBinks joined, theorbtwo joined, tewk joined, spinclad joined, zamolxes joined, scw_ joined, gnuvince joined, wolverian joined, Juerd joined, Eidolos joined, cj joined, nnunley joined 16:47 tewk_ joined, cj_ joined
agentzh okay, i've reproduced exactly the same error message got by jerrylee when he tried out mp6-perl5-boot.pl with his activeperl 5.8.8 b820 on my ubuntu system. 16:47
"Can't locate v6.pm in @INC (@INC contains: /etc/perl /usr/local/lib/perl/5.8.8 /..." 16:48
someone must have changed the 'make install' logic :(
16:48 sreeram joined 16:57 amnesiac joined
riffraff miserably fails to understand where "ant" comes from 16:58
(want)
16:59 amnesiac_ joined
riffraff is there a smarter way of doing (0..$word.chars-1) .map { dostuff($word,$char_index) 17:00
?
17:01 dmpk2k joined
riffraff basically I need something like "n times" but I don't know how to do that 17:02
b_jonas well, ^$word.chars or even ^$word might work instead of the range
riffraff what is the ^ 17:04
?
I can see it generates a useful list of elements :) 17:05
even if they are floats, for some reason, I wonder what is the real meaning of ^ ? 17:06
rindolf Hi riffraff
Hi b_jonas
riffraff hi rindolf
17:06 iblechbot joined 17:07 Psyche^ joined
b_jonas hi 17:07
well, ^ is the iota operator according to the Synopsis
agentzh Don't run "make install" for Pugs! | moritz.faui2k3.org/irclog/ | pugscode.org | sial.org/pbot/perl6 | ?eval [~] <m oo se> | We do Haskell, too | > reverse (show (scanl (*) 1 [1..] !! 4)) | "Perl 6 Today" video from YAPC::Asia: xrl.us/v6op 17:07
b_jonas that is, ^6 is supposed to generate the list 0, 1, 2, 3, 4, 5 or something like that 17:07
riffraff doesn0t really look like a iota
but which synopse?
b_jonas does -> look like lambda
the operators one
riffraff very true 17:08
b_jonas that's S03
"prefix:<^>, upto operator" 17:09
"Constructs a range or locates a metaclass. See "Range semantics"."
riffraff very nice thanks 17:10
17:10 marmic joined
rindolf riffraff: how's life treating you? 17:11
agentzh how can i reinstall modules under /usr/local/share/perl/5.8.8/ via debian please? 17:12
riffraff good wuestion 17:13
i'm actually torn. I feel fine but I have this stupid sneeze
and tomorrow I will start my last period in the university, which is bad bad not so bad, cause I think I have one nice class 17:14
agentzh really wants to cry... 17:15
i'm removing things...
riffraff agentzh, I did: remove perl, reinstall perl, 17:16
probably --reinstall would work
17:16 nperez joined
agentzh riffraff: good idea 17:16
removing perl will also remove hundreds of other debian packages... 17:18
riffraff yes that'0s why reinstall should work
agentzh --reinstall doesn't work for me.
riffraff mh
that's bad
agentzh *nod*
riffraff I'm quite sure I did that
agentzh pretty bad.
636 packages will be removed if i remove perl. :) 17:19
riffraff tr gpind down to the dpkg level maybe 17:23
17:24 Psyche^ is now known as Patterner
agentzh riffraff: will try 17:24
working now 17:25
after removing the problematic modules from /usr/local/share/perl/5.8.8/
interesting 17:26
riffraff mh
that would explain why I have a 5.8.8 dir in the pugs working directory
17:29 unobe joined 17:32 b_jonas_ joined, b_jonas_ is now known as b_jonas
unobe hi agentzh 17:33
i changed the make install code a bit about a month ago
agentzh unobe: hello
unobe: really?
agentzh is still in a big pain.
unobe it was supposed to add more config variables to pugs when running
when moritz built w/debian, he said it still worked for him 17:34
and so i thought i didn't royally screw up
agentzh but now my perl is totally broken.
and i believe jerrylee and laye sufferred from similar problems too. 17:35
unobe there should be a file called current.build.yml
in the pugs build dir
giving the build configuration for pugs 17:36
agentzh looking
unobe: which installation path is for p6 modules? 17:37
unobe i believe it's sitelib
but i'm not on a debian system
agentzh sitelib: /usr/local/share/perl/5.8.8
my god...
unobe the subdir debian
it has the debian config options no? or is that just for bundling? 17:38
agentzh ?
unobe nm, that's for building a dpkg 17:39
agentzh unobe: want to have a look at my .current.build.yml?
unobe sure
paste?
pasteling "agentzh" at 221.131.66.145 pasted "my .current.build.yml" (42 lines, 1.2K) at sial.org/pbot/24724
riffraff uff, I want gather-less take for expressione with modifiers 17:42
is that already ruled out from decades?
17:44 bernhard joined
unobe agentzh: what's the sitelib for debian supposed to be? 17:54
agentzh i think that's for p5 modules, not p6.
unobe what did perl -MConfig -e 'print $Config{sitelib}, "\n"' say?
yeah, but the Makefile for pugs uses the perl5 config vars
util/PugsConfig.pm
agentzh /usr/local/share/perl/5.8.8 17:55
unobe on my machine, i always remember pugs installing into the sitelib of Perl5 17:56
agentzh but p6 versions of File::Find, Test::Simple, Getopt::Std will lead to a disaster for perl 5. 17:57
unobe right, but i believe they get installed in their own subdir, perl6 under the sitelib
agentzh sitelib is in perl 5's @INC, no? 17:58
unobe yes, it is 17:59
agentzh my perl 5 is no longer usable now.
unobe but i thought that pugs created subdir under the sitelib which perl5 was installed into during installation 18:00
agentzh it's keeping complaining v6.pm is not found or something like that
unobe: but it didn't.
unobe i'm untarring the pugs i bundled and seeing how it would be installed 18:02
agentzh i've just removed /usr/local/share/perl/5.8.8 with anger. now i'm reinstalling everything related to perl. alas. 18:03
18:06 kanru joined
unobe i'm sorry for totally fucking up your system, agentzh 18:07
agentzh unobe: it may not be your fault ;)
avar so how about fixing `make install' ?:) 18:08
agentzh won't try to do that. 18:09
it's dangerous :)
especially when testing it.
18:13 SamB joined
unobe i found the problem 18:18
it's in util/PugsConfig.pm
agentzh unobe++
unobe it was my fault :-/
agentzh :)
agentzh needs to do regular backup for his own machine. :) 18:19
unobe I'm svk up'ing and then I'll submit
agentzh cool 18:20
Think twice before running "make install" for Pugs | moritz.faui2k3.org/irclog/ | pugscode.org | sial.org/pbot/perl6 | ?eval [~] <m oo se> | We do Haskell, too | > reverse (show (scanl (*) 1 [1..] !! 4)) | "Perl 6 Today" video from YAPC::Asia: xrl.us/v6op 18:20
unobe should put "Think thrice before changing 'make install'" 18:22
agentzh lol 18:23
18:23 jrockway joined
agentzh end of day for me & 18:26
18:26 agentzh left 18:29 amnesiac_ is now known as amnesiac 18:41 ruoso joined
svnbot6 r16203 | agentz++ | PCR - added a smartlink to 09-ratchet.t 18:41
r16205 | unobe++ | fixed 'make install' problems 18:44
unobe @tell agentzh the current.build.yml should now show good paths to install pugs into. 18:45
lambdabot Consider it noted.
Tene notes that 'find /usr/lib/perl5 -cmin -30 -print0|xargs -0 sudo rm -f' is mostly effective at cleaning up after a broken 'make install' 18:51
19:02 ruoso joined
riffraff is it normal thgat doing make smoke-upload takes ages? 19:04
Khisanth does make smoke normally take a long time for you? (it does for me ...) 19:07
or rather it did, I haven't tried it recently
maybe one or two releases ago
svnbot6 r16206 | agentz++ | PCR - more smartlinks to 09-ratchet.t
riffraff no I mean the smoke.html is built 19:14
I just can't upload :)
19:14 lichtkind joined
svnbot6 r16207 | agentz++ | PCR - trimmed trailing spaces in 09-ratchet.t and added several more smartlinks 19:31
19:37 nperez joined
diakopter agentz: working past 'end of day'? ;) 19:39
19:41 amnesiac joined 19:43 penk joined, [particle] joined 19:44 jisom joined 19:47 crashmatrix joined 19:51 araujo joined
svnbot6 r16208 | agentz++ | PCR - 09-ratchet.t - even more smartlinks 19:51
19:53 benny joined, sapir joined 19:55 ruoso_ joined 19:56 explorer joined 19:57 ruoso joined
svnbot6 r16209 | agentz++ | PCR - minor cleanup in 09-ratchet.t 20:06
20:19 REPLeffect joined 20:24 REPLeffect_ joined 20:34 Psyche^ joined 20:35 _bernhard joined 20:41 lichtkind joined
rhr anyone know why ext/DateTime/t/basic.t would start failing since yesterday? 20:43
pugs> my Date $date = Date.new( :string<now> )
*** No such subroutine: &Date
that worked last night...
whoops, forgot to use Date;... but, having fixed that, $date.year is undef and $date.month and $date.day are 1 20:49
20:50 ruoso joined, Psyche^ is now known as Patterner 20:51 jisom joined 21:04 SubStack joined 21:09 ruoso joined 21:14 Aankhen`` joined
rhr $date.year should be set to localtime(time).year, which correctly returns 2007, but it's undef 21:19
and apparently it works on agentzh's machine 21:20
it looks like the wrong BUILD multi is being used, but why would it do that? 21:23
21:25 ozo joined
rhr aha, I smell a ::Rat. I still don't understand why it worked yesterday and not today... 21:27
pasteling "rhr" at 65.94.38.10 pasted "strange Date.pm behavior" (8 lines, 237B) at sial.org/pbot/24733 21:35
rhr how could $epoch be undecalred? and why does it work the second time?
21:41 explorer joined
pasteling "rhr" at 65.94.38.10 pasted "similar but different weirdness after a make clean" (13 lines, 222B) at sial.org/pbot/24734 21:53
rhr it's calling the wrong BUILD multi the first time, but the right one the second time
I give up for now 21:54
22:04 larsen_ joined 22:06 weinig joined 22:14 pbuetow joined 22:44 mithraic joined 22:53 DarkWolf84 joined 22:59 ticallion joined 23:26 offby1` joined
Yaakov buu 23:28
23:31 offby1` is now known as offby1