Check your feather email | spec.pugscode.org | paste: sial.org/pbot/perl6 | pugs.blogs.com
Set by TreyHarris on 9 October 2006.
stevan but like I said,.. I am not a Java programmer so I very well may have no idea what I am talking about ;) 00:00
but right now it's time to go play Lego StarWars with my daughter :)
Limbic_Region stevan - I am optimizing for runtime speed and calling constructors/destructors fewer times is the win I am looking for 00:02
I had considered String btw
00:10 mako132_ joined
TreyHarris so someone enlighten me--what's "duck typing" if not dispatch purely by method name? 01:11
TreyHarris has returned from a couple days battling a nasty flu 01:12
01:24 mako132_ joined 01:27 buubot joined 01:35 Limbic_Region joined, pomi1 joined 01:49 Aankh|Clone joined 01:59 mako132_ joined 02:08 nothingmuch joined 02:13 pomi1 left 02:57 [particle] joined 03:24 SubStack joined 03:41 root4o joined
root4o hi 03:41
if I follow installation procedure
in Pugs::Doc::Hack
it will install pugs on its own dir... not in my distro dirs.. i.e. /usr... 03:42
03:42 mdiep joined
root4o what's the difference between : use v6 AND use v6-alpha 04:04
anybody here? 04:11
04:21 root4o left 05:12 rick_ joined 05:15 BooK_ joined 05:23 rick_ joined 05:32 penk joined 05:34 Pomin8654 joined 05:51 scsibug joined 05:59 Pomin1929 joined 06:02 Pomin1929 joined 06:20 justatheory joined 06:41 root4o joined
root4o anyone here ? 06:41
06:41 Pomin1929 joined 06:43 Pomin1929 joined 06:48 kcwu joined 06:56 Pominx joined 07:00 kcwu joined 07:01 Psyche^ joined 07:07 Pominx joined 07:12 Psyche^ is now known as Patterner, root4o left 07:22 diotalevi_ joined 08:15 Pominx joined 08:17 iblechbot joined 08:33 Pominx joined 08:46 buetow joined 09:05 buetow joined 09:06 xinming joined 09:50 buetow joined 10:18 chris2 joined 10:19 buetow joined 10:49 explorer joined, elmex joined 11:01 pmurias joined
pmurias hi root4o, 11:01
use v6 requires perl 6.0.0, use v6-alpha is trickery to use the v6.pm module
he left 7 hours before :( 11:02
time--
well maybe he backlogs 11:03
:) 11:06
11:27 foo\ joined 11:51 stennie joined 11:53 spoop joined 12:25 shillo joined
shillo Hullo all! 12:27
Just wanted to say (when anyone is awake) that I was testing lazy lists, and it seems that (1..Inf).map:{ $_/2 }[0] runs itself into infinite loop in the current SVN pugs. 12:29
(and I'd assume that the lazy list test times out or something)
12:41 stennie joined 12:56 Psyche^ joined 13:00 weinig is now known as weinigbbl 13:11 Psyche^ is now known as Patterner 13:22 Limbic_Region joined 13:32 swaroop joined 13:34 snowstalker joined, snowstalker left 13:36 snowstalker joined
Limbic_Region salutaitons all 13:40
pmurias Limbic_Region: hi 14:07
i looked at the perl5 script you were porting to java and it seems rather bloated 14:08
Limbic_Region pmurias - the Perl or the java? 14:10
pmurias the perl5 one
Limbic_Region the perl was something I whipped up in a jiffy to explain the functional equivalence of what I was trying to accomplish
it was not optimized 14:11
but since I don't know what you mean by bloat, I will need more context to comment further
pmurias i mean code size 14:12
Limbic_Region *shrug* - enough to get the point across 14:13
in any case - my Java version works great except I run out of heap
so now I have to make the code even more complex to help manage that
pmurias i didn't realise it was a java backport 14:14
it was quite fun striping it down though
i reduced it to 26,20,13 lines changeing the algorithm and removeing strictness in the latter case 14:16
how do you like java?
s/changeing/changing/ 14:17
the java version seems rather advanced, you must be a very fast lerner 14:21
svnbot6 r14335 | audreyt++ | * prior.t: unTODO. 14:24
14:28 plural_ joined
Limbic_Region pmurias - not really a fast learner 14:31
I took a 4 day professional course
I think Java isn't nearly as evil as some would have us believe
I think I found as many things to like about the language as I did to not like 14:32
so I came out really neutral
14:36 foo\bar joined
pmurias catching typos at compile time is propably the only thing i like about the language 14:37
i would propably use Scala if i wanted to target the JVM 14:39
Limbic_Region *shrug* - different strokes for different folks 14:42
in any case, it was a 3 language project that should have just been one - Perl 6 14:43
Limbic_Region can't wait for Xmas
pmurias what where the two other languages? 14:47
Limbic_Region Perl, C (via Inline::C) and Java
the bulk in Perl of course 14:48
pmurias that's good :)
Limbic_Region I was hoping to have a working online demo of my results by Monday but I forgot to bring home my data files
SamB oops! 14:49
pmurias are you using your power-set script at work?
Limbic_Region no, I was playing at work 14:50
Limbic_Region is allowed to work on personal projects that expand his knowledge that is applicable for real work related tasks
SamB cool
Limbic_Region yep 14:51
and since I am an "Information Technology Specialist", just about anything falls into that category
Limbic_Region checked the IRC logs and is amazed how quiet it has been 14:52
is there a conference or something going on?
SamB heh
pmurias Limbic_Region: what is the upper string-size limit the java script can handle?
Limbic_Region pmurias - don't know as my longest string is 26 chars long 14:54
the issue is that the hashset works across all strings
and there are millions of those 14:55
so I run out of heap
Limbic_Region knows how to get around this but has yet to make the necessary modifications
came to me as I woke up this morning
14:56 norageek joined
Limbic_Region basically I am going to make the hashset a HoH in Perl terms - where level 1 is the length of the string 14:56
once all possible strings for that level have been seen - gc it 14:57
I just then need to remember not to recurse below that depth
and provided I process my millions of strings based of increasing length I should be fine
pmurias pmurias is planing a none recursing version 14:58
Limbic_Region pmurias - good luck, I tried very unsuccessfully to make the optimizations with an iterative version 14:59
myself and many of the monks at the Monastery came to the conclusion it was impossible
pmurias - see perlmonks.org/?node_id=576101 for details 15:00
lambdabot Title: Powerset short-circuit optimization
Limbic_Region fwiw, I have written extremely fast and memory efficient powerset iterators in the past 15:01
I have even written such iterators that were possible to "skip" portions of the powerset in certain situations
I was unsuccessful in finding one here
pmurias Limbic_Region: you were unable to make the hash garbage collection optimazation with the iterative version 15:05
?
Limbic_Region oh, no that's not the problem
the problem with the iterative version is this 15:06
say string one was ABCD, string two is ABCE, and string three is ABCFG
they all share ABC, AB, AC, BC, A, B, C
so why generate them at all in strings 2 and 3
pmurias in the recursive one you use a hash too 15:07
Limbic_Region yes, but the hash prevents recursion all together
in an iterative version you would still touch every node and just not include it
IOW - ABC would be seen in strings 2 and 3 but not AB, AC, BC, A, B, C
once it sees it had seen this it would stop recursing 15:08
pmurias i see
Limbic_Region in fact, the recursive version is suboptimal for a single powerset
it only has its advantages over many strings that have subsets in common
which I designed specifically for my dataset
pmurias (by the way i was planing to use,was already used by Data::PowerSet) 15:09
Limbic_Region oh, well I have written things even better than the bitmap binary counting
not to take anything away from Data::PowerSet 15:10
but it does not know how to "skip" at all
I have written iterators that can skip - under the right circumstances
basically I had to generate the powerset in progressive order such that 15:11
if your first 3 items exceeded a certain amount
continuing to follow this powerset getting progressively larger was pointless
yadda yadda
so do you see how the gc optimization I am going to make is going to work now? 15:12
pmurias yes 15:13
Limbic_Region fwiw, the project is already finished - I am just trying to optimize the runtime a bit now 15:14
16:04 foo\ joined, weinigbbl is now known as weinig, larsen joined
svnbot6 r14336 | audreyt++ | * util/yaml_harness.pl: Remove the smoke-concurrent 16:07
r14336 | audreyt++ | reduce-number-of-first-process bias again as the recent re-enabling
r14336 | audreyt++ | of t/regex/ made ext/ not as heavy by comparison.
r14337 | audreyt++ | * Remove Judy from the Pugs build process. 16:09
r14337 | audreyt++ | This is a pervasive change to avoid GC issues completely
r14337 | audreyt++ | before the release, as the nature of addForeignPtrFinalizer's
r14337 | audreyt++ | weird interaction with GHC is not fully diagnosed.
r14337 | audreyt++ | Also, TVar/IORef based storage, which was exceedingly slow
r14337 | audreyt++ | in GHC 6.4, now actually compares competitively with Judy maps.
r14337 | audreyt++ | Judy is still more memory efficient and may be faster on some
r14337 | audreyt++ | architectures, but its re-incoporation will have to wait until
r14337 | audreyt++ | after release and a fuller understanding of its use of FFI/GC.
16:18 pmurias left
audreyt hm, today was rather swamped with nonpugs tasks, mostly sleeping :/ 16:18
audreyt goes to rest some more and hopefully tomorrow will be more productive... 16:19
svnbot6 r14338 | audreyt++ | * Set/t/overload.t: Mark as :todo<bug>, instead of showstopping. 16:48
masak good night audrey 16:58
17:05 mako132_ joined 17:19 Aankhen`` joined 17:25 penk joined, frankg joined 17:30 weinig joined 17:40 Ikarus joined 17:51 Southen joined
svnbot6 r14339 | fglock++ | v6/emitter(2) - passes isa.t, for.t 18:16
18:44 lisppaste3 joined 18:53 weinig is now known as AwayWeinig
TimToady smoking r14339, after much hassle getting recompiled. 19:27
there's something out there in some subdirectory that still wants HsJudy that isn't cleared out by realclean. 19:28
don't know offhand which subdirectory, but src is a likely suspect.
anyway, blowing it away and re svning up seems to have fixed it. 19:29
also commented out some Judy stuff in Makefile.PL, not sure whether that was necessary or not. 19:30
lunch & 19:32
avar wrt the sub prototypes in p6, is there a mechanism for naming arguments in side lists in the argument list 19:45
like def sub((a, b), c); and x = (1, 2); y = 3; sub(x, y)
a = 1, b = 2, c = 3;
(python example)
TimToady avar: sure, see S06:1128 "Unpacking array parameters" 20:07
svnbot6 r14340 | lwall++ | Makefile.PL: more Judy commentage 20:08
TimToady yay, my 1st 100% 20:12
20:15 penk joined 20:20 AwayWeinig is now known as weinig, penk joined 20:29 andara joined 20:33 andara left
avar TimToady: neato 21:00
21:09 Sarta2 joined, Sarta2 is now known as Eidolos 21:15 nperez joined
masak I get this when running make: 21:22
ghc-6.4.1: ghc-6.4.1: panic! (the `impossible' happened, GHC version 6.4.1): unknown exception
not too reassuring...
21:22 avarab joined
masak maybe it really is time to make that GHC upgrade 21:22
21:23 Limbic_Region joined
SamB now that 6.6 has been released, it might be easier 21:24
Limbic_Region salutations all 21:28
avarab eval: $^O 21:31
buubot avarab: OS/2 Warp 21:32
21:32 avarab is now known as avar
Limbic_Region ?eval $*OS 21:32
21:32 evalbot_r14334 is now known as evalbot_r14340
evalbot_r14340 user error (Incompatible version number for compilation unitā¤Consider removing blib6/lib/Prelude.pm.yml and make it againā¤)ā¤pugs: user error (*** *** Can't modify constant item: VUndefā¤ at <prelude> line 784, column 9-46ā¤ <prelude> line 784, column 9-46ā¤ at)ā¤ 21:32
Limbic_Region well, that doesn't look so good 21:33
anyone with sudo privs on feather about?
21:34 weinig is now known as weinig|food
avar !eval: sub foo (Int $x) { $x } foo(5); 21:56
?eval sub foo (Int $x) { $x } foo(5);
evalbot_r14340 user error (Incompatible version number for compilation unitā¤Consider removing blib6/lib/Prelude.pm.yml and make it againā¤)ā¤pugs: user error (*** *** Can't modify constant item: VUndefā¤ at <prelude> line 784, column 9-46ā¤ <prelude> line 784, column 9-46ā¤ at)ā¤
SamB eek 21:57
fake newlines
21:59 ajs joined
avar Anyway I was wondering about something, S06 shows: sub prefix:<moo> and sub prefix:{"moo"} as being equivalent and sub circumfix:<a b> equivalent to sub circumfix:{"a", "b"} 22:00
then later macro circumfix:Ā«<!-- -->Ā» ($text) is parsed / .*? / { "" } is shown as an example
"Whitespace may never be part of the name (except as separator within a <...> or Ā«...Ā» slice, as in the example above)."
22:01 Southen_ joined
avar Isn't the <> there just a quote operator, then saying that there is some whitespace excemption is a bit misleading since it would just be equivalent to macro circumfix:{"<!--", "-->"} 22:01
Or am I completely wrong here? 22:02
I.e. it's not at all part of the operator name any more than " " is part of one of the array items in @a if my @a = <a b c d>; 22:03
22:29 mauke joined 22:33 larsen joined 22:55 ilogger2 joined
nothingmuch @seen cmarcelo 23:00
lambdabot Last time I saw cmarcelo was when I left #darcs, #gentoo-haskell, #ghc, #haskell, #haskell-blah, #haskell-overflow, #haskell.es, #haskell.it, #haskell.se, #haskell_ru, #oasis, #perl6 and #scannedinavi
an 4d 20h 37m 13s ago, and .
nothingmuch very specific =)
Limbic_Region yeah 23:01
so is anyone else unable to build pugs
or is it just once again Win32/MinGW?
jdv79 strawberry? 23:10
Limbic_Region no, MinGW 23:14
TimToady I had a lot of trouble rebuilding.
Limbic_Region or rather
TimToady on Linux
Limbic_Region I have no idea why I said MinGW
that should have nothing to do with the Pugs build 23:15
TimToady my error was that it was still trying to link in HsJudy
Limbic_Region tries again
TimToady eventually I blew away src/ and several other subdirs and re svnup'd
Limbic_Region ah
I was wondering about that 23:16
is there a way to tell svn update every file anyway
TimToady something isn't getting realcleaned I think
but I didn't have the patience to chase it down.
Limbic_Region ok, will just get liberal with rmdir /s 23:17
TimToady well, I don't keep anything of my own in the subdirectories as a matter of policy so that I can do that.
it was *probabl& src
*probably* even
Limbic_Region neither do I 23:18
TimToady but I also blew away several other dirs so I can't say for sure, and it kinda sucks to get all the way through and have it fail again
I think "realclean" fails the DRY principle. 23:19
just as I think keeping a list of force_todos in a separate file is a recipe for disaster.
I know there have been systems that actually tracked what files were produced by the various subprocesses in a build so they automatically know what to clean up. Unfortunately make has always been a bit too simpleminded that way. 23:21
Limbic_Region well, I got really liberal with what got removed 23:22
TimToady we'll I've done that, and regretted it. :)
*well
Limbic_Region I left t examples and a few others I didn't think could possibly have anything to do with building alone 23:23
TimToady I did discover that if you blow away blib6 but not ext, you end up having to go through and remove all the pm_to_blib "touch" files. :/ 23:24
another example of braindead makery...
Limbic_Region heh, it thinks I don't have ghc 6.6 23:28
TimToady did you install ghc 6.6 in one of your subdirectories? :) 23:30
Limbic_Region no, I have been building ghc 6.6 from source prior to it being released as 6.6
so it reports 6.5 23:31
TimToady I see.
Limbic_Region building from source (not bootstrapping) isn't as painful as some might lead you to believe
23:32 gnuvince joined
Limbic_Region will your work project (or parts thereof) ever see the light of day? 23:35
TimToady not likely--it's in support of some proprietary hardware. 23:36
Limbic_Region oh
was hoping to see a sample representation of your code applied to a real problem 23:37
TimToady unfortunately the people paying me to work on a real problem would have a real problem with that... 23:38
Limbic_Region right
gnuvince TimToady: where do you work? 23:39
Limbic_Region 92 of 92 compiled....
TimToady across the street from Google.
links okay?
Limbic_Region hugs TimToady
TimToady ...cant...breathe... 23:40
Limbic_Region releases the bear hug
how well does it smoke?
TimToady huh huh huh huh
well, mine went 100%
Limbic_Region nice - speed?
TimToady couple minutes slower than my previous run, prolly statistically insignif 23:41
given there were other things running on my computer
Limbic_Region audreyt still hoping to release before Monday her time? 23:42
TimToady well, she's got one day if so. 23:43
it seems pretty solid from the Linux perspective at least.
Limbic_Region I didn't think there was a lot left on her list 23:44
TimToady probably most of the remaining work is changelogging...
Limbic_Region well most of the Win32 issues aren't real issues
yeah, she always asks for help on that and I always feel bad for not helping more
TimToady the eqv thing
or maybe it's just not reading the Perl 6 code in text mode... 23:46
Limbic_Region hard to say how many are due to the eqv thing but more than 1
TimToady isn't Windows supposed to translate crlf to \n for you?
buubot correct syntax is: translate to/from language string
Limbic_Region running smoke now 23:47
TimToady - I don't think so
23:47 Southen joined
Limbic_Region certain files opened up in notepad look like everything is jammed on one line 23:48
opened in wordpad look normal
TimToady then why did we add binmode to Perl?
well, but that's the opposite problem.
Limbic_Region well, perl does the right thing 23:49
meaning a file with \r\n or \n both work as expected with <$fh>
I don't try and understand windows - I just try and make it get out of my way and stay there 23:50
TimToady but if you look at the string in Perl 5, it just as a newline.
s/as/has/
Limbic_Region well, I know $/ is "\n" by default and "\n" takes on the meaning of the platform 23:51
but
TimToady but it seems like Haskell isn't maintaining a \r free zone.
Limbic_Region perlbot nopaste
perlbot Paste your code here and #<channel> will be able to view it: sial.org/pbot/<channel>
pasteling "Limbic_Region" at 24.35.57.240 pasted "example of the eqv problem" (21 lines, 448B) at sial.org/pbot/20438 23:52
TimToady looks like slurp is defaulting to binary rather than text. 23:55
whatever it's defaulting to, it's not the same as the parser... 23:56
Limbic_Region right 23:57
TimToady is the \r on the expected or on the actual?
Limbic_Region let me think about this 23:58
the \r is in the .t file
so expected
TimToady you can't deduce this from whether it's in the .t file
Limbic_Region I mean if I open the .t file in notepad it is fubar
TimToady oh, so the newline policy is wrong on the file maybe? 23:59
Limbic_Region that's a possibility
TimToady hmm.

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!