svn switch --relocate svn.openfoundry.org/pugs svn.pugscode.org/pugs/ | run.pugscode.org | spec.pugscode.org | paste: sial.org/pbot/perl6 | pugs.blogs.com
Set by avar on 16 November 2006.
TimToady audreyt: thanks! I have to have my -Ofun too, ya know. :) 00:13
dduncan I am also happy with that overhaul of file-test operators ... I never liked their -foo form 01:55
putter didn't *mean* to start a long thread about Inf... 02:29
audreyt TimToady: one of the tests had 02:30
audreyt $file ~~ :s > 42 02:30
unfortunately, that is now chaincomp 02:31
and eval as
$file ~~ :s and :s > 42
which is false
unintended consequence?
allbery_b saw that go by and is somewhat glad the stupid parser issues are hopefully history now 02:32
frustrating
putter re Inf, spec is IEEE. and customization is needed because IEEE has had slosh, and because not everything has been IEEE.
(but I was just asking a question about perl 5 existing implementation pragmatics) 02:33
gaal: pong.
gaal: replied. no problem.
ashleyb: see svn.pugscode.org/pugs/misc/pX/Commo...ex_engine/ especially README. 02:36
lambdabot Title: Revision 15164: /misc/pX/Common/yet_another_regex_engine, tinyurl.com/2d96yh
putter Code is buried down in svn.pugscode.org/pugs/misc/pX/Commo...ModuleA.pm as is p5 current (unfortunate imho) practice.
lambdabot tinyurl.com/2fyhqj
ashleyb putter: thank you 02:39
putter goals: create a usuable p6 regex engine - we need one; create the first non-pugs parser - the pace of pugs's parser development is holding us up a bit; ideally have something which could be transliterated from p5 to p6 to get the parser part of a p6 implemented in p6.
support parser development. 02:40
I think that's about it.
Written in p5. I try to keep a "ways one can help" at the top of the README.
There is also the main pugs "ways you can help" in svn.pugscode.org/pugs/project_planning/TASKS 02:41
ashleyb looking... 02:43
putter history: the code is basically just the regex engine from the... Jan '06(?) regexp_engine_and_parser_spike, which got cleaned up slightly in... Fall '06? in the hope of making it more usable/attractive, and then overhauled with a new ast (instead of japhy's Regexp::Parser) starting a few days ago. 02:45
ashleyb what's wrong with fglocks Pugs::Compiler::Regex? 02:45
putter slow. 02:46
ashleyb I've been using that in production for a few months now (though I've got a few bugs I've been working around)
k
putter neat! it would be interesting to hear of your experiences.
The cleanup was done in the hope that PCR might switch backends, but no one had the tuits or interest or something. 02:47
ashleyb my needs are more for ease of implementation and of maintainability over speed 02:48
so I chose P::C::R over regular perl 5 regexs
(all my code is in perl 5)
but yes I do find it slow, but tolerable for what I want. 02:49
ashleyb the funny bug with it is, is that instead of crashing (or throwing an error) it always hangs. 02:50
putter good. i imagine at some point it will simply start running faster.
ashleyb which makes debugging interesting, but once you get the knack you can figure it out
putter ah. :/ hopefully that will improve too. 02:51
audreyt TimToady: I'm going to assume that :d>10 will eventually raise a warning somehow and goahead committing the implementation.
ashleyb putter: so how does P::C::Regex and yet_another_regex_engine go together? is there something different of value in each? 02:53
putter: that they can be combined at one point?
putter ashleyb: if you were interested in helping out this evening, I'm curious where the computes are going, but I think I'll spend tonight nudging the % test coverage up a bit. If you wanted to run a profile...
PCR runs p6 regex, yare only knows about p5 so far. 02:54
ashleyb a profile of PCR?
putter PCR's emitters could be changed to use the yare engine core (backtracking protocol, anonymous subs, and their factories) 02:55
a profile off yet_another 02:56
re combined, perhaps. p6 is largely a bootstrapping game. given n parts sort of working, how can we puzzle them together to make progress. 02:57
re profile, eg dprofpp or whatever on perl lib/Regexp/Module.pm --test 02:59
ashleyb how do I get the yet_another code? 03:01
putter eg, pwd misc/pX/Common/yet_another_regex_engine; perl -d:DProf lib/Regexp/ModuleA.pm --test > deleteme; dprofpp ...
do you have a copy of pugs?
ashleyb nope. 03:02
putter have you used svn or cvs?
ashleyb cvs a long time ago.
p4 mostly now
no svn/svk
putter basically just $ svn co svn.pugscode.org/pugs/ 03:03
lambdabot Title: Revision 15164: /
putter (someone check me on that)
ashleyb hmmm that seems to be working :D 03:05
tene you're right.
putter then if you express even the slightest interest, someone give you a commit bit (to write back to the repository)
tene Just as a warning, any response to that statement might be considered interest in having a commit bit. 03:06
putter lol :)
so true
ashleyb I think I had one once -- but I'll ask again as soon as I have something to check in.
tene Accidentally using an anagram of "commit" or "patch" can get a commit bit in your inbox.
ashleyb haha
allbery_b looking at the word "commit" or "patch" can land you a commit bit around here... 03:07
SamB heh
but, you can't get a commit bit in your inbox without someone at least knowing your email address, can you? 03:08
SamB is safe because he already has one
ashleyb the password is another matter...
:)
putter from the top of README: prove ... tests and provides a nice summary; --test lets you watch the individual tests go by; and --repl lets you interactively play. 03:11
svnbot6 r15165 | audreyt++ | * Filetest operators are now removed; use smartmatch against Pairs instead:
r15165 | audreyt++ | -e 'README'
r15165 | audreyt++ | !-e 'README'
r15165 | audreyt++ | becomes:
r15165 | audreyt++ | 'README'~~:e
r15165 | audreyt++ | 'README'~~:!e
ashleyb ok done check out. 03:16
fglock hi
putter oo, wow. -e gone away. it's been a lot of years. a moment of quiet contemplation.
ashleyb hi :D
putter hi fglock :)
ashleyb putter:ok so how do I run the profiler? 03:17
\
putter man dprofpp? 03:18
ashleyb where is lib/Regexp/Module.pm 03:19
putter if you haven't looked at profiler output before, it can be rather a learning curve.
ashleyb do I need to do a make?
putter don't need make. just cd misc/pX/Common/yet_another_regex_engine 03:20
fglock update on kp6: lexical subs, metaclass, and env work; I'm now trying some possible container implementations 03:31
putter :) 03:34
pasteling "ashleyb" at 66.201.51.66 pasted "bug in yet_another_regex_engine --test" (3 lines, 338B) at sial.org/pbot/22774 03:35
putter a month can't have passed already... are you still on vacation, but typing... (like audreyt's sleeping but committing;) 03:36
oh, there are a whole lot of bugs in --test, that's the current exercise... :) 03:37
fglock putter: I came back to the city for a wedding - I'm travelling again tomorrow :)
ashleyb ah -- so fix the bugs and get the profile...? 03:38
putter fglock: ah. enjoy. 03:39
ashleyb: probably most efficient for me to fix stuff. though feel free. warn me and I'll sync more often than once an evening. if you wanted to profile it now, it's running slower than I expected, and am curious where the time is going. 03:41
though for a first task (and a commit bit:), you might look at the project_planning/TASKS list. some of those have much lower barriers to entry than doing a profiling exercise of code you haven't seen before. 03:43
ashleyb what's svn for sync?
(i.e. how do I sync in svn?)
tene svn up
to send changes back, use 'commit' 03:44
'up' is short for 'update'
putter re bugs, thanks for reminding me, i meant to edit README to mention the objective is 100%, or no worse than the underlying p5 implementation version. older versions, more re_tests failures. as we'll keep updating re_tests from the most recent p5.
there's something about s/\\([\\\'])/\\\\$1/g I just find humorous. 03:47
re 'commit', neat, didn't know that. I always use svn ci 03:49
ashleyb putter: it's about 8PM here. and I've gotta decommute, so maybe I'll catch you tomorrow. I've got revesion 15165 -- but it doesn't seem to give me any changes to misc/pX/Common/yet_another...
(when I svn up)
& 03:51
putter goodnight 03:52
err, s/\\([\\\'])/\\\\\\$1/g; sigh 03:57
wolverian audreyt++ # hacking in absentia 04:06
putter oh, fussbucket. I'd forgotten you don't know what \10 is without nonlocal info. 10 preceding captures, it's a backref, else, a char. sigh. 04:25
dduncan fyi, while I 'make' new pugs versions semi-daily, I haven't smoked for months, so I started one now ... 04:59
allbery_b thinks that backref thing is a serious wart 05:00
putter p5
a very old and wrinkled wart 05:01
allbery_b yeh
putter with a reason for being too
dduncan hm, I notice that a lot of tests are spitting out "# Actual: undef" but no "Expected" lines
allbery_b that's been true ever since I started with pugs 05:02
putter ok() vs is() perhaps?
svnbot6 r15166 | putter++ | misc/pX/Common/yet_another_regex_engine - very minor progress. p5 re_tests at 82% ok (only a few % up from yesterday). Even slower. As always, see README. 05:08
putter good night all & 05:20
TimToady audreyt: I'm wondering whether the base form is not .TEST(:X) but just .:X and we simply allow a type to define methods starting with : 06:43
alternately, just make .:X a shorthand for .TEST(:X)
then we don't have a precedence problem with $x.:s > 1024 06:44
or maybe there's some other way to map a pair test to methodname space. 06:46
the .TEST(:X) approach would tend to encourage people to use switch statements.
the method :X approach would of course encourage method defs instead 06:47
the latter would require a CANDO thing if you wanted to handle arbitrary pairs, while .TEST could handle that more easily.
another possibility is :s(1_000_000..*) selectors, but that doesn't work so well for <=> comparisons 06:50
dduncan the smoke of r15165, repository snapshot / darwin is now done, taking about 117 minutes using 4 threads on 2 CPUs ... result is 98.27% ... 17717 test cases: 17411 ok, 306 failed, 4498 todo, 706 skipped and 0 unexpectedly succeeded 07:24
er, that's the default Haskell backend, I believe 07:25
mncharity @tell fglock People have been asking for ways they can help. Any suggestions? 08:02
lambdabot Consider it noted.
svnbot6 r15167 | lwall++ | More ~~. 09:21
r15167 | lwall++ | More whitespace around ~~.
r15167 | lwall++ | Several eval'd tests of .:X syntax
IllvilJa Hello, I know that Perl6 is under development but what is the situation of processing XML docs under Perl 6? Are there any XML modules available in, say, Pugs? 13:37
wolverian you can use perl5 modules from pugs - 'use perl5:XML::LibXML;' for an example. 13:39
IllvilJa Oh! That sounds delicious!
I'll give it a try then.
Thanks for the help!
wolverian you're welcome :)
coke waves. 14:58
ayrnieu coke waves? Only on Planet Coca-Cola! 14:59
Coke So, how close is @larry to finalizing the spec? =-) 15:20
cognominal babel easel 15:24
oops
rindolf Aloha everyone. 15:24
I mistakenly thought that the NL hackathon is today, rather than a month from now.
But the good guys on #parrot corrected me.
BooK rindolf: yes, it's march 2-4 15:28
rindolf BooK: yes, I know now.
We might have another hackathon in Israel sometime as well.
BooK do you plan to come?
rindolf BooK: to the NL hackathon? 15:29
BooK oh, yes, isreal may be a bit far :)
rindolf BooK: not for me, I live here.
BooK I meant, Israel is a bit far from NL for you to come
rindolf BooK: yes. 15:30
BooK not much people registered yet, for the NL hackathon 15:30
rindolf BooK: well, I just started a new job, so I'm not sure if they'll be fond of me taking a trip for a few days so soon. 15:32
BooK yeah, if there are plans for one in israel, it's certainly simpler to go to that one 15:38
I also start a new job, but it's in Amsterdam :-)
rindolf BooK: doing Perl? 15:41
BooK yep 15:42
rindolf BooK: nice. 15:50
gaal @tell putter we moosed a connection and all's well, expect 1.02 up soon, and 2.0 RSN 17:00
lambdabot Consider it noted.

Powered by Cro and the Raku® Programming Language.

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

Thank you!