pugscode.org | sial.org/pbot/perl6 | ?eval [~] <m oo se> | We do Haskell, too | > reverse (show (scanl (*) 1 [1..] !! 4))
Set by agentzh on 17 April 2007.
00:17 redear joined 00:31 ludan joined 00:41 thorat left 01:07 Psyche^ joined 01:22 Psyche^ is now known as Patterner 01:35 buetow joined 02:12 offby1 joined 02:31 lb2 joined 02:44 lambdabot joined 02:48 agentzh joined 02:49 agentzh joined
agentzh Oops, auto-smoke failed again :( 02:50
pasteling "agentzh" at 221.131.66.71 pasted "auto-smoke failed" (211 lines, 14.4K) at sial.org/pbot/24393 02:51
offby1 my auto smokes sometimes. I think I need to get the pistons re-bored 02:55
agentzh decides to work on Smart::Links for CPAN. 03:04
03:07 kanru joined 03:08 justatheory joined
agentzh offby1: should i "make clean" before "make smoke"? 03:11
offby1: or should i "make" before "make smoke"?
running "./auto-smoke" on feather manually didn't have any problem. 03:12
weird.
offby1 I have no idea; sorry. I was just making a dumb joke
agentzh offby1: okay
finally get the pun in your joke ;) 03:13
offby1 wasn't really worth the effort, was it? :-) 03:15
agentzh hehe 03:16
03:22 deq` joined 03:34 thorat joined, thorat left
gaal agentzh: which ghc is invoked when you say "ghc --version" on your interactive shell? Which one when you autosmoke from cron? 04:35
I'm thinking perhaps the environments are different, and you end up with mismatching versions of the haskell networking library.
04:35 Psyche^ joined
agentzh gaal: it's feather. 04:37
GHC 6.6
gaal: you mean cron and i are using different GHC? 04:38
checking now
agentzh@feather:~$ ghc --version 04:39
The Glorious Glasgow Haskell Compilation System, version 6.6
04:44 buetow joined
agentzh ah, "make" no longer work for me. i'm getting the same output from interactive shell :( 04:49
*works
should i wipe out third-party/installed ? 04:51
04:51 Psyche^ is now known as Patterner
agentzh Upgrade to YAML::Syck 0.85 before smoking | pugscode.org | sial.org/pbot/perl6 | ?eval [~] <m oo se> | We do Haskell, too | > reverse (show (scanl (*) 1 [1..] !! 4)) 04:52
gaal agentzh: I have to leave to work right now, but yes, in general tasks started from cron get their own environments, different paths etc. 04:55
agentzh gaal: okay
gaal I'll update Task::Smoke later to indicate the Y:S recommendation 04:56
agentzh cool
gaal &
05:23 jisom joined 05:38 atp joined 06:13 BooK joined
agentzh Hmm, interesting, my interactive shell on feather uses /usr/local/bin/ghc while cron uses /usr/bin/ghc 06:22
is there a simple way to fix this?
06:22 dduncan left 06:25 rissy joined
gaal agentzh: put PATH=/usr/local/bin:$PATH in your crontab 06:26
agentzh gaal: okay :)
06:29 xinming_ joined
svnbot6 r16059 | gaal++ | * avoid generating bad YAML in the harness by depending on 06:33
r16059 | gaal++ | YAML::Syck 0.85 or falling back on YAML.pm
r16060 | agentz++ | [auto-smoke] prepended '/usr/local/bin:' to PATH
06:34 penk joined
agentzh it seems that YAML.pm fails to read the stuff generated by YAML::Syck 0.85. :( 06:39
06:41 amnesiac joined
jjore was incredibly surprised to learn why YAML is impossible to write by hand: it disallows indentation with tabs. 06:42
So much of the PAIN I remember from it is explained by just that. 06:43
agentzh thinks YAML is designed to be human-readable rather than human-writable.
jjore Its not difficult to come up with non-tab indentation but with diagnostics in YAML.pm that don't detect that and tell me about my goofs, I can go years without knowing why I can't write something that looks so simple. 06:44
rgs but aren't tabs The evil ? 06:45
jjore Hee. Yes, they are.
I had to go to significant effort to get my Emacs to do tab *only* indentation at work.
You know, I wanted "tab" to mean tab everywhere except in my work's source code where it means "insert a tab or trigger autocompletion" 06:47
svnbot6 r16061 | agentz++ | [auto-smoke] added some diagnostic code regarding /usr/bin/ghc
agentzh gaal: cool, cron now picks up /usr/local/bin/ghc as the interactive shell. hmm, i think putting a PATH definition into /etc/default/cron might help too ;) 07:21
gaal don't do that 07:22
agentzh why?
gaal that'll modify root's cron environment
agentzh k
gaal /usr/local shouldn't in generally be in root's path
agentzh okay, already removed /etc/default/cron
devbot6 planet6: Audrey Tang: Weekly Perl 6 mailing list summary for 08-14 April, 2007 <pugs.blogs.com/pugs/2007/04/weekly_...4.html>
07:43 xinming joined 07:51 kane_ joined
svnbot6 r16062 | kudra++ | Now contains all messages of the week 08:00
08:00 kanru joined 08:01 franck__ joined, elmex joined 08:11 bernhard joined 08:26 kanru joined
svnbot6 r16063 | kudra++ | This commit covers all threads which spanned this week and last week, as 08:26
r16063 | kudra++ | well as a few small threads. I am checking in more often since I may soon
r16063 | kudra++ | be sharing this summary writing duty :)
08:36 elmex joined 08:52 elmex joined 08:53 araujo joined 09:01 marmic joined, iblechbot joined 09:11 Southen joined 09:24 ludan joined 09:30 Muixirt joined 09:47 ilbot-test joined 09:48 kane_ joined 10:00 laye joined 10:01 laye joined
agentzh laye: problem solved? 10:05
10:11 Belaf_sleeping joined 10:45 ludan joined 10:47 agentzh left
moritz in p5, can I somehow match a string literally that is interpolated into a regex? 11:30
I want something like $str = "foo...bar"; m/something... $str/ so that the dots in $str are treated literally? 11:31
gugod try \Q$str\E 11:33
moritz gugod: thanks 11:34
gaal moritz: see also the related quotemeta builtin 11:35
moritz gaal: I thought about that, but does it what I want?
gaal: is \' defined in regexes?
the dot was just an example 11:36
gaal /\Q$str\E/ is the same as $str_1 = quotemeta $str; /$str_1/
moritz so if I use $str multiple times in a regex, the quotemeta version is faster?
gaal if you run the quotemeta once, yes :) 11:37
moritz gaal: I do, thanks ;)
gaal it's unlikely to have any significant impact in your application, but hey you probably know better :)
11:37 agentzh joined
moritz well, it does not have any impact ;) 11:38
agentzh is enjoying moritz++'s #perl6 logs. 11:41
moritz I'm currently working on a more flexible color management
agentzh nice 11:42
moritz now I assign colors to the nicks who speak most frequently... 11:52
should I do that on a per-day basis, which gives nicer output, or on a "global" basis, so that a nick has the same color every day?
anyway, I'm off & 11:54
12:10 buetow joined 12:11 qmole joined 12:14 qmole joined, Muixirt left 12:26 pstu joined 12:57 OwlEye joined 13:05 marmic joined 13:08 chris2 joined
agentzh moritz: please don't do coloring globally since i may dislike the current one painted on myself ;) 13:15
moritz: or it can be customized by ourselves. for example, when agentzh says "ilbot: red!" in the channel, it will put red on agentzh. ;) 13:18
agentzh thinks this feature is kinda cool. 13:19
13:21 Limbic_Region joined
jiing mabye add an option to switch this feature 13:22
agentzh hehe
lumi Just assure uniqueness of colours for nicks of the same length and starting with the same letter, everything else is bonus 13:24
Limbic_Region huh what? 13:25
agentzh Limbic_Region: moritz.faui2k3.org/irclog/out.pl?ch...2007-04-22 13:26
lambdabot Title: IRC log for #perl6, tinyurl.com/23setw
Limbic_Region iblech solved the coloring problem a long time ago 13:31
moritz re 13:32
agentzh: I think that's overkill, irc logs are overrated ;)
Limbic_Region perlmonks.org/?node_id=384347 13:33
lambdabot Title: Coloring IRC logs nicely
Limbic_Region and here is a screen shot m19s28.vlinux.de/iblech/iblechbot-2.png 13:34
moritz kicks svnbot6 ;) 13:57
svnbot6 r16064 | moritz++ | added the irclog bot and cgi scripts
moritz agentzh: if you want to setup the bot and cgi scripts on feather, just tell me so, and I'll provide a sql dump of the data I it collected on my server 13:58
pasteling "evalbot_r16063" at 194.145.200.126 pasted "Pugs build failure" (314 lines, 17.1K) at sial.org/pbot/24400 13:59
svnbot6 r16065 | franck++ | r21438@franck-mbp: franck | 2007-04-22 15:31:29 +0200 14:00
r16065 | franck++ | better translation
14:08 agentzh joined
agentzh moritz: please do :) 14:08
14:09 irclogbot joined
moritz agentzh: allright... I'm currently working on moving the config to different files 14:09
agentzh no hurry :) 14:10
svnbot6 r16066 | moritz++ | moved bot config to separate file
agentzh moritz++
moritz agentzh: moritz.faui2k3.org/tmp/irclog.sql.bz2 14:11
agentzh thanks 14:12
i'll first try to set it up on my own machine :) 14:13
moritz agentzh: wait half an hour and you have to modify less code, and more config files ;)
agentzh cool
14:16 irclogbot joined 14:18 agentzh joined
svnbot6 r16067 | moritz++ | irclog: most configuration is now done in the corresponding config files 14:25
agentzh is caught by weird svk conflicts again. 14:31
to be honest, i really won't say "svk --" since i really like it.
moritz agentzh: then it's SVN that sucks ;) 14:32
agentzh but...
moritz if you don't want to blame your implementation, the protocol sucks ;)
agentzh svn seldom gives me troubles.
apologies if i've done something wrong to the pugs repos. 14:35
i'll revert the last commit if it's not empty. 14:36
14:42 rindolf joined 14:52 fridim joined
diakopter agentzh: I was doing some thinking about smartlinks yesterday 15:00
agentzh diakopter: yes?
diakopter and then I saw your msg about Smart::Links
agentzh that's Audrey and jerry's idea :) 15:01
diakopter oh, I didn't see their messages I guess
agentzh (long long ago)
agentzh is lazy. 15:02
diakopter oh. my thoughts were these:
agentzh listens. 15:03
15:04 kanru joined
diakopter Perl6::Spec as a regular CPAN release would be nice. I was envisioning it as containing all the synopses in both Perldoc and POD(5) (so that CPAN could render them as html), the (future) yaml-organized official Perl 6 test suite, and of course the smartlinks tools, to generate the interactive versions. 15:07
agentzh Perldoc == Pod6? 15:08
diakopter yeah
agentzh the cpan toolchain doesn't support Pod6, does it? 15:09
diakopter no, not yet
other grandiose thoughts from that: 15:10
agentzh what do you mean by "yaml-organized test suite"?
moritz yaml is overrated ;)
agentzh has the feeling too.
diakopter Perl6::Implementation::Smoke(::Interpreter,::Compiler,::Emitter) that uses Perl6::Spec to smoke arbitrary Perl 6 implementations 15:11
agentzh for "interactive version", do you mean the synopses at perlcabal.org/syn ?
lambdabot Title: Official Perl 6 Documentation
diakopter yes
agentzh Hmm 15:12
moritz @karma agentzh
lambdabot agentzh has a karma of 6
agentzh the idea of using specs to smoke implementation is funny.
*implementations
@karma agentz
lambdabot agentz has a karma of 10
agentzh 10 + 6 = 16
diakopter *someone* put "yaml test suite" in the wide project dependency graph
agentzh :)
diakopter: you mean the pictures produced by putter++? 15:13
diakopter I wonder who that could have been.... ;)
I meant this: dev.pugscode.org/browser/project_pl...format=raw
lambdabot tinyurl.com/yr6gyp
agentzh looking 15:14
diakopter "YAML-based test framework"
agentzh yes, that was drawn by putter and others 15:15
right, the current pugs smoking mechanism is based on YAML.
there's no doubt about it. 15:16
agentzh has been fighting against YAML::Syck's bug these days.
we use YAML to serialize the smoke results as well as other related configuration settings.
15:16 penk joined
diakopter right, but I assumed that the test suite would additionally need to become more structured in order to be useful for Parrot-Perl6 and others 15:17
agentzh we need a good harness.
the current harness is mostly Perl 5's Test::Harness. 15:18
particle has introduced a neutral TODO tricks to parrot
and we believe it's useful for the Pugs test suite too. 15:19
# TODO: parrot 0.4.5, pugs 6.28.1
or something like that.
the harness may or may not use YAML.
to the best of my understanding, YAML is just a data-serialization tool. 15:20
avar it's a serilization format
agentzh avar: agreed.
:P 15:21
diakopter: a standalone CPAN module named Perl6::TestSuite makes much more sense to me ;) 15:22
diakopter of course. What I am trying to get you to mind-read from me :) is that perhaps the test suite could be *in* Perldoc, with the individual tests inline with the specs that explain them.
avar It doesn't always make sense to test things in the order that they're explained and vice-versa 15:23
just look at the correlation between the spec and the smartlinks
agentzh diakopter: putting test suite into perldoc? wow 15:24
diakopter: i think i've got what you're saying.
moritz agentzh: did you get the irclogger running?
agentzh moritz: sorry, not yet. fixing svk issues right now. 15:25
moritz agentzh: no hurry
agentzh ;)
moritz agentzh: just ask me when you have questions
agentzh diakopter: you're suggesting bringing smartlinks to ordinary CPAN modules, nright?
*right
moritz: yes, of course! 15:26
diakopter: or bringing smartlinks to Perl 6 implementations other than the Haskell Pugs?
diakopter avar: true. Note the trac ticket about reorganizing the test suite to (possibly) match the specs
15:27 stevan_ joined
diakopter agentzh: yes, enabling smartlinks for other Perl 6 implementations, and hopefully allowing further linking to other (all?) implementations' results/progress. 15:28
(from the "official" hosted version of the specs)
agentzh diakopter: *nod*
i think it can already achieved by perhaps setting a environment in the current pugs smoke system? 15:30
*be achieved
avar diakopter: there's a trac?
agentzh diakopter: i think this is our goal :) 15:31
diakopter avar: dev.pugscode.org/query?status=new&a...r=priority 15:32
lambdabot Title: Custom Query - Pugs - Trac, tinyurl.com/yujl4y
agentzh that ticket is great. 15:33
diakopter which ticket? 15:34
agentzh to be more specific, we need to update "./pugs" to "$*EXECUTABLE_NAME" or something like that.
diakopter: "Rearrange t/ for reuse with multiple Perl 6 implementations."
I think the test harness needs some treatment too. 15:35
diakopter agentzh: re Perl6::TestSuite, since the Perl 6 test suite is authoritatively defined as *the (machine-readable) specification* for Perl 6, I thought Perl6::Spec was not a misnomer 15:36
then of course the synopses are the human-readable versions.
agentzh are you sure our test suite is *only* machine-readable? 15:37
a lot of our tests are very good code examples ;)
diakopter right, I mean the specs (with the inlined tests via smartlinks or Perldoc) were human-readable. 15:38
synopses.
agentzh your explanation of "spec" is quite interesting though.
gotcha
diakopter if I'm wrong, correct me please... that's how I understood it.
agentzh diakopter: you're suggesting bundling everyting (the spec, the perldoc tool, and the test suite) into a whole? 15:41
diakopter yeah; one CPAN distribution that's released often, but that includes many packages.
agentzh and making regular CPAN release with whatever it might be called?
*releases
diakopter yes 15:42
agentzh wow, that will be a HUGE module ;)
diakopter then pugs would be released without it, but with it as a build dependency
agentzh diakopter: i think this has been discussed on #perl6 and p6l last summer :) 15:43
diakopter oh.
do you happen to remember what the conclusion was, if any?
agentzh the consensus is keeping the Perl 6 test suite in the Pugs SVN repos. 15:44
*was
diakopter okay; I wasn't suggesting changing that
agentzh i know :)
and making the test suite as general as possible 15:45
diakopter: the problem is we are still not at the point where there's a second Perl 6 implementation that can run most of the test suite like Pugs. 15:46
so testing the independency of our tools is rather difficult. 15:47
I think Pugs is still the *only* active user of the test suite since v6.pm is no longer under active development 15:48
(but v6.pm may come up again soon)
Limbic_Region actually, pmichaud's perl 6 compiler in parrot using PGE is only failing 4 of the sanity tests 15:49
moritz agentzh: but p6-in-p6 (on parrot) is evolving as well
Limbic_Region once it passes all those tests, it will be open to the entire test suite
minus any implementation specific tests of course
unfortunately, he has been plagued with family issues and it has been sitting at those 4 failing tests for some time now
to me, that's the most exciting thing going on these days 15:50
there is nothing like thousands of potentially failing tests to drive development
agentzh the perl 6 design team is looking for more people working on perl6-on-parrot. 15:51
Limbic_Region in any case, I am off to scrub toilets before my wife gets home tomorrow
avar agentzh: perl6.pir ?
agentzh avar: yeah
diakopter to me, the only problem with my suggestion is that (pmichaud told me) one of parrot's goals is to eliminate the perl5 build/smoke dependency (eventually).
Limbic_Region in the parrot repo, look at languages/perl6
agentzh diakopter: coding that in perl 6? 15:52
or PIR?
diakopter they didn't say 15:53
agentzh oh
diakopter it sounded strange to me.
I don't know who set that goal.
I suppose I might have misunderstood. 15:54
agentzh i suspect it will be PIR or something that can be compiled to PIR.
hehe 15:56
it's not very important from my point of view.
Limbic_Region The goal for parrot has always been to eliminate the perl 5 build dependency 15:57
especially for probing the system for Configure.PL
diakopter what's not important - that they want to lose the perl5 dependency?
Limbic_Region WRT smoking
agentzh adding more tests and removing explicit "pugs" depencency from the test suite is the #1 goal right now.
Limbic_Region presumably all that needs to have happen is p5 on parrot
and then port the p5 to p6
and done
diakopter how far is p5 on parrot from there? 15:58
agentzh these things are too far to vision...
Limbic_Region dunno
actually, they should have been kept in mind from the very beginning
it wouldn't be so hard to divorce them now
parrot has been around a lot longer than pugs
but again, I should be scrubbing toilets 15:59
TTFN
agentzh bye, Limbic_Region
15:59 toshinori joined
diakopter Limbic_Region: right, but at the time, it was not dreamt that parrot would accede to pugs' test suite as "official" 15:59
Limbic_Region diakopter - you are missing the point 16:00
which I haven't done a good job of making
when parrot first started development, it should have always always always had a HLL targeting it
to keep it straight
the most obvious choice there being perl 5
then anything written in p5 could have been used by parrot natively rather than as a dependency 16:01
having a high level language that was more accessible by more people would have meant finding bugs sooner
more better tests
potentially more contribution, etc
not to try and take anything away from nick and company with ponie
but I think things would have turned out a lot differently if p5 on parrot would have been there from the beginning 16:02
of course, it may have turned parrot internals into p5 internals too
so who knows
speculation speculation speculation
Limbic_Region puts on his rubber gloves (which will make typing and avoiding those toilets harder)
agentzh perl 5 is very hard to target for parrot. 16:03
diakopter agentzh: re "adding more tests and ..." - that's what I was suggesting with the separate release
agentzh given that allison is still working on perl 1.
diakopter agentzh: I thought I read that she gave up on it?
agentzh oh! 16:04
avar writing a perl1 in pir is totally different from changing perl5 to run on the parrot VM
agentzh really?
avar: *nod*
XS nightmare, source filters, and etc etc etc 16:05
geoffb_ diakopter: I thought she handed punie over to someone else, not dropped it entirely 16:06
agentzh diakopter: what's the (direct) benefit of releasing a CPAN module for the test suite? 16:07
diakopter agentzh: making it separate from pugs CPAN release.
agentzh what's it good for? 16:08
diakopter direct usage by parrot, kp6, pyrl6?
agentzh okay 16:09
how can you test it?
diakopter test the test suite?
agentzh yeah
diakopter include a dummy interpreter/compiler that fails all tests 16:10
agentzh okay, point taken.
i think that's a good idea.
at last v6.pm can pass hundreds of tests right now. 16:11
*least
diakopter another direct benefit:
this would allow much further restructuring/modularization of the current pugs svn repo
agentzh true
the current pugs test harness is a mess 16:12
i'm really looking forward to such things happen, to be honest ;)
diakopter all the various implementations with all their circular dependencies could be hierarched.
agentzh *nod* 16:13
diakopter: are you going to work on that?
diakopter well I'm certainly not going to block/stop anyone else by saying yes, so... "yes" 16:14
agentzh cool
diakopter++
diakopter unless someone with better ideas or good criticism pipes up in the near term 16:15
or beats me to the namespace
or whatever
agentzh but do choose a good name for the modules :)
diakopter okay; what do you suggest
agentzh diakopter: you can start coding under msic/pX.
diakopter draw out the tree
agentzh we certainly don't want to break everything at once ;)
diakopter: midnight here, i may not know what i'm acutally saying ;) 16:16
i'm sure you'll get a lot of suggestions from others quite soon. 16:17
diakopter: take a close look at Pugs's Makefile.PL and possibly util/prove6. 16:19
they may be a good start.
diakopter agentzh: thanks
agentzh night, all &
16:19 agentzh left 16:38 ayrnieu joined 16:50 Lunchy joined 16:51 cognominal joined 17:15 amnesiac joined
diakopter @karma diakopter 17:19
lambdabot You have a karma of 2
diakopter @karma audreyt
lambdabot audreyt has a karma of 41
diakopter I guess it got reset at some point 17:20
moritz @karma autreyus
lambdabot autreyus has a karma of 0
moritz wrong spelling I guess
diakopter @karma autrijus
lambdabot autrijus has a karma of 6
svnbot6 r16069 | agentz++ | reverted the last bogus commit. (svk bug)--
pasteling "evalbot_r16067" at 194.145.200.126 pasted "Pugs build failure" (309 lines, 17K) at sial.org/pbot/24401 17:22
17:34 SamB joined 17:44 ludanolo joined 18:02 rhr joined 18:04 moritz joined, Psyche^ joined 18:05 rindolf joined
rindolf Hi all. 18:05
18:08 ilbot6 joined
diakopter Hi rindolf 18:08
rindolf Hi diakopter
18:09 moritz_ joined 18:15 Psyche^ is now known as Patterner 18:43 larsen_ joined 18:56 sapir joined 19:12 moritz_ is now known as moritz
rindolf moritz: here? 19:18
19:18 dduncan joined
moritz rindolf: depends on your notion of "here" ;) 19:18
19:20 ozo joined 19:22 ozoo joined 19:23 ilbot-test joined
moritz rindolf: anyway, what's up? 19:28
rindolf moritz: I'm fine.
Started working on my guide to perl5 extending and embedding.
19:46 devogon joined 19:50 jisom joined 19:54 macroron joined 19:57 lb joined 19:59 lb joined 20:06 OwlEye joined 20:11 Aankhen`` joined 20:16 penk joined 20:48 whooosh joined 21:00 rlb3 joined 21:24 lisppaste3 joined 21:39 jisom_ joined
moritz did somebody implement something like a diff utility for smoke.yml files? 21:44
that gives a list of tests that passed in the last but failed in the second-to-last, and the other way round
21:53 demq joined 23:20 ilogger2 joined
svnbot6 r16071 | moritz++ | yaml-diff.pl: changed order of args to be the same as that of ordinary diff 23:31