Perl 6 - the future is here, just unevenly distributed

IRC log for #perl6, 2006-02-02

Perl 6 | Reference Documentation | Rakudo

| Channels | #perl6 index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:03 rhesa Congratulations on the new release folks!
01:07 dduncan I Nth that notion
01:16 pjcj joined perl6
01:37 justatheory joined perl6
01:38 vel joined perl6
01:55 jroc joined perl6
02:02 vel joined perl6
02:15 Khisanth joined perl6
02:18 revdiablo joined perl6
02:20 vel6608 joined perl6
02:31 vel joined perl6
02:36 dduncan I have download and am now smoking the official 6.2.11 CPAN release
02:36 dduncan er, I'm making it right now, will smoke
02:37 vel joined perl6
02:46 vel joined perl6
02:55 Amnesiac joined perl6
02:58 xinming joined perl6
03:03 justatheory joined perl6
03:14 drbean joined perl6
03:14 SamB joined perl6
03:21 azuroth "I dropped out of school at the age of 3" "why?" "because all the teachers tried to player hate on me"
03:36 dduncan now smoking the 6.2.11 release
03:36 dduncan this time, the prelude was compiled, so speed is normal
03:37 dduncan should be done in about 135 minutes
03:37 putter joined perl6
03:39 putter I'm trying to think of a name for a file.  It's like READTHEM, but more transient, less curated.  Basically, papers that went by on #perl6, that someone thought highly enough about (whether they read them or not) to add to a file.  So I don't again have to grovel over weeks of irc log trying to find a paper.  Thoughts?
03:40 xinming joined perl6
03:40 putter perhaps ideally matches READ[U-Z].*
03:42 rhesa READXTRAS? bleh
03:42 rhesa READWEITER ;)
03:43 mlh_ READTOO?
03:45 putter weiter?
03:46 putter READTOO... hmm...
03:46 putter READUP?
03:47 putter ok, READTOO seems the best so far... any other ideas?
03:47 rhesa weiter is german for further. but i agree that READTOO is the best
03:47 azuroth READYRC ;-)
03:48 putter *groan* ;)
03:48 putter oh, thanks all
03:48 putter s/oh/ok/
03:53 putter "readyrc" grumble grumble groan
03:53 svnbot6 r8938 | putter++ | READTOO: New file for papers recently mentioned on #perl6.
03:56 rantanplan_ joined perl6
04:07 kanru joined perl6
04:34 justatheory joined perl6
04:35 ozone joined perl6
05:33 rantanplan_ joined perl6
05:44 lypanov obviously it should have been READON .. :P
05:53 Daveman :p
06:04 Kattana READALSO or FURTHERREADING sound better/make more sense imo
06:06 wolverian easier to find with READ<tab>
06:06 wolverian (than FURTHERREADING)
06:07 dduncan okay, I just posted a smoke of the release version of Pugs 6.2.11, haskell runcore, darwin
06:07 dduncan no errors, as expected
06:12 revdiabl1 joined perl6
06:12 revdiabl1 is now known as revdiablo
06:30 vel joined perl6
06:32 mj41_ joined perl6
06:33 shapr left perl6
06:41 vel joined perl6
06:43 vel joined perl6
06:45 vel joined perl6
06:49 KingDiamond joined perl6
07:22 Debolaz joined perl6
07:22 integral joined perl6
07:24 rantanplan_ joined perl6
07:31 scook0 joined perl6
08:19 beppu joined perl6
08:41 offby1 joined perl6
08:42 offby1 bah.  Built r8934 from http://svn.openfoundry.org/pugs on Debian sarge; "make test" passed (first time I ever saw that -- great work!).
08:42 offby1 but now, when I run ./pugs, it dies with a segfault ...
08:42 wolverian ouch..
08:42 offby1 I _did_ try "fakeroot make -f debian/rules binary".
08:42 offby1 I wonder if that hosed something.
08:42 offby1 (probably did ...)
08:42 wolverian try without? and report back :)
08:43 offby1 well, I expect I'll have to "make clean all", which will take ... long enough that I'll fall asleep.
08:43 offby1 oh well
08:43 offby1 my fault
08:44 wolverian maybe it's faster to download again? :)
08:44 offby1 yeah, there's likely an x86 binary somewhere.
08:44 offby1 I've never done that; always built from svn, for no particularly good reason.
08:44 wolverian it's actually in debian, but I don't know if 6.2.11 is yet
08:44 wolverian (unstable, naturally..)
08:45 offby1 yeah, I'm running stable.
08:45 offby1 stuff from unstable tends to want to pull in a new libc6, which scares me.
08:45 wolverian are you sure you want to do anything at all on it? isn't stable meant for an environment that doesn't change, at all? :)
08:46 lypanov win 2
08:46 wolverian lose 3
08:46 offby1 I assume it doesn't change except for security updates.
08:46 offby1 cool zippy stuff I usually build from source, and keep well separated from the OS.
08:46 elmex joined perl6
08:47 szbalint heh, I see not only me uses irssi here. ;)
08:47 wolverian yeah. I generally steal another box from somewhere to use new things on
08:47 wolverian and keep the old one running stable
08:48 mj41_ left perl6
08:48 offby1 hmm, the "Download" link on http://search.cpan.org/dist/Perl6-Pugs/ gets me a 404
08:49 wolverian it's most likely not on your mirror yet
08:49 offby1 "make unoptimized"-ing
08:51 christo joined perl6
08:52 christo left perl6
08:54 lypanov wolverian: ;)
08:54 wolverian just realistic :)
08:59 kane_ joined perl6
09:17 G2 joined perl6
09:18 flw joined perl6
09:35 frodo72 joined perl6
09:43 gaal I'm pretty sure 6.2.11 hasn't hit unstable yet.
09:57 cognominal joined perl6
09:59 r0nny joined perl6
10:02 dada joined perl6
10:10 dduncan fyi, corresponding to Pugs 6.2.11 release on cpan, the perl 5 counterparts of my perl 6 modules bundled with it (Locale-KeyedText, Rosetta, Rosetta-Engine-Native) have been uploaded to cpan in 3 distros
10:10 dduncan after some pause, you should see them as versions 1.72.1, 0.720.0, 0.1.0 respectively
10:10 dduncan you heard it here first
10:11 dduncan those are the first post-rewrite versions of Rosetta on cpan
10:15 binary42 joined perl6
10:15 j0sephi joined perl6
10:22 Debolaz2 joined perl6
10:25 lypanov dduncan: is there a good summary of rosetta lying around anywhere?
10:25 nnunley joined perl6
10:26 dduncan there are several, of different levels of accuracy
10:26 dduncan one is in ext/Rosetta/docs
10:26 dduncan that one should take 4 minutes to read aloud
10:27 dduncan it was written 2 months before the rewrite, so some details are now false, but the general idea is still true
10:27 dduncan I hope to have better unified docs over time
10:28 lypanov thx. i'll take a look at that :)
10:28 dduncan a more up to date summary is that ...
10:29 dduncan Rosetta is a federated truly relational database whose query language is derived from Tutorial D, which is like relational algebra, and not very much like SQL
10:29 dduncan it gets its name because it is also good for making database engine independent apps
10:30 dduncan I'm also positioning it as a good tool to make object relational mappers with
10:30 lypanov how's it doing database backend wise?
10:30 dduncan the current post-rewrite version is mostly documentation, and doesn't implement any backends yet
10:30 feng joined perl6
10:31 dduncan that should happen over the next month, at least one backend
10:31 lypanov neat :)
10:31 lypanov unit tested?
10:31 dduncan even the documentation needs a lot added or changed, but I figured I should push what I had now as a snapshot
10:31 dduncan it is very component based, so it should be easy to unit test
10:32 dduncan unlike DBI, which just provides database driver independence, Rosetta provides database language independence
10:33 dduncan besides that docs/ file, look at the various pod in the lib/ directory for various pieces of info
10:43 Debolaz joined perl6
10:44 lypanov dduncan: is it a sane design that won't make me want to cry?
10:44 lypanov dduncan: dbi is awful in both api design and implementation...
10:45 lypanov dduncan: my last experience with hacking up a workaround for a bug in the db we used ended up with me writing the patch for ruby dbi and then porting as i just couldn't code otherwise :P
10:52 Debolaz2 joined perl6
11:06 dduncan lypanov, you shouldn't have any of those problems with Rosetta
11:06 dduncan it is designed to be sane, and elegant, and more or less to be mathematically/logically sound
11:07 dduncan of course, I welcome all possible assistence to make it better, including design
11:07 dduncan but I have to bed now, so that's for another day
11:08 dduncan the design is intended to be both correct and unambiguous
11:12 dduncan lypanov, who are you outside #perl6?
11:13 dduncan are you in the pugs AUTHORS file, or have a cpan id ?
11:14 dduncan I saw no "lypanov" in either place
11:15 dduncan actually, I may have found you ...
11:15 dduncan is this yours: http://www.lypanov.net/ ?
11:16 dduncan I'll assume so, since it talks about Ruby
11:16 dduncan well, good night
11:28 wilx joined perl6
11:41 wilx joined perl6
11:45 wilx joined perl6
12:00 vytautas joined perl6
13:00 iblechbot joined perl6
13:05 chris2 joined perl6
13:08 lypanov mornin' chris2
13:09 chris2 hi lypanov
13:13 chris2 lypanov: i got my toy compiler to work today =)
13:21 audreyt chris2: ooh, url?
13:22 lypanov chris2: yay!
13:22 lypanov heya audreyt :)
13:24 szbalint morning.
13:24 chris2 audreyt: nothing to see yet
13:25 Debolaz joined perl6
13:29 kolibrie joined perl6
13:29 sahadev joined perl6
13:34 pdcawley joined perl6
13:35 audreyt :)
13:35 * audreyt is still recovering after a 15hr sleep
13:38 * clkao has been sketching new svk internals with perl6 classes etc. breezy!
13:39 audreyt indeed? nice
13:39 audreyt I slept till 8pm
13:39 audreyt clkao: are you using the full set of features S12 offers?
13:39 audreyt in particular "handles"
13:39 clkao we might still have a chance to meet tomorrow. i also want to make the js backend prelude compilation not suck
13:39 clkao ya, handles. and method signatures
13:39 audreyt cool
13:40 clkao ( --> SVK::Root)
13:40 audreyt woot
13:40 audreyt ok... I may have some time tomorrow afternoon
13:40 audreyt not sure yet
13:40 clkao can one say ( --> (SVK::Root, SVK::Foo, Hash of something)) ?
13:40 audreyt characteristically of me, my cellphone is already lost
13:41 clkao i am just using it to help myself thinking.  as we know, a real language helps a lot finding out stupidness
13:41 audreyt and the answer is yes, you can write that
13:41 clkao whoot
13:41 audreyt method foo ((A, B, C) --> (D, E, F)) ought to be legal
13:42 audreyt though you'll need to use := instead of = to take advantage of the types
13:42 clkao := in ?
13:42 audreyt otherwise it distributes using normal list distribution semantics (ala p5)
13:42 audreyt my ($d, $e, $f) := $obj.foo($a, $b, $c);
13:42 clkao ah.
13:43 audreyt which is good, as it involves no copying
13:43 audreyt and can reuse the same heap objects (or registers, or boxes, or whatever the underlying runtime uses)
13:43 clkao do you still remmeber the context issue in pil? i don't see them map correctly to what Want should know.
13:43 clkao if only we start to care about performance :P
13:43 audreyt ah, but that's SEP
13:43 * clkao grins
13:44 audreyt someone else's (in particular, Leo, Nicholas and Brendon's) problem
13:44 audreyt or maybe s/Nicholas/$&+Rafael/
13:44 audreyt PILN is reasonably optimized for performance... but only when clarity is not at risk
13:45 audreyt I'm not going to pull the -optc-O3 tricks the #haskell people are doing for alioth
13:45 chris2 joined perl6
13:46 clkao how about the context thingy?
13:47 kolibrie my svk update is only pulling down up to r8820.  how do I unstick it?
13:48 audreyt svk sy -s 8821 //mirror/pugs
13:48 audreyt try that
13:48 * kolibrie tries
13:48 audreyt clkao: the static context inferencer is not what you want... JS needs its own dynamic context info
13:49 audreyt clkao: do you just want item/slurpy distinction or down to types as well?
13:49 clkao no no, Want.Scalar, Want.Void etc
13:50 audreyt so just the same as perl5
13:50 clkao perl5 backend already have want?
13:50 audreyt er no, I mean the same granularity as perl5's watnarray
13:50 audreyt wantarray, even
13:50 kolibrie audreyt: Retrieving log information from 8821 to 8938
13:50 clkao sure, but i suppose pil should give me the inforamtion?
13:50 kolibrie audreyt: thanks
13:51 Limbic_Region joined perl6
13:51 audreyt clkao: so there are static contexts and dynamic contexts
13:52 audreyt clkao: before we have "proto" support, there can be no useful static context information really
13:52 audreyt except maybe for invoking closure literals or lexical scoped subroutines
13:53 audreyt the reason for this is that the user is free to insert new variants at runtime, or rebind existing (multi)subs away into something that yields a different context
13:53 clkao er? i am talking about things like the differences between calling is(not(), undef) and (not())
13:54 audreyt clkao: yeah, (not()) is guaranteed void
13:54 audreyt but is(not(), undef) doesn't guarantee anything for not()'s context, statically, at this moment
13:55 audreyt unless &is is early bound with a type signature
13:55 audreyt docs/notes/context_*.pod attempts to fix this
13:55 audreyt as is docs/notes/mmd*
13:55 audreyt but neiter are implemented at this moment, so you have to rely on dynamic information instead of looking at the PIL tree
14:01 clkao ok.
14:07 cdpruden joined perl6
14:08 elmex joined perl6
14:25 Grrrr joined perl6
14:42 Southen_ joined perl6
14:53 stevan joined perl6
14:59 tewk Morning Stevan
14:59 stevan morning tewk
15:01 tewk Whats on the radar now the release is out the door, pug wise that is.
15:03 BakedB joined perl6
15:09 rodi joined perl6
15:10 vel joined perl6
15:12 vel joined perl6
15:12 theorbtwo joined perl6
15:13 Limbic_Region tewk - I would imagine objects objects objects for 6.28.  IIRC, audreyt wants to make that the next release
15:14 stevan Limbic_Region: yup, 6.28 is the "Object" release and will be the next one we do
15:14 stevan it should also have the new PIL2 runcode
15:17 tewk stevan: So I have an hour or two to burn, what in PILN needs tests or something like that?
15:18 stevan tewk: I think the PILN classes more needs a serious reworking
15:18 stevan we have to fix the representation type issue
15:18 stevan resolve how &bless works with &CREATE
15:19 stevan or doesnt work with it,.. I am not 100% sure yet
15:19 vel joined perl6
15:20 eric256_ joined perl6
15:20 eric256_ left perl6
15:20 * stevan is just waiting for $work to slow up a bit so he can get back to pugs
15:23 tewk Where can I find irc logs for #perl6?
15:23 putter joined perl6
15:23 putter howdy all
15:24 putter http://colabti.de/irclogger/irclogger_log/perl6
15:25 stevan heya putter
15:29 putter heya.  re piln, I have only the fuzziest picture of it myself, but perhaps there's more to it that is stable than uncertain?  on the one side, there's lots of stuff beside the type rep, and on the other, client side, a lot of the object prelude is independent of the details of the metamodel.  or am I missing the point?
15:30 szbalint hey
15:30 putter I guess my real question is "What is piln?" :)   What should I read/look at?
15:31 stevan putter: what do you mean by "a lot of the object prelude is independent of the details of the metamodel"?
15:35 Limbic_Region tewk - in general, if you want to find information about Pugs or #perl6 check the pugs homepage - the links are there
15:35 putter I was thinking in terms of  class String is Item { method mumble....}  or whatever.  where CREATE/bless decisions perhaps only affect a line or two.  but now that you ask, another concept is
15:35 stevan yeah those are the boxed types
15:36 stevan String, etc
15:36 Limbic_Region stevan - I am waiting for life to slow up so I can think about getting involved with Pugs again, write a book, learn C and Haskell up skydiving - err the list goes on and on
15:36 putter putting together an oo interface to piln code is also something which could likely be done now (assuming it hasnt been already)
15:36 stevan putter: OO interface?
15:36 putter it would be nice to be able to manipulate piln trees in p6
15:37 Eimi joined perl6
15:37 stevan it has basic method dispatch,  and a fully bootstrapped metamodel already
15:37 stevan putter: that would be nice,..
15:37 putter so how would you suggest someone get a handle on piln?  start at the test suite and work backward?
15:38 stevan putter: the best way to understand piln is to understand the Perl6-ObjectSpace stuff
15:38 putter looking...
15:39 stevan putter: the basic idea is to create a core set of "things" that the runtime must implement
15:39 stevan those core things are then used to construct the metamodel
15:39 stevan then the metamodel is used to construct the contianer types
15:40 stevan then the boxed types are created or wrapped with the metamodel
15:40 stevan then we have the containers, the boxed types and the metamodel
15:41 stevan then we build some kind of bridge between this runtime system and Perl 6
15:41 stevan that bridge is PIL2
15:45 stevan the PILN VM is basically those "core things"
15:45 stevan the fuzzy hand waving part of all this is the "bridge to Perl 6" part,.. but I think audreyt has that worked out in her head, just not in bits yet
15:45 stevan of course there are other fuzzy details too like containers and boxed types,.. but that is what the whole repr debate/issue is about
15:45 stevan basically asking the question "what is the best way to handle these"
15:45 * stevan has to run and grab some more coffee bbiab &
15:46 putter oh, drat.  ah well. ;)
15:51 Debolaz2 joined perl6
15:52 binary42_ joined perl6
15:53 putter so, my first question is, wasn't there a concept that pil-mumble was going to be "a language, like scheme, but with additional constructs to handle p6's richer semantics"?  So I was imagining I would be saying "ok, there's lambda, there's the dynamic environment, there's exceptions and continuations, ...".  no?
15:54 gaal so, trying to add exports for vars, but ruleVarDeclaration is full of distractions
15:55 hexmode joined perl6
15:55 gaal my $a is AAA, $b is export; # what's exported?
15:55 gaal my $a, $b is export # for that matter
15:55 gaal of course there are crazier combos:
15:56 gaal my $a := $b is export = 42;
15:56 putter eeep
15:56 integral ah, C's *'s problem
15:57 * putter has always been a bit puzzled by the "is export" concept.
15:57 binary42_ is now known as binary42
15:57 gaal const * const :-)
15:57 gaal putter: by itself it isn't that bad
15:57 putter Though I suppose this is a general "how do you apply....attributes(?)" to variables.
15:57 gaal it's a decoration on a var marking it as exportable
15:57 theorbtwo What's confusing about the product of two constants?  <G>
15:57 putter :)
15:57 theorbtwo Anyway, I think the answer is that is binds tightly.
15:58 gaal theorbtwo: no, actually it's an OCaml tuple of two consts :-p
15:58 theorbtwo my $a, $b is export, I think, just exports $b.
15:58 cognominal in rules what the diff between <null> and <?null>
15:58 putter is there anyone to distribute an "is mumble" over a bunch of vars?
15:59 gaal my ($a, $b, $c) is export; #??
15:59 putter null?  you mean nothing?
15:59 gaal clearly sanity is a precious thing
15:59 gaal because there's so little of it
15:59 putter (or is putter remembering the name wrong..)
16:00 vytautas joined perl6
16:00 cognominal putter: see in S06 the part with the header "nothing is illegal"
16:00 putter oh, right.  <null>.  nothing is a sub.
16:01 stevan putter: PILn is a language like scheme
16:01 stevan but a very minimal one
16:01 * gaal is very glad for that rule - I was actually bitten by null| in p5 just a few days ago
16:01 putter cognominal: <null> creates a capture (not entirely uninteresting - has position info after all), but <?null> doesnt?
16:02 stevan putter: there is lambdas, a dynamic environment, lists, hashes, numbers, strings,... no exceptions and continutations though
16:02 cognominal ok, thx
16:02 cognominal it is a long time I have not read the synopsis. MAny thingd have changed
16:03 * putter had to grep http://dev.perl.org/perl6/doc/design/syn/S05.html
16:03 putter stevan: non-determinacy?
16:03 cognominal ho that was S05 not S06, sorrt about that.
16:04 theorbtwo Mmmpf.  Is the tree not compiling now, or is my checkout toasted again?
16:05 putter r8933 was fine for me
16:05 stevan putter: what do you mean?
16:09 szbalint Is there somewhere a list of what is currently implemented in Pugs aside the source code and the verbose changelog? I'm looking for a shorter summary.
16:09 cognominal $<0 1 2> is equivalent to $/[0,1,2]  # should not that read  $/<0 1 2> is equivalent to $/[0,1,2]  ?
16:19 offby1 bah.  Built r8934 from http://svn.openfoundry.org/pugs on Debian sarge; "make test" passed (first time I ever saw that -- great work!). but now, when I run ./pugs, it dies with a segfault ...
16:19 offby1 wonder if there's something odd about not passing it any command-line arguments.
16:19 offby1 unfortunately I don't get a meaningful stack trace if I invoke it under gdb.
16:21 offby1 ... looks like it's in some sort of recursive loop -- the stack has hundreds of frames.
16:22 justatheory joined perl6
16:25 putter offby1: embedded parrot?
16:25 offby1 you know -- I don't know.
16:25 offby1 how do I find out?
16:25 offby1 (oddly, 'make test' passes.)
16:26 putter set your PATH so parrot can't be found, and try "a" ~~ /a/  ?
16:26 offby1 one moment.
16:26 putter oh, better also not have parrot where PARROT_PATH said it would be
16:27 * putter thinks it would be nice to have a $PUGS_mumble variable saying what's embedded.  or is that info available via config?
16:27 offby1 well, I "hid" parrot, so it's no longer on my PATH, but the problem persists.
16:28 offby1 the problem doesn't happen if I give it some command-line arguments: in gdb I typed <<run -e 'say "what"'>>, and that worked fine
16:28 offby1 not surprisingly
16:28 putter cognominal: $0 $1 $2  all work.  their aliases for $/[0] $/[1] $/[2].  so that may be valid?  though perhaps unnecesary
16:28 putter s/their/they're/
16:29 gaal offby1: embedded or external parrot?
16:29 offby1 gaal: I don't know.
16:29 gaal offby1: is there an old parrot installed too?
16:30 offby1 gaal: I don't have a "parrot" executable on my PATH, for what that's worth.
16:30 gaal offby1: when you ran perl Makefile.PL, did you have PUGS_EMBED include "parrot" ?
16:30 offby1 gaal: again, I don't know.  The command I typed was "perl Makefile.PL".  
16:31 cognominal putter: oh, ok. I got it. it is also equivalent to $/<0 1 2> but that was not the point that was made in the doc.
16:31 offby1 gaal: so I didn't set PUGS_EMBED by hand, but I have no idea what its default value (if any) is.
16:31 gaal It's not to embed.
16:31 offby1 OK, then I didn't ask to embed parrot.
16:31 gaal but to make sure: please ldd pugs|grep parrot
16:31 putter stevan: sorry.  an amb() or choice() operator.  perhaps with a fail() and maybe commit() to go with it.   amb(exp1,exp2), evaluate one of the exp, it decides which, and maybe you get to ask it to choose another (fail) and to cease accepting such requests (commit).
16:32 cognominal TSMWTDI!  # There are so many ways to do it!
16:32 offby1 gaal: ldd shows no "parrot".
16:32 szbalint How long should "perl Makefile.PL" take btw?
16:32 offby1 not very.
16:32 gaal offby1: okay. please strace -o /tmp/something ./pugs -e "something that segfaults"
16:33 gaal and find the last mention of parrot in /tmp/something
16:33 marmic joined perl6
16:33 offby1 gaal: one moment.
16:34 gaal oh wait! did you say just "./pugs" segfaults for you?
16:34 offby1 gaal: yes.
16:34 gaal then it's probably similar to the readline trouble people were having on MacOSX
16:34 offby1 gaal: and there's no mention of parrot whatever in the output from stracce.
16:34 gaal offby1: right, that was a red herring, sorry.
16:34 * offby1 nods
16:34 gaal you have trouble with the interactive shell.
16:35 gaal ldd pugs | grep readline ?
16:35 putter could someone with an embedded parrot look at ./pugs -V and see if there is any sign of it?
16:35 gaal putter: there is not.
16:35 offby1 gaal: libreadline.so.5 => /lib/libreadline.so.5 (0x40017000)
16:36 nothingmuch joined perl6
16:36 gaal offby1: okay, same version as mine. but i don't have the problem :(
16:36 offby1 :( indeed
16:36 gaal offby1: hmmm, how big is your src/Pugs/PreludePC.hs ?
16:37 offby1 $ ll src/Pugs/PreludePC.hs
16:37 offby1 -rw-r--r--  1 erich erich 3311450 Feb  2 01:18 src/Pugs/PreludePC.hs
16:37 gaal okay, so you have a standard precomp prelude... is the segfault immediate or after the banner shows?
16:37 offby1 what's weird is that when it segfaults, not only does pugs die (of course), but the shell from which I invoked it also dies.  I don't know if I've ever seen behavior like that before.
16:37 offby1 gaal: after the banner
16:38 kulp joined perl6
16:38 gaal I suspect the readline stuff.
16:38 offby1 me too
16:38 offby1 seems reasonable
16:38 offby1 wonder if I can re-build without readline.
16:39 offby1 I wouldn't miss it.
16:39 gaal when was your last dist-upgrade? (you said debian unstable yes?)
16:39 offby1 gaal: no, debian Sarge, which is stable.
16:39 stevan putter: in what context would that be in?
16:40 gaal We have lots of knobs in the build system, but I don't think it's possible to configure pugs to build w/o readline :(
16:40 offby1 any way to get gdb to work?
16:40 gaal well, you can trace the output of the probing and hack it to suppress it...
16:40 offby1 gaal: probing?  Are you talking to me, or to someone else?
16:40 svnbot6 r8939 | putter++ | TASKS +=1; ./pugs -V should mention whether parrot is embedded or not.
16:41 gaal gdbing a haskell program is... sanitaxing.
16:41 gaal offby1: I was.
16:41 offby1 I don't know what "probing" is in this context, and ... I don't know what "sanitaxing" is, but I can guess, and I agree :-|
16:41 gaal offby1: when you create the makefile, there's some automagic detection of your system
16:41 offby1 ah
16:41 offby1 so I can hide the readline headers and re-rerun Makefile.PL :-)
16:41 * offby1 hides stuff a lot
16:42 offby1 "sanitaxing" -- taxing of one's sanity
16:42 gaal chmod -r may be sufficient :) too bad there's no "temp" directive for unix commands
16:42 offby1 or "local"
16:42 gaal offby1: indeed, we like portmanteaus here
16:42 gaal offby1: "temp" is p6 for "local"
16:42 offby1 that is to say: you're portmantliking
16:43 offby1 gaal: ah
16:43 nothingmuch hola
16:43 offby1 see, I know so little of p6.
16:43 offby1 guess I'll just write little scripts for now.
16:43 gaal aloha mr nm sir!
16:43 Amnesiac joined perl6
16:43 offby1 is there some official way to report this as a bug?
16:43 gaal consider it reported :) but you can write to p6-c if you prefer.
16:43 offby1 nah
16:43 offby1 I'd rather consider it reported; you sound authoritative
16:44 offby1 (I'll believe anyone who sounds authoritative; it's a weakness :-| )
16:44 gaal stop believing me!
16:44 gaal (heh heh heh.)
16:44 offby1 I don't believe you when you say you're not believable.
16:44 gaal stop believing just anyone who sounds authoritative!
16:45 offby1 I'll try, I promise.
16:45 * stevan suggests gaal say that in a more authoritative tone of voice
16:45 * gaal is reminded of Smullyan riddles
16:45 offby1 Does the barber shave himself?
16:45 szbalint you're not alone offby1. Most people are like that. :)
16:45 offby1 szbalint: that's why we have the government we do :-|
16:45 offby1 it's all my fault.
16:46 * offby1 sulks
16:46 szbalint Since I don't know you personally and can't see your host I can only wonder which government you mean.
16:46 szbalint It is probably true to every one of them though. :)
16:46 putter stevan: in piln?  another, perhaps more useful construct, would be a "I don't care what order these happen in".   I don't remember the name for that.  the semantics of let (vs let*) in scheme.   its a way of avoiding adding incedental ordering constraints.
16:47 * offby1 is USA-ian
16:47 putter "$x should be set to 1, and $y to 2, but I dont care in what order this occurs".
16:47 stevan putter: PILN doesnt work on that level really, I think that will be the territory of PIL2, and it will just compile the PILN as appropriate
16:48 gaal darn, I was about to suggest to offby1 to use examples/obfu/l33t.p6 as a replacement REPL till pugs' is fixed, but I notice it's accrued a bug :-(
16:48 putter or more usefully,  @a[0..2000] = @b[2..2002] and I don't care in what order the cell assignments occur.
16:48 stevan putter: that might be handled with a lazy generator anyway
16:48 offby1 well, I'm not l33t myself, so it'd be pretentious to use it.
16:49 * stevan sees that as two lazy slices, and a lazy assignment
16:49 offby1 (plus my eyes would cross)
16:49 gaal offby1: you magically become l33t when you use it
16:49 * putter looks at the looong backlog, decides to ask
16:49 offby1 gaal: but I don't -want- be become l33t.
16:49 offby1 then I'd have to play video games.
16:49 offby1 and drink Mountain Dew.
16:49 putter gaal: so what's the one-liner summary of offby1's issue?
16:49 offby1 putter: pugs segfaults when invoked with no arguments.
16:49 szbalint Anyone who has a number in his nick, is by definiton, l33t. :)
16:49 gaal putter: interactive repl segfaults on debian stable.
16:49 offby1 is now known as dummy
16:50 putter ouch
16:50 * dummy feels better
16:50 szbalint lol.
16:50 dummy is now known as offby1
16:51 putter platform info?  ghc version?  parrot version?  gcc version (3or4)?
16:51 gaal putter: we're suspecting readline is the culprit. offby1, what does dpkg -l libreadline say?
16:52 offby1 putter: Debian sarge (stable), ghc 6.4, not using parrot, gcc (GCC) 3.3.5 (Debian 1:3.3.5-13)
16:52 gaal grrr, or wahtever package provides it
16:52 putter stevan: ok.  I was just going down my fuzzy mental list of constructs I've seen kernel languages support.  wondering what pil... "piln", yes?  supports.
16:52 offby1 libreadline5-dev                 5.0-10
16:52 stevan putter: its basically lambda calc + a few literals
16:53 gaal FWIW, on unstable I have 5.1-5
16:53 stevan most of the fancy stuff you are talking about will likely be handled in PIL2
16:53 putter offby1:  mac (version?)? linux? ...?
16:53 putter cpu?
16:53 offby1 putter: Debian linux "Sarge" (aka "Stable"), x86
16:53 offby1 model name: Intel(R) Pentium(R) 4 CPU 2.40GHz
16:53 offby1 if that matters
16:53 * stevan waves his hands to show the fabulous, mysterious and magical PIL2 ;)
16:54 putter :)
16:54 * stevan now wanders off to do some $work :(
16:54 * offby1 squints, but sees not
16:54 putter offby1: ok.  (x86_64 needs 6.4.1..)
16:55 gaal offby1: for the sake of the experiment... how hard would it be for you to try 5.1-5? I haven't been running stable for a while now so I don't remember how likely libs are likely to pull in half the world with them as dependencies
16:55 offby1 gaal: probably not too hard.
16:56 offby1 I'd uninstall the libreadline5-dev package, and install 5.1-5 from source.
16:56 gaal offby1++
16:56 * offby1 checks packages.debian.org
16:56 szbalint isn't it still down?
16:56 offby1 which is dead, per usual :-(
16:56 offby1 jinx!
16:57 szbalint heh
16:58 gaal there may be a newer 5.1 actually, I haven't updated in a... week? :)
16:58 svnbot6 r8940 | putter++ | TASKS +=0.5;  It would also be nice if ./pugs -V answered the "usual questions one asks someone with a pugs problem" - what os, cpu, ghc version, gcc version, etc.  But some of that could be hard.
16:59 putter ralf: note the surrounding discussion of a debian stable x86 "pugs segfaults on startup" issue.
16:59 * putter is starting to really like the new TASKS file ;)
16:59 gaal putter: a lot of the info in pugs -V is stolen from perl -V
17:00 putter right
17:00 putter time to add a bit of our own perhaps
17:00 putter can one see the -V info from within pugs?
17:00 gaal and fixup archname, which on my pugs says something as funny as i486-linux-gnu-thread-multi
17:01 gaal ?eval %?CONFIG.perl
17:01 evalbot_8937 is now known as evalbot_8939
17:01 evalbot_8939 "\\undef"
17:02 putter gaal: add a TASKS entry.:)  hmm, this is starting to sound like a TASKS section!
17:02 gaal oh, it's hidden in safe mode. but it's there.
17:02 putter ok.  I wonder if that's mentioned in pugsrun.pod
17:02 putter or whatever it's called
17:03 gaal ::run
17:03 putter or someplace...  where should info like that go?
17:04 putter (I was thinking of lib/pugs/run.pod, but I'm not sure that's the right place... maybe not.)
17:04 sapper joined perl6
17:05 putter ah, it's already in ./docs/quickref/var
17:09 putter gaal: so, in what way is  i486-linux-gnu-thread-multi  funny?
17:09 svnbot6 r8941 | gaal++ | Remove warning about embedded parrot as requested by TASKS.
17:09 svnbot6 r8941 | gaal++ | Add more information about `pugs -V` TASKS.
17:09 gaal putter: it's there because my perl was compiled by debian
17:09 gaal but my box is not i486
17:10 putter ahhh
17:10 putter err, but if your box isnt i486 compatible, how are you running it?
17:10 gaal of course it's i486 compatible
17:10 gaal but my pugs probably isn't
17:10 gaal even though my perl is.
17:10 * putter thinks iX86 is more a comment on what instructions the binary uses, rather the cpu of the platform...
17:11 putter ie, a i486 binary doesnt use any i586-only instructions...
17:11 gaal yeah, but pugs probably does use them.
17:12 gaal i686 ones too, even :)
17:13 gaal as for -thread-multi-, these things sound perl5ish too; doesn't multi mean I can have multiple perl5 interpreters in one process?
17:15 putter might mean threads-compatible, ie, `-mthreads'
17:15 putter     Support thread-safe exception handling on `Mingw32'.  Code that
17:15 putter     relies on thread-safe exception handling must compile and link all
17:15 putter     code with the `-mthreads' option.  When compiling, `-mthreads'
17:15 putter     defines `-D_MT'; when linking, it links in a special thread helper
17:15 putter     library `-lmingwthrd' which cleans up per thread exception
17:15 putter     handling data.
17:15 putter from gcc info
17:15 theorbtwo gaal: You're right on the definition of multi.
17:16 putter s/might mean/might also mean/
17:18 * theorbtwo wonders where mulitplicity is supposed to be documented.
17:18 putter though if pugs always uses threads (yes? maybe?), and the gcc standard libraries always supports them, it may not be useful information.
17:18 gaal I need to go now, catch y'all later! &
17:18 putter &
17:22 offby1 OK, installed newer readline library; rebuilding pugs; will let y'all know how it works when it's done (which will probably be an hour or so ...)
17:23 svnbot6 r8942 | putter++ | TASKS: tweak.
17:27 putter offby1: thanks :)
17:28 pasteling "putter" at 66.30.119.55 pasted "pugs files which match /\bPIL/i (excluding perl5/(PIL2JS|PIL-Run)/)" (115 lines, 3.5K) at http://sial.org/pbot/15713
17:29 putter stevan: could do a quick summary of what's what with the /pil/ pugs files?
17:37 putter ./ext/Test/lib/Test/PIL is used by ./t/pil/*...
17:38 awwaiid joined perl6
17:39 putter ./ext/Perl-Compiler/ is currently inactive...
17:39 putter the docs are relevant...
17:40 putter and the PIL2 src/Pugs files...
17:42 putter and ./misc/Parser-Mini...
17:42 putter and the piln examples...
17:43 putter ./src/PIL/misc is the beginnings of a prelude to run on top of piln...
17:44 putter and ./src/PIL/ ...
17:44 putter and that's it/
17:44 stevan joined perl6
17:46 putter lib/pugs/hack.pod, slightly out of date, says
17:47 putter    |   |-- Perl-Compiler      A Perl 6 port of Pugs
17:47 putter    |   |-- PIL                PIL2 implementation
17:48 putter hey stevan.  just trying to figure out what all the /bPIL/i files are...
17:49 kulp is now known as kulp|afk
17:50 * putter tries  make pil  unsure what it does...
17:50 stevan putter: src/PIL/Native/* is where everytghing is
17:51 stevan putter:  make pil will create a pil executable
17:51 stevan which evaluates pil code
17:51 stevan piln code to be more precise
17:51 * stevan wanders off the lunch
17:52 bernhard joined perl6
17:52 putter ./pil
17:52 putter ### Welcome to PIL2 REPL, the Pugs Again Shell! ###
17:52 putter pugs>
17:52 putter :)
17:52 putter thanks stevan
17:53 * putter wonders if we have a pil evalbot? ;)
17:53 stevan nope
17:53 stevan but I bet you could make one easily
17:53 * stevan really wanders off to lunch
17:54 putter :)
17:54 putter though there's the whole pesky "is safe" security thing... no?
17:55 putter hmm, pugs> 3  and pugs> say 3  and pugs> say "3"  all give errors.
17:55 * putter goes to t/ to find stuff which works...
17:56 * putter wonders if doing this online is not exactly achieving its objective of tempting people to play with piln...?
17:58 putter drat.  it seems unfortunate that trivial Makefile.PL changes force a pugs recompile...
18:02 Mister_Ed joined perl6
18:06 frodo72 left perl6
18:07 * putter wonders if Rule is(does?) Method for the sole purpose of participating in inheritance.  the minute you have multiple engines, a Rule looks more like a blob of DSL info for multi's, rather than something you would actually ever call.
18:10 putter might be nice to have an... Inheritable role... "is inheritable"...?  So random stuff can participate in class/role inheritance?   perhaps a bit evil, but...
18:11 Southen_ joined perl6
18:11 azuroth joined perl6
18:11 mlh_ joined perl6
18:17 * putter tries to picture how you would do a macro  <a=2pi*r**2> () { "L(a,2) = 2*pi*L(r)**2" }  so you can { use SymbolicCalculator; use Eqns :circle_stuff; } or { use NumericCalculator; use Eqns :circle_stuff }.  One copy of the equation, which depending on context, is used differently.
18:18 putter the "one copy of the equation" being the objective of the exercise.
18:27 putter stevan: so, I get almost a complete failure of the t/pil/ tests.  a couple of bindings work, and a couple of Class etc are defined.  all the rest fail.  is this current state, or did I do something wrong?
18:28 Mister_Ed left perl6
18:31 stevan putter: that might be my doing
18:31 stevan i checked in too much when i commited yhe changelog
18:31 gaal is there a way to protect whitespace from the <<>> quoter?
18:32 * stevan resolves to devote some time to pugs this weekend
18:32 gaal I was reminded that it does handle pairs.... <<:animal(moose)>>
18:33 gaal but can you do something like <<{"alces alces"} caribou deer elk moose>> ?
18:34 Ovid joined perl6
18:35 gaal and back to crazy var declarations: is this legal? my $a = 42, $b = 54;
18:36 gaal I'm still trying to get a sense of var decl and there's so much stuff
18:37 cognominal the comma bind less than the = I think
18:37 cognominal I would write it like in Perl5
18:37 nothingmuch joined perl6
18:37 gaal my ($a, $b) = (42, 54) I can understand
18:41 xinming joined perl6
18:41 svnbot6 r8943 | putter++ | t/syntax/parsing/numbers.t: added failing tests, :todo<bug>, for adverbial based numbers with invalid digits.  Eg, :2<9>.
18:42 putter stevan: ok, thanks.  I'll try using an earlier version.
18:43 cognominal gaal I am very surprised that it works in Perl5 too
18:43 cognominal perl -e 'use strict; my $a = 42, $b = 54; print qq|"$a""$b"|'
18:44 cognominal new idiom to me.
18:44 cognominal and with -MO=Concise, we indeed see the pad slot creation
18:45 cognominal why should it be different in Perl6
18:46 putter gaal: looking at S02, <<"alces alces" caribou deer elk moose>>  should work.
18:46 putter (grep for french)
18:47 putter The implicit split is done after interpolation, but respects quotes in a shell-like fashion, so that «'$foo' "@bar[]"» is guaranteed to produce a list of two "words" equivalent to ('$foo', "@bar[]").
18:48 gaal oh, thanks putter
18:49 gaal oh, weird, cognominal
18:49 gaal :)
18:49 putter gaal: re declarations, how about \(my $a = 2), \(my $b = 3); ;)
18:49 putter that is, an expression which returns a list of two refs
18:49 gaal GetOptions \our %Cmdline, qw(....) is an idiom I use a lot
18:50 hcarty joined perl6
18:50 putter ah.  I was quite surprised the first time i saw such.  had been thinking of my in a very non-expression (statement) kind of way.
18:51 * cognominal reads the synopsis in constant time. Each time I read them there is so much adds and changes that it takes as much time
18:52 * putter really likes google:  site:dev.perl.org  mumble mumble ;)
18:52 offby1 joined perl6
18:53 offby1 OK, about to try the new shiny pugs-with-newer-readline, to see if it too segfaults immediately
18:54 offby1 oh hell, it still seems to be linking with the stock libreadline, despite my having purged the libreadline5-dev package.
18:55 offby1 hmph.
18:55 offby1 doesn't help.
18:56 offby1 $ LD_LIBRARY_PATH=/usr/local/stow/readline/lib/ ldd pugs
18:56 offby1 libreadline.so.5 => /usr/local/stow/readline/lib/libreadline.so.5 (0x40001000)
18:56 offby1 looks good, right?
18:56 offby1 $ LD_LIBRARY_PATH=/usr/local/stow/readline/lib/ ./pugs => Welcome to Pugs -- Perl6 User's Golfing System
18:56 offby1 Type :h for help.
18:56 offby1
18:56 offby1 Segmentation fault
18:56 offby1 :-(
18:58 gaal putter: you know about \($x, $y) in p5?
18:58 gaal curious feature :)
18:59 cognominal I hate it.
18:59 gaal offby1: too bad... maybe the tail of strace has a hint?
19:00 xinming gaal: \($a, $y) is the same as [$x, $y]
19:00 gaal xinming: not in p5 :)
19:00 cognominal no!
19:00 xinming ??
19:00 xinming ...
19:00 gaal xinming: it's (\$x, \$y)
19:01 xinming ah... right. I forgot.
19:01 cognominal the \ is distributes amoung the element of the list
19:01 cognominal s/distributes/distributed/
19:02 cognominal on the other hand, if what you want to say is [$x, $y ], why using the more convoluted syntax? so that may the rational for giving it another meaning
19:03 cognominal s/may/may be/
19:04 vel joined perl6
19:04 * xinming is confused with perl 6 as he doesn't use perl 5 for a long time.
19:06 putter offby1: :(
19:06 putter gaal: re \($x,$y), eeep!
19:07 putter offby1: you could try compiling pugs without... oh, you're not compiling pugs are you?
19:07 cognominal unix always fingers this poor Segmentation guy. Wonder why he has never been fired :)
19:08 putter lol
19:09 putter offby1: binary distribution?
19:09 putter offby1: rafl, the debian... "packager"(?), will be by... at some point.  he might have thoughts.
19:11 * putter backlogs to see what the strace results were...
19:12 putter ghc 6.4.1 isnt in stable yet?
19:20 putter gaal: the mac readline issue seems unlikely to arise on linux.  the problem was mac had an out-of-date netbsd version (not quite a fork, but significantly modified) of readline.  so that's probably not it.
19:21 putter googling for ghc and gcc 3.3.5, I didn't see any obvious reports of incompatibility.
19:22 DaGo joined perl6
19:22 putter hmm... if the debian distribution is binary... is stuff compiled with ghc 6.4.1 backwards compatible with a 6.4 library (assuming one is used... it seems unlikely but possible that a debian binary is statically linked).
19:25 G2 joined perl6
19:28 putter xinming: did you send of a patch with your => to -> change?
19:29 putter (xinming noticed one of the synopsis was saying => in places where -> seemed clearly intended)
19:30 xinming putter: No, I haven't, I just asked in p6l,
19:31 putter ah, ok.  that should take care of it too.  tnx :)
19:33 * putter again notes that www.nntp.perl.org doesnt do so well with some encodings.  eg http://www.nntp.perl.org/gro​up/perl.perl6.language/24611  is a blob of encoding. :/
19:33 * putter goes to look at google's copy...
19:34 putter s/again notes/again notices/   haven't note'ed it before ;)
19:35 putter but google is out of date.  ah well.
19:36 putter maybe proofreading the synopsis should be added to TASKS as a reoccurring task?
19:36 putter s/sis/ses/
19:36 xinming putter: how to create a patch by the way? :-/
19:36 xinming diff S05.pod nS05.pod > patch.txt ?
19:36 putter diff -u  , but yes. :)
19:38 putter If it's a subdirectory, still run diff in the top pugs directory, so it's easy to apply.   diff -u foo/bar foo/bar.new should do it
19:39 putter you can test it out by running patch < patch.txt   or perhaps patch -p1 < patch.txt .
19:40 xinming putter: thanks,
19:40 xinming I'll send the patch
19:41 offby1 putter: I am indeed building pugs myself.  I'd gladly disable readline support if I knew how to.
19:42 offby1 it appears to die immediately after printing the banner:
19:42 offby1 write(1, ".", 1)                        = 1
19:42 offby1 write(1, "\n", 1)                       = 1
19:42 offby1 write(1, "\n", 1)                       = 1
19:43 offby1 --- SIGSEGV (Segmentation fault) @ 0 (0) ---
19:43 offby1 +++ killed by SIGSEGV +++
19:43 offby1
19:43 putter ah.  turned out we didnt add something as simple as an env flag, but you can just edit Makefile.PL:      if (has_ghc_package('readline')
19:43 putter        and  try_compile("import System.Console.Readline\n"
19:43 xinming offby1: what os you use?
19:43 putter if(0 and has_ghc...  should do it.
19:43 xinming offby1: I ever in debian testing, and encountered this problem.
19:44 xinming putter: Do I copy the content of the patch file to the mail. or I attach a file to send to p6l ?
19:44 putter xinming: debian stable x86.
19:45 putter how did you deal with it?
19:45 xinming offby1: upgrade the ghc to 6.4.1 in debian. apt-get install -t unstable ghc6. these 2 problems will go.
19:45 putter xinming: re patch, attach. ;)
19:45 putter xinming++
19:45 xinming offby1: hmm, also, please. use testing in debian if the system isn't a server. :-P
19:47 putter luqui: re the complexity cost of permitting 6.4.0... ;)
19:47 * putter _likes_ stable.
19:47 xinming hmm, If you wish to compile ghc 6.4.1 from source. I'd tell you "don't do that", There is a bug in make which comes with debian make the compliation fail.
19:48 offby1 xinming: my system _is_ a server, so I don't want to upgrade to testing.
19:49 xinming I tried that, and waste 2 days... (I ever thought How can the compliation progress take so long? over a day. It is still compiling...)
19:49 xinming offby1: hmm, check your make version. if it is `3.81beta4`, then, I think You should also compile make yourself.
19:50 offby1 xinming: make --version => GNU Make 3.80
19:50 putter offby1: so it looks like you have two options: edit Makefile.PL to disable readline (simple, but leaves you without it - not a big deal); or get 6.4.1 from testing (but my fuzzy understanding is grabbing things from testing can be problematic/contagious).
19:50 xinming offby1: I don't know if this version of make contains the bug I encountered.
19:51 nothingmuch joined perl6
19:51 offby1 putter: I'll try disabling readline.  Thanks.
19:51 putter luqui: ok, I note that requiring 6.4.1 would have here caused even _more_ problems. ;)
19:52 putter offby1: np
19:52 xinming putter: hmm, I think he'd better install ghc 6.4.1 as in 6.4 , the complied program might be buggy.
19:54 putter perhaps, but it's (pugs) not being used in production, there aren't yet reports of such, and, correct me if I'm wrong, mixing debian testing into stable is not something to be done lightly.
19:54 putter no?
19:55 xinming mixing testing with unstable is much a better choice than stable + testing. :-)
19:55 offby1 not by me, anyway
19:55 putter (well, the readline would be such a report... how about any _other_ reports;)
19:58 putter stevan: the r8912 did modify a bunch of src/PIL/ files
19:58 putter should they be reverted?
19:59 stevan putter: yeah I have a copy and so will the VCS
19:59 putter k
20:05 Debolaz joined perl6
20:11 lightstep joined perl6
20:12 offby1 joined perl6
20:14 svnbot6 r8944 | putter++ | src/PIL/Native/Bootstrap: revert various files unintentionally modified by r8912.
20:17 gaal "omniscient debugger" google tech talk: http://video.google.com/videoplay?docid=3​897010229726822034&amp;q=Google+techtalks
20:18 nothingmuch omniscient debugging is cool
20:18 offby1 sounds like domestic surveillance.
20:20 KingDiamond joined perl6
20:23 putter hmm, can't watch the video just now.  but debuggers that can slide around in time have been around for a long time... the simplest approach being schemes which roll forward to the t of interest, using checkpoints to avoid having to keep going back to the beginning of the program.
20:24 putter wonder if nifty new stuff has been added?  or maybe just "the tools we're using still suck compared to what we know how to do".
20:25 putter :/
20:26 offby1 :\
20:26 offby1 that makes a "K".
20:27 frederico_ joined perl6
20:30 chris2 joined perl6
20:31 offby1 hooray -- disabling readline (by nixing the relevant stuff in Makefile.PL) fixed my problem on Debian.
20:32 offby1 now I wonder if I can get the functionality back, without provoking the bug, by just using "rlwrap" ... :-)
20:34 offby1 hee hee!  rlwrap works fine.
20:34 putter *laugh*
20:34 offby1 strange but true.
20:34 offby1 thanks folks.
20:34 offby1 left perl6
20:36 putter stevan: the reversion fixed all the tests but those in t/pil/container/  which entirely failed.  expected?
20:37 frodo72 joined perl6
20:42 offby1 joined perl6
20:43 gaal putter, what does the tweak in r8942 mean? %?CONFIG has whatever -V does, it's just not available in safe mode hence for evalbot
20:43 offby1 left perl6
20:44 gaal -V is basically $_.perl.say for %?CONFIG, @*INC
20:44 putter I just wanted to mention it's existence.
20:45 Shillo joined perl6
20:45 gaal okies
20:45 putter one of the ideas for TASKS is for people who have the "big picture" to brain dump that which is needed for someone without it to solve the problem.  the existence of %?CONFIG seemed like such a thing to mention
20:46 gaal yes, good idea. I wish I could have a chat with someone with a big picture re: ver decl
20:46 gaal *var decl
20:47 gaal but for now I must afk! see you.
20:47 putter bye & :)
20:47 * SamB wonders why he is now considered an admin
20:48 putter yeah, right now our only "request for clarification" mechanism is p6l posts... I wonder what else might be done...?
20:49 gaal putter: get larry and audreyt together in the same room as oneself for a week, as in fact is going to happen in three weeks :-)
20:49 putter SamB: because everyone who contributes is a committer, and everyone who.... becomes an admin?
20:49 gaal *ppf*
20:49 SamB but... but... I didn't do anything!
20:49 putter gaal: that sound like great fun :)
20:49 putter lol
20:50 putter then it logically follows that mere existence is sufficient... ;)
20:50 SamB well, I mean, I did get commit access somehow
20:51 SamB I don't remember doing anything useful
20:51 SamB I think I broke building on non-GHCi platforms
20:52 integral well now you can recruit someone to unbreak it ;-)
20:52 putter My fuzzy impression is the idea behind spreading around the admin bit is so there is always someone here who can add new committers.
20:52 SamB it was fixed months ago
20:53 Shillo Hmm, one question, about specs. S06 mentions tokens, which should be a special kind of rules. Where are they described?
20:53 Shillo Don't remember seeing them on A05 or S05.
20:54 stevan putter: yes, the container tests should still fail
20:56 Shillo Hmm, also, does anyone else get insta-crash running PUGS interactively? I suspect readline.
20:57 putter Shillo: re tokens, I don't know.  perhaps
20:59 putter they are related to the as yet undocumented(?)  top-down/operator-precedence/top-down parser sandwitch ... but it may also jut be historical artifact?
20:59 putter re insta-crash, bet you have ghc 6.4.0 not 6.4.1, yes?
21:00 putter that if you're running linux.  on mac, readline.
21:00 * Shillo facepalms.
21:00 putter facepalms?
21:00 Shillo And I -did- read that change.
21:01 Shillo Yeah. As in, cover my face with the palms of my hands.
21:01 * putter is confued... and my s-key is intermittant...
21:01 putter ah.  which change?
21:03 * putter rediscovers "is parsed" in S06... decides he really really wants it... ;)
21:04 Shillo 6.4 -> 6.4.1 required
21:04 Shillo Although, pugs works for me just fine if I just run scripts.
21:05 putter actually, while that was talked about, 6.4.1 is only required on x86_64, and, well, strongly encouraged on mac osx 10.4.  the debian dependency on 6.4.1 was not expected.
21:05 Shillo I think it's readline, then.
21:05 Shillo pugs itself does work.
21:06 putter what platform?  mac?  linux?
21:06 theorbtwo SamB: email address?
21:06 putter Multivac: can entropy be reversed?
21:06 * theorbtwo wonders if he can figure out again how to give people committer bits.
21:06 theorbtwo (WHile drunk.)
21:06 SamB theorbtwo: you want to know my email address?
21:07 Shillo putter: Ubuntu
21:07 theorbtwo I want to give you a comitter bit.
21:08 SamB theorbtwo: oh, what happened to my old one?
21:08 theorbtwo i thought you just said you didn't get one...
21:08 theorbtwo nevermind.
21:08 putter theorbtwo: SamB was just saying he unexpectedly got an admin bit... o I assume he ha (d*mn s) a commit bit
21:09 lightstep left perl6
21:10 putter Shillo: ah.  right.  that is readline related too.  yes, the options are upgrade to 6.4.1 or fiddle with Makefile.PL to build without readline support (and then you can rlwrap to get it back)
21:10 Shillo putter: Hmm. :)
21:10 Shillo I'm fine with running noninteractive for now, actually.
21:10 Shillo I kinda want SLIME for p6.
21:11 putter there may be other options... there's only been one case so far...
21:11 putter or two
21:11 putter SLIME for p6 ++
21:14 * Shillo grins!
21:14 putter I started to look into being a slime backend... http://common-lisp.net/project/slime/  ... swank...  But I guess I got distracted.  Next step would be to grab the source and see if the protocol is something we can easily handle.
21:15 Shillo And in-language support. SWANK introspects everything.
21:15 Shillo Like Intellisense on steroids.
21:16 Shillo Protocol is just a socket.
21:16 Shillo Probably textual, beneath that.
21:17 Odin- Hey, once that gets working, how about integrating Perl6 into emacs? :p
21:17 theorbtwo A while back I looked at smart syntax hilighting, but didn't get terribly far.
21:17 * putter goes looking for ghc debugger...  we may be too "compiled" to make this easy until there is more of a pure p6 implementation...
21:18 Odin- (After that, I think the only remaining thing would be to turn emacs into a Mozilla-y thingamajig. Whoo, all the big swiss army knives rolled into one.)
21:18 Debolaz2 joined perl6
21:18 putter theorbtwo: the sticking point was... not enough position annotations?
21:18 Shillo Hehe. Well, I suspect something specs enough introspections for things to work.
21:18 theorbtwo Even at the Parse-YAML level, we don't have enough position information, or possibly enough syntax left, for it to be very um, um, informative.
21:18 theorbtwo putter: That, and 90% of them were just "app".
21:19 putter ah
21:19 ingy hola
21:19 theorbtwo I'll put my code somewhere if you want it.
21:19 Odin- ingy: You know, that translates into "a hole" in Icelandic.
21:19 putter Odin-: :)
21:20 theorbtwo It's kind of in a half-done state, where it extrats information, but doesn't actually do anything with it.
21:22 putter theorbtwo: instead of using the pug parser, I wonder if pge is far enough along that one could start using a grammar for p6 to grovel over the file... no, I guess that's a bad idea.  need a more informative parser then...
21:22 Shillo Does parrot support shared substrings?
21:22 putter hi ingy
21:22 Shillo If yes, you get position info for free from the rules matches, you only need to add methods to substrings to tell you what they're pointing to.
21:23 ingy hi putter
21:23 theorbtwo The problem with using the rules engine for this is that pugs doesn't use the rules engine, so you can get into situations where what your hilighter says is going on doesn't match reality, and avoiding that was the whole point.
21:24 Shillo theorbtwo: Isn't that a self-solving problem, since it will have to use rules eventually?
21:25 putter Shillo: the gotcha with just using rules to grovel over p6 source is that p6 is too much a live thing.  not so much now, because macros and such are barely implemented, but increasingly.  you basically cant parse p6 without a live p6 system.
21:26 Shillo Hmm.
21:26 theorbtwo Well, rules should be a live perl6 system, when done.
21:26 theorbtwo In the meantime, however...
21:27 putter theorbtwo: though one idea would be to start using rules (putting together a p6 grammar is a worthwhile goal in and of itself), which gives you position info, which you can pretend came from the parser.  which lets you get the rest of the system up.  then when the parser provides more info, you can just switch over..?
21:27 theorbtwo Or when your grammar is good enough, pugs can switch over.
21:27 putter ;)
21:28 Shillo The main problem, actually, are string-returning macros. AST-returning macros would be nice and neat.
21:28 xinming In fact, every time you face this excited idea of perl 6. You'll be surprised..
21:29 putter one of the questions I want to ask audreyt is what her vision for the mix of haskell and p6 is.  ie, should we be adding statement-control:<while> or however that's specced, and yanking it from Parser.hs?  or not?
21:29 Limbic_Region /quit tiuq\
21:31 putter theorbtwo: actually, you could argue you want _both_ sources of info - heuristic patterns and the parser tree.  because you still want to highlite etc code even when it's broken, when the parser will be of little help.
21:33 putter have you thought about how say emacs(?) and this analyzer interact?
21:33 Khisanth joined perl6
21:33 dduncan joined perl6
21:34 putter if i did slurp ~~ /^(sub|class|package)\s+(\w+)/  how much more would I need to do basic syntax coloring?
21:35 Shillo slurp can be overloaded. So can ~~. /.../ can be redefined (right?)
21:36 Shillo sub prefix:slurp ...; :)
21:37 putter I'm just wondering if one can usefully get started immediately, with a relatively small analyzer, get it plugged into emacs or whatever, and start gaining experience.
21:38 Shillo Er, yeah. Now I reread what you said and figured I misunderstood.
21:41 putter http://www.gnu.org/software/emacs/elisp-m​anual/html_mono/elisp.html#Font-Lock-Mode
21:41 frederico_ joined perl6
21:46 putter bbiab
21:46 Shillo Laters, all.
21:46 Shillo left perl6
22:10 hcarty left perl6
22:29 dudley_ joined perl6
22:32 rantanplan_ joined perl6
22:38 Shachaf joined perl6
22:43 rafl putter: You rang the bell?
22:50 Debolaz joined perl6
22:55 putter hey rafl
22:56 putter just to note that some debian stable folks running 6.4 were getting segfaults on pugs startup.
22:56 rafl putter: Which pugs?
22:56 putter works if readline support is disabled by munging Makefile.PL
22:56 rafl putter: And I don't even provide packages for debian stable.
22:57 * putter backlogs...
22:58 putter ok, it looks like compiling pugs on debian stable or testing with 6.4 causes the described behavior.
22:59 putter which I guess is not your problem ;)
22:59 putter any suggestions in case it comes up again?
23:00 putter on non-stable, one can upgrade ghc.  on stable, one can edit the Makefile.PL.  any other options come to mind?
23:00 nothingmuch rafl: how's the smoke server coming along?
23:00 rafl nothingmuch: I waited on YAML 0.5 a while. Now that it is out I could continue, but it's exam time.
23:01 rafl nothingmuch: I'll continue work in 1.5 - 2 weeks.
23:01 putter rafl: anyway, that was the ping.  problem+debian=>consult-ralf
23:01 putter arg
23:01 putter rafl
23:02 putter thanks
23:02 rafl putter: Who had the problem?
23:02 szbalint offby1
23:03 nothingmuch rafl: =)
23:04 rafl Ah. I think he should contact me so I can get some additional info.
23:05 putter offby1 http://colabti.de/irclogger/irclogger_log​/perl6?date=2006-02-02,Thu&amp;sel=31#l70  down thru 20:34.
23:06 weinig joined perl6
23:06 putter he had the stable.  xinming reported (briefly) the same problem on testing.
23:07 putter offby1: if you ever see this, ping rafl
23:08 Debolaz joined perl6
23:15 Limbic_Region joined perl6
23:16 justatheory joined perl6
23:24 feng123 joined perl6
23:27 cdpruden joined perl6
23:41 azuroth joined perl6
23:41 svnbot6 r8945 | putter++ | misc/IDE-support: a silly little demo of emacs calling a p6 program which then highlights the emacs buffer from which it was called.  The idea is to have a p6 server which provides source code analysis and refactoring services to editors and IDEs.
23:44 putter theorbtwo: here's a toy.  if you're interested in pursuing it, let's talk.
23:44 putter or if anyone else is ;)

| Channels | #perl6 index | Today | | Search | Google Search | Plain-Text | summary

Perl 6 | Reference Documentation | Rakudo