Camelia, the Perl 6 bug

IRC log for #parrot, 2008-04-09

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:05 kid51 Infinoid:  What does 'brute force' mean in this context?  (the problem seems to be with make test as much as Perl::Critic)
00:10 kid51 On another subject:  Do you have anything to add re http://rt.perl.org/rt3/Tic​ket/Display.html?id=52154 -- the ticket having to do with pulling 'svk' and 'git' related code out of Parrot::Revision?
00:11 kid51 Coke will have to make a decision, because it's one of those things where we can't make everybody happy.
00:15 TonyC joined #parrot
00:25 Infinoid kid51: by "brute force", I meant: attack the problem of "why can't it find my module" head-on, by verifying the environmental things that could affect module loading
00:26 tetragon joined #parrot
00:27 Infinoid does bitcard have any "don't forget about my login" related preferences?
00:27 nopaste "kid51" at 71.247.48.190 pasted "'prove' finds Perl::Critic with no problem; why not 'make test'?" (19 lines) at http://nopaste.snit.ch/12662
00:28 kid51 Infinoid:  Don't know; unfamiliar with Bitcard except it simplest usage.
00:28 kid51 Would that have a bearing on problem at hand?
00:29 Infinoid just an annoyance at being bounced to a page that doesn't have "Reply" buttons, no big deal
00:30 Infinoid hmm.  I don't really feel too strongly about #52154.  if the concensus is to keep svn around, I won't challenge it, as long as everything else is simplified
00:32 paco joined #parrot
00:32 Infinoid you could pretty easily parse the svn rev out of "svn info".  but the same argument could be made of Configure.pl doing so
00:32 Infinoid I think the main issue was the pain of having to support svk, git, git-svn, and that is being addressed
00:35 Infinoid so ... do you want me to post that to the ticket?
00:37 kid51 Yes, please post whether you agree with the patch or not and what any outstanding issues may be.  (Asking you 'cause I remember our conversation on IRC 2 weeks ago about this.)
00:37 Infinoid re: Data::Dumper, ok, that's bizarre.
00:42 kid51 Now, something equally bizarre ... but a pattern is developing:
00:42 nopaste "kid51" at 71.247.48.190 pasted "Where t/*.t files do an 'eval {require ...}' to locate a Perl module" (12 lines) at http://nopaste.snit.ch/12663
00:44 Infinoid ok, I've cast my vote
00:45 kid51 Ooh, I've got a hunch!
00:45 Infinoid it almost looks like the harness is doing something that breaks the *first* use, or require, or whatever, in interesting ways
00:45 Infinoid I got a weird warning from YAML::Syck in that patch I nopasted earlier, too.
00:45 Infinoid something about a barename only being used once, possible typo
00:46 Infinoid what's the hunch?
00:46 kid51 My hunch:  Note that there's no semicolon after Perl::Critic -- nor is there one after Moose in that other test.
00:46 Infinoid hmm.  do BEGIN blocks get concatenated together before execution?
00:47 kid51 But why 'make test' would gag there when 'prove' and 't/harness' don't, I cannot explain.
00:47 chromatic The semicolon issue is a red herring.
00:47 chromatic The final statement in a block does not require a semicolon in Perl 5.
00:47 kid51 Okay, then what color is the herring?
00:47 chromatic Red.
00:48 * kid51 set himself up for that one.
00:48 chromatic Dump @INC to STDERR from the test file and see what's different between runs.
00:48 Infinoid chromatic: http://nopaste.snit.ch/12662  (standalone)
00:49 Infinoid kid51: does "perl t/harness t/codingstd/perlcritic.t" look the same?
00:50 kid51 It looks same as 'prove' i.e., it locates Perl::Critic and runs the file without complaint.
00:50 Infinoid so, the harness is fine, something from the Makefile maybe?
00:51 kid51 for use of perl t/harness, see my post to list of 17:18:09 PDT
00:52 Infinoid on my box, "make test" actually runs: /usr/bin/perl5.10.0 t/harness --gc-debug --running-make-test
00:52 chromatic `which perl`
00:53 Infinoid /usr/bin/perl  (symlink to perl5.10.0)
00:54 chromatic Are you seeing this craziness too?
00:54 Infinoid no.
00:56 kid51 chromatic:  When I include the semicolon, 'make test' runs t/codingstd/perlcritic.t without complaint!
00:57 chromatic That's unique.
00:57 chromatic $ perl
00:57 chromatic eval { require CGI::Cookie };
00:57 chromatic print $@;
00:57 chromatic print $INC{'CGI/Cookie.pm'}
00:57 chromatic /usr/local/share/perl/5.8.8/CGI/Cookie.pm
00:57 kid51 Note:  I realize that I don't actually have Moose installed
00:59 kid51 Well, I guess it can't hurt to put that semicolon after 'require Perl::Critic' -- correct?
00:59 chromatic Correct, but I fear it's covering up some sort of weirdness elsewhere.
01:00 kid51 Shall I commit, or do you want to explore further?
01:00 Infinoid c, can you reproduce?
01:00 chromatic No, I can't.
01:01 Infinoid like I said, I don't see this craziness, but I do see other craziness.  like, warnings from modules that normally load silently.
01:01 kid51 Infinoid:  You see such warnings ... when?
01:01 chromatic I only get that from Text::Balanced.
01:01 Infinoid Name "YAML::Syck::ImplicitBinary" used only once: possible typo at /usr/lib64/perl5/vendor_perl/5.10.0/x​86_64-linux-thread-multi/YAML/Syck.pm line 65.
01:02 Infinoid I get that under perl t/harness, but not under prove.
01:02 * kid51 doesn't have YAML::Syck installed anywhere.
01:02 chromatic prove -v
01:02 * Infinoid can't even begin to explain this.
01:02 chromatic I mean prove --version
01:03 Infinoid no warning under prove -v
01:03 Infinoid oh
01:03 Infinoid prove v2.64, using Test::Harness v2.64 and Perl v5.10.0
01:03 chromatic TAP::Harness v3.02 and Perl v5.8.8
01:04 chromatic Hm, I thought 5.10 included TAP::Harness 3.
01:04 nopaste "kid51" at 71.247.48.190 pasted "patch to assure that 'make test' locates Perl::Critic" (27 lines) at http://nopaste.snit.ch/12665
01:05 kid51 Guys, I gotta step away.  If you want to (literally) patch the problem, there's a suitable patch.
01:06 Infinoid 5.10 does not seem to include TAP::Harness
01:10 chromatic Maybe it just missed the cutoff.
01:10 chromatic Do you see the weirdness with TAP::Harness?
01:22 Ademan_ joined #parrot
01:38 pmichaud anyone think we can close rt#51732?
01:38 chromatic I'm sure it was a lacking realclean.
01:40 Infinoid chromatic: yes, I still see the weirdness with TAP::Harness v3.10
01:40 Infinoid but I still don't know whether my weirdness is related to kid51's weirdness.
01:44 chromatic The missing semicolon is bizarre.
01:47 Infinoid what do you think our possibilities of coming up with a oneliner that reproduces it are?
01:47 chromatic Pretty low, unless it includes an -I
02:02 kid51 joined #parrot
02:02 Coke (makefile) I'm aiming to get all the dependencies worked out. having -j work is a nice side effect there.
02:04 Coke kid51: are you sure you're using prove and perl from the same distro?
02:06 kid51 My Perl is 5.10, compiled from source the day it came out.  My prove is that associated with most recent Test::Harness (3.10)
02:06 kid51 Darwin:  [jimk] 503 $ prove --version
02:06 kid51 TAP::Harness v3.10 and Perl v5.10.0
02:07 Coke was that test::harness installed against the same perl?
02:07 kid51 Linux:  [li11-226:~] 501 $ prove --version
02:07 kid51 TAP::Harness v3.09 and Perl v5.10.0
02:07 Coke (just that would explain the library discrepancies.)
02:07 kid51 Yes.  Because TAP::Harness 3.10 is more recent than Perl 5.10.
02:08 kid51 But note that this error only occurred during 'make test' and only when perlcritic.t read:  eval { require Perl::Critic };
02:09 kid51 But it occurred, albeit with different output, on 2 different OSES.
02:09 kid51 OSES.
02:09 kid51 OSes.
02:10 kid51 Also, note that since perlcritic.t was only very recently included in 'make test', I wouldn't have noticed this problem, say, 3 months ago.
02:12 pmichaud kid51: do you think we can close rt#51732?  (I'm reviewing tickets in my inbox and happened across this one.)
02:12 Infinoid are you using perl 5.10 on both OSes?
02:12 kid51 Yes.  But here's some new data.
02:13 nopaste "kid51" at 71.247.48.190 pasted "Using the 'vendor' Perl and prove on Darwin to run perlcritic.t" (8 lines) at http://nopaste.snit.ch/12666
02:13 chromatic That depends on where you have P::C installed and what's in @INC.
02:13 kid51 That's the first case where I can use a version of 'prove' to simulate the results with make test.
02:14 kid51 Err, let's see.
02:15 Infinoid that one's probably because P::C is installed in /usr/local/lib/perl5/site_​perl/5.10.0/Perl/Critic.pm
02:15 Infinoid I doubt that's in 5.8.6's search path.  do you have another copy installed?
02:15 kid51 Yes.  the @INC for 5.8.6 does not include the directory where P::C is installed.
02:15 Infinoid sorry
02:16 tetragon I'm installing P::C right now on my Darwin install
02:16 tetragon (it used to be there, but was lost in a system upgrade)
02:16 kid51 tetragon:  All those dependencies are fun, right?
02:16 tetragon hehe
02:16 tetragon I think people grumble about LedgerSMB's dependencies more
02:17 kid51 I met Elliot, the current maintainer of P::C, when I was at the mini-hackathon in Chicago in December.
02:17 kid51 Nice guy, but I still hate modules where the cpan shell forces me to say yes 3 times.
02:17 kid51 Mainly 'cause it's sitting on PPI.
02:18 Theory joined #parrot
02:18 kid51 So, would my 'make test' -- on either OS -- for some reason not know to look in the @INC associated with 5.10?
02:18 kid51 And why would a semicolon make a difference?
02:19 kid51 Perhaps with 5.10 you need the last semicolon in a block???
02:23 tetragon Incidentally, vendor Perl on OS X 10.5 picks up on the P::C I just installed with prove
02:23 tetragon And that test is swamping my system
02:24 kid51 which test?
02:24 tetragon t/codingstd/perlcritic.t
02:24 chromatic Unless something changed the parser in a backwards-incompatible way, the semicolon isn't the problem.
02:24 kid51 Really?  It's not particularly speedy on my iBook G4, but there are others slower.
02:24 tetragon It has finished
02:25 chromatic It's pretty IO intensive.
02:25 tetragon And subtest 3 failed
02:25 tetragon It was consuming all CPU
02:25 kid51 The question we've been discussing is that 'make test' on my boxes fails to locate Perl::Critic and therefore SKIPs that test.
02:25 tetragon The failed subtest looks like it was my modifications
02:25 kid51 Oooooh.
02:26 tetragon I patched config/init/defaults.pm so that I wouldn't have to remove all the -arch flags from all the Makefiles
02:26 kid51 Ah, yes, *that* patch :)
02:29 dalek r26864 | coke++ | trunk:
02:29 dalek : Add a TODO test for #46499: [RFE] Allow comment lines in PIR .param list
02:29 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26864
02:54 Eevee joined #parrot
03:11 silug joined #parrot
03:16 dalek r26865 | infinoid++ | trunk:
03:16 dalek : [t] Make sure parrot returns non-zero on parse failure, and doesn't generate
03:16 dalek : an output file.
03:16 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26865
03:19 Infinoid pmichaud: so far, parrot's return value looks good.  but my test case is pretty simple, and might need expanding.
04:14 pmichaud Infinoid++
04:36 allison joined #parrot
04:52 chromatic http://www.cs.dartmouth.edu/r​eports/abstracts/TR2001-404/
04:58 dalek r26866 | pmichaud++ | trunk:
04:58 dalek : [pct]:
04:58 dalek : * Allow --encoding=fixed_8 to work for HLLCompiler.
04:58 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26866
05:12 silug joined #parrot
05:12 dalek r26867 | pmichaud++ | trunk:
05:12 dalek : [rakudo]:
05:12 dalek : * Since actions.pm doesn't have any non-ascii characters in it yet,
05:12 dalek :   we can tell NQP to compile using fixed_8 instead of the utf8 default.
05:12 dalek :   This gives us a 28% (101 versus 141 sec) speed improvement on
05:12 dalek :   my box.
05:12 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26867
05:18 chromatic Nice.
05:18 chromatic I'll take temporary victories until we can fix the real things.
05:43 iblechbot joined #parrot
06:14 uniejo joined #parrot
06:25 Theory joined #parrot
07:08 Ademan_ joined #parrot
07:12 silug joined #parrot
07:34 cosimo joined #parrot
07:38 AndyA joined #parrot
07:57 AndyA joined #parrot
08:08 dalek r26868 | chromatic++ | trunk:
08:08 dalek : [Rakudo] Fixed tests temporarily broken in r26813 with type hierarchy
08:08 dalek : rearrangement; TAP doesn't support marking failing tests with the SKIP
08:08 dalek : directive.  These tests are now TODO tests, which allows them to fail without
08:08 dalek : penalizing the test run.  We know they'll fail, and we'll fix them.
08:08 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26868
08:23 ruoso joined #parrot
08:35 AndyA joined #parrot
08:37 IllvilJa joined #parrot
08:41 cognominal joined #parrot
08:47 AndyA_ joined #parrot
09:12 silug joined #parrot
09:30 AndyA joined #parrot
09:32 slightlyoff joined #parrot
09:44 AndyA joined #parrot
09:44 mj41_ joined #parrot
10:10 iblechbot joined #parrot
10:14 ask_ joined #parrot
10:17 AndyA joined #parrot
10:27 AndyA joined #parrot
10:38 AndyA_ joined #parrot
10:39 ask_ joined #parrot
10:42 AndyA joined #parrot
10:56 AndyA joined #parrot
11:23 Psyche^ joined #parrot
11:23 tetragon joined #parrot
11:42 Psyche^ joined #parrot
11:57 contingencyplan joined #parrot
12:30 Ademan_ joined #parrot
12:34 Coke seen ambs?
12:34 purl ambs was last seen on #parrot 16 hours and 5 minutes ago, saying: :D
12:34 * Coke thwaps ambs. =-)
12:36 AndyA joined #parrot
12:50 IllvilJa joined #parrot
12:57 iblechbot joined #parrot
12:58 Coke ooooh. we may be able to get someone much more familiar with smolder to help out.
13:26 particle that'd be lovely
13:27 rdice joined #parrot
13:32 Coke http://trac.macports.org/projects/macport​s/changeset?old_path=trunk%2Fdports%2Flan​g%2Fparrot&old=35891&new_path=tru​nk%2Fdports%2Flang%2Fparrot&new=35890
13:32 shorten Coke's url is at http://xrl.us/bi5si
13:32 Coke yay.
13:33 Coke I think this means that the 0.5.2 port for macports will get applied shortly.
13:34 particle you wanna skip 0.6.0 and go straight for 0.6.1 next week?
13:35 Coke yah.
13:35 Coke (as far as the macport goes.)
13:36 * particle wonders when ubuntu 8.04 is due out
13:36 particle ...obviously, this month sometime
13:37 dalek r26869 | coke++ | trunk:
13:37 dalek : [ports]
13:37 dalek : Update port to reflect change in [maintainer] in portfile-current
13:37 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26869
13:41 particle coke: do you get a failure in t\01-sanity\07-simple-multisubs.t if you run make test in perl 6?
13:42 Coke particle: svn upping
13:47 Coke particle: no.
13:47 particle crappity.
13:48 Coke ... but then, it doesn't seem to run.
13:48 particle i get 'push_pmc' not implemented in 'Closure'
13:48 Coke oh, wait, 07. no.
13:48 Coke (it does run)
13:48 Coke try with -G?
13:49 particle i've gotta look up how to do that
13:49 particle ah, right.
13:49 particle push_pmc() not implemented in class 'Closure'
13:49 particle errorlevel = 1
13:49 particle with -G
13:53 Coke I would expect TEST_PROG_ARGS=-G make test to work.
13:53 Coke particle: huh. I ran my test on svn latest with osx/x86
13:54 particle yeah, jonathan doesn't see the error, either
13:54 particle i don't have any mods to the Closure pmc
13:56 Coke can someone on a mac tell me what "port list parrot" shows?
13:56 Coke (someone with macports. =-)
13:57 gryphon joined #parrot
14:09 Andy joined #parrot
14:16 pmichaud good morning
14:16 purl Is it morning again?  YAWN...
14:16 pmichaud particle:  I'm running kubuntu 8.04 beta, if that helps
14:20 particle pmichaud: i think i'll dl the ubuntu beta now
14:20 pmichaud particle:  so far I really like 8.04
14:20 particle my 7.10 vm stopped accessing the network. can't figure out why
14:20 pmichaud I really liked 7.10 also, but 8.04 just feels even more solid than 7.10 did
14:20 pmichaud (and that's surprising, because I recall being impressed with how solid 7.10 felt.)
14:21 pmichaud I tried out 8.04 beta on my laptop, and after a day decided I wanted it for the desktop as well
14:21 uniejo joined #parrot
14:24 Tene Coke: http://www.macports.org/ports.​php?by=name&substr=parrot
14:24 Tene @0.4.0          lang/parrot
14:26 Coke bah!
14:29 Tene What, you didn't like release 0.4?
14:30 Coke see my mail to the mailing list saying that 0.5.2 was available. =-)
14:33 particle maybe it takes a bit to propagate
14:36 Coke The selfupdate command installs the latest port revisions from subversion (at a slight delay),
14:36 Coke ... but no word how long slight is. =-)
14:41 Coke no more than 1/2 a day, based on feedback from #macports.
14:41 AndyA joined #parrot
14:43 rotty joined #parrot
14:43 Coke it would be most awesome if we could get the install of the perl6 faux-cutable done for 0.6.1
14:47 nopaste "particle" at 24.19.3.148 pasted "pmichaud: how do i get this working?" (14 lines) at http://nopaste.snit.ch/12667
14:48 pmichaud does rakudo know how to do mmd yet?
14:48 pmichaud as opposed to arity-based multisubs?
14:48 pmichaud (short answer:  I haven't studied jonathan's class patches enough yet to know for sure)
14:49 particle i'm trying to find out
14:49 particle it's not parsing atm
14:49 particle Statement not terminated properly at line 6, near "(Thing    "
14:49 particle maybe we can't do types in params yet....
14:49 pmichaud yes, I'm still not caught up on all of the changes to rakudo yet as far as classes/types are concerned
14:50 particle i'll ping jonathan
15:00 Coke particle: is there anything you're specifically looking for in 0.6.1 this month?
15:06 particle coke: ease of release
15:07 uniejo joined #parrot
15:07 particle it's maint, so no big feature requests
15:12 paco joined #parrot
15:19 pasty joined #parrot
15:24 rotty where can I find docs about PCT::HLLCompiler API?
15:27 rotty ah, already found http://search.cpan.org/~bschmal/parrot/run​time/parrot/library/Parrot/HLLCompiler.pir
15:27 shorten rotty's url is at http://xrl.us/bi5w7
15:33 rotty hmm, where exactly should private global variables live? The PAST->PIR transformation emits get_hll_global, but the docs say (and TCL code uses) set_root_global, so the variables can not be found...
15:34 particle you mean, sticking private global vars in ['_tcl']
15:34 rotty particle: yes
15:34 particle i don't think hllcompiler supports that yet
15:34 rotty how would i construct a PAST node that loads from that?
15:34 particle you can embed pir
15:35 particle :inline
15:35 rotty ah, right
15:39 pmichaud we could probably create a 'root' scope for PAST::Var
15:40 particle it may be worth supporting a 'private' scope
15:40 pmichaud ...private?
15:40 particle since the namespace pdd talks about it
15:40 particle each language namespace gets its own private namespace at the root level
15:40 pmichaud isn't that just the (e.g.)  ['_tcl']  scope from the root namespace?
15:40 particle tcl gets _tcl
15:40 particle yes
15:40 pmichaud so, 'root' should cover it
15:41 pmichaud as PAST doesn't really know what hll namespace it's in
15:41 pmichaud we can have the thing that builds the PAST tell it
15:41 pmichaud for example:
15:41 particle hrmm
15:41 particle yeah, you're right
15:41 pmichaud PAST::Var.new(:scope('root'), :namespace('_tcl::Some::NameSpace'), ... )
15:42 particle root++
15:44 rotty is get_root_global fast? I need to store e.g. the boolean 'true' and false value in a global, as they must be 'issame'
15:46 particle it's fast
15:46 particle it's a single parrot op
15:47 rotty yeah, but it must do a hash lookup and more, doesn't it?
15:48 particle vim src/ops/var.ops +273
15:57 Dave joined #parrot
15:58 cotto_work is there a way to pass -D options to gcc through dynpmc.pl?
16:08 particle cotto_work: you can modify the pl file manually, or modify the template file under config/gen/makefiles
16:08 particle it uses the same flags that parrot uses to build
16:08 cotto_work you mean dynpmc.pl
16:08 particle our $CFLAGS = q[@ccflags@ @cc_shared@ @cc_debug@ @ccwarn@ @cc_hasjit@ @cg_flag@ @gc_flag@];
16:09 cotto_work the idea is that I can do "make pmc" for a normal build and "make pmc-debug" for a debugging build
16:09 cotto_work s/can/could/
16:09 Infinoid well, you could hack it to add extra cflags from an environment variable...
16:10 nopaste "Infinoid" at 96.238.213.50 pasted "Something like this." (13 lines) at http://nopaste.snit.ch/12668
16:11 Infinoid then pass the extra environment variable in the pmc-debug rule
16:13 Theory joined #parrot
16:20 cotto_work Infinoid++
16:20 cotto_work that's a good starting point
16:20 Theory joined #parrot
16:26 sjansen joined #parrot
16:31 jonathan pmichaud: not done type-based MMD yet.
16:31 jonathan I want to tidy up my types patches a tad and stuff too.
16:31 particle jonathan: did you see my nopaste?
16:31 jonathan Found something that doesn't work right today. :-(
16:32 jonathan particle: no, only back from Stockholm just a few moments ago...
16:32 particle hurry!!! :P
16:33 jonathan multi sub defeats(Thing    t1, Thing    t2) { false };
16:33 jonathan Erm, don't t1 and t2 need sigils?
16:33 particle ook!
16:33 * particle blames the bad error message
16:33 jonathan I *think* that should parse, but don't expect it to do anything useful.
16:34 jonathan should parse once you put the sigils in, anyway.
16:34 jonathan Only MMD now is arity based.
16:34 particle indeed, it parses now
16:34 jonathan It's a bit of work to get full type-based dispatch to really work right.
16:34 particle jonathan++ particle--
16:34 jonathan What do you actually need?
16:35 particle my $paper = Paper.new;
16:35 particle my $rock  = Rock.new;
16:35 particle say defeats($paper, $rock);
16:35 particle i just want that to work
16:35 jonathan Maybe I can do that.
16:35 particle see chromatic's last post for a great link
16:35 jonathan Post where?
16:35 purl i heard Post was Parrot Opcode Syntax Tree or for changing data and uncacheable
16:36 particle here
16:36 particle http://www.cs.dartmouth.edu/r​eports/abstracts/TR2001-404/
16:36 Ademan_ joined #parrot
16:37 jonathan Oh, the patch to fix my inability to write tests? :-)
16:37 jonathan Ah
16:37 jonathan Looks interesting.
16:37 jonathan Why do I sense that's going to make my head hurt...
16:37 particle yes, i translated the example they use
16:37 jonathan OK, going to grab a shower...sticky from the train, flight, train...
16:38 particle as a first step to getting an implementation
16:38 jonathan I like that approach. Show me code, and hope I make it work. :-)
16:38 jonathan back soon
16:38 rotty seen barney?
16:38 purl barney was last seen on #parrot 17 hours and 58 minutes ago, saying: The PAST generating code can be one big nested constructor
16:47 barney joined #parrot
16:52 dalek r26870 | bernhard++ | trunk:
16:52 dalek : [Eclectus]
16:53 dalek : Add copyright indicators in t/*.pl
16:53 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26870
16:55 dalek r26871 | bernhard++ | trunk:
16:55 dalek : #52616: [PATCH] Eclectus: Small portability fixlets
16:55 dalek : Courtesy of Andreas Rottmann
16:55 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26871
17:00 dalek r26872 | bernhard++ | trunk:
17:00 dalek : #52618: [PATCH] Eclectus: Gauche support
17:00 dalek : Courtesy of Andreas Rottmann
17:00 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26872
17:06 dalek r26873 | bernhard++ | trunk:
17:06 dalek : #52662: [PATCH] Eclectus: Implement BEGIN
17:06 dalek : Courtesy of Andreas Rottmann
17:06 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26873
17:12 dalek r26874 | bernhard++ | trunk:
17:12 dalek : #52664: [PATCH] Eclectus: Implement proper quoting
17:12 dalek : Courtesy of Andreas Rottmann.
17:12 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26874
17:13 dalek r26875 | bernhard++ | trunk:
17:13 dalek : Fix SVN properties for new files.
17:13 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26875
17:13 barney rotty++    for being amazing
17:15 jonathan particle: Need to deal with some email backlog from $DAYJOB, and find some place to give me food.
17:15 jonathan But will try and have a crack at the MMD thing for you later.
17:33 rotty karma rotty
17:33 purl rotty has karma of 2
17:34 rotty barney: did you have a look at the macro system post to the chicken ML yet?
17:35 barney Not yet, what was the URL ?
17:36 rotty http://lists.gnu.org/archive/html/c​hicken-users/2008-04/msg00013.html
17:36 shorten rotty's url is at http://xrl.us/bi556
17:40 rotty AFAICS, the Pair PMC is not yet implemented, right?
17:40 * barney being still confused by Scheme macros
17:41 rotty barney: join #scheme @ irc.freenode.org and ask there ;)
17:41 barney As I read up on Macros some months ago, I remember that there a Scheme-Implementations of Macros,
17:41 pmichaud there is a Pair PMC in Parrot, but I don't know that it does what you want.
17:42 barney with rely on only very specific hooks provided by the systems
17:42 barney rotty: I'm lurking on #scheme
17:43 rotty pmichaud: http://svn.perl.org/viewvc/parrot/trunk/src/​pmc/pair.pmc?revision=26309&view=markup -- note the "#if 0"
17:43 barney pmichaud: EclectusPair is extending Pair
17:43 shorten rotty's url is at http://xrl.us/bi567
17:44 * particle finally ordered his laptop
17:46 rotty barney: if I understand you correctly, riaxpander is such a system
17:47 pmichaud rotty:  yes, it's incomplete.  But there are also tests for the Pair PMC   t/pmc/pair.t
17:47 pmichaud particle:  what did you get?
17:48 particle t61 14.1" widescreen
17:48 particle 2.6GHz core 2 duo, 4GB ram
17:49 particle with shipping and tax, ~$1800
17:59 barney rotty: Is that correct: As 'define-macro' can be implemented in terms of 'lamba', it should be no problem to support 'define-macro' in Eclectus?
18:09 ambs joined #parrot
18:10 rotty barney: I don't think we want to support define-macro (because it has no way to preserve hygiene), but in case we want, it can be implemented easily on top of explicit renaming macros
18:11 rotty (and riaxpander does syntactic closures, ER macros and syntax-rules)
18:11 rotty barney: do you want to target R6RS eventually?
18:12 pmichaud anyone have a comment on http://www.nntp.perl.org/group/perl.p​erl6.internals/2008/03/msg44294.html >
18:12 barney My naive answer is yes
18:12 shorten pmichaud's url is at http://xrl.us/bi6a9
18:12 pmichaud ?
18:12 * rotty would rather target R5RS, as for R6RS we'd have to implement syntax-case (which is quite baroque and can do nothing that can't be done with syntactic closures or ER)
18:14 particle pmichaud: can it be defined in HLLCompiler, rather than in PAST?
18:14 pmichaud not easily.
18:15 pmichaud because at the time we're compiling PAST, we might not be coming from a HLLCompiler
18:15 pmichaud besides, it seems easier to do it within a language's  TOP action method than to do it in HLLCompiler
18:15 particle true. TOP is a fine place
18:16 barney R5RS is fine with me.
18:16 * pmichaud feels like he's in Star Wars with R5RS, R6RS, R2D2, C3P0, etc.
18:16 ruoso joined #parrot
18:17 rotty fine
18:17 barney As somebody outside the Scheme community I find it strange that R6RS was finalised, while there seems much dispute about it
18:17 pmichaud barney:  they're just following the ooxml model :-)
18:18 rotty or rather ooxml follows the R6RS model (while IMO R6RS is much better than ooxml, as a standard)
18:18 pmichaud particle:  think I should use '*' or '' for the "default" values entry?
18:18 pmichaud particle:  or something else?
18:18 purl i think something else is probably wrong
18:19 rotty well, R6RS adds a module system which is quite ok, IMHO, but syntax-case, case-sensitivity, and other stuff was pretty controversial
18:19 particle pmichaud: what about * (no quotes)
18:19 pmichaud particle:  it's a string (key)
18:20 pmichaud i.e., normally one would have a variable name there
18:20 rotty barney: have you formed an opinion on riaxpander yet?
18:20 particle where nqp will interpret the syntax of * (Whatever) to a default var name behind the scenes
18:20 pmichaud this is a pct/past specification
18:20 pmichaud so I need to know what the default value will be in pct/past
18:20 * rotty already has a patch that replaces the PREPROCESS procedure with a call into riaxpander
18:21 pmichaud by way of example...
18:21 pmichaud in rakudo, the TOP action method would contain
18:22 barney What is the licensing of riaxpander ?
18:22 pmichaud my $past := $( $<statement_block> );
18:22 pmichaud $past.symbol('*', :viviself('Undef'), :scope('package'));
18:22 rotty barney: I think one of the next things we have to add to eclectus is closure conversion
18:23 pmichaud or, we could give PAST::Block an extra method (and perhaps attribute) for this
18:23 pmichaud $past.defaults(:viviself('Undef'), :scope('package'));
18:24 particle is this the only thing that will need to be defaulted?
18:24 pmichaud there may be more later, which is why it's a hash
18:24 slightlyoff joined #parrot
18:24 pmichaud oh, you mean we might want to do defaults in something other than PAST::Block?  I doubt it
18:25 diakopter what version of netbsd do folks want smoked?  4.0?
18:25 pmichaud diakopter: any version would be fine with me
18:26 particle pmichaud: then i like PAST::Block.default(s)
18:27 pmichaud yes, I think I'll go with that for now.
18:30 pmichaud particle: tnx
18:30 diakopter pmichaud: NetBSD/amd64 or i386?
18:30 pmichaud i386 to start with, I think
18:30 pmichaud one of each would be nice :-)
18:31 pmichaud the 64-bit builds often find oddities that don't show up on the i386 ones -- at least in the linux world
18:32 diakopter pmichaud: I've set up VMs to smoke on amd64-linux-cc and amd64-linux-gcc nightly
18:32 diakopter since I heard about MSFT setting up windows smokers, I haven't done any of those yet
18:33 Infinoid diakopter: is your cc non-gcc?
18:33 diakopter Infinoid: I don't know :D
18:33 pmichaud afk # walk
18:33 Infinoid on my amd64-linux box, the only cc I've got is gcc.
18:34 diakopter one of them is debian sid; the other is centos5.1
18:35 Infinoid oh, ok
18:35 diakopter seems unlikely it would be anything but gcc... I've just used the default packages
18:35 Infinoid diakopter++
18:36 diakopter the solaris VM is 32-bit and is autosmoking as well.  I have yet to succeed at getting a 64-bit install.
18:37 diakopter I haven't decided if I want to donate the electricity to languages-smoke each nightly
18:38 barney rotty: Wouldn't using riaxpander be a hindrance to bootstrapping ?
18:39 Infinoid I'd like to be able to smoke specific languages
18:39 particle i'd love to see icc smoking on solaris / linux / windows
18:39 rotty well, the code is not very demanding - it uses srfi-1, srfi-9, and srfi-8
18:40 diakopter Infinoid: which languages do you want smoked?  or you do mean you want a make smoke target for each of the languages?  couldn't that be generated?
18:40 rotty and not much more AFAICS
18:40 Infinoid diakopter: the latter, and yes
18:40 rotty barney: perhaps we should switch to #scheme?
18:40 jonathan I'd like to see a cuban cigar smoked.
18:40 Infinoid cigar smoking is a spectator sport.
18:40 diakopter yeah, I'd gladly set them all to smoke rakudo nightly
18:41 diakopter or even bihourly or something
18:41 barney cd languages ; perl t/harness --html --languages='pheme,perl6'
18:41 Infinoid make rakudo-smoke lolcode-smoke cigar-smoke
18:41 barney should work now
18:41 rotty but we should drop petite support, it doesn't support any SRFIs AFAICS, and its command-line interface is too limited to load code before execting a script
18:42 particle http://www.savethebottles.co​m/labels/smoking_parrot.jpg
18:55 pmichaud particle:  I think I've decided to just go with '' for now
18:55 pmichaud i.e.,   PAST::Block.symbol('', :viviself('Undef'), :scope('package'))
18:55 pmichaud for one, it means that if we ever do implement something like * in nqp, it would just work
18:55 pmichaud PAST::Block.symbol(*, :viviself('Undef'), :scope('package'))
18:56 particle ok. you're in charge, and '' works.
18:56 pmichaud I may regret it, but we can change it later.  I'm thinking that explaining '' as a special symbol will be easier than explaining a separate 'defaults' method
18:56 pmichaud although 'symbol_defaults' might be easier
18:56 pmichaud hmmm
18:56 pmichaud okay, I've switched my mind again.  'symbol_defaults' is easier.  :-P
18:57 particle symbol( :defaults, :viviself...
18:57 pmichaud nah, I don't like :defaults as a flag, because then a symbol can't have a 'defaults' attribute.
18:57 particle ok
18:57 particle symbol_defaults is good
18:57 pmichaud PAST::Block.symbol_defaults(...)  seems to read well
18:58 pmichaud and makes clear the relation with .symbol(...)
19:01 particle lunch &
19:05 dalek r26876 | pmichaud++ | trunk:
19:05 dalek : [pct]:
19:05 dalek : * If we're just looking up a nonexistent symbol for PAST::Block,
19:05 dalek :   don't use it as an opportunity to store a new empty hash.
19:05 dalek :   Only store the attribute hash when an attribute is explicitly
19:05 dalek :   given.
19:05 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26876
19:27 dalek r26877 | bernhard++ | trunk:
19:27 dalek : [Eclectus]
19:27 dalek : Use the deprecated syntax for "run-command",
19:27 dalek : keep Gauch 0.8.8 happy.
19:27 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26877
19:30 Coke would anyone here be interested in an ack macport?
19:42 ambs joined #parrot
19:44 muixirt joined #parrot
19:50 ambs_ joined #parrot
19:59 TimToady phone
20:02 dalek r26878 | bernhard++ | trunk:
20:02 dalek : [Eclectus]
20:02 dalek : Use Gauche as the bootstrapping scheme.
20:02 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26878
20:31 Coke particle: cc me on that one. =-)
20:32 particle wilco
20:39 dalek r26879 | pmichaud++ | trunk:
20:39 dalek : [pct]:
20:39 dalek : * Fix problem where PAST compilation aborts and leaves $?SUB unclean.
20:39 dalek : * Resolves RT#49758, patch courtesy Stephane Payrard.
20:39 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26879
20:58 jan joined #parrot
21:04 ambs joined #parrot
22:00 rdice joined #parrot
22:09 mj41 good night ... http://test.perl6.cz/ttserv/show.pl
22:15 mire joined #parrot
22:18 dalek r26880 | jonathan++ | trunk:
22:18 dalek : [rakudo] Some more cases for handles trait verb, including pairs.
22:18 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26880
22:25 jonathan (plane hacking)++
22:25 Limbic_Region joined #parrot
22:25 Infinoid I have a machine running linux/x86 which can't link miniparrot.  says it can't find symbols like Parrot_new
22:26 Infinoid if I run gcc by hand, I can get it to link with libparrot.a, but not with libparrot.so
22:43 cognominal joined #parrot
22:47 jonathan cognominal: Any free slots at the FPW? It's just possible I will be free on those dates...a bit 50/50 at the moment.
22:50 kid51 joined #parrot
23:08 * diakopter fights with netbsd
23:10 dalek r26881 | jonathan++ | trunk:
23:10 dalek : [rakudo] Slight refactoring of type parameters so I can have a crack at a basic implementation of multi.
23:10 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26881
23:10 * kid51 remembers hearing/reading somewhere that the test of whether you can be a NetBSD hacker is whether you can install it or not.
23:11 jonathan Heh. Objective C's interface header syntax looks almost like an approximation to ASCII-art UML.
23:15 Limbic_Region joined #parrot
23:40 diakopter kid51: well, it's installed and I've installed some packages.  looking for subversion...
23:43 tetragon joined #parrot
23:44 dalek r26882 | jkeenan++ | trunk:
23:44 dalek : Refactor declaration of, and assignments to, @srcopsfiles, $num and $skip out
23:44 dalek : of individual test files and into testlib/GenerateCore.pm, from which they are
23:44 dalek : exported on demand.
23:44 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26882
23:45 kid51 Did you have utter incantations to Theo?
23:45 kid51 Or, perhaps, imprecations?
23:49 particle joined #parrot
23:49 dalek r26883 | jkeenan++ | trunk:
23:49 dalek : Renumber one test file so that each test gets a unique two-digit prefix.
23:49 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26883
23:52 dalek r26884 | jkeenan++ | trunk:
23:52 dalek : Update references to name of file within file.  Small POD corrections.
23:52 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=26884
23:52 Eevee wow, busy today
23:57 mire_ joined #parrot

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

Parrot | source cross referenced