Camelia, the Perl 6 bug

IRC log for #parrot, 2008-06-03

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:02 ank joined #parrot
00:03 ank joined #parrot
00:10 AndyA joined #parrot
00:12 bacek_ joined #parrot
00:26 workbench joined #parrot
00:40 bacek_ joined #parrot
00:41 bacek_ morning everyone
01:09 japhb joined #parrot
01:27 cotto_work pmichaud++ for the P6object post
01:40 scrottie pmichaud-- # for being an idle-butt
02:36 TiMBuS joined #parrot
02:45 Zaba_ joined #parrot
03:24 dalek r28041 | coke++ | trunk:
03:24 dalek : [codingstd]
03:24 dalek : make t\codingstd\trailing_space pass.
03:24 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28041
03:29 cotto_home karma pmichaud
03:29 purl pmichaud has karma of 1324
03:29 cotto_home seen scrottie
03:29 purl scrottie was last seen on #parrot 1 hour and 50 minutes ago, saying: pmichaud-- # for being an idle-butt
03:29 cotto_home karma pmichaud
03:29 purl pmichaud has karma of 1324
03:46 ank karma police
03:46 purl Arrest this man!
03:48 cotto_home (karma police)++
03:48 cjfields joined #parrot
03:49 cjfields left #parrot
03:53 niteice joined #parrot
04:06 tetragon joined #parrot
04:08 Infinoid karma chameleon
04:08 purl It comes and goes.
04:09 Auzon instant karma?
04:09 purl Auzon++
04:09 Auzon woah
04:11 Infinoid JIT karma
04:17 cotto_home test #9 in t/examples/tutorial.t is failing
04:27 dalek r28042 | coke++ | trunk:
04:27 dalek : [docs] point more accurately at LWN
04:27 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28042
05:14 TonyC joined #parrot
05:16 nopaste joined #parrot
05:31 nopaste "bacek" at 211.29.157.151 pasted "Fix of List.reduce for pmichaud to commit" (35 lines) at http://nopaste.snit.ch/13148
06:16 uniejo joined #parrot
06:28 nopaste "bacek" at 211.29.157.151 pasted "Refactored version List.reduce for pmichaud to review." (103 lines) at http://nopaste.snit.ch/13149
06:29 scrottie left #parrot
06:36 UltraDM joined #parrot
06:45 Zaba joined #parrot
07:01 Zaba_ joined #parrot
07:08 masak joined #parrot
07:20 mj41 joined #parrot
07:30 dalek r28043 | chromatic++ | pdd25cx:
07:30 dalek : [IMCC] Made IMCC more rentrant.  When compiling code with IMCC from within a
07:30 dalek : PIR program, it was possible to switch the constant table for the currently
07:30 dalek : executing parent context to an empty constant table created for the context of
07:30 dalek : the code being compiled.  This does bad things from within continuations.
07:30 dalek : The fix is to save the parent context *before* potentially overwriting its data.
07:30 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28043
07:30 dalek r28044 | chromatic++ | pdd25cx:
07:30 dalek : [src] Tided some code, while I was debugging them.  No behavioral changes.
07:30 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28044
07:36 mj41 joined #parrot
07:42 mj41_ joined #parrot
07:45 mj41 joined #parrot
09:45 dalek r28045 | kjs++ | trunk:
09:45 dalek : [tutorial] add episodes 5 to 9 to languages/squaak/doc directory
09:45 dalek : Manifest is updated.
09:45 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28045
09:48 ruoso joined #parrot
09:51 dalek r28046 | kjs++ | trunk:
09:51 dalek : [tutorial] fix a missing =head2.
09:51 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28046
09:57 bacek joined #parrot
10:44 bacek <bacek> rakudo: my $a = 1 + undef; say $a; say $a.WHAT;
10:44 bacek <exp_evalbot> rakudo r28046 OUTPUT[1?Failure?]
10:44 bacek pmichaud, I found why List.map failing with Failure.
11:02 clunker3 joined #parrot
11:12 rdice joined #parrot
11:14 davidfetter joined #parrot
11:48 nopaste "bacek" at 202.7.166.169 pasted "Fix src/pmc/integer.pmc +/- ops with undef (pmichaud/particle/chromatic to review)." (20 lines) at http://nopaste.snit.ch/13150
11:49 bacek anyone alive here?
11:49 * moritz drops dead
11:50 * bacek empty AK-47 to moritz. Just to be sure...
11:51 moritz ouch, that hurt! I can't even begin to imagine how much a full AK-47 would hurt ;)
11:53 bacek moritz, empty is about few kilogramms. Full is little bit more :)
11:54 moritz in Soviet Russia Kalachnikov empties *you*
11:56 bacek moritz, no. Its usually fills you with some small metal things :)
11:57 moritz bacek: and empties your guts in the progress
12:03 bacek moritz, probably :)
12:04 jonathan Seems I picked an interesting time, to appear on #parrot...
12:07 moritz ;-)
12:07 bacek hi jonathan :)
12:08 cognominal make: *** No rule to make target `src/glut_callbacks.o', needed by `runtime/parrot/dynext/libglutcb.bundle'.  Stop
12:08 cognominal do I need a new version of the glut library?
12:09 bacek cognominal, make realclean; Configure.pl; make
12:09 cognominal I ded
12:09 cognominal I did!
12:09 bacek can anyone test S29-list/map.t via 'make localtest'?
12:10 bacek cognominal, works for me...
12:10 davidfetter anybody have an example of using parrot to replace lexx/yacc ?
12:11 moritz bacek: All tests successful, 27 subtests skipped.
12:11 bacek moritz, hmm...
12:11 davidfetter well, the relevant parts of parrot, anyhow
12:11 moritz bacek: fudged, of course
12:11 bacek davidfetter, languages/squaak/src/parser/
12:12 * davidfetter looks at this
12:12 bacek moritz, yeek... I didn't expect it...
12:12 bacek moritz, is it clean checkout of parrot?
12:13 moritz bacek: actually map.t is not in spectest_regression.data because it had unclear tests (that I cleaned in the mean time)
12:13 moritz bacek: no, slight modifications
12:14 Whiteknight joined #parrot
12:14 moritz the runcore=gcdebug path, mostly
12:16 bacek moritz, can you 'say (1+undef)+undef' result?
12:16 cognominal I suspect this patch :  http://rt.perl.org/rt3//Publ​ic/Bug/Display.html?id=55074
12:16 moritz bacek: same as evalbot
12:17 davidfetter bacek, um, that's not quite what i had in mind. i'm thinking of a walk-through where somebody replaces lexx/yacc with parrot in an existing code base
12:17 dalek r28047 | Whiteknight++ | gsoc_pdd09:
12:17 dalek : [gsoc_pdd09] updating to trunk r28046
12:17 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28047
12:17 cognominal does not seem so dangerous a patch :(
12:18 bacek bacek@icebolt:~/src/parrot/languages/perl6$ ../../parrot perl6.pbc
12:18 bacek > say (1+undef)+undef
12:18 bacek 1
12:18 bacek I don't understand why S29-list/map.t passes without my patch...
12:18 moritz bacek: which specific test are you worrying about?
12:18 bacek davidfetter, ouch... I don't think that someone did it
12:19 bacek moritz, n-ary functions
12:20 moritz bacek: they are fudged
12:20 tetragon joined #parrot
12:21 bacek moritz, i'mm stupid... I forgot to commit unfuged version...
12:23 bacek >  say  ~(1,2).map({ $^a+$^b+$^c+$^d+$^e })
12:23 bacek 3
12:23 bacek ok. Look's like this is right patch for integer.pmc
12:23 bacek summon jonathan
12:26 * jonathan appears
12:27 bacek http://nopaste.snit.ch/13150 this is another small patch for src/pmc/integer.pmc
12:27 bacek It fix (1+undef)+undef problem in parrot
12:28 jonathan Hmm
12:28 jonathan Basically it just makes the return type that of the LHS of the operator rather than the RHS.
12:29 bacek LHS is always Int.
12:29 bacek And it consistent with all other places in this file :)
12:30 cjfields joined #parrot
12:30 bacek ouch...
12:30 bacek mmd--
12:30 bacek I hate it...
12:31 cjfields left #parrot
12:33 bacek jonathan, ok... mmd is weird
12:34 bacek this patch will fix part of the problem anyway.
12:35 bacek another part is properly implement 'undef.pmc' with redirecting 'undef of PMC' to 'PMC op undef'
12:48 * bacek cast resurrect on jonathan
12:55 Whiteknight watch out! zombie jonathan will eat your brains!
12:56 pmichaud pong
12:57 bacek pmichaud, I found why List.map doesn't work with Failure.
12:57 bacek http://nopaste.snit.ch/13150
12:58 bacek Actually other VTABLE_methods use patch's approach
12:59 pmichaud interesting.  looking.
12:59 bacek and question is: is it good idea to implement undef + 1 in src/pmc/undef as redirecting it to 1 + undef
12:59 pmichaud bacek:  can't do that for subtraction, though.
12:59 pmichaud (so, no)
13:00 bacek pmichaud, yes of  cause. But for commutative methods it will work :)
13:00 pmichaud well, we need to find a fix for all of them :-)
13:00 pmichaud and I'd rather not have some do it one way and others do it another
13:00 pmichaud I think I'm going to fix it all using a higher mmd anyway
13:01 bacek pmichaud, pugs says that 'undef - 1' is -1.
13:01 moritz that's right, plus ideally a warning
13:01 bacek Looks like pugs morph undef to RHS type
13:01 bacek may be this is 'way to go'
13:02 * bacek still trying to build bugs...
13:02 pmichaud no, because   undef - "123"   shouldn't morph to a string.
13:02 pmichaud I have the fix in mind -- just haven't had a chance to write it.
13:02 pmichaud just a sec, I'll fix it now.
13:02 bacek pmichaud, ok
13:03 pmichaud (need to svn up, rebuild, etc first)
13:04 pmichaud but I think the real solution is to have infix:+ (_,_)  cast its values into num registers, then do the addition.
13:04 pmichaud and for an optimization we can have   infix:+ (Int,Int)   do an integer addition
13:04 pmichaud and then we don't have to deal with Parrot's MMD at all.
13:05 bacek pmichaud, is it good???
13:05 pmichaud (er, Parrot's opcode MMD)
13:06 bacek pmichaud, anyway, it's next step. My patch just fix some inconsistency.
13:06 moritz bacek: it's the only generic solution
13:06 moritz bacek: because infix:+ should work for all things that can be used in numeric context
13:06 moritz bacek: even user defined classes
13:06 bacek moritz, why it should be on 'language level' not 'VM level'?
13:07 pmichaud one thing I haven't figured out in Parrot's opcode MMD is how one adds new MMD combinations to an opcode
13:07 pmichaud for example, if I want to have a special version of 'add'  for   types Foo and Bar, how do I do that?
13:07 moritz bacek: because if you have N types that can be used in numeric context, you'll have to add O(N²) inifix:+ multis
13:08 pmichaud although I guess it's done using a vtable override of the 'add' opcode
13:08 pmichaud sorry, vtable override of the add vtable method
13:09 pmichaud anyway
13:11 bacek pmichaud, infix:+(Foo,Bar) :)
13:11 bacek I can't find commutative handling in MMD too...
13:12 gryphon joined #parrot
13:12 pmichaud bacek:   infix+(Foo,Bar)  would be a multi sub, not another MMD entry in Parrot's opcode table
13:13 pmichaud I mean the case where I have
13:13 pmichaud $P0 = new 'Foo'
13:13 pmichaud $P1 = new 'Bar'
13:13 pmichaud n_add $P2, $P0, $P1
13:14 bacek pmichaud, hmm...
13:14 pmichaud as I said, I think it would need to be a vtable method on Foo
13:15 pmichaud anyway, I'll go ahead and fix it for rakudo
13:24 pmichaud making spectest_regression
13:27 pmichaud Hash needs get_number, adding
13:33 Whiteknight true or false: "Parrot has six basic stacks, each for a different purpose: The control stack, the pad stack, and the four register backing stacks"?
13:33 pmichaud false
13:33 pmichaud we no longer have register backing stacks, I don't think
13:33 Whiteknight I already removed "user stack" from the list
13:33 Whiteknight okay
13:33 pmichaud and i'm not sure we have a pad stack, either.
13:34 Whiteknight I'm working on /docs/book, and a lot of stuff in here is about the stacks and needs to get deleted.
13:35 Whiteknight And I know the number of stacks has been reduced, but I dont know the full extent of the reduction
13:37 kj joined #parrot
13:38 Whiteknight pmichaud, would you know who I could ask about the existence of the pad stack?
13:39 Whiteknight or, where I could find the information myself?
13:39 pmichaud chromatic or allison are the best bets
13:39 pmichaud you might also try the mailing list, or perhaps at #parrotsketch later today
13:39 Whiteknight okay, thanks!
13:39 paco joined #parrot
13:41 dalek r28048 | pmichaud++ | trunk:
13:41 dalek : [core]:
13:41 dalek : * Add get_number vtable method to Hash PMC.
13:41 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28048
13:42 rdice joined #parrot
13:44 Whiteknight well, the opcodes for new_pad, push_pad and pop_pad are all missing from ops.num. That's not a good sign in favor of existence
13:44 gmansi joined #parrot
13:44 Whiteknight even if it does exist, if you can't manipulate it directly, there's no sense talking about it in the book
13:45 dalek r28049 | pmichaud++ | trunk:
13:45 dalek : [perl6]:
13:45 dalek : * Fix a number of operators to do proper casting.
13:45 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28049
13:45 pmichaud bacek:  okay r28049 should do what you want.
13:45 bacek pmichaud, rebuilding...
13:47 bacek pmichaud, (while building parrot) http://nopaste.snit.ch/13149
13:48 pmichaud the multis don't look right
13:48 bacek yes.
13:49 bacek I think we should drop all :multi(_,List)
13:49 bacek because in sub(foo, 1, (1..5)) List is not generated anymore AFAIK
13:51 pmichaud shouldn't it be  :multi(Sub)   ?
13:51 bacek pmichaud, probably.
13:52 bacek no. definitely :)
13:52 bacek pmichaud, (last patch) it woks!
13:52 bacek pmichaud++
13:53 pmichaud how does pugs report an error in reduce if given a sub with arity < 2 ?
13:54 moritz it calls Code.arity?
13:56 bacek pmichaud, yes.
13:56 bacek sting is copy-pasted from evalbot
14:00 nopaste "bacek" at 202.7.166.166 pasted "replacing :multi subs with :multi(Sub)" (49 lines) at http://nopaste.snit.ch/13151
14:00 * bacek running spectest_regression for  http://nopaste.snit.ch/13151
14:04 Zaba_ did perl5's $0 go in perl6? what is in its place?
14:05 pmichaud (reduce)  what if the argument is a multisub?
14:05 moritz Zaba_: probalby $*PROGRAM_NAME or some such
14:05 dalek r28050 | Whiteknight++ | trunk:
14:05 dalek : [book] Updated chapter 5 by removing references to "user stack" and "pad stack", and other updates/additions.
14:05 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28050
14:06 bacek pmichaud, Result: PASS with http://nopaste.snit.ch/13151. You are totally right
14:06 Zaba_ aha
14:07 iblechbot joined #parrot
14:07 pmichaud also, fwiw, I'd like to have parameter names in the implementation match the parameter names in the spec.  :-)
14:07 bacek pmichaud, ok :)
14:08 bacek btw, may be we should made 2 subs: :multi(Sub) and :multi(Sub, List) for 'premature optimisation'
14:09 pmichaud nah, don't do the premature optimization
14:09 pmichaud there's little point.
14:09 pmichaud :multi(Sub) should be sufficient.
14:10 bacek ok :)
14:10 bacek S29?
14:10 purl S29 is http://svn.pugscode.org/pugs/​docs/Perl6/Spec/Functions.pod
14:12 jonathan OK, I'm back
14:12 jonathan Had Slovak lesson.
14:12 pmichaud wb, jonathan
14:15 nopaste "bacek" at 202.7.166.166 pasted "replacing :multi subs with :multi(Sub) once again (for pmichaud)" (94 lines) at http://nopaste.snit.ch/13152
14:17 pmichaud it's not necessary to create a separate list
14:17 pmichaud simply use   values.'map'(expression)
14:17 jonathan pmichaud: Was it my parameter handling patch that caused issues?
14:17 pmichaud jonathan: both patches caused issues
14:17 pmichaud I was going to fix them yesterday but got sidetracked onto other items
14:18 pmichaud well, I should rephrase that
14:18 pmichaud getting Perl6Scalar to accept  infix:= is going to be tricky, because find_method on Perl6Scalar is always using find_method of the value
14:18 jonathan I know the gc one caused issues...
14:18 pmichaud and yes, the gc one caused issues too :-)
14:19 jonathan Ah, yes. The MMD stuff gets...nasty...there. :-(
14:19 jonathan Oh, wait, it's not about MMD
14:19 jonathan Hmm, but yes, you're right.
14:21 pmichaud the params patch hard-coded Perl6Scalar instead of using $?PERL6SCALAR
14:22 pmichaud so switching $?PERL6SCALAR at the top didn't quite work
14:22 pmichaud but I really want to refactor parameter handling anyway
14:22 jonathan ok
14:22 pmichaud you were just a day or so ahead of me
14:22 jonathan ok
14:23 pmichaud I'm also planning to get rid of a lot of the inline PIR surrounding placeholder params, since that's not needed any more
14:23 pmichaud and try to remove some of our flag variables.
14:31 bacek sleep time
14:32 bacek g'night everyone
14:51 moritz maybe we should introduce a "committest" that runs rakudo's test, spectest_regression and parrot's codetest
14:53 particle i think that should be called 'test' so it's the default
14:53 particle -Odevelopment
14:53 * jonathan wouldn't dislike what particle suggests.
14:53 particle then we have targets like spectest, fulltest, etc
14:54 moritz I think it's useless to run the code tests when you do a 'make languages_smoke'
14:54 particle languages_smoke should run 'smoketest' then
14:55 particle every language should define smoketest, and most will point to test
14:55 particle smoketest : test
14:55 moritz good point
14:55 purl Thanks, moritz, I'm glad you agree with me.
14:55 moritz we just need to make sure that we don't fail when sombody doesn't have svn installed
14:55 particle or that we fail gracefully
14:56 NotFound You can add make installsvn X-)
14:56 moritz NotFound: naughty boy
14:57 * moritz is surprised that purl doesn't have a clever comment on that one ;)
14:59 davidfetter open the pod bay doors, purl
14:59 purl I'm sorry, davidfetter, I'm afraid I can't do that.  I still have the utmost confidence in the mission.
15:02 * NotFound is considering the possibility of creating a text adventure game by feeding purl with selected texts.
15:03 moritz NotFound: purl *is* the text adventure
15:04 NotFound I loose.
15:04 davidfetter itym lose
15:05 pmichaud purl, the mission?
15:05 purl well, the mission is probably pretty inexpensive.
15:05 pmichaud no, purl, the mission is to complete Perl 6.  :-)
15:05 purl okay, pmichaud.
15:09 Whiteknight the mission?
15:09 purl the mission is probably to complete Perl 6.  :-)
15:09 Whiteknight excellent use of the qualifier "probably"
15:09 pmichaud probably, purl?  Are you sure you're unsure?  Are you certain it's not certain?
15:10 * pmichaud picks on the bot.
15:13 NotFound joined #parrot
15:22 japhb cognominal: I have not had time to read the backlog (and I'm about to step out), but when I get back I can take a look at your OpenGL build problems.  In the mean time, can you email me a tarball of your OpenGL/GLUT header files?  (I'm surprised you're having problems on Mac OS, I thought I nailed these to the ground.)
15:24 pmichaud moritz: is #55164 still an issue for you-- i.e., does t/spec/S29-array/push.rakudo still segfault?
15:25 moritz pmichaud: checking...
15:25 purl hmmm... checking is not the hard part here. ;-)
15:26 cognominal japhb : your email address?
15:27 moritz pmichaud: it doesn't segfault, I'll now check the reduced example with --runcore=gcdebug
15:29 cotto_home japhb?
15:29 purl somebody said japhb was Geoffrey Broadwell, mailto:geoff@broadwell.org
15:31 moritz pmichaud: no segfault, but still errors with --runcore=gcdebug
15:31 pmichaud interesting.
15:31 pmichaud okay, thanks, that's what I needed to know.  I'll try it on my box.
15:32 NotFound #55710 has been solved?
15:32 NotFound I mean, #55170
15:33 pmichaud I don't know if it's been solved or not
15:33 NotFound I'v tested now with a c++ build, and it works.
15:34 NotFound Rebuliding with c.
15:42 NotFound Interesting, it segfaults in the c build but works in the c++ build.
15:43 NotFound Both builds after a realclean, in r28050
15:45 jhorwitz joined #parrot
15:47 IllvilJa joined #parrot
15:52 cotto_home NotFound, where does the c build for 28050 explode for you?
15:53 Theory joined #parrot
15:54 cotto_home nm
15:58 NotFound cotto_home: make perl6, when executing the hello.
16:00 NotFound Buliding with --optimize, as #55170 explains.
17:16 davidfetter joined #parrot
17:24 Theory joined #parrot
17:31 avar 1/w 22
17:33 grim_fandango joined #parrot
17:33 moritz purl: ps?
17:33 purl ps is postscript or process status
17:33 moritz purl: #ps?
17:33 purl somebody said #ps was where I can say "okay, these issues are really troubling/annoying/blocking" or #parrotsketch, where committers can talk and bystanders can listen
17:35 Tene Oh, #ps time and coke isn't around.
17:37 gmansi joined #parrot
17:38 particle isn't parrotsketch in an hour?
17:38 particle 1830utc
17:38 particle currently 1738
17:39 moritz aye
17:39 Tene Gah, timezones.
17:39 moritz I hoped that purl would tell me the time
17:39 moritz purl, #ps is also at 1830 UTC
17:39 purl okay, moritz.
17:39 chromatic joined #parrot
17:40 chromatic Is there a SoC meeting?
17:40 particle no, parrot soc students report during parrotsketch
17:40 Tene I thought I was on central time again.
17:40 chromatic So there's no special SoC meeting?
17:41 moritz chromatic: and Auzon reports on #perl6-soc in irc.freenode.net
17:41 particle there's a meeting for the perl 6 soc project, tomorrow at 1930utc
17:41 Auzon but that's on Wednesday.
17:41 Tene " Wednesday 19:30 UTC weekly meetings" -- from #perl6-soc on freenode
17:41 particle auzon misreported in his blog a week or so aga
17:41 particle *ago
17:41 Auzon I did?
17:41 particle you misreported something...
17:41 * moritz didn't notice
17:42 particle perhaps it's that you said it's before the weekly parrot meeting
17:42 moritz let's throw a few stones, just to be sure ;)
17:42 particle that's incorrect, it's 30m before the weekly perl 6 design meeting conference call
17:42 Auzon OK, I'll change that.
17:42 chromatic Okay, that sounds familiar.
17:42 particle so i'd like to keep that meeting to 30m ;)
17:42 moritz particle: I think it will usually be shorter, it was just the first time that we had so much to say
17:43 particle moritz: agreed
17:43 particle moritz: do you log #parrotsketch?
17:43 moritz particle: no. Should I?
17:43 particle would you mind? it'd be nice in case tweety goes down
17:44 particle and your log interface is much nicer
17:44 ilbot2 joined #parrot
17:44 Topic for #parrotis now Parrot 0.6.2 "Reverse Sublimation" Released | http://parrotcode.org/ | 18/672 new/open tix | logged in http://irclog.perlgeek.de/parrot/today
17:44 moritz particle: thanks ;-) it logs now
17:45 particle moritz++
17:45 Auzon OK, I fixed my blag. Sorry about the confusion.
17:51 pmichaud purl: ps?
17:51 purl it has been said that ps is postscript or process status
17:51 pmichaud purl: parrotsketch?
17:51 purl rumour has it parrotsketch is a status meeting for parrot core committers held every Tuesday at 18:30 UTC in #parrotsketch
17:51 pmichaud ps is also see "parrotsketch"
17:51 purl okay, pmichaud.
17:52 pmichaud no, #ps is a status meeting for parrot core committers held every Tuesday at 18:30 UTC in #parrotsketch
17:52 purl okay, pmichaud.
17:54 * pmichaud begins preparing his #ps report
17:54 Theory joined #parrot
17:56 NotFound #55170 is the same problem as #55000, at least in my system.
17:57 chromatic Does the patch for #55000 fix it for you?
17:58 NotFound chromatic: I tested a variant of that patch, that checks both calls to really_destroy, and fix both tickets.
17:58 chromatic Sounds good to me.
17:58 NotFound I've posted the patch in #55170.
17:58 chromatic Let's see how dirty my tree is.
18:10 chromatic Yep, it's a fix.
18:11 chromatic Wow, with an optimized build, that Rakudo benchmark builds four times faster than it did when I started optimizing it.
18:11 scrottie joined #parrot
18:11 chromatic And the file is three times larger.
18:11 NotFound A lot of inlining, maybe.
18:12 chromatic Not too much... dropping PARROT_ASSERT helps a little and optimizing away useless NULL checks.
18:12 scrottie pmichaud, wake up.
18:12 chromatic As I understand it, GCC is terrible about cross .o inlining and optimization.
18:12 chromatic I'd like to see how clang handles this at some point.
18:12 Whiteknight chromatic, do we still have a pad stack?
18:12 NotFound Ah, sorry, I think you talked about the library file size.
18:13 Whiteknight i was under the impression that the pad stack is caput
18:15 chromatic I meant the size of the input file, not the .o file... so the build process takes three times as much work to process the file.
18:15 dalek r28051 | chromatic++ | trunk:
18:15 dalek : [src] When destroying an interpreter, only destroy its exceptions if there are
18:15 dalek : any present.  Fixes RT #55000 and RT #55170 (NotFound, Vasily Chekalkin, and
18:15 dalek : Andrew Whitworth).
18:15 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28051
18:15 pmichaud scrottie: pong
18:15 chromatic Whiteknight, if you mean the Stack_Chunk_t thing that pmichaud and I talked about a few times, it's gone in the pdd25cx branch now.
18:16 pmichaud whiteknight means whatever the book mean by "pad stack"
18:16 pmichaud I assumed the pad stack was the lexical pad stack
18:16 Whiteknight yeah, lexical pad stack. The "push_pad" and "pop_pad
18:16 Whiteknight " instructions are all gone
18:16 scrottie pmichaud:  yo.  not sure if you replied to my email or if you got it.  a lot of people block me for lack of forward and reverse DNS matching.
18:17 jhorwitz joined #parrot
18:17 pmichaud either pobox or gmail may have blocked it.
18:17 scrottie pmichaud:  you doing YAPC this year?
18:17 pmichaud yes.
18:17 scrottie ahh, excellent.  arriving the night before?
18:17 pmichaud arriving saturday before
18:17 allison joined #parrot
18:17 pmichaud (parrot hackathon)
18:19 scrottie i get in sunday eve.  would you mind just throwing Larry in your suitcase and then I'll try to get in touch?  what are your Sunday eve plans?
18:19 pmichaud I'm thinking of attending the pre-conference dinner
18:19 pmichaud other than that, I have no plans at present.
18:19 grim_fandango joined #parrot
18:19 * scrottie checks his flight info again
18:20 pmichaud I was a little worried about transporting Larry because I'm only doing carryon (no real "suitcase").  But I think I've found a spot for it.
18:20 pmichaud and I have it pretty well protected at the moment :-)
18:20 scrottie oh.  right.  the fascists decided to come up with more clever ways to make security screening take forever so that people miss their flights so they can overbook een more
18:21 scrottie haven't had my coffee yet... just saw you were only 15 min idle and tried to sieze the chance... so I'm a bit dopey...
18:21 pmichaud no prob :-)
18:21 pmichaud can I get my karma back?  ;-)
18:21 chromatic That would assume competence on the part of both parties in the conspiracy, which when applied to both airlines and the TSA stretches credulity.
18:22 pmichaud yes, I'm not sure what American Airlines was thinking when they started their "pay to check any bags" bit.  Fortunately I'm flying Southwest :-)
18:27 NotFound About library size: 16M in normal build, 4.3M with --optimize
18:27 jonathan Pay to check any bags is done by various European budget airlines, but I'd never really have considered AA a budget airline...
18:28 pmichaud AA just announced the new policy a week or so ago
18:28 pmichaud $15 for first bag, $25 for additional bag, each way
18:29 chromatic Wow.  That's a lot of debugging symbols.
18:29 moritz airlines mostly suck
18:29 pmichaud (for tickets booked after June 15)
18:29 scrottie chromatic:  I'm assuming fascism and working backwards from that assumption.
18:29 pmichaud scrottie: anyway, I'll have Larry with me at the conference.
18:30 chromatic That's fine.  Just remember competent fascism (such as Mussolini or Franco) implies timely public transportation systems.
18:30 particle oh, crap, gotta type up my report
18:30 chromatic #ps time
18:30 scrottie pmichaud:  alrighty, I get in 2:23.  sorry to be itchy about getting him back... i have plans for him that involve having time before the con.  but I did intend for him to travel and I'm glad you had him.
18:31 pmichaud scrottie: no problem -- sorry we didn't get together before now to take care of it.  had been meaning to do so but it kept slipping down my list of 'to-do'
18:31 scrottie no, this is awesome.  I'd have coordinated earlier but I didn't have this particular plan in effect before ;)
18:31 NotFound chromatic: looks like you never travelled in the Spain of Franco.
18:31 wknight8111 joined #parrot
18:32 chromatic He predated me by a bit.
18:32 scrottie pmichaud++ # not being an ide-butt
18:32 scrottie idle
18:35 particle scrottie-- # misplleling karma comments
18:36 Zaba joined #parrot
18:36 particle scrottie++ # keeping karma in balance
18:36 scrottie =P
18:38 scrottie pmichaud:  by the way, Richard Dice had Larry the previous year, and when he (Larry) came back, he was all grubby.  Apparently people like to touch Larry.
18:40 pmichaud hasn't been touched a lot here.  Mostly been a decoration on my desk :-)
18:41 TimToady .oO(whew!)
18:42 Theory joined #parrot
18:51 chromatic Hmm, looks like get_free_buffer is a candidate for GC-related optimization.
18:52 Ivatar joined #parrot
18:53 pmichaud maybe we should put a parrot hackathon announcement on parrotblog ?
18:53 particle good idea!
18:53 purl particle: Good Idea: Playing cops'n'robbers in the park. Bad Idea: Playing cops'n'robbers in the bank.
18:53 particle wish i remembered my rakudo.org password
18:53 pmichaud there's not a way to recover passwords?
18:53 chromatic It's just a Blogger password, right?
18:53 particle not obvious, since i didn't set up the account, i don't have the magic phrase
18:54 particle gotta poke andy
18:54 chromatic Oh no, I'm thinking of parrotblog.org
18:55 pmichaud rakudo.org is movable type
18:55 particle too many domains
18:55 particle i'm confuddled and befused
18:55 particle ...and i work here!
18:55 pmichaud it's interesting that on this project we get to use just about every blogging application there is.
18:56 pmichaud (at least the major ones.)
18:56 Whiteknight chromatic, get_free_buffer seems pretty bare-bones already. how could it be optimized?
18:56 pmichaud having so many different blogging systems seems.... suboptimal.
18:56 moritz pmichaud: just write THE ONE blogging system ;)
18:57 pmichaud moritz: I did -- in fact, I'm now using it to post to the others.  :-P
18:57 pmichaud (I write my articles in my wiki, then I have a filter that translates them into suitable formats for each blog application.)
18:58 moritz pmichaud: can it stand on its own? (I mean as "real" blog)
18:58 pmichaud well, it needs a few more features to be a "real" blog, although lots of people use it for that already
18:58 pmichaud but yes, my wiki app definitely stands on its own.
18:58 japhb msg DietCoke I responded to your bug RT #55238.  It's one of those "should not be possible" things.
18:58 purl Message for dietcoke stored.
18:59 japhb cognominal: I have not received email from you yet, nor have I had a chance to read backlog (damn $real_life).  I will check back after lunch and try to help you then.
19:04 * scrottie enjoyed his visit to #parrot very much (thank you!) but he's going to slink off back to that other chan now.
19:04 scrottie cheers!
19:04 scrottie left #parrot
19:04 chromatic Whiteknight, before I read the code I thought it might be pulling from a bufferlike pool.
19:05 Whiteknight isn't it? it is calling pool->get_free_object()
19:05 chromatic One of the sized bufferlike pools.
19:05 chromatic Not a string header pool.
19:06 Whiteknight oh, okay
19:07 sjansen joined #parrot
19:11 cognominal japhb, I updated the ticket
19:13 cjfields joined #parrot
19:15 jhorwitz_ joined #parrot
19:15 cxreg joined #parrot
19:25 allison joined #parrot
19:26 Ademan joined #parrot
19:27 Andy joined #parrot
19:40 smash joined #parrot
19:40 smash hello everyone
19:41 moritz hi smash ;)
19:41 japhb cognominal, OK, I'm back
19:51 ruoso joined #parrot
20:01 japhb cognominal: what ticket did you update?
20:01 japhb Man, I just cannot manage to be here at the same time as him ...
20:06 NotFound Someone can tell what Parrot_exec_rel_addr and Parrot_exec_rel_count are used for? I've deleted all his usages and test passed, make hello works and make perl6 works.
20:07 japhb cognominal: Oh, I see, 55228.
20:10 chromatic NotFound, make testj
20:11 NotFound Doing...
20:11 purl rumour has it doing is better than talking
20:12 NotFound PASS
20:13 NotFound TODO passed: 8
20:13 chromatic Hm, and you're on a platform with working JIT?
20:14 NotFound Ubuntu 8.04 i386
20:15 chromatic Okay, let's get someone on Mac OS X 10.4 on PPC to try the patch with make testj, and I'm happy.
20:15 NotFound Those variables are only in 386 jit files.
20:16 NotFound Maybe we must test in win 386
20:16 chromatic Oh yeah.  I keep forgetting.
20:24 dalek r28052 | chromatic++ | trunk:
20:24 dalek : [config] Resolved an undefined value warning when building Parrot without exec
20:24 dalek : core (reported by Will Coleda in RT #55244).
20:24 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28052
20:29 davidfetter joined #parrot
20:30 Zaba_ joined #parrot
20:32 particle meeting &
20:34 cotto_work joined #parrot
20:40 mj41 Hi, probaly today TapTinder can help a bit - http://tt.perl6.cz/
20:42 * japhb is having an extremely interrupt-driven day
20:43 japhb Hmmm, can make depend on a success or failure exit code from a script, rather than on the existence and timestamp of a file?
20:43 chromatic Doubt it.
20:44 particle wrap the script in another that generates a file upon successful exit of the inner script
20:44 particle really gone &
20:45 japhb particle: yep, I was thinking that's what I'd have to do.
20:45 chromatic Although if you had a rule that just ran the script and relied on its exit code....
20:45 dalek r28053 | coke++ | trunk:
20:45 dalek : [docs] Update list of upcoming release managers based on RSVPs.
20:45 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=28053
20:45 japhb Hmmmm.  Can the timestamp of a file be set arbitrarily in a cross-platform way?
20:45 Zaba joined #parrot
20:47 chromatic You probably don't want to get into that.
20:50 * japhb wishes make had a programmatic API, for about the hundredth time
20:58 Zaba_ joined #parrot
21:04 smash pmichaud: i'm getting a bit confused browsing svn, in PCT, the T stands for Tools or Toolkit in nowadays ?
21:06 cotto_work joined #parrot
21:07 NotFound Patch sended to #54602 with the cleaning of the exec_rel_addr things.
21:22 spinclad smash: Toolkit as i recall, but my #parrot logs gives for both
21:22 spinclad s/gives/give hits/
21:23 smash hehe
21:23 spinclad pct?
21:23 purl pct is probably a good solution, but heavyweight or the Parrot Compiler Toolkit
21:24 smash hmm, how updated are you purl ?
21:25 spinclad 2008-03-31: <@pmichaud> At the conference I'm wanting to do 2 or 3 talks regarding Rakudo (the Parrot implementation of Perl 6) and the new Parrot Compiler Toolkit. ...
21:25 smash ok, i guess that settles it
21:25 smash thks spinclad
21:28 Zaba joined #parrot
21:36 sjansen joined #parrot
21:49 gmansi joined #parrot
21:50 chromatic joined #parrot
21:51 cotto_work no purl, pct is the Parrot Compiler Toolkit
21:51 purl okay, cotto_work.
21:51 cotto_work pct?
21:51 purl rumour has it pct is the Parrot Compiler Toolkit
21:52 Zaba_ joined #parrot
22:16 Zaba joined #parrot
22:17 teknomunk joined #parrot
22:23 ank joined #parrot
22:43 Zaba_ joined #parrot
22:54 Limbic_Region joined #parrot
23:04 tetragon joined #parrot
23:05 japhb Anyone around who can discuss the current state and future direction of the config/build processes in Parrot?
23:13 Limbic_Region well at one point, the plan was not to use perl 5 - but the build/config is extremely dependent on perl 5 ATM
23:14 chromatic We'll likely stick with Perl 5 for Parrot 1.0.
23:14 Limbic_Region purl nopaste
23:14 purl nopaste is at http://nopaste.snit.ch/ (ask TonyC for new channels) or http://rafb.net/paste or http://paste.husk.org/ or http://poundperl.pastebin.com/ or http://paste.scsys.co.uk/ or don't bother me while I'm eating or App::Nopaste or tools/dev/nopaste.pl
23:16 nopaste "Limbic_Region" at 68.49.236.220 pasted "Parrot fails to finish building on Win32/MinGW due to PGE" (997 lines) at http://nopaste.snit.ch/13154
23:16 Limbic_Region purl msg pmichaud parrot fails to finish building on Win32/MinGW due to pge - see http://nopaste.snit.ch/13154
23:16 Limbic_Region grrr
23:16 chromatic boticide
23:16 Limbic_Region yep
23:17 smash hahahah
23:17 smash Limbic_Region: you bad person you  :-)
23:17 smash making poor bot crash !!
23:23 PerlJam Limbic_Region: If someone would volunteer to create a build/config system from miniparrot, that would be nice  ;-)
23:23 chromatic Yeah, but once you have miniparrot, you have most of libparrot built.
23:24 Limbic_Region PerlJam - as chromatic said, the trouble is getting there.  IMO, the real trouble is portability
23:25 chromatic Portability isn't that awful.
23:25 Limbic_Region no, it is a great thing really
23:25 chromatic Most of what configure needs to do is compile and run some C programs and fill in some templates.
23:26 PerlJam it would be interesting if we could elide the "compile some C programs" step in lieu of a table lookup of some sort.
23:27 Limbic_Region so here is my proposed solution - write a configure.c - step 1.  compile configure.c  step 2, run the resulting configure.exe  step 3 - run make :-)
23:27 chromatic A table lookup?
23:28 Limbic_Region oh, yeah - in addition to the PGE build failure on Win32/MinGW - the rakudo spectest suite seems to hang in an infinite loop on Win32/Cygwin
23:28 Limbic_Region t/spec/S29-conversions/ord_and_chr..................  # ran for 4 hours the other night
23:28 PerlJam chromatic: as part of the config step.  But I guess we need to do that anyway to get parro. Never mind me; I'm a little slow this evening
23:29 Limbic_Region should I submit a ticket for that - or is mentioning it here sufficient?
23:29 PerlJam lack of sleep and sick kids spreading their disease is enough to make  anyone slightly crazy :)
23:30 chromatic Someone should invent an antiseptic toddler.
23:30 cjfields joined #parrot
23:33 Zaba joined #parrot
23:36 japhb sorry all, got pulled away myself.
23:36 japhb I was thinking at a higher level than implementation ...
23:37 Infinoid if you think at a high enough level, you stop thinking altogether :)
23:38 japhb What do we want configure to do?  Right now, it conflates several concepts: auto-detection for features, platform details detection, user overrides, generation of a number of files, etc. ...
23:39 japhb We certainly need *something* to do each of these steps, but is configure the right place for all of them?  (My biased answer: no)
23:39 japhb I was honestly surprised that configure generated *anything* other than a file listing the detected and overridden settings.
23:40 PerlJam Thsi sounds like it would have made a good google SoC  :)
23:40 japhb I created config::gen::opengl and config::gen::call_list because they needed to fit into the current config framework, but it's always struck me as wrong.
23:41 japhb PerlJam: Too bad I can't even remotely afford to live on SoC wages.  Plus I'm not a student.  ;-)
23:41 smash cya all tomorrow
23:41 PerlJam At the very least it sounds like something that someone should apply to TPF for a grant for.
23:42 PerlJam (assuming someone has an idea how it should work)
23:43 japhb Too bad we can't do retroactive TPF grants:  "OK, I just did Feature X.  It took me 43 hours.  I'm applying for a grant for the Feature X results, so that I can afford to do that again with Feature Y."
23:44 PerlJam japhb: just do the work but don't release  it until after you've gotten the grant  ;>
23:44 japhb World's most detailed grant application.  :-)
23:45 japhb "I know exactly how this would have to work ...."
23:47 japhb Anyway, re: config and build system ...
23:48 PerlJam japhb: anyway .. just do the work  :)
23:49 japhb PerlJam: If I keep getting bug reports from people who haven't run 'make realclean', I just might.  It's frustrating to my work get spurious bug reports because the whole procedure doesn't DWIM for most people.
23:50 japhb Wow, that was mangled grammar.
23:50 japhb OK, what is the DWIM thing?  Why are people so adverse to 'make realclean'
23:51 japhb ?
23:52 PerlJam perhaps it's just because it's  an extra step to have to remember.
23:53 japhb From other projects that people are on, how are the makefiles generated?  Is it expected that Configure does this?
23:55 PerlJam 'make fromscratch' that did make realclean; perl Configure.pl && make  # one step, easy to remember  :-)
23:56 japhb PerlJam: that assumes that we have a 'local prefs' file that tells Configure.pl about it's overrides, so that when it is rerun there, it produces a matching config to what was done by hand
23:56 japhb ... which I was thinking made sense anyway.
23:57 japhb perl Configure.pl <options> should really just write <options> to the prefs file (or a temp one, whatever) and then run normally

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

Parrot | source cross referenced