parrot.org/ - clean up those smolders for the release!
Set by moderator on 20 October 2008.
00:10 AndyA joined 00:26 jhorwitz joined 00:31 davidfetter joined 00:37 davidfetter joined 00:47 jimmy joined 00:49 rhr joined
jonathan bacek++ # patch 00:51
dalek r32681 | jonathan++ | lex2: 00:52
: [rakudo] Update trans to check for Sub now that Closure is gone away.
diff: www.parrotvm.org/svn/parrot/revision?rev=32681
Coke hurm. how can I make TEST_JOB magic work on tcl? 00:54
hurm. nevermind. 00:55
01:00 rhr joined 01:05 jimmy joined
bacek_ jonathan: around? 01:08
jonathan: there is also registering Closure class in src/classes/Code.pir on line 22. It should be removed, AFAIU. 01:09
rakudo: say eval "my $x = 3" 01:18
polyglotbot OUTPUT[No handler to delete.␤current instr.: 'parrot;PAST;Compiler;as_post' pc 6052 (src/PAST/Compiler.pir:1651)␤called from Sub 'parrot;PAST;Compiler;post_children' pc 1779 (src/PAST/Compiler.pir:367)␤called from Sub 'parrot;PAST;Compiler;pirop' pc 3048 (src/PAST/Compiler.pir:791)␤called from Sub
..'parrot;PAST;Compiler;post_children' pc 1779 (sr...
bacek_ rakudo: say %*VM<version>
polyglotbot OUTPUT[Use of uninitialized value␤␤]
bacek_ rakudo: say %*VM.k
polyglotbot OUTPUT[Method 'k' not found for invocant of class 'Perl6Hash'␤current instr.: '_block11' pc 24 (EVAL_12:15)␤called from Sub 'parrot;PCT;HLLCompiler;eval' pc 866 (src/PCT/HLLCompiler.pir:501)␤called from Sub 'parrot;PCT;HLLCompiler;evalfiles' pc 1141 (src/PCT/HLLCompiler.pir:631)␤called from Sub
..'parrot;PCT;HLLCompiler;command_line' pc 1320 (src/...
bacek_ rakudo: say keys %*VM
polyglotbot OUTPUT[␤]
bacek_ hmm
Coke that's a lot of failure output. 01:19
bacek_ Coke: for 'eval "my $x = 3"'? 01:20
Coke ... just now. from the bot.
bacek_ 5 lines. 01:21
purl hmmm... 5 lines is pushing it :)
bacek_ purl: bad girl
purl bacek_: what?
01:22 klapperl joined
cotto perl6: say %*VM<config><revision> 01:31
polyglotbot OUTPUT[get_pmc_keyed() not implemented in class 'Undef'␤current instr.: '_block11' pc 40 (EVAL_12:23)␤called from Sub 'parrot;PCT;HLLCompiler;eval' pc 866 (src/PCT/HLLCompiler.pir:501)␤called from Sub 'parrot;PCT;HLLCompiler;evalfiles' pc 1141 (src/PCT/HLLCompiler.pir:631)␤called from Sub
..'parrot;PCT;HLLCompiler;command_line' pc 1320 (src/PCT/HL...
Coke src/exceptions.c has a reference to the old list. 01:35
Coke wonders why ^C is not working on feather. wtf.
01:39 allison joined 01:44 chromatic joined
Coke cries over his tcl memory explosions. :| 01:44
Tene Did we put "no memory explosions" on the board? 01:45
01:45 bacek_ joined
chromatic 4.0 01:54
01:57 silug joined
Whiteknight How often are we planning to have a major version change? Every year? 01:57
chromatic Yes. 01:58
If we use x.x release numbers.
Whiteknight okay 01:59
I have to put the 1.0 release into my hiveminder now 02:00
Who is the November release manager? 02:02
I got a branch here that needs a-mergin', but I don't want to do it so close to the tuesday release without permission 02:04
dalek r32682 | pmichaud++ | lex2: 02:07
: [rakudo]: There's no longer a Closure class.
diff: www.parrotvm.org/svn/parrot/revision?rev=32682
Coke next week is c. 02:10
and no, please wait until after release for trunk merging.
Whiteknight done
Coke ... unless it's teeny or super awesome.
Coke supposes he should cast his GC votes.
jonathan votes that GC gets faster
Coke wrong gc. 02:11
jonathan ;-)
Coke would love if that other gc was faster, though. :|
chromatic Which branch is that you want to merge? 02:14
Whiteknight calling_conventions. I'm waiting till wednesday to do it
I may make a few more changes to it before then, since I have time now
want to rename Parrot_PCCINVOKE to Parrot_pcc_invoke_method_from_c_args 02:15
Coke asks what folks here think of news.perlfoundation.org/2008/11/200...egrat.html
chromatic I'd like to ask Allison what she thinks of merging that branch tonight. 02:16
... provided you're at a stable point. 02:17
Whiteknight I'll do it whenever people say to do it. The branch is stable now
pmichaud ...I'm a little concerned it might conflict with the lexicals branch merge.
Whiteknight at least, it's stable until I get my grubby hands back onto it
chromatic Is the lexicals branch merge ready for tonight?
Or will it be?
pmichaud it will be.
Whiteknight what does the lexicals branch do?
pmichaud I need to discuss over dinner, assuming we're doing a group dinner. 02:18
I think the lexicals branch might break lua.
lua seems to do some odd things with Closure PMCs, which don't really exist in the lexicals branch.
(at least 'ack' reports quite a few Closure-related items in lua.)
Whiteknight: the lexicals branch is a re-implementatio of Parrot's lexical handling in contexts 02:19
chromatic Lua has its own closure type.
Whiteknight oh, okay
pmichaud chromatic: yes, but it's a subclass of Closure
which imcc no longer generates.
(at least I think I saw that it was a subclass of Closure)
I haven't tried it yet -- that's just what a quick 'ack' pointed out 02:20
dalek r32683 | chromatic++ | trunk:
: [ops] Added the box opcode, which takes a primitive and turns it into an
: HLL-appropriate PMC in one step. This invalidated bytecode, so 'make
: realclean' before you report any failing tests.
diff: www.parrotvm.org/svn/parrot/revision?rev=32683
pmichaud my question, of course, is this a blocker to performing a trunk merge and/or release?
(i.e., if we can't easily fix lua.)
bacek_ pmichaud: no... Lua doesn't build at all. Because of ".return" usage. 02:21
pmichaud a-ha!
so it already broke before me. That's good enough for me :-)
jonathan has a working prefix:| for arrays
bacek_ I can try to fix it
Who is Lua maintainer? 02:22
Coke does languages/LANGUAGES* say?
chromatic Francois and kjs 02:24
dalek r32684 | chromatic++ | trunk: 02:25
: [t] Added tests for non-constant primitive boxing for box opcode too.
diff: www.parrotvm.org/svn/parrot/revision?rev=32684
jonathan SEGV! 02:26
Damm.
Coke no whammies!
jonathan All I did was :named :flat...
pmichaud ...:named? 02:28
why :named?
prefix:| on arrays should just be :flat, yes?
jonathan pmichaud: |%h
I've got the array case working.
Then I was stupid enough to try and do the hash case too...
02:28 jimmy joined
pmichaud oh, that's going to be a bit ugly, isn't it? 02:29
how can | know if it has an array or hash?
jonathan I was faking it for now, in the end we should probably make :flat work on captures maybe.
I was for now checking the sigil. But it's sick.
I think I'm just not going to bother...
Coke that's some sick sigil, yo.
pmichaud If we're doing that, I prefer that we just do arrays for now and save hashes for later. 02:30
jonathan Yeah
Works for me - I don't have to find/fix a segv...
nopaste "bacek" at 123.243.38.218 pasted "Big patch to replace .return with .tailcall in Lua" (479 lines) at nopaste.snit.ch/14596
jonathan ooh, dinner time...
bacek_ Can someone apply s/\\.return/.tailcall/ patch from nopaste.snit.ch/14596 ? 02:31
pmichaud bacek_: does lua pass with that patch?
bacek_ Lua still broken, but at least it compiles...
jimmy hello, there is editor.in, and there is a line: pir.vim: pir_vim.in $(OPS_DIR)/*.ops ops2vim.pl , it is gened to: pir.vim: pir_vim.in $(OPS_DIR)\\\\*.ops ops2vim.pl,but i can't file where is gen code. can anybody help me?
pmichaud dinner time here
bbiaw
Whiteknight bacek_: I'll take a look at that patch for you
since all the other parroteers are out eating dinner and having a good time without us 02:32
bacek_ Whiteknight: thanks. Patch is pretty straight forward
jimmy it is in Parrot::Configure::Compiler,but i don't know wich line!
bacek_ :)
Coke gets that abms monkey off his back.
ambs, even.
Coke stares at the box opcodes. 02:33
bacek_ jimmy: hm. Can you rephrase your question?
is TGE totally obsoleted??? 02:34
Coke nope 02:35
Whiteknight bacek, what command should I use to apply this patch?
Coke tge is used by pheme & tcl, among others.
Whiteknight I'm not good at patching from the commandline
bacek_ Whiteknight: patch -p3 < file
Whiteknight: from languages/lua
Whiteknight: or patch -p1 < file from top-level parrot dir 02:36
jimmy bacek: I want which line do the changes
here is a line: pir.vim: pir_vim.in $(OPS_DIR)/*.ops ops2vim.pl , it is gened to: pir.vim: pir_vim.in $(OPS_DIR)\\\\*.ops ops2vim.pl
bacek_ Coke: ../../parrot ../../compilers/tge/tgc.pir --output=src/PASTGrammar_gen.pir src/PASTGrammar.tg
Null PMC access in get_string()
current instr.: 'parrot;TGE;Rule;__onload' pc 0 (TGE/Rule.pir:20)
Coke bacek_: what are you running to get that? 02:37
jimmy i means which is gen code line
bacek_ Coke: make in languages/lua
jimmy: next one
purl hmmm... next one is good, "times arrow"
bacek_ This line is make's dependencies 02:38
Coke bacek_: I just did an svn up, realclean, build, cd languages/lua, and make complets with no errors. 02:39
bacek_ Coke: I'm in lex2 branch...
Coke .... that's a very important bit of information. =-)
jimmy what next one ? 02:40
purl next one is probably good, "times arrow"
bacek_ Coke: no. It's actually "master" which is "trunk"
Coke bacek_: trunk works here.
r32684
Whiteknight does lua build?
bacek_ jimmy: "pir.vim: pir_vim.in $(OPS_DIR)/*.ops ops2vim.pl" declares dependencies, how to build "pir.vim"
Coke in trunk? yes. and passes at least the first few tests. 02:41
bacek_ looks like I have very old checkout. Like 2 days old...
Whiteknight well lua isn't building for me, with the patch that bacek_ gave me, or in a pristine checkout I keep 02:42
nopaste?
clunker3 pasta.test-smoke.org/ or paste.husk.org/ or nopaste.snit.ch:8001/ or rafb.net/paste or poundperl.pastebin.com/ or paste.scsys.co.uk/
purl nopaste is at nopaste.snit.ch/ (ask TonyC for new channels) or rafb.net/paste or poundperl.pastebin.com/ or paste.scsys.co.uk/ or App::Nopaste or tools/dev/nopaste.pl or at www.extpaste.com/
nopaste "Whiteknight" at 69.136.75.101 pasted "Lua failure" (6 lines) at nopaste.snit.ch/14597 02:43
Whiteknight that's in a pristine repo without the patch. With the patch fails very similarly
Coke and what does your line 21 look like? 02:44
Whiteknight line 21?
purl i think line 21 is open INFILE or die "Couldn't open '$INFILE\\ that you asked me to.n";
Coke ... the line that is reported in your failure?
jimmy bacek_: i need the gen code 02:45
bacek_ time for "realclean"...
jimmy not the make file means
Coke Whiteknight: sounds like you need to rebuild parrot.
Whiteknight .const 'Sub' entry = 'luaopen__helpers' 02:46
Coke the .const syntax changed at some point in the last month.
Whiteknight ok
Coke have you done the standard realclean, rebuild dance?
jimmy here is a line: pir.vim: pir_vim.in $(OPS_DIR)/*.ops ops2vim.pl , it is gened to: pir.vim: pir_vim.in $(OPS_DIR)\\\\*.ops ops2vim.pl 02:47
i just want to which code do the change
bacek_ jimmy: ah, ok. I'll try to find
Whiteknight okay, I'll rebuild. Shows you how much I've been working in trunk lately
jimmy i try to find the code, but i can not find it
Whiteknight bacek.needs(commit_bit).priority(high); 02:48
Coke kicks off a tcl spec test run to see how bad the memory leak is killing him. :|
jimmy go out now. please left message here.
thanks bacek_ 02:49
bacek_ Whiteknight: not yet.
Whiteknight >:( 02:50
bacek_ I managed to built Lua.... Few segfaults in "make test"...
bacek@icering:~/src/parrot/languages/lua$ make 02:51
../../parrot --output=src/lib/glut.pbc src/lib/glut.pir
error:imcc:No such file or directory
in file 'src/lib/glut.pir' line 69
Is it expected???
Or my "git svn" just goes insane?
Coke I see that as well, but I don't have glut.
bacek_ Coke: me either... 02:52
# './parrot languages/lua/lua.pbc languages/lua/t/functions_34.lua ' failed with exit code [SIGNAL 11].
Lot of them...
purl lot of them are from double-bounce@
Coke lua? 02:54
purl i heard lua was at www.lua.org/
Coke lua/MAINTAINER
03:10 chromatic joined
Coke chromatic, my tcl is borked. halp. 03:11
Whiteknight my lua is borked too. rebuilding parrot doesn't help 03:13
bacek_: somebody else is going to have to deal with this patch for you, my system is too stupid to do it 03:14
bacek_ Whiteknight: no problems.
polyglotbot: help 03:15
polyglotbot bacek_: Usage:
..<(lua.past|ruby.paste|lolcode.pir|php.past|pynie.pir|smalltalk.past|cardinal.parse|pheme.pir|smalltalk.parse|python.paste|cardinal.pir|php.paste|bc.paste|pynie|pheme.paste|punie|pipp|squaak.pir|tcl|ruby.past|squaak.parse|bf|lua.paste|cardinal.past|bc.pir|rakudo.parse|punie.past|pheme.parse|APL.pa
..rse|squaak.paste|APL.paste|chitchat.past|bc.parse|abc.paste|tcl.paste|python|bc.past|pipp.paste|php.pir|squaak.past|pynie.past|tcl.pir|rakudo.paste|pynie.paste|perl6.pir|smalltalk.paste|php.parse|tcl.past|pipp.past|lua|pynie.parse|python.parse|lua.pir|bf.parse|perl6.parse|php|lolcode.paste|chitch
..at|abc.parse|bf.paste|python.past|bf.pir|chitchat.paste|lua.parse|cardinal.paste|rakudo|APL|pheme|APL.past|punie.parse|bf.past|abc.past|APL.pir|perl6.past|ruby.parse|abc|tcl.parse|pipp.parse|perl6.paste|smalltalk.pir|smalltalk|punie.pir|lolcode.past|bc|rakudo.past|ruby.pir|punie.paste|squaak|lolc
..ode.parse|python.pir|perl6|cardinal|ruby|rakudo.pir|pipp.pir|pheme.past|lolcode|chitchat.parse|chitchat.pir|abc.pir): $perl6_program>
Coke smacks polyglotbot
chromatic: aha. I'm only getting the memory errors on feather. :| 03:16
Whiteknight punie.pir
<punie.pir>
bacek_ lua: f = function () return 1 end; print(f .. "end")
Coke those test files complete on my intel/osx box.
polyglotbot RESULT[lua.pbc: _._:1: unexpected symbol near '( ( do { f = function () return 1 end'␤]
bacek_ lua: f = function () return 1 end print(f .. "end") 03:17
polyglotbot RESULT[lua.pbc: _._:1: 'eof' expected␤]
bacek_ lua: f = function () return 1 end
polyglotbot RESULT[lua.pbc: _._:1: 'eof' expected␤]
03:31 Psyche^ joined 03:35 s1n joined 03:45 jimmy_ joined
jimmy_ hello bacek_ 03:48
04:36 chromatic joined
Coke chromatic: how goes on the wrong coast? 04:36
04:38 Andy joined
chromatic Coasterific. 04:42
Coke way to keep google from getting damp. 04:43
04:43 particle joined
Coke ... Do you have any idea how much damage this bulldozer would suffer if I let it run right over you? 04:44
chromatic None at all? 04:45
Tene Coke: yeah, feather is kinda hosed. parrot was failing badly there for me too. 04:47
04:55 allison joined
jimmy_ i don't know why 05:17
'pir.vim: pir_vim.in $(OPS_DIR)/*.ops ops2vim.pl
$(CP) pir_vim.in pir.vim
$(PERL) ops2vim.pl $(OPS_DIR)\\\\*.ops >> pir.vim
' can work and
'pir.vim: pir_vim.in $(OPS_DIR)\\\\*.ops ops2vim.pl
$(CP) pir_vim.in pir.vim
$(PERL) ops2vim.pl $(OPS_DIR)\\\\*.ops >> pir.vim' can't work
Coke you don't need to use \\\\ in perl on win for dir separators. 05:18
jimmy it a makefile
Coke if the / works, why change it?
3rd line is a perl invocation.
jimmy because the gen code do this change.
Coke in the makerule, you don't need to escape then \\
should be: pir.vim: pir_vim.in $(OPS_DIR)\\*.ops ops2vim.pl 05:19
jimmy so i wan't find the gen code.
yes
so i wan't find the gen code line where does the change.
want
Coke "want to"
jimmy yes
Coke somewhere in config\\ 05:20
hurm. or lib\\
particle # ex: set ro ft=make:
# DO NOT EDIT THIS FILE
# Generated by Parrot::Configure::Compiler from config/gen/makefiles/editor.in
Coke particle uses vi on windows. perhaps he can be movtivated to help you track this down.
-> sleep
05:21 kj joined
dalek r32685 | tewk++ | trunk: 05:24
: Refactor to eliminate code duplication
diff: www.parrotvm.org/svn/parrot/revision?rev=32685
r32686 | kjs++ | trunk: 05:29
: [pirc] some cleanups, docs etc.
diff: www.parrotvm.org/svn/parrot/revision?rev=32686
Tene pmichaud: I have a question about HLLCompiler when you get a few minutes. 05:31
05:32 Andy joined 05:43 stockwellb joined
jimmy particle: there may be something in Parrot::Configure::Compiler 05:44
the makefile gened from editor.in is not passed in windows 05:45
pmichaud Tene: go ahead
jimmy there may do a wrong replace in Parrot::Configure::Compiler 05:46
Tene pmichaud: In HLLCompiler, when I'm registering a class and putting the protoobject in the appropriate namespace, where do I get the HLL from?
pmichaud isn't there a hll option to register?
Tene Is there? Lemme check.
jimmy can anybody fix the problem? 05:47
Tene erm, P6Object, not HLLCompiler. 05:48
Yeah, looks like you're right.
I wonder what my problem was the last time I tried this...
stockwellb In a test file, with a forced failure like 0k (0, 'some comment'). Should *some comment* describe what failed ore what was tested? 05:51
s/ore/or 05:52
pmichaud Tene: so, are you asking about HLLCompiler or P6object? 05:56
in HLLCompiler there's no option. Instead, you just put a .HLL directive at the top of the master .pir file
e.g., in perl6, we would put .HLL 'perl6' at the top of perl6.pir
Tene Right. I was trying to remember the problem I ran into last time, but apparently failed badly. 06:00
06:01 apeiron joined
pmichaud issue #1 for .hll: we don't have a way to import the PAST::* classes into the HLL namespace 06:01
Tene Was there some other prerequisite to making PCT hll-aware that you wanted me to finish first?
pmichaud soooo
Tene Particle's export stuff was relevant to that, wasn't it?
dalek r32687 | pmichaud++ | lex2:
: [pct]: Update to properly close named inner subs.
pmichaud yes, but that only applies to stuff written in Rakudo
dalek diff: www.parrotvm.org/svn/parrot/revision?rev=32687
pmichaud in P6object, when we create a protoobject and bind it into the namespace, I want it to also bind into EXPORT::ALL 06:02
Tene so when I new_class 'Foo::Bar', also put the proto in Foo::Bar::EXPORT::ALL ? 06:03
pmichaud for example, if P6object creates a class named Foo::Bar::Baz, I want it to create ['Foo';'Bar'], 'Baz' and ['Foo';'Bar';'EXPORT';'ALL'], 'Baz' 06:04
dalek r32688 | allison++ | trunk:
: [cage] Deprecation note for src/pic.c.
diff: www.parrotvm.org/svn/parrot/revision?rev=32688
pmichaud (both of which are bound to the protoobject)
chromatic stockwellb, what was tested. 06:05
Tene Okay.
P6object currently has a set of loops to deal with the fact that get_root_namespace on an RPA won't create a namespace if it doesn't already exist. I believe you told me that get_root_namespace should be creating namespaces if they don't exist? 06:07
Does that sound familiar?
pmichaud I did tell you that, but I think we determined I was wrong.
Tene Ah, okay.
pmichaud however
you should just be able to do set_hll_global
*that* automatically creates the namespace 06:08
(or set_root_global)
Tene Looks like we're getting the namespace so we can run newclass on it.
pmichaud set_root_global ['hll';'Foo';'Bar';'EXPORT';'ALL'], 'Baz', $P0 # automatically creates namespace
currently we have 06:09
set_root_global ns, $S0, protoobject
(line 396)
all that is needed after that is
push ns, 'EXPORT'
push ns, 'ALL' 06:10
set_root_global ns, $S0, protoobject
stockwellb chromatic: thank you.
Tene P6Object.pir:485 is the end of those loops.
... right, that's why I was in here.
pmichaud that's if we're creating a class
but we already have the protoobject -- we just need to bind it into an additional location 06:11
Tene Yeah, I was off on a tangent there.
pmichaud ...in fact, we could (and should) use set_root_namespace to create the namespace we need.
instead of the loop.
Tene What do we set in it? 06:12
Undef? Failure?
pmichaud looking.
stockwellb chromatic: may a post a test snippet for you to look at?
s/post/nopaste 06:13
chromatic Of course!
nopaste "stockwellb" at 70.160.222.145 pasted "try catch finally" (22 lines) at nopaste.snit.ch/14598
pmichaud $P0 = new 'NameSpace' 06:14
Tene Clever.
stockwellb chromatic: right track?
pmichaud $S0 = pop base_ns
set_root_global base_ns, $S0, $P0
but even easier might be (for now) 06:15
set_root_global base_ns, '', $P0
pretty much anything will work for $P0 there -- the point is simply that we want Parrot to create the namespace.
chromatic stockwellb, that's reasonable, though for clarity's sake I'd set $I0 to 1 in the non-exception case and call ok() only once per unique test (with $I0 instead of a constant first parameter).
That makes it easier for someone skimming the code to count how many tests pass or fail.
stockwellb I see, it looks like three ok and one is test! 06:17
chromatic: thanks for the help. 06:18
chromatic You're welcome.
A good name for a new language: PIMPMOBILE 06:22
A bad name for a good language: POPEMOBILE
dalek r32689 | tene++ | trunk: 06:27
: [P6Object]: Add created protoobjects to the default export namespace
diff: www.parrotvm.org/svn/parrot/revision?rev=32689
r32690 | tene++ | trunk:
: [P6Object]: Replace some ugly loops with a simple workaround.
diff: www.parrotvm.org/svn/parrot/revision?rev=32690
06:32 iblechbot joined
dalek r32691 | allison++ | pdd22io: 06:36
: [pdd22io] Un-todo readline test.
diff: www.parrotvm.org/svn/parrot/revision?rev=32691
pmichaud tene++ 06:41
do we have tests for those?
I (we?) also need to clean up t/library/p6object.t -- there's a lot of commented-out tests there
Tene No tests for the former. No functional change to test for the latter. 06:43
pmichaud we should probably add a test for the EXPORT::ALL binding.
07:03 TiMBuS joined
particle s/probably/definitely/ 07:29
07:30 Zaba joined
Tene Where does the contents of the EXPORT::ALL namespace get imported? I'm having a bit of trouble tracking that down. 07:30
07:33 johbar joined
dalek r32692 | allison++ | pdd22io: 07:37
: [pdd22io] Make write buffering work.
diff: www.parrotvm.org/svn/parrot/revision?rev=32692
Tene ... okay, looks like it doesn't happen. 07:43
pmichaud: can you confirm that this should work: gist.github.com/25429 ? 07:45
dalek r32693 | allison++ | pdd22io:
: [pdd22io] The interpreter's STDERR is still an old-style I/O object during the
: temporary migration, so uses old-style function.
diff: www.parrotvm.org/svn/parrot/revision?rev=32693
Tene pmichaud: and if so, does that mean that I need to export the EXPORT::ALL::* entries into the calling namespace from rakudo's use or from HLLCompiler's evalfile, or where? 07:46
dalek r32694 | chromatic++ | trunk: 07:47
: [IMCC] Added documentation for all undocumented functions in this file.
: Tidied code (very minor).
: Improved IMCC_INFO(interp)->ghash clearing from r32633; it looks unlikely that
: resetting global symbol colors will work correctly, at least with IMCC as it is
: now.
diff: www.parrotvm.org/svn/parrot/revision?rev=32694
07:54 tewk_ joined
Tene ... or is it supposed to happen in NS's export_to ? 07:55
particle tene: you don't need to worry about the import end of things 07:56
Tene particle: explain please?
pmichaud I'm going to fudge the import side
particle as long as it's in ::EXPORT::ALL, it's ready for pmichaud's use
pmichaud and the import that we use for PAST won't be the same as 'use' 07:57
because it's NQP and actions.pm that needs to know about the symbols
(and neither NQP nor actions.pm use 'use')
Tene the docs for the export_to method on the NameSpace pmc mention a default set of items that isn't yet implementes.
pmichaud correct
Tene Is that related?
pmichaud I think that's probably bogus 07:58
(the namespace pdd needs a serious rethink)
otoh, I would like to see us go ahead and create the make_root_namespace operation
(and make_hll_namespace)
I can think of a couple of other places where we'll want to use it soon. 07:59
Tene pmichaud: Can you vaguely describe an architecture for this sometime soon so I can start working on it? 08:03
particle for make_*_namespace? it's straightforward: 08:04
if the namespace exists, do nothing
if not, create it 08:05
Tene No, for the export stuff we need for dealing with HLL issues.
particle ah
pmichaud I'm simply following what's described in PDD13
sorry, S13
Tene ... oh. :)
pmichaud so, we put all of the symbols to be exported into an EXPORT::ALL namespace 08:06
then, to import the symbols, we iterate over the namespace rebinding the names we find in EXPORT::ALL to the namespace we're wanting to export them to
in the case of trying to get .HLL to work, we'll bind PAST::EXPORT:ALL into the PAST namespace in Rakudo's HLL namespace 08:07
Tene Are we adding real support for this somewhere, or just faking it somewhere?
pmichaud that way actions.pm can find them by using PAST::Var.new(whatever) 08:08
"real support" depends on what we mean by "real support
"
the way that rakudo handles imports won't likely be the same as the way we do it in NQP
sorry, I was wrong again -- Synopsis *11* 08:09
particle bed & 08:10
Tene So is some part of PCT going to be binding the PAST::EXPORT::ALL namespace into the caller's HLL namespace, or will the individual HLLs need to do this themselves? 08:12
pmichaud it's going to be a part of PCT.
I haven't entirely decided where yet.
but I think I can write the EXPORTALL module that I want
Tene Also, won't this make the PAST namespace available to all user programs in all pct-using hlls? Is this an issue we need to consider? 08:13
pmichaud yes, it does that for now 08:14
Tene Okay.
pmichaud but it's more than that, because the grammar and the action methods are also "available to all user programs..." 08:15
I suspect that they'll eventually go into each HLL's private namespace
well, except for Rakudo, which will have them in the public namespace
Tene I remember Coke also talking about HLL private namespaces. Is that specced anywhere?
pmichaud pdd21_namespaces 08:16
Tene :)
cotto would this be a good time to ask for a resolution to #59810 (using a non-default hash seed causes brokenness)? 08:19
pmichaud I think most everyone's headed off to bed
cotto so probably not then 08:20
pmichaud I can say that we decided today that PARROT_API doesn't mean what it says
and it's going to change shortly after the next release 08:21
essentially, things that say PARROT_API today really mean "parrot export" -- i.e., those labels are there to indicate that these symbols have to be available to the dynamic loader/linker 08:23
so they'll probably change to read PARROT_EXPORT
cotto The function should be made less surprising either way.
pmichaud then, we'll be doing a review of all of the functions to decide which are really part of the "API" -- i.e., those that we consider to be "user facing" functions
I don't know that string_hash will fall in that category.
beyond that, I'm not familiar enough with parrot guts to say anything intelligent on the matter 08:25
Tene: think you'd be able to take a crack at writing EXPORTALL ?
cotto pmichaud, is there a good place I could add a note about that specific function so it'll come up during the review? 08:26
pmichaud there's already a note in the source, yes?
Tene pmichaud: is it different from NameSpace's export_to method? 08:27
pmichaud Tene: checking.
(probably yes.)
cotto pmichaud, there is
pmichaud Tene: I'm not looking to necessarily implement the PDD, because I think (and hope) it will change in a lot of ways 08:28
I'd like to have a method on HLLCompiler called 'EXPORTALL'
it should take a source and destination namespace as parameters
(you can assume they are already namespace PMCs for now) 08:29
it should then loop through all of the symbols in the source's 'EXPORT';'ALL' subnamespace, and for every symbol it finds it should bind it into the destination namespace with the same name.
so, for example: 08:30
$P0 = get_root_namespace ['parrot';'PAST']
$P1 = make_root_namespace ['perl6';'PAST']
hllcompiler.'EXPORTALL'($P0, $P1)
and all of the symbols that are in ['parrot';'PAST';'EXPORT';'ALL'] then go into ['perl6';'PAST'] 08:31
Tene Sure. I can do that. 08:32
pmichaud cotto: since the note is there I suspect it'll get some review.
Tene: we may need to adjust the interface a bit in the future, but I think we can make this work for now.
cotto In that case, I'll leave it at that. 08:33
pmichaud cotto: is pipp blocking on this particular item? 08:34
cotto nope. It's just something that should eventually be solved. 08:35
pmichaud then yes, it'll probably come up in the normal course of things.
dalek r32695 | pmichaud++ | lex2: 08:50
: [core]: Get capture_lex to work on MultiSub PMCs.
diff: www.parrotvm.org/svn/parrot/revision?rev=32695 08:51
r32696 | pmichaud++ | lex2:
: [rakudo]: Add VTABLE_get_iter to Perl6MultiSub (needed for capture_lex)
diff: www.parrotvm.org/svn/parrot/revision?rev=32696
pmichaud time for sleep 09:09
actually, past time for sleep.
back tomorrow.
dalek r32697 | pmichaud++ | lex2: 09:21
: [core]: Turn off the capture_lex check for an outer sub.
diff: www.parrotvm.org/svn/parrot/revision?rev=32697
09:23 iblechbot joined 09:41 Ademan_ joined
cotto is barney at pds? 09:48
dalek r32698 | cotto++ | trunk: 10:05
: [t] add a TODO'd test for #60556
diff: www.parrotvm.org/svn/parrot/revision?rev=32698
r32699 | cotto++ | trunk: 10:08
: [t] add RT ref to previous commit
diff: www.parrotvm.org/svn/parrot/revision?rev=32699
10:12 klapperl_ joined 10:15 jimmy joined 10:22 ruoso joined
dalek r32700 | fperrad++ | trunk: 11:03
: [forth]
: - update exception usage
: - TokenStream : overload set_native_string instead of new_from_string
: - all tests pass
diff: www.parrotvm.org/svn/parrot/revision?rev=32700
jimmy # replace \\* with \\\\*, so make will not eat the \\
this comment sometimes is not believable 11:04
for example: 11:05
pir.vim: pir_vim.in $(OPS_DIR)\\*.ops ops2vim.pl is availabe
but pir.vim: pir_vim.in $(OPS_DIR)\\\\*.ops ops2vim.pl is not availabe
maybe t\\\\*.t t\\\\*\\\\*.t t\\\\*\\\\*\\\\*.t: all is availabe but i don't know if 't\\*.t t\\*\\*.t t\\*\\*\\*.t: all' whether is availabe too. 11:07
dalek r32701 | fperrad++ | trunk: 11:11
: [Lua]
: - install some extension libraries
diff: www.parrotvm.org/svn/parrot/revision?rev=32701
11:11 klapperl joined
jimmy hello cotto 11:38
nopaste
clunker3 pasta.test-smoke.org/ or paste.husk.org/ or nopaste.snit.ch:8001/ or rafb.net/paste or poundperl.pastebin.com/ or paste.scsys.co.uk/
jimmy nopaste?
clunker3 pasta.test-smoke.org/ or paste.husk.org/ or nopaste.snit.ch:8001/ or rafb.net/paste or poundperl.pastebin.com/ or paste.scsys.co.uk/
purl i guess nopaste is at nopaste.snit.ch/ (ask TonyC for new channels) or rafb.net/paste or poundperl.pastebin.com/ or paste.scsys.co.uk/ or App::Nopaste or tools/dev/nopaste.pl or at www.extpaste.com/
jimmy cotto: hello 11:39
11:39 masak joined, barney joined
jimmy hello masak 11:41
nopaste "jimmy" at 219.134.167.104 pasted "[PATCH]'make vim-install' works in Windows NT series now." (44 lines) at nopaste.snit.ch/14599
jimmy now vim plugin can be installed by 'make vim-install' in windows now. 11:42
stop 11:43
wrong patch
masak zaochen hao, jimmy 11:53
jimmy 早上好啊，masak，我们这里是晚上8点了，呵呵 11:54
masak 是，当然。:) 11:55
that's because the Earth is round :)
while people are all "OMG!" about the 1.0 release being scheduled for March, I'm all "OMG!" about lex2 being about to be merged back into trunk. 11:57
it feels unreal, somehow.
pmichaud++ jonathan++
jimmy yes. 11:58
cognominal Christmas is in March this year?
jimmy lex2 being about to be merged back into trunk? where? 11:59
masak jimmy: right here, before our eyes. 12:00
jimmy: but I don't know exactly when.
last estimate I heard was that it's be merged by yesterday
12:03 bacek joined
bacek good night 12:03
purl sleep fast, sleep tight, don't let the bed bugs bite
jimmy masak: i will take a look.
dalek r32702 | bernhard++ | trunk: 12:04
: [codingstd] remove trailing space
diff: www.parrotvm.org/svn/parrot/revision?rev=32702
bacek masak: lex2 isn't merged yet. 12:08
But it looks great
dalek r32703 | bernhard++ | trunk:
: [codingstd] Satisfy c_parens.t
diff: www.parrotvm.org/svn/parrot/revision?rev=32703
masak bacek: no, that's right. that's what I meant by 'about to be merged' 12:09
I'm glad it looks great. I hear it's about to be merged. :) 12:10
dalek r32704 | bernhard++ | trunk: 12:26
: [PCT] add test case where an 'our' var is set in called action
diff: www.parrotvm.org/svn/parrot/revision?rev=32704
12:28 gmansi joined
dalek r32705 | bernhard++ | trunk: 12:29
: [PCT] remove unnecessarg attribute 'pasttype' in tests
diff: www.parrotvm.org/svn/parrot/revision?rev=32705
12:34 purl joined 12:46 rurban joined 13:09 gmansi joined 13:22 particle joined 13:27 Whiteknight joined 13:42 particle joined 13:54 register joined
dalek r32706 | bernhard++ | trunk: 14:03
: [Pipp] Set predefined constant PHP_OS
diff: www.parrotvm.org/svn/parrot/revision?rev=32706
pmichaud (lex2) I'm still running into spectest failures.... every time I think "ah, *this* will finish it" some other segfault or bug appears. 14:07
particle is your latest committed? 14:08
pmichaud except for the change I made just 2 minutes ago, yes. 14:09
now committed, r32707 14:10
dalek r32707 | pmichaud++ | lex2:
: [pct]: We also capture_lex when taking a reference.
diff: www.parrotvm.org/svn/parrot/revision?rev=32707
particle rebuilding now
spectest is fine so far, in S02-literals now 14:14
Whiteknight pmichaud: That's how my GC work went: everytime I fixed one segfault, another one appeared.
pmichaud four spectest files failing 14:30
Whiteknight: yes. In this case the problems are compounded because of some limitations in the current parrot design (limitations we're being quite effective at redesigning this weekend :-)
particle i get two failures 14:31
pmichaud s/compounded/complex/ 14:32
which files?
purl which files are burnt at the stake.
pmichaud purl, forget which files
purl pmichaud: I forgot which files
particle S02-polymorphic_types\\subset-code.
S12-class\\declaration-order.t
damned declaration-order
pmichaud declaration-order.t is a known failure from trunk
particle can we regress that?
known failures suck
pmichaud yes, we'll do that tomorrow if we haven't fixed it by then.
particle ah, right, release.
perfect. 14:33
pmichaud (I'd prefer to keep it "visible" during the summit in case we get a chance to look at it)
ah, the tests pass with -G 14:35
that makes me think it's a garbage collection issue
...does -G affect garbage collection of contexts, though? 14:36
Whiteknight I'm looking forward to seeing all your changes!
Infinoid maybe something is freed that shouldn't be? 14:37
pmichaud it's likely -- handling the reference counts on contexts is tricky (and afaict broken in a few key cases)
Infinoid hmm. is that something we can test directly? 14:39
pmichaud I don't know, I haven't looked into it. allison also says that we're likely to make contexts into PMCs so they become part of normal garbage collection 14:40
Whiteknight (become part of normal garbage collection)++
Infinoid yeah, refcounts suck. sounds good to me :)
I'm curious about the current refcounting stuff, though. where's the code for this? 14:41
pmichaud it's only for contexts, so primarily src/sub.c and src/gc/register.c
particle spectests are run with -G by default? i thought that had changed 14:42
pmichaud no, they aren't run with -G by default. that's why they're failing
particle sorry, i don't get it. 14:44
you have different tests failing, i guess
pmichaud right
particle which files fail for you?
pmichaud $ cat t/localtest.data 14:45
t/spec/S02-polymorphic_types/subset-code.t
t/spec/S02-polymorphic_types/subset-range.t
t/spec/S12-class/attributes.t
t/spec/S12-class/inheritance.t
particle ok, i'll run those with a few permutations of options and flags after a shower 14:48
Coke makes sure lex2 doesn't break tcl, JIC.
masak cheers on the lex2 branch 14:50
Infinoid ah, t/op/calling.t seems to show a context freeing failure. pmichaud, that's the one you're talking about? 14:53
Failed test 'Tail call without arguments should not free the context when a closure depends on it'
pmichaud Infinoid: no, we're doing tests in the spectest branch 14:54
sorry
test in rakudo, "make spectest"
but yes, there may also be tests in t/op/calling.t that I don't know about.
Infinoid well, if parrot doesn't work, I wouldn't expect rakudo to work either, so I figured I'd start with the basics. :)
pmichaud I had parrot "make test" working before this last batch of changes, so apparently something regressed.
Infinoid ok, that's a starting point. thanks 14:55
dalek r32708 | bernhard++ | trunk:
: [Pipp] Add some predifined constants
diff: www.parrotvm.org/svn/parrot/revision?rev=32708
pmichaud r32709 has some context changes. 14:56
dalek r32709 | pmichaud++ | lex2:
: Make sure we mark our outer contexts as live, or their contents get GCed.
diff: www.parrotvm.org/svn/parrot/revision?rev=32709
Infinoid gah, git has spoiled me. I keep expecting "svn blame" to automatically pipe itself to less 14:57
Coke particle: "wontfix" isn't the same as "resolved" 14:58
(configure ticket.)
(hurm. rt spells it as "rejected". There we go.) 14:59
Infinoid r32709 change didn't affect t/op/calling.t 15:00
Infinoid works on figuring out which context was freed 15:02
15:06 Theory joined
dalek r32710 | bernhard++ | trunk: 15:08
: [Pipp] Add test using the constant syntax
diff: www.parrotvm.org/svn/parrot/revision?rev=32710
particle coke: "for now" isn't the same as "resolved", either
Coke particle: the ticket was marked as resolved. 15:10
I did not verify who resolved it, though. 15:11
particle ah, didn't realize the ticket was resolved, just responded to an email in my inbox
Coke kid51.
wow. even more awesome, it was closed in august. wtf. 15:12
15:12 jimmy joined
Coke you also don't seem to have responded to the ticket. 15:12
Coke re-resolved the ticket, assuming it actually was. 15:13
tcl fails horribly in lex2, but only because lex2 hasn't been updated from trunk recently enough.
dalek r32711 | bernhard++ | trunk: 15:14
: [Pipp] Normalize here doc delimiters
diff: www.parrotvm.org/svn/parrot/revision?rev=32711
Coke er, well, I don't know if that's the ONLY reason, but it does mean I can't test it.
particle i responded to a private email from you to me and c
Coke ... ah 15:15
jimmy nopaste.snit.ch/14601
hello cotto or coke?
Coke Sorry; last email in a 50 line thread, didn't check the headers.
bacek pmichaud: I have own t/02-regression in rakudo (for few lexical issues). Is it feasible to submit patch for it? 15:16
Coke jimmy: ?
jimmy >\tnopaste.snit.ch/14601
i created a patch
Coke jimmy, we need to get you submitting patches via RT. =-)
jimmy i am in blacklist.
Coke particle: can you review that patch since you actually use fancy vi stuff on windows? 15:17
(I merely use straight vim)
Coke is very frustrated at not being at the pds. 15:18
jimmy and makefile something like: @echo "http:/smoke.****.org\\"
it is wrong
pmichaud *sigh* well, looks like I broke something in lex2 core -- guess I have to go back and figure that out. :-(
Coke I can't test the calling conventions branch for the same reason; it doesn't have enough of the latest mods from trunk. 15:19
jimmy and somebody can rescue me from the RT blacklist?
Coke jimmy: please send me an email describing your problem, and I will forward it to the RT admins. 15:20
(I'm assuming that if you try to mail them directly, the same blacklist will apply.)
basically "I sent an email to this address and it never arrived."
jimmy zhuomingliang@yahoo.com.cn
i am in the RT blacklist 15:21
Coke ... fine. Never mind the email.
you sent to parrotbug@parrotcode.org, correct?
jimmy i had sent to. 15:22
I had sent to there 3 email
and 1 email to rt-admin@perl.org
Coke email sent. 15:24
I'll let you know if I hear anything.
jimmy thanks coke
and this patch? nopaste.snit.ch/14601 ?
Coke I'll open a ticket so someone who knows about it can do something about it. 15:25
jimmy coke: thanks again :) 15:26
bacek pmichaud: looks like r32709 broke lex2.
jimmy and hello bacek, i fixed the problem now. thanks 15:27
bacek pmichaud: hmm... If I remove 'use v6' from my test it start working...
jimmy: great :)
pmichaud bacek: broke lex2 in _what_? 15:28
(what are you testing that it broke?)
and does it work in r32708?
Coke jimmy: you have an RT login; when I create a ticket and plugin your address, it knows you are. whee. 15:29
jimmy yes
i am jimmy
Coke that patch is rt 60584 15:30
anyone here can do the same thing by opening the ticket with a patch, then adding you as a requestor to the ticket, and removing themselves.
pmichaud rant: working on lexicals in parrot (and the lex2 branch) is becoming critically -Onofun for me. 15:31
jimmy thanks
dalek r32712 | fperrad++ | trunk:
: [PIR]
: - update syntax : use .tailcall
: (all tests fail)
diff: www.parrotvm.org/svn/parrot/revision?rev=32712
Infinoid reboots his swap-dead laptop, sets a process memory quota and tries again 15:32
nopaste "bacek" at 211.29.157.151 pasted "58392.t for pmichaud" (16 lines) at nopaste.snit.ch/14602 15:33
pmichaud bacek: we already have that in the spectest suite.
bacek pmichaud: this is my test. It works. But if I add 'use v6' it crashes with segfault
pmichaud bacek: noted. 15:34
15:39 klapperl joined
dalek r32713 | Whiteknight++ | calling_conventions: 15:44
: [calling_conventions] While I'm waiting for the release on tuesday... Rename Parrot_PCCINVOKE to Parrot_pcc_invoke_method_from_c_args.
diff: www.parrotvm.org/svn/parrot/revision?rev=32713
masak can one hope for a lex2 merge before the release? 15:46
Infinoid I have no idea. is the context freeing issue the only known blocker? 15:48
masak Don't know.
Infinoid hmm. -G doesn't affect t/op/calling_95.pir - I'm not so sure invalid frees are the cause of this failure 15:49
jimmy i found that some .pir files uses PASM style like 'I0' 15:50
register is there some documentation about Parrot::Configure internals?
Infinoid those modules include POD documentation, "perldoc lib/Parrot/Configure.pm" is a good start 15:51
register thx 15:52
Infinoid "perldoc docs/configuration.pod" is the other half
jimmy i found that some .pir files uses PASM style like 'I0' and it is [DEPRECATED]
Infinoid patches welcome. :) 15:53
jimmy but i can't test it
it is ok？
search such as ' I0', and found it. 15:54
Infinoid it's better than nothing, I suppose. you can send a patch to parrotbug and mention that it's untested 15:55
that way, the issue is tracked, even if the patch isn't perfect.
Whiteknight we can always test it for you
Infinoid yep
I think the motivation behind deprecating that was to try to move toward named variables in PIR, by making the non-named ones to be as ugly as possible 15:56
s/to be //
pmichaud masak: yesterday I was completely expecting that lex2 merge would happen before release... today I'm very uncertain.
masak pmichaud: ok.
pmichaud the number of things I'm having to change/fix/workaround to get it to work is growing quite rapidly 15:57
(and that's not a good sign.)
masak pmichaud: I understand this is a difficult fix. I'm hoping you solve it eventually. Just wondering when it's time to pop the champagne.
Infinoid ok, compiling Test.pm to Test.pir is using an insane amount of memory... think that's what killed my laptop last time 15:59
pmichaud yes, I think we're not reclaiming contexts.
i.e., we're leaking contexts.
Infinoid (this time, I killed it at the 2GB marker)
pmichaud I need to check out of the hotel and then will work on this a bit more in the lobby. 16:02
Infinoid no problem.
r32708 does not use nearly as much memory
pmichaud right
wait, it doesn't use as much memory as r32709 ?
I think I'll need to go chase down all the ref_count modifiers and make sure they're in the right places. 16:03
Infinoid that's what it seems like
r32709 runs my laptop out of memory, r32708 completes normally
Infinoid double checks that 16:04
pmichaud okay
I'm a bit surprised but that, but it's possible.
s/but/by/
the core failures are in t/op, yes? 16:05
Infinoid yes
pmichaud there's not much point in building rakudo until we get those fixed.
Infinoid though my test for this is just "make" in languages/perl6
for some reason, generating Test.pir triggers this leak
pmichaud Test.pir is the first program that is generated from something that uses lexicals.
Infinoid ok, double checked 16:06
bacek wanders why parrot needs refcount?
Infinoid bacek: because contexts aren't tracked by the normal GC yet
pmichaud bacek: I think the idea was that it would be more efficient to do refcounting on contexts instead of normal GC
bacek why?
pmichaud having played with this a bit, I somewhat agree. 16:07
(I'm not saying we should keep it that way, just that I can see the argument.)
why refcount: context creation happens a lot at well defined points, so it's realtively easy to know "when to allocate" and "when to release" 16:08
also, at the time it was created contexts were more of a stack than a tree
bacek and make mistake easily :)
pmichaud (er, context creation was more of a stack than a tree)
that's one of the things the lex2 branch is having to take care of -- it's having to remove all those places where it was assumed that outer contexts were part of the call stack 16:09
Infinoid pmichaud: specifically, the core failures are t/op/calling.t and t/op/lexicals.t
pmichaud Infinoid: okay, I just wanted to check that I could do "prove t/op" to check for core failures instead of a full "make test" 16:10
Infinoid yep
pmichaud okay, checkin g out of hotel -- bbiab
Infinoid some of the output from lexicals.t looks like a recursive exception 16:11
pmichaud aha, I see why we weren't freeing contexts 16:16
and also why we were losing them. 16:17
oh wait, maybe not. hrm. 16:18
16:20 chromatic joined
dalek r32714 | pmichaud++ | lex2: 16:28
: [GC] Switch Parrot_(push|pop)_context to use refcounting.
diff: www.parrotvm.org/svn/parrot/revision?rev=32714
register i have a question about ticket 38929 16:29
I believe to have understood how to modify the build system so that it is possible to separate the file jit_emit.h in two 16:30
nopaste "bacek" at 211.29.157.151 pasted "Infinite recursion segfault in lex2 for pmichaud" (17 lines) at nopaste.snit.ch/14604 16:31
register one containing the code and the other one consisting only of the declarations
i see that this ticket is assigned on whiteknight
i wonder what is the state of the ticket
i offer to help with this one
is whiteknight online now? 16:32
pmichaud seen wknight 16:33
purl I haven't seen 'wknight', pmichaud. Close matches are: wknight-away [268d, 17:55:16], wknight-work [120d, 20:54:52], wknight8111_ [32d, 21:44:01], wknight8111 [1d, 22:04:36]
16:33 gmansi joined
pmichaud seen wknight8111 16:33
purl wknight8111 was last seen on #parrot 1 days, 22 hours, 4 minutes and 46 seconds ago, saying: I'm more worried about coming up with a clever name :) [Nov 14 18:28:25 2008]
pmichaud oh wait, he's right there
seen Whiteknight
purl Whiteknight was last seen on #parrot 38 minutes and 19 seconds ago, saying: we can always test it for you
bacek pmichaud: ../../parrot perl6.pbc --target=pir --output=Test.pir Test.pm
make: *** [Test.pir] Segmentation fault
it's on r32714 16:34
pmichaud 16:05 <pmichaud> there's not much point in building rakudo until we get those fixed.
bacek pmichaud: ah, ok
pmichaud (the tests in t/op)
register i'm wondering if he has already filtered the source code in the file 16:35
if yes i will take care of modifying just the configure system
else i can take care of the whole issue
i will send hime a line
jonathan morning all
purl morning, jonathan
bacek running 'prove t/op'
pmichaud hello, jonathan 16:36
masak oh hai, jonathan
jonathan is just waiting for kjs to return from nom, then will come to check out
OH NO! lex2 segfaults now? 16:38
masak :/
Tene pmichaud: what should EXPORTALL do if the EXPORT::ALL namespace doesn't exist? Return fialure, exception, etc?
pmichaud it should do nothing. 16:39
Tene Thanks.
pmichaud we could throw an exception for now, though. it shouldn't occur.
and an exception would let us see that.
Infinoid trunk fails to build for me on amd64, miniparrot aborts with an out of memory error 16:41
#3 0x00007f9138743548 in Parrot_new_debug_seg (interp=0x14c3080, cs=0x15ec040, size=140260746985849) at src/packfile.c:2521
that's a big size.
bacek Infinoid: good size :)
pmichaud jonathan: lex2 is having serious problems at the moment. 16:42
jonathan pmichaud: Ah. :-( 16:43
pmichaud: Leaking contexts? 16:44
pmichaud and not marking them as live 16:46
so some are being gc'ed early
jonathan ah, sucky 16:47
jonathan is coming down
pmichaud I have a question about src/pmc/sub.pmc
okay, it can wait.
bacek and some of contexts have loops. So mark_context eat all stack and crashes... 16:49
16:55 elmex joined
nopaste "tene" at 67.152.86.163 pasted "EXPORTALL for pmichaud. There must be an easier way to do this..." (50 lines) at nopaste.snit.ch/14605 16:56
16:57 elmex joined
Tene Ooo... syntax errors... 16:57
16:58 samlh joined 17:01 barney joined
Whiteknight pmichaud, you need me? 17:02
bacek pmichaud: src/sub.c + 563 looks very suspicious to me. 17:03
Infinoid hmm. chromatic's r32694 broke IMCC in trunk on linux/amd64 17:06
where'd he go? :)
17:08 gmansi joined
Whiteknight register, I would love your help with ticket 38929 17:08
if you know how to make the separation happen, I will gladly apply any patches you send
oh, he already left 17:09
Infinoid oh, he mixed int and size_t pointer types, so the top 32 bits of the size variable was uninitialized. 17:15
Infinoid fixes 17:16
17:16 rob joined
Infinoid tests 17:20
pmichaud: if it helps, those coretests started failing in r32695 17:21
dalek r32715 | tene++ | trunk: 17:24
: [HLLCompiler]: Add an EXPORTALL method.
diff: www.parrotvm.org/svn/parrot/revision?rev=32715
17:24 allison joined 17:25 particle joined 17:27 chromatic joined
bacek got answer for refcount question. #60564 :) 17:28
pmichaud Infinoid: yes, it helps a lot. 17:29
Infinoid ok. adding back that refcount++ at the end seems to fix those tests again 17:30
and here comes a fix for trunk!
svn-bisect++ 17:31
dalek r32716 | infinoid++ | trunk: 17:32
: [IMCC] Fix builds on amd64.
: r32694 changed some variables to size_t, but now a pointer to a size_t was
: passed to get_codesize() as an int*. On linux/amd64, int is 32 bit and size_t
: is 64 bit, thus, the high 32 bits were uninitialized and miniparrot crashed
: immediately when trying to allocate a 140TB buffer.
diff: www.parrotvm.org/svn/parrot/revision?rev=32716
r32717 | jonathan++ | trunk:
: [rakudo] First cut of prefix:| for flattening arrays; doesn't work with hashes yet, but should be enough for what mod_parrot needs.
diff: www.parrotvm.org/svn/parrot/revision?rev=32717
particle Infinoid++ # great message 17:35
Infinoid heh. good thing I don't have 200TB of RAM, otherwise I never would have noticed that bug.
(a man can dream) 17:36
pmichaud: should I check in the bit that adds ref_count++ back in? (the coretests are still working when I apply it to -rHEAD) 17:38
I'm not really sure what else it breaks.
17:38 dmknopp joined
Infinoid and Test.pir no longer runs me out of memory, but I'm not sure whether that was r32714 or my local change 17:46
chromatic It's probably a combination. 17:48
Infinoid no, it was r32714. adding the "ctx->ref_count++" back in fixes the parrot coretests but doesn't seem to make any difference when building rakudo (which works now) 17:50
rakudo test passes... I'm trying a spectest now. 17:53
17:55 kj joined
dalek r32718 | kjs++ | trunk: 17:55
: [pirc] add the new :nsentry flag for subs.
diff: www.parrotvm.org/svn/parrot/revision?rev=32718
r32719 | kjs++ | trunk: 18:05
: [pdd19] add description of the new :nsentry subroutine flag.
diff: www.parrotvm.org/svn/parrot/revision?rev=32719
Infinoid spectest memory usage still spirals out of control, and the process memory quota I set didn't help. poor thinkpad's had a rough day. 18:08
dalek r32720 | kjs++ | trunk: 18:13
: [pirc] :method can take an optional string argument.
diff: www.parrotvm.org/svn/parrot/revision?rev=32720
r32721 | tewk++ | nsentry:
: [nsentry] branch creation
diff: www.parrotvm.org/svn/parrot/revision?rev=32721
r32722 | bernhard++ | trunk: 18:30
: [perl] Normalize the path returned by Parrot::Test::path_to_parrot()
diff: www.parrotvm.org/svn/parrot/revision?rev=32722
r32723 | bernhard++ | trunk: 18:45
: [Pipp] Simplify Makefile by setting @INC in t/harness
diff: www.parrotvm.org/svn/parrot/revision?rev=32723
r32724 | kjs++ | trunk: 18:51
: [pirc] change :lexid into :subid.
diff: www.parrotvm.org/svn/parrot/revision?rev=32724
r32725 | bernhard++ | trunk: 19:06
: [Pipp] fixing unified languages testing
: PHP and PIR test files are not submitted
diff: www.parrotvm.org/svn/parrot/revision?rev=32725
r32726 | kjs++ | trunk:
: [pdd19] :lexid will be changed into :subid.
diff: www.parrotvm.org/svn/parrot/revision?rev=32726
r32727 | kjs++ | trunk: 19:07
: [DEPRECATED] deprecate :lexid in favor of :subid
diff: www.parrotvm.org/svn/parrot/revision?rev=32727
19:08 PacoLinux joined 19:17 Limbic_Region joined
dalek r32728 | kjs++ | trunk: 19:29
: [NEWS] add 'box' opcode and ':subid/:lexid' change
diff: www.parrotvm.org/svn/parrot/revision?rev=32728
19:56 davidfetter joined 19:58 masak joined 20:02 gmansi joined 20:17 rdice joined
dalek r32729 | tewk++ | nsentry: 20:25
: [nsentry] Implementation. :method and :vtable still add sub to namespace
diff: www.parrotvm.org/svn/parrot/revision?rev=32729
r32730 | bernhard++ | trunk: 20:31
: [Pipp] Allow running tests from 'languages/pipp'.
diff: www.parrotvm.org/svn/parrot/revision?rev=32730
r32731 | bernhard++ | trunk: 20:38
: [Pipp] Remove trailing whitspace
diff: www.parrotvm.org/svn/parrot/revision?rev=32731
r32732 | chromatic++ | trunk: 20:40
: [JIT] Switched JITted NCI functions to storage in ManagedStruct PMCs, which
: will free the memory allocated for the JIT function when the PMC gets
: destroyed.
diff: www.parrotvm.org/svn/parrot/revision?rev=32732
r32733 | jonathan++ | trunk: 21:09
: [rakudo] Fix initialization of attributes of array and hash types (@ and % sigils). Resolves RT#60572.
diff: www.parrotvm.org/svn/parrot/revision?rev=32733
bacek rakudo: say eval "my $x" 21:10
polyglotbot OUTPUT[No handler to delete.␤current instr.: 'parrot;PAST;Compiler;as_post' pc 6052 (src/PAST/Compiler.pir:1651)␤called from Sub 'parrot;PAST;Compiler;post_children' pc 1779 (src/PAST/Compiler.pir:367)␤called from Sub 'parrot;PAST;Compiler;pirop' pc 3048 (src/PAST/Compiler.pir:791)␤called from Sub
..'parrot;PAST;Compiler;post_children' pc 1779 (sr...
bacek jonathan: can you check this one?
rakudo: say eval "my $x = 1"
polyglotbot OUTPUT[No handler to delete.␤current instr.: 'parrot;PAST;Compiler;as_post' pc 6052 (src/PAST/Compiler.pir:1651)␤called from Sub 'parrot;PAST;Compiler;post_children' pc 1779 (src/PAST/Compiler.pir:367)␤called from Sub 'parrot;PAST;Compiler;pirop' pc 3048 (src/PAST/Compiler.pir:791)␤called from Sub
..'parrot;PAST;Compiler;post_children' pc 1779 (sr...
21:20 lu_zero joined
lu_zero hi 21:20
masak lu_zero: y0 21:27
21:36 lu_zero joined 21:46 rob joined 22:08 Andy joined 22:16 kj joined 22:22 kj joined
jonathan bacek: I don't get that locally... 22:26
22:32 Theory joined 22:49 Limbic_Region joined
pmichaud I think that polyglotbot is not a recent copy. 22:59
23:04 Psyche^ joined
masak rakudo: say $/ 23:04
polyglotbot OUTPUT[␤]
masak why doesn't using $/ before it has been defined produce a warning?
shouldn't it?
(I realize I'm asking for more 'use of undefined value' warnings here. Just curious why $/ would be an exception.) 23:05
nopaste "chromatic" at 216.239.45.19 pasted "kjs: Parrot::Test Patch for pirc" (59 lines) at nopaste.snit.ch/14608 23:11
23:12 kj joined
chromatic kj: nopaste.snit.ch/14608 23:12
masak um, never mind. ordinary variables don't give a warning on that either. 23:15
but shouldn't they?
23:26 s1n joined
dalek r32734 | chromatic++ | trunk: 23:27
: [lib] Added pirc_2_pasm_* functions to Parrot::Test to improve testing of pirc.
diff: www.parrotvm.org/svn/parrot/revision?rev=32734
r32735 | kjs++ | trunk: 23:39
: [pirc] add a first test file.
: + fix output of pirc a bit to make it testable.
: + update MANIFEST
diff: www.parrotvm.org/svn/parrot/revision?rev=32735
lu_zero question: where is a tutorial considering parrot installed in the system? 23:40
(maybe stupid question)
chromatic I'm not sure I understand the question. 23:41
lu_zero chromatic I'm updating the parrot ebuild in gentoo
and I'm trying to install it so it won't spew random directories in /usr 23:42
dalek r32736 | kjs++ | trunk:
: [pirc] fix properties of test file and add coda and copyright statement.
diff: www.parrotvm.org/svn/parrot/revision?rev=32736
lu_zero I sort of managed to get it done but now beside feeding parrot with hello world I didn't do much to make sure it works correctly 23:43
chromatic reini knows more about that. 23:45
lu_zero how to contact him?
chromatic I'm looking at the draft PDD right now. 23:47
Although sending an email to parrot-porters@perl.org is a good step too.
dalek r32737 | tewk++ | trunk:
: [tools] nci_test_gen.pl nci tests generator
diff: www.parrotvm.org/svn/parrot/revision?rev=32737
chromatic lu_zero, see /docs/pdds/draft/pdd30_install.pod for the current thouhts.
thoughts
purl "I'm never here, I'm never here, I'm never here / never a bird, or a flower in the tree, or the pain of the respect thereof..."
dalek r32738 | chromatic++ | trunk: 23:51
: [pirc] Added t/harness to run pirc tests; added code to run the harness to the
: Makefile.
diff: www.parrotvm.org/svn/parrot/revision?rev=32738
lu_zero it's quite enlightening =) 23:53
Tene reminds particle to write something about ticket dependancy graphs eventually. 23:56

Powered by Cro and the Raku® Programming Language.

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

Thank you!