Parrot 0.8.2 Feliz Loro Released | parrot.org | 45 TT | 538 RT
Set by moderator on 17 December 2008.
kid51 Alright, I see that if you click on Original Format, you get something that you can do a "Save As..." on. 00:04
sort of
But that's all a GUI operation. What I really want to do is to fetch the patch to a directory on a server which I do not access via a GUI. 00:06
So I first have to save the patch to my laptop, then scp it to my server.
GeJ there's a plug-in to add XML-RPC capabilities to Trac. From the project's page, it looks like you can download attachments from wiki pages. 00:08
I have no idea if that also applies to tickets though.
trac-hacks.org/wiki/XmlRpcPlugin 00:09
00:09 AndyA joined
GeJ Just telling as FYI. I didn't use it myself, so I can't recommend it. 00:10
kid51 I frankly have no idea why we moved from one ticket system to another. It would have been nice to have all the functionality we had in RT. 00:11
00:14 Whiteknight joined, TonyC joined
kid51 Switching to a new ticket tracking system always seems like a great idea to those promoting the switch -- but it imposes a learning cost on everyone else. 00:26
pmichaud kid51: the switch was because of a desire to move tickets off of the perl.org server 00:27
at which point it was a question of trying to set up rt on the new server, or using a different ticking system
(this is my understanding of things, I was not at all involved in the decision or in influencing the outcome)
00:28 Zaba_ joined
kid51 Well, I was never clear why we moved off perl.org. Was it inadequate server performance? Was it the rush to make everything Parrot Foundation rather than Perl Foundation? 00:29
And I don't know why we didn't replicate RT on parrot.org. 00:30
pmichaud I don't have any insight into those decisions. But there are a lot of ways in which RT is _very_ frustrating to use. 00:31
kid51 I can't spend all day on #parrot, which is what I suppose I would have to do to participate in those decisions -- or even find out their rationale.
pmichaud: Perhaps I had adapted enough to RT that what I needed to do with it wasn't frustrating to me. 00:32
We use Bugzilla at $job 00:33
But I gather that Trac is the Shiny New Thing in ticketing systems. 00:34
pmichaud I've been doing rt with parrot and perl6 for going on four years now and I still find pieces that are frustrating. I'm sure I'll find frustrations with Trac, too, but so far I kinda like what I've seen. Its reporting facilities are _very_ nice.
00:37 davidfetter joined
dalek r34163 | Whiteknight++ | trunk/src/jit/amd64 (2 files): 00:37
: [JIT] remove executable code from src/jit/amd64/jit_emit.h
link: svn.perl.org/viewsvn/parrot/?hideat...63#dirlist
shorten dalek's url is at xrl.us/t3fwn
kid51 Something is wrong with this directory in our repository: t/compilers/imcc/syn/ 00:55
I'm trying to do 'svn blame' on one of the files in this directory, but what I get is: 00:56
$ svn blame t/compilers/imcc/syn/macro.t
Skipping binary file: 't/compilers/imcc/syn/macro.t'
Why does SVN think this is a binary file?
pmichaud probably because of an incorrectly set property 00:57
just a sec, I get the same error.
kid51 This is on a fresh svn co
svn pl t/compilers/imcc/syn/macro.t
Properties on 't/compilers/imcc/syn/macro.t':
cvs2svn:cvs-rev
svn:keyword 00:58
svn:mime-type
Copyright
purl hmmm... Copyright is at www.benedict.com/contents.htm or strong in the US. And getting stronger. Copyright owner now holds for life plus 200 years?
kid51 svn:keywords
purl i heard svn:keywords was all one property, so doing it twice would overwrite the setting. You have to set it to 'Rev Id'
kid51 svn:eol-style
00:58 apple-gunkies joined
pmichaud I don't know why svn thinks it's binary. 00:59
nopaste "kid51" at 70.107.17.133 pasted "Comparison of 'svn pl -v' on two files" (16 lines) at nopaste.snit.ch/15050 01:02
pmichaud I'm suspicious of that "svn : pl" line and the "svn:mime-type" 01:03
kid51 Some but not all of the other files in the same directory have this problem. Let's examien.
examine.
01:04 Ademan joined
kid51 For example: svn blame t/compilers/imcc/syn/bsr.t 01:05
Skipping binary file: 't/compilers/imcc/syn/bsr.t'
That "svn : pl" was something local. But even clearing that up, I get the "Skipping binary file" message 01:12
01:13 Zaba joined
kid51 But other files do 'svn blame' perfectly: svn blame t/compilers/imcc/syn/regressions.t 01:31
pmichaud according to subversion.tigris.org/faq.html#binary-files it's the svn:mime-type property 01:32
checking.
try removing that property and see if it fixes it. 01:33
kid51 But I recall that at one point we set all .t files to 'svn:mime-type text/plain' -- which is, of course, what they are -- to get them to show up properly as attachments.
pmichaud: Will try.
pmichaud maybe even though it appears to us that it's "text/plain" perhaps there are some spaces or characters making svn think otherwise.
kid51 svn pl -v t/compilers/imcc/syn/regressions.t 01:34
Properties on 't/compilers/imcc/syn/regressions.t':
svn:mime-type : text/plain
svn:keywords : Author Date Id Revision
svn:eol-style : native
purl well, text/plain is the bane of satan or a lynx users choice
kid51 regressions.t does 'svn blame' correctly.
It too, is svn:mime-type: text/plain 01:35
dalek r34164 | jkeenan++ | trunk/t/compilers/imcc/syn: 01:47
: Deleting svn:mime-type property in attempt to see whether this fixes the 'Skipping binary file' problem when you try to 'svn blame' this file.
link: svn.perl.org/viewsvn/parrot/?hideat...64#dirlist
shorten dalek's url is at xrl.us/uck7o
kid51 No, that didn't work. 01:48
dalek r34165 | jkeenan++ | trunk/t/compilers/imcc/syn: 01:49
: Restoring previous properties, as deletion of 'svn:mime-type' did not clear up
: the 'Skipping binary file' problem.
link: svn.perl.org/viewsvn/parrot/?hideat...65#dirlist
shorten dalek's url is at xrl.us/ucuj8
Infinoid the first rev of bsr.t had a mime-type of application/x-troff 02:26
so it may be complaining about the old mime-type, not the current one.
that said, svn blame does have a --force option. 02:27
kid51 See trac.parrot.org/parrot/ticket/73 02:31
dalek r34166 | Whiteknight++ | branches: 02:32
: Creating temporary branch to try and deal with #RT#38929 for the i386 platform
link: svn.perl.org/viewsvn/parrot/?hideat...66#dirlist
shorten dalek's url is at xrl.us/uhoiq
Infinoid it looks like the original mime-type was cvs2svn's fault 02:33
kid51 Infinoid: You are correct.
Infinoid its definitely a valid issue, but I think its going to require some intervention from the perl.org admins (dump, edit and re-import the database), or else lose those files' histories 02:34
kid51 $ svn blame --force t/compilers/imcc/syn/macro.t | head -5 10826 ambs #!perl
26163 bernhard # Copyright (C) 2001-2008, The Perl Foundation.
7594 bernhard # $Id$
7594 bernhard
3886 leo use strict;
svn: Can't write to stream: Broken pipe
Infinoid or else fix svn to not do such silly things. :)
kid51 So --force works 02:35
Yeah, it's probably more of an SVN bug.
dalek r34167 | pmichaud++ | trunk/languages/perl6/src/pmc:
: [rakudo]: Add radix support for string-to-number conversions (RT #59222, s1n++)
: * Based on a patch courtesy s1n <jswitzer@gmail.com>
link: svn.perl.org/viewsvn/parrot/?hideat...67#dirlist
shorten dalek's url is at xrl.us/uh2fg
Infinoid actually, I think svn's behavior makes sense... if a file was originally binary and then became text at some later point, and "blame" worked, I'd get a screwed up terminal full of binary jibberish 02:36
kid51 Apart from what may have happened when it was imported from CVS, I can't see any pattern of our committers doing something wrong.
Infinoid err. wait, no I wouldn't, nevermind
kid51 Infinoid: Could you post comments in that Trac ticket I just opened? Thx.
Infinoid ok :)
for that matter, troff isn't a binary format anyway (I think). so getting it added to their whitelist might be an easy win 02:40
dalek r34168 | jkeenan++ | trunk/config/init/hints: 02:41
: Deleting AIO-related code, per discussion in rt.perl.org/rt3/Ticket/Display.html?id=57920.
link: svn.perl.org/viewsvn/parrot/?hideat...68#dirlist
shorten dalek's url is at xrl.us/uimea
kid51 When did dalek start displaying these viewsvn URLs? When did the parrotvm.org links go away? 02:43
Infinoid I switched the RSS feed it uses earlier today, to get the branch names displaying correctly again. 02:44
and I was just now in the process of switching the diff links back
02:45 dalek joined
Infinoid there. now we wait for a commit to see if it worked. 02:45
kid51 The parrotvm.org links were often slow to display ... but they were diffs -- which I think are more useful immediately after someone commits.
Infinoid that's true
you can click a couple of times to get a diff from those links, but parrotvm is less work
kid51 The viewvc is more of a snapshot of the last-revision status of a given directory in the repository.
Infinoid yeah, I agree. the links in perl.org's RSS aren't as useful as I expected 02:46
anyway, it should be switched back now.
Plus, this should shut shorten up.
kid51 is tired from dealing with our infrastructural problems. 02:47
kid51 thinks of food and beer
Infinoid I'm just trying to help
food and beer can be helpful, too :)
kid51 should probably think of Xmas shopping as well 02:48
dalek r34169 | Whiteknight++ | branches/jit_h_files/src/jit/i386 (2 files): 02:50
: [jit_h_files] first step in the i386 fix, moving function definitions and global vars to the .c file
diff: www.parrotvm.org/svn/parrot/revision?rev=34169
Infinoid sees if that link worked 02:51
o...kay. apparently that doesn't work so well on branches 02:52
kid51 I got "Safari can�t open the page �www.parrotvm.org/svn/parrot/revision?rev=34169� because the server unexpectedly dropped the connection, which sometimes occurs when the server is busy."
... which I *often* get with parrotvm.org. 02:53
... which I believe is feather
Infinoid yes, it is
kid51 I wonder: Is the problem that it's running a script to display that SVN diff? And only one person at a time can view it?
Infinoid a race condition? I don't think so 02:54
I get "Missing @@ line before @@ -1 +1,2722 @@ at line 5 of string"
...but 34168 works fine
kid51 Infinoid: Yes, that's the *first* page I saw. When I refreshed, I got the dropped-connection message.
Infinoid beautiful.
purl rumour has it beautiful is a love thing
Infinoid I'll see if I can get a direct diff link out of svn.perl.org then 02:55
kid51 feather?
purl feather is feather.perl6.nl/ or a perl6 community development server
dalek r34170 | Whiteknight++ | branches/jit_h_files/src/jit/i386: 02:56
: [jit_h_files] some random cleanups to the .h file, removing unneeded macros and fixing weird whitespace
diff: www.parrotvm.org/svn/parrot/revision?rev=34170
Infinoid is svn.perl.org/viewvc/parrot?view=rev...sion=34169 a decent place to link to? it doesn't give you a diff, but it gives you a list of files with diff links ("changed text") for each file. 02:58
shorten Infinoid's url is at xrl.us/ukyhd
Infinoid I suppose I'll have to update all of this yet again when svn moves to trac 02:59
03:02 PacoLinux joined
dalek r34171 | infinoid++ | trunk/languages/perl6/src/pmc: 03:12
: [cage] Fix up some t/codingstd/check_toxxx() failures.
diff: www.parrotvm.org/svn/parrot/revision?rev=34171
03:13 dalek joined
dalek r34172 | infinoid++ | trunk/languages/perl6/src/pmc: 03:19
: [cage] Fix up some t/codingstd/check_isxxx.t failures.
: Reformat some if/else chains to make it a little more readable.
review: xrl.us/uoe6u
Infinoid brain not working, need food. afk
03:49 rurban_ joined
Whiteknight I thought GC was the devil, but JIT is THE DEVIL 03:57
dalek r34173 | Whiteknight++ | branches/jit_h_files/src/jit/i386 (2 files): 03:58
: [jit_h_files] move function prototypes to the .h files and fix them up all pretty
review: xrl.us/utkbn
Whiteknight Infinoid++
04:01 elmex_ joined
dalek r34174 | Whiteknight++ | branches/jit_h_files/src/jit/i386: 04:08
: [jit_h_files] move function definitions lower in the file to prevent conflicts with some macros
review: xrl.us/uushg
kid51 Whiteknight: Do you know anything about this file? cat src/jit_defs.c 04:09
Contents: /* Stub file for RT#38929 fixes */
Whiteknight yes, I've added it. It's a stub now but will eventually not be 04:10
kid51 I have it showing as not under version control. Is that correct?
Whiteknight It's generated from the files in src/jit/ARCHITECTURE/jit_defs.c
it's not under version control itself, it's generated during the build
kid51 If it's generated, shouldn't it be under svn:ignore (or something like that)? 04:11
dalek r34175 | Whiteknight++ | branches/jit_h_files/src/jit/i386: 04:12
: [jit_h_files] add a missing header file include
review: xrl.us/uu3ds
Whiteknight I don't even know what svn:ignore is. How do I make that happen? 04:13
dalek r34176 | Whiteknight++ | branches/jit_h_files/src/jit/i386 (2 files): 04:19
: [jit_h_files] fix multiple function declaration error
review: xrl.us/uvgbe
r34177 | jkeenan++ | trunk/t/perl: 04:21
: Applying patch submitted by Ron Schmidt in
: rt.perl.org/rt3/Ticket/Display.html?id=57320. All tests continue to
: pass. Let's see if this resolves temp directory problem.
review: xrl.us/uvjkp
04:21 Zaba joined
dalek r34178 | Whiteknight++ | branches/jit_h_files/src/jit/i386: 04:24
: [jit_h_files] removing some #undef commands are are messing with my jauns
review: xrl.us/uvqs8
r34179 | chromatic++ | trunk/languages/perl6/src/pmc: 04:28
: [Rakudo] Turned a C++-style comment into a C-style comment so as not to confuse
: old, broken, vendor-supplied compilers stuck in the twentieth century.
review: xrl.us/uvzio
r34180 | Whiteknight++ | branches/jit_h_files/src/jit/i386 (2 files): 04:35
: [jit_h_files] fixing up some macros which cause global conflicts by renaming them to JIT_*
review: xrl.us/uwgi7
r34181 | Whiteknight++ | branches/jit_h_files/src/jit/i386 (2 files): 04:40
: [jit_h_files]
review: xrl.us/uwzxi
r34182 | jkeenan++ | trunk: 04:54
: Add 'jit_defs.c' to list of files targeted for 'svn:ignore' in directory src/.
review: xrl.us/uyu9d
04:54 Andy joined
kid51 must sleep 05:04
purl $kid51->sleep(8 * 3600);
Andy evening 05:05
dalek r34183 | petdance++ | trunk/lib/Parrot/Pmc2c: 05:07
: Cleaning up quotes, removed unnecessary imports, updated copyright
review: xrl.us/u2sw5
r34184 | pmichaud++ | trunk/languages/perl6/src/parser: 05:18
: [rakudo]: Fix initialization of variables in signatures (RT #61300, cspencer++)
: * Patch courtesy cspencer <cspencer@sprocket.org>
review: xrl.us/u3796
r34185 | petdance++ | trunk/lib/Parrot/Pmc2c: 05:22
: simplified a regex
review: xrl.us/u4hfn
06:01 tetragon joined 06:16 nute joined
dalek r34186 | pmichaud++ | trunk/languages/perl6/docs: 06:22
: [rakudo]: spectest-progress.csv update: 261 files, 5645 passing, 0 failing
: Yes. Over 420 new passing tests in one day.
review: xrl.us/vb5p6
r34187 | petdance++ | trunk (9 files): 06:26
: explicitly include File::Spec where it is used, rather than assuming another module will pull it in
review: xrl.us/vcokd
TiMBuS hmm.. is call_list.txt some kind of temporary hack for nci or is it a permanent thing? 06:29
chromatic It's useful for platforms without a JIT that can build NCI thunks.
TiMBuS so is it not used on platforms with a jit 06:31
chromatic More or less right, yes. 06:32
There's a configuration setting to use JIT to build call frames, but you get the idea.
TiMBuS ah, well im trying to let the language im implementing import c functions, which i thought i could just pass down to loadlibrary/dlfunc, but parrot does not like that 06:36
maybe i should think something up in the form of a pmc 06:37
chromatic You may need to add a signature to call_list.txt. 06:38
06:51 pdcawley joined
GeJ pmichaud: is the 420-ish bump due to a specific feature implemented or the sum of several ones? 06:53
szabgab chromatic: have you seen the Parrot::Interpreter related issue I sent to the mailing list? 07:09
chromatic I saw it, but haven't had time to look at it. 07:10
szabgab shall I create similar test scripts for some of the other languages?
chromatic Can you reproduce it with other languages?
szabgab Which should be the most important ones?
chromatic Pheme is easiest to debug. 07:11
szabgab I have only tried with Perl6 and there I can load the pbc file
chromatic Rakudo, Pynie, Cardinal, and Lua are probably the most significant.
szabgab but I cannot find the main becasue of the nested namespaces
I think,
at least that's what we discussed back a month or two ago 07:12
so I'll try to write more test scripts, shall I send them to the mailing list or add them to trac?
chromatic Trac's best. 07:13
szabgab ok,
and "make languages" is supposed to build all the languages, right? 07:14
chromatic In theory, but it may have suffered bitrot.
szabgab I'll see if it builds the ones I test 07:16
07:41 Khisanth joined 07:51 Zaba joined 07:52 ask_ joined
szabgab trac.parrot.org does not let me create tickets. I registered as szabgab but when I try to create a ticket I get 07:56
TICKET_CREATE privileges are required to perform this operation
on the padre trac we allow any registered and logged in user to create a ticket, would that be possible here too? 07:57
besides I keep getting the following warning:
Warning: <acct_mgr.web_ui.MessageWrapper object at 0x810d8f0c>
ok, I take back the first one, I can create a ticket after I verifyed my e-mail addres 07:59
s
08:18 Theory joined 08:35 clunker3 joined 09:03 pdcawley joined 09:10 iblechbot joined 09:30 barney joined
dalek bernhard.schmalhofer@gmx.de | Pipp: 09:56
link: www.perlfoundation.org/parrot/index.cgi?pipp
barney www.parrotcode.org/docs/submissions.html has no content 10:01
szabgab barney: you might be familiar with pipp, right ? 10:03
I am trying to use it with Parrot::Interpreter
I can load the pipp.pbc but I cannot find the main code, $interp->find_global( 'main', 'pipp' ); 10:04
returns undef
dalek r34188 | bernhard++ | trunk (5 files): 10:07
: [Pipp] Add support for 'elsif'
: Courtesy of Daniel Keane.
review: xrl.us/v64xt
barney Do you mean the sub 'pipp' in languages/pipp/src/common/pipp.pir ? 10:11
szabgab I am actually not sure, I think I shuld find the "main" of Pipp 10:16
$interp->find_global( 'main', 'Pipp' ); also returns undef
$interp->find_global( 'Pipp' ); returns a Parrot::PMC but I don't seem to be able to call ->invoke on it 10:17
barney szabgab: The sub 'pipp' is the ':main' of pipp.pbc. But I can rename that to 'main', in order to be more in line with other languages. 10:18
szabgab let me try to find that name first 10:19
10:19 [Tux] joined
barney $interp->find_global( 'Pipp' ) might be a namespace 10:19
[Tux] seen pmichaud
purl pmichaud was last seen on #parrot 8 hours, 45 minutes and 49 seconds ago, saying: maybe even though it appears to us that it's "text/plain" perhaps there are some spaces or characters making svn think otherwise.
szabgab ok, I can find the pipp sub 10:20
dalek r34189 | fperrad++ | trunk/languages/lua/src (2 files):
: [Lua]
: - now, use :opt_flag
review: xrl.us/v749f
szabgab ?paste 10:22
barney szabgab: Today I'll start on removing the other Pipp variants. concentrating on using PCT.
szabgab that does not mean to me much yet :-)
10:23 [Tux] left
szabgab sial.org/pbot/33969 10:24
that includes the code I am trying to execute and the error message I get when callin invoke
maybe I have to give a filename as a param? 10:26
can I pass a string containing PHP code to the pipp sub ? 10:27
barney Yes. It expects a filename
szabgab so is there another sub that already gets the content of the file? 10:28
barney Not yet. pipp_compiler.'command_line'( ) is called later. This also takes a filename, but might have support for strings. 10:32
dalek r34190 | fperrad++ | trunk (6 files): 10:35
: [Lua] bit
: - an another bitwise operation library
review: xrl.us/v9o5u
szabgab I can't seem to get it working even if I pass a path to a .php file to it 10:36
my $pmc = $pipp->invoke( 'S', '/home/gabor/work/parrot/languages/pipp/a.php' ); tells me
too few arguments passed (0) - 1 params expected
and if I call it with 'PS' sugnature then I get the shift_string() not implemented in class 'String' 10:37
barney try f=/home/gabor/work/parrot/languages/pipp/a.php 10:38
or '-f' '/home/gabor/work/parrot/languages/pipp/a.php'
szabgab same error 10:40
(the latter could not work at all so I tried '-f /home/gabor/work/parrot/languages/pipp/a.php'
10:43 alvar joined
barney the sub pipp expects a ResizableStringArray() containing the script and the command line params 10:47
szabgab that would be \\@xx in Perl 5 I guess 10:48
barney I'll add a sub that simply accepts a string with code
\\@ARGV 10:49
But first I'll do some simplifications. 10:50
Up to now Pipp has support for alternative parsers, like PHC and Antlr3. I'll remove that and contrateon PCT 10:51
szabgab that's great, I'd really like to be able to write functions in PHP 10:52
and then call them from within Perl5 using Parrot::Interpreter
Basically Inline::PHP on Pipp
Shall I open a ticket for this on trac? 10:53
barney Go ahead.
I have an issue with with variable scopes in Pipp. I plan to make all, besides the superglobals, lexical. 10:55
This should allow to write closures in PHP, and using them in Perl-land 10:56
szabgab so for now I can add a Document::PHP that will allow execution of external PHP script from Padre 10:58
but will have to wait with the capability to write plugins for Padre in PHP
barney should check out Padre 10:59
szabgab btw I could not invoke it with \\@ARGV either
11:00 pdcawley joined
dalek bernhard.schmalhofer@gmx.de | Pipp: 11:08
link: www.perlfoundation.org/parrot/index.cgi?pipp
szabgab barney: array() is not supported yet? 11:09
barney No array() yet 11:10
dalek r34191 | bernhard++ | trunk (5 files): 11:17
: [Pipp] Remove support for the Antlr3 variant.
review: xrl.us/wfknx
r34192 | bernhard++ | trunk (8 files): 11:41
: [Pipp] Remove support for antlr3 variant.
review: xrl.us/whxoh
11:49 rurban_ joined
dalek r34193 | bernhard++ | trunk (5 files): 11:50
: Remove no longer needed test modules.
review: xrl.us/wiwzz
12:05 pdcawley joined
dalek r34194 | bernhard++ | trunk/languages/pipp: 12:07
: [Pipp] Update ROADMAP.
review: xrl.us/wkpk6
12:10 cognominal joined 12:11 bacek joined 12:26 Zaba joined
szabgab barney: trac.parrot.org/parrot/ticket/76 12:35
dalek r34195 | bernhard++ | trunk (5 files): 12:52
: Add dummy for 'eval_string'.
review: xrl.us/wpm4u
barney szabgab: Started on that. The subname 'eval_string' might change. 12:54
12:54 masak joined
szabgab barney: I am trying the same things now for cardinal and pynie and submit test scripts for each one 12:56
oh and of course rakudo
barney szabgab++
dalek r34196 | bernhard++ | trunk (12 files): 12:59
: [codingstd] set svn props
review: xrl.us/wqdbn
13:13 masak joined
dalek r34197 | bernhard++ | trunk (3 files): 13:17
: [codingstd] trailing space, C++ comments, CPP indention
review: xrl.us/wse5v
13:18 Hadi joined, Hadi left 13:47 Whiteknight joined 13:57 masak joined
dalek r34198 | Whiteknight++ | branches/jit_h_files/src/jit/i386 (2 files): 14:04
: [jit_h_files] remove reliance on Parrot_end_jit since it's a static function in another file
review: xrl.us/wxmca
14:05 Theory joined 14:10 tetragon joined
dalek r34199 | bernhard++ | trunk (5 files): 14:11
: [Pipp] Rename 'eval_string' to 'eval' and move the function into eval.pir.
review: xrl.us/wyfaf
14:12 tetragon joined 14:13 johbar joined 14:14 nute joined 14:17 kid51 joined 14:27 tetragon_ joined
masak 5645 spectests! O_o 14:32
we'll hit 6000 before we know it.
Whiteknight rakudo++ 14:33
masak pmichaud++ jonathan++ 14:37
14:39 workbench joined
dalek r34200 | jkeenan++ | branches: 14:40
: Creating smoke2smolder in svn.perl.org/parrot//branches
review: xrl.us/w2sd9
r34201 | jkeenan++ | tags:
: Tagging trunk at r34199 so that the smoke2smolder can later be synched to it.
review: xrl.us/w2sfg
kid51 Creating a branch to see if I can get Coke's patch to work.
Whiteknight nice 14:41
14:42 Infinoid joined
Coke kid51: ping 14:50
dalek r34202 | Whiteknight++ | trunk/config/gen/platform/win32: 14:52
: [win32] Take a stab at resolving RT#48276, + a few small upgrades
review: xrl.us/w34ap
15:12 elmex joined
Coke kid51: ping 15:16
(I can just apply my work in progress to your branch for smoke2smolder, if you like.) 15:17
Coke saves that checkout in case kid51 wants that. 15:22
kid51 Hi, Coke.
I'm just now running 'make test' in that branch.
Have you done more patching than what you posted yesterday? 15:23
Coke no; just figured that if you couldn't apply it cleanly, I could. 15:32
but if you've got it working, you're where I was.
15:36 Hadi1 joined 15:50 contingencyplan joined
Coke pmichaud: you still suffering from the infinite "class registered" exception loop? 15:52
I think it's blocking the two regex spec tests in partcl.
15:56 Zaba joined
kid51 Coke: I should have this merged in about half an hour. 15:59
dalek r34203 | jkeenan++ | branches (13 files): 16:05
: Implement Coke's patch in branch.
review: xrl.us/xcdo7
pmichaud Coke: yes, the class registered exception loop still exists 16:06
but I suspect it's true for any exception thrown from C, not just that one.
16:10 Theory joined
Coke kid51: ok. 16:11
pmichaud: hurm. if I go through the trouble of breaking down the problem a bit, it looks like tclsh is misparsing the code. 16:12
the command in question is:
regexp ($|^X)* {}
that should result in a call to &regexp with two parameters, the raw strings there.
but if I compile that to PIR and emit the PIR (kind of a PITA for tcl atm, but doable), I get: 16:13
nopaste "coke" at 72.228.52.192 pasted "why is infix:concat getting invoked?" (23 lines) at nopaste.snit.ch/15051
Coke it seems to endup the same thing, just confused why concat is getting invoked at all. 16:14
once I try to compile the rule ($|^X)* using the P5 regexp compiler, that's when it goes off into the weeds. 16:15
what's the RT for the class registered exception loop?
pmichaud looking 16:16
16:17 iblechbot joined
pmichaud what is generating the code in your nopaste? 16:17
that doesn't look like anything PGE generates.
RT #59778 16:19
Coke that's from the tcl compiler. 16:21
old school pge/tge
pmichaud okay, it must be tge that is generating that then.
Coke pmichaud: that code works in parrot current. 16:24
(at least, something very similar to it.)
nope. direct cut and paste works. 16:25
pmichaud I'm a bit confused, overall, because I don't see why ($|^X)* is two separate strings. 16:26
Coke me either.
pmichaud is it supposed to be one regex?
Coke red herring, though, as the failure occurs inside the call to &regexp.
let's back up one more step: the RT you pointed me at seems to be closable.
pmichaud oh? 16:27
Coke run that code. it works. =-)
pmichaud works as in "runs to completion"?
Coke yes.
pmichaud or works as in "runs and doesn't show the long backtrace when you stop it in the middle?"
Coke spits out the debug information, properly gives only one warning...
pmichaud okay, checking.
Coke I didn't step through with gdb, because the way I've seen this present before is that after foo(1), the call to foo(2) would just hang. 16:28
pmichaud no, the calls never hang for me.
Coke (generating errors.) 16:29
pmichaud: then how do you know the exceptions aren't handled?
pmichaud they are handled -- that's not the issue.
from a parrot perspective, everything appears fine
Coke ok. then your error isn't my error.
pmichaud but internally, we end up with a long stack of exception handlers, because the handlers never "clear"
Coke because when i see this, the process hangs. if I happen to be running gdb, then I can see the enormous backtrace. 16:30
pmichaud if I call foo(...) 1000 times, then all of them will run, but I have 999 exception handlers in the C calling stack
Coke maybe it is the same, and I'm just 15m into a running parrot process so it's more noticable. 16:31
pmichaud anyway, if it's a PGE issue, I'd probably need to know what string is being passed to P5Regexp 16:32
Coke should be "($|^X)*"
but that works fine on its own. 16:33
pmichaud that could end up in an infinite loop, though.
probably does.
Coke (having already tried to narrow things down.) Though It may be that it fails the same way yours does and didn't notice.
(I didn't)
Oh! 16:34
shouldn't your code have end: before pop_eh ?
pmichaud yes, but that doesn't fix it.
I should write an update to that ticket (which was filed before we did the pop_eh change)
Coke rant: gdb on mac osx doesn't default to paging. 16:35
so 'bt' inside screen sucks. 16:36
pmichaud ($|^X)* is likely to run into RT #37745 ("handle zero-length repeated captures")
Coke gives up and focuses his efforts elsewhere in the tcl suite. 16:40
dalek r34204 | jkeenan++ | (13 files):
: Merging smoke2smolder branch into trunk. Implement's Coke's patch
: deactivating old-style smoke testing in favor of Smolder. Cf.:
: trac.parrot.org/parrot/ticket/69 and
: rt.perl.org/rt3/Ticket/Display.html?id=49276.
review: xrl.us/xezgx
r34205 | jkeenan++ | branches: 16:41
: Branch has been merged into trunk; no longer needed at HEAD.
review: xrl.us/xe3m4
r34206 | jkeenan++ | tags: 16:42
: Branch to which tag corresponded has been merged into trunk; deleting tag.
review: xrl.us/xe44e
Coke kid51: svn 1.5 should obviate the need for the tag marking, fyi.
kid51: thanks for the merge. (Coke should have led with that, sorry) 16:43
pmichaud: wrote down the rt #'s code.google.com/p/partcl/wiki/ParrotIssues so I wouldn't lose them. thanks. 16:44
dalek r34207 | jkeenan++ | branches:
: No further need for this branch, as smoke has been superseded by smolder.
review: xrl.us/xe9va
r34208 | jkeenan++ | tags: 16:45
: No further need for this tag, as smoke has been superseded by smolder.
review: xrl.us/xfbpv
kid51 Coke: re svn1.5: I haven't studied that at all, but the tag is a holdover from a Perl script I wrote when we first started doing branches. 16:46
Coke: Since I never have attempted to file a Smolder report on any languages, you may want to double check that area. 16:47
nopaste "pmichaud" at 72.181.176.220 pasted "+/-Inf works (for cspencer)" (12 lines) at nopaste.snit.ch/15055 16:49
16:49 AndyA joined
Coke kid51: I'm ripped out languages-smoke in my patch; the goal for 1.0 is to have the languages out of our repository; so it's more of a nice to have than a must have, I think. 16:54
s/I'm/I'd/
kid51++ getting the smoke.html ignore in also.
dalek r34209 | jkeenan++ | trunk/tools/util: 16:55
: Contents of this file are now out-of-date, as old-style smoke testing has been
: superseded by smolder.
review: xrl.us/xgf54
16:55 tetragon joined
dalek r34210 | jkeenan++ | trunk/lib/Parrot/Docs/Section: 16:56
: Delete reference to now-deleted file.
review: xrl.us/xgkjj
16:57 Hadi1 left
kid51 Coke: If goal for 1.0 is to have languages out of repository, then what does this imply for language-related functions in Parrot::Test? (This is the big remaining question for rt.perl.org/rt3/Ticket/Display.html?id=46893.) 16:58
dalek r34211 | jkeenan++ | trunk (2 files): 17:00
: Updating MANIFEST to reflect recent file deletion.
review: xrl.us/xg3it
Coke I would expect parrot to at most provide a framwork for languages to extend. 17:01
(but not actually host any of their extensions)
perhaps this issue can be addressed as more languages trac.parrot.org/parrot/wiki/LeaveTheNest 17:02
hurm. after your application of my patch, my working copy is now conflicted. 17:06
dalek r34212 | coke++ | trunk/config/gen/makefiles: 17:10
: last item in the list doesn't need a continuation-slash.
review: xrl.us/xh68o
Coke remaining conflicts (aside from those two) seem unimportant. 17:11
dalek r34213 | coke++ | trunk/languages/t:
: rescue slight verbage change from original smoke 2 smolder patch.
review: xrl.us/xid7w
r34214 | pmichaud++ | trunk/languages/perl6/src/builtins: 17:15
: [rakudo]: Add +Inf and -Inf to min and max (RT #60868, cspencer++)
: * Patch courtesy Cory Spencer.
review: xrl.us/xiu5y
r34215 | bernhard++ | trunk/languages/pipp (2 files): 17:19
: [Pipp] Add simple implementation of 'eval'
review: xrl.us/xi8op
Coke runs into another parrot bumper on unicode this time and redirects again. 17:23
no wonder I spend so much time treading water. =-)
dalek r34216 | bernhard++ | trunk/languages/pipp/t/embed: 17:25
: [Pipp] Call function defined in 'eval' in the embedder.
review: xrl.us/xjsie
barney Shouldn't get_hll_global [] 'no_args' be allowed ?
pmichaud it's just get_hll_global 'no_args' 17:26
barney Yes, but shouldn't an empty key list be valid. Like in .namespace [] 17:28
pmichaud I don't know -- that's up to other designers. Originally .namespace [] didn't exist either. :-) 17:31
17:45 leto joined
dalek r34217 | bernhard++ | trunk/languages/pipp/t/pmc (3 files): 17:48
: [Pipp] Remove useless '.HLL' declaration in PIR test scripts.
review: xrl.us/xmcg4
Coke yay, I think I unstuck another tcl test file.
raye++ # bringing me coffee!! 17:49
tcl has a lot of 'ensemble' commands, where the command itself does nothing but dispatch to a subcommand, e.g. "info tclversion" and "info foobar" both are invoked via 'info', but are basically entirely separate commands. Be nice if I could write some of these subcommands in tcl. "tclversion" just does "return $::tcl_version" ; but I had some handrolled PIR that missed an edge case. 17:52
Guess I could do that at compile time without too much heartache. 17:54
17:57 tetragon_ joined
barney Ah the default 'HLL' namespace is 'parrot'. Is that documented somewhere? 17:58
Coke it should be documented near '.HLL' in pdd19 18:00
(but I don't see it)
when you say default, you mean "in the absense of any previous .HLL directive" ?
barney yes
Coke barney: hey, did you remove the original verbage on my TT regarding languages/status ?
pmichaud it might be in the namespaces pdd
Coke docupatch greatly welcome on pdd19 on that one, though.
18:01 tetragon joined
barney Maybe, I'm a noob with Trac. 18:01
Coke fyi, there's a 'comment' window and an "original message" window. 18:02
figured that's what it was. 18:03
pmichaud RT slow.
dalek r34218 | pmichaud++ | trunk/languages/perl6/src (2 files): 18:04
: [rakudo]: Add support for Inf and NaN (RT #58014)
review: xrl.us/xmfak
pmichaud imo, rt.perl.org is rapidly becoming unusable as a issue tracking system, because of its slowness and frequent timeouts. 18:08
Coke has anyone complained to rbrt?
barney Yes, I wanted to comment and changed the original. Now repaired. 18:09
Coke barney++ 18:12
barney Pipp package vars and subs seems to be all stuffed in the 'parrot' namespace. I'll try to make that more sane tomorrow. 18:14
18:17 tetragon joined 18:19 tetragon_ joined 18:24 tetragon joined 18:31 tetragon joined 18:35 tetragon joined 18:41 tetragon_ joined 19:05 Theory joined 19:06 tetragon joined
Coke finds muddy catprints all over his home office desk. :| 19:34
lu_zero O_o 19:37
Coke I have 3 cats, so it's not inexplicable. It's just annoying. =-) 19:39
lu_zero ah ^^ 19:40
I was thinking about the muddy ghost-cat
19:50 rurban_ joined 19:51 elmex_ joined
Coke summons mdiep to hack on parrot blockers for partcl. 19:55
19:56 Hadi joined 20:06 ffwonko joined 20:07 alvar joined, kid51 joined
pmichaud irclog 20:13
20:14 Hadi left
Coke irclog? 20:16
purl irclog is irclog.perlgeek.de/parrot/today or see also: infrared clogs
dalek r34219 | pmichaud++ | trunk/languages/perl6/src/pmc:
: [rakudo]: Recognize numification of "Inf" and "NaN" strings.
: * Also fix strlen() bug introduced in r34167 patch.
review: xrl.us/xtspm
20:23 kid51 joined 20:30 kid51 joined
Coke wow. I actually needs someone who speaks tcl. 20:45
20:48 Theory joined
Coke Hurm. Or I could cheat. 20:49
20:57 donaldh joined
wknight-away Anybody around here running Win32? 20:58
Coke I have a windows notebook I'm running on 20:59
donaldh I have cygwin here.
wknight8111 if you have a moment, could you check out the jit_h_files branch and make testj?
I'm getting some test failures where the right output is created but Perl is returning an exit status of 255 21:00
actually, I'm getting a lot of such errors
Coke is strawberry ok? 21:01
wknight8111 I'm sure it is
I dont even know what my other computer has, although it might be activestate 21:02
if I could figure out what was causing these test failures, I would be able to merge the branch back into trunk 21:13
Coke wknight8111: building now 21:18
wknight8111: ok. I'm seeing tons of test failures. now what? =-) 21:26
GeJ Good morning everyone 21:29
leto mornin'
wknight8111 now what? I have no idea. do the tests that fail look like they are creating the correct output? 21:35
donaldh wknight8111: I get lots of Exited with error code: [SIGNAL 139] but the output matches the expected 21:39
Coke yup.
C:\\research\\jit_h_files>echo %ERRORLEVEL% 21:40
-1073741819
purl -1073741819
Coke I'm guessing something is wrong in your global cleanup. <Shrug>
strawberry doesn't seem to have 'gdb'. If you had a debugger, you'd be able to have your last opcode be something unusual, put in a breakpoint on it, and then just step through after that to see where things go wierd. 21:43
21:45 Eevee joined
wknight8111 unfortunately the code in question is i386 specific, and I don't have an i386 'puter here with GDB on it 21:56
Coke I'd put out a call to the list. 21:57
the only other x86 box I have /has/ gdb on it, but doesn't support JIT.
21:59 Theory joined
donaldh wknight8111: The stackdump is reporting STATUS_ACCESS_VIOLATION 22:03
wknight8111 STATUS_ACCESS_VIOLATION? wtf? 22:04
donaldh ah, probly the norm for that file. Not the most useful. 22:05
I have gdb, what args should I run parrot with, just -j ? 22:06
Coke you'll need a .pir file that dies.
wknight8111: any of the .pir leftovers in t/ that failed their test run will do. 22:07
22:08 ask_ joined 22:11 apeiron joined
ask_ git.develooper.com/?p=parrot.git;a=summary 22:15
Coke ask++
ask_ hmn, not sure what happened to all the branches. 22:16
Will figure that out some other day; have to sleep now.
ah, there we go I think... 22:19
dalek r34220 | Whiteknight++ | branches/jit_h_files/src/jit/i386 (2 files): 22:36
: [jit_h_files] moving a function definition back into the .h file because it's conditionally defined in two different ways. Will find a better solution later
review: xrl.us/yb6ch
22:38 Theory joined
Infinoid wknight8111: mingw's gdb works with strawberry 22:42
wknight8111 I'm testing my last commit now, or I would be if my internet connection wasn't craptastic 22:43
22:58 ruoso_ joined
wknight8111 okay, that didn't work 23:00
purl That's because you can't just stick someone else's code into your program and expect it to do what you want, Retardo!
wknight8111 ...that's what my wife calls me. Now I'm sad :(
no, she doesn't call me retardo 23:02
23:03 mj41_ joined
GeJ wasn't that from mjd's sigsnatures? 23:04
Coke hai. 23:06
Coke reclaims a few more tcl spec tests today. whee.
Coke got a very helpful suggestion from dgp on #tcl@freenode, even. 23:13
23:28 Hadi joined 23:29 Hadi left
dalek r34221 | pmichaud++ | trunk/languages/perl6/src/classes: 23:31
: [rakudo]: Clean up .Complex method; eliminate unnecessary Complex objects.
review: xrl.us/yixcb
23:41 ffwonko joined 23:46 bacek_ joined 23:51 ffwonko joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!