Parrot 0.6.0 "P&P" released | Please mentor for SoC | parrotcode.org/ | YAPC::NA talks deadline is Mar 31 | tinyurl.com/2pmnlq
Set by moderator on 18 March 2008.
00:03 teknomunk joined
svnbotl r26507 | chromatic++ | trunk: 00:06
: [JIT] Applied CONTEXT() macro changes to PPC JIT, which I somehow missed in
: r26491. This patch fixes RT #51912.
diff: parrotvm.org/svn/parrot/revision/?rev=26507
00:16 Theory joined 00:34 amoore joined 01:31 kid51 joined
Infinoid wknight8111: they both sound very useful 01:33
wknight8111 I'm thinking of applying for SOC, but I want a project that is going to be challenging and beneficial
Infinoid well, both of the ones you mentioned (GC and PCT optimizers) will definitely be beneficial 01:34
it's all gotta get done, after all :)
wknight8111 What does parrot do now, in terms of garbage collection? 01:36
kid51 purl seen chromatic? 01:37
purl chromatic was last seen on #parrot 1 day and 18 hours ago, saying: Or maybe just code.google.com/soc/2008/mentor_step1.html [Mar 19 00:14:04 2008]
kid51 purl seen rdice? 01:38
purl rdice was last seen on #perl 9 days and 13 hours ago, saying: rjbs, I highly recommend picnicface. To the point where my guys wish I stopped inflicting the videos on them. [Mar 11 06:06:16 2008]
Infinoid I'm pretty sure it's a mark & sweep system
I'm not too familiar with the internals, unfortunately
wknight8111: please see www.parrotcode.org/docs/pdd/pdd09_gc.html 01:40
wknight8111 yeah, thanks 01:41
01:42 particle joined
wknight8111 ugh, I haven't updated the SVN in a week, and now it's going to suck up all my computer resources 01:47
note to self: don't work on parrot on an old computer
Infinoid I build it on a 266MHz powerpc machine, about once a month. it takes a while :) 01:49
wknight8111 For some reason, TortoiseSVN crashes after updating about 500 files. So i have to stop, cleanup, and try updating again 01:50
Infinoid hmm. does tortoisesvn execute a command-line svn process, or handle things internally? does the command-line tool drop out? 01:55
wknight8111 it's all visual, integrated into the Win32 shell 02:00
i think my repository is completely borked. I may need to check out again 02:02
Infinoid on win32, I've only used the command-line tool, but it worked fine for me. 02:03
wknight8111 yeah, i should probably get something more primitive. I'm a glutton for this kind of nonsense
Infinoid the shell integration sounds really useful, but the "working" feature is kinda important to me :) 02:05
Eevee tortoise is pretty solid
02:06 Theory joined
Infinoid ok, so what's different between you two? 02:06
wknight8111 It may be, and I'm sure my computer has some kind of problem. I'm overdue for a reformat anyway. 02:07
I just don't want to reformat until I get my thesis finished, for obvious reasons
Infinoid makes sense 02:08
wknight8111 I get errors saying that my "access denied" to move one of the folders, and the update cancels 02:10
so I have to run a cleanup, and then try again
Infinoid blinks 02:11
02:12 Andy joined 02:45 cotto_home joined, cotto_home left 02:46 cotto_home joined
kid51 purl seen allison 03:00
purl allison was last seen on #parrot 9 days and 7 hours ago, saying: jonathan: yes, it's there now [Mar 11 12:56:49 2008]
03:00 c9s joined 03:07 Andy joined, c9s_ joined
wknight8111 literally, the svn just finished updating 03:08
I wanted to devote about an hour or two to hacking tonight, and that's out the window 03:09
Infinoid wow. might be more efficient just to ssh into a linux box and develop remotely 03:13
sorry to hear you're having issues...
03:14 c9s joined 03:16 teknomunk joined 03:21 c9s joined 03:26 Ademan joined
wknight8111 it's okay, I'll persevere. I have most of the day tomorrow to jump into this 03:30
Infinoid ok. I'll be around most of the day tomorrow, so please speak up if you need anything else 03:32
PerlJam Infinoid: I need a working perl 6 ... can you get me one of those? ;) 03:33
Infinoid heh, so do I :) 03:34
wknight8111 I'm building now. I'm going to log all the compile warnings, and try to fix a bunch tomorrow 03:39
Infinoid wknight8111++ 03:41
PerlJam is drawing a complete blank on how to debug a change he made to rakudo 04:09
Infinoid are there tests for it? 04:13
PerlJam I don't need a test. I know it doesn't work :) 04:14
Infinoid well... what's the change?
PerlJam I was attempting to add the loop { ... } construct 04:15
I thought I'd start of implementing only the infinite loop in actions.pm, but when I execute some code with a loop {...} in it, I get an error "No result object" from get_scalar in PCT/Grammar 04:16
Infinoid "* no result object" 04:18
"This is the case when you try to retrieve the result object from a subrule, but the subrule's action didn't set a result object using the "make" command. Check whether there's an action invocation token "{*}" in the subrule and whether that subrule's action has a "make" command." 04:19
I found that in docs/pct/gettingstarted.pod when I grepped for the error message 04:20
but I don't know if it helps you or not... you obviously know more about PCT than I do :)
PerlJam Infinoid: I have both of those things. 04:24
Infinoid well, apparently the value you're "make"ing is null 04:27
PerlJam It's probably my constant PAST::Val for the infinite conditional that's causing me grief 04:32
and it's probably that I'm really tired that I can't think straight :) 04:35
04:56 cotto_home joined 05:32 Psyche^ joined
spinclad Eevee: AIUI: docs/art/pp002-pmc.pod is out of date. the faq is correct, the VM has as many as it wants of each type. i don't see offhand which of the pdd's covers or should cover this, but in general go to the pdd's for definitive answers. 05:36
followup question for all: which pdd(s) cover(s) the basic architecture of the vm, like number/allocation of registers, frame and interpreter structure/attributes/properties, ... ? 05:39
05:40 contingencyplan joined
spinclad it may be sitting there staring at me, but i haven't stumbled on it yet. i know the book chapters cover this, but (a) were/are they definitive, and (b) they got grossly outdated before getting released to us, and aiui their revision is In Progress. 05:42
cotto_home why does are there references to 78-character lines when the coding standard and associated test says 100? 06:45
obvious answer, rhetorical question 06:48
06:54 barney joined
cotto_home anyone knowledgeable about PMC internals awake? 06:55
bah. I need to get Parrot to not puke with -DFREEZE_ASCII 06:57
it's too mysterious as is 06:58
barney is awake, but has to leave in 10m 07:01
cotto_home you're more of a hll guy anyway, right?
07:02 grim_fandango joined
barney yep. not deep into internals 07:03
07:06 slightlyoff joined 07:13 peepsalot joined 08:20 jjuran joined
svnbotl r26508 | fperrad++ | trunk: 08:20
: [Lua]
: aligned with Lua 5.1.3 (part 4)
diff: parrotvm.org/svn/parrot/revision/?rev=26508
08:51 wknight8111 joined 08:56 turrepurre_ joined 10:33 c9s joined 12:02 c9s joined 13:01 jrt4 joined 13:04 skids joined 13:17 cdm joined 13:22 slightlyoff joined 13:42 wknight8111 joined 13:45 Andy joined 13:53 particle joined 14:05 jhorwitz joined 14:12 gryphon joined
svnbotl r26509 | fperrad++ | trunk: 14:12
: [imcc]
: fix version message when building a release :
: - DEVELOPING is removed
: - so Parrot_revision & Parrot_config_revision give 0
: (see lib/Parrot/Revision/Util.pm)
: with the latest release, the current message is :
: This is parrot version 0.6.0 (r26483) built for i386-MSWin32.
: Warning: runtime has revision 0!
: Warning: used Configure.pl revision 0! 14:13
: Copyright (C) 2001-2008, The Perl Foundation.
diff: parrotvm.org/svn/parrot/revision/?rev=26509
14:49 particle joined 15:00 sjansen joined 15:07 particle left 15:16 slightlyoff joined
wknight8111 I want to clean up the snprintf macros and put them all in the same place to avoid redefinitions 15:38
should they go into parrot/misc.h (where one of them is now), or should I move them to some place more appropriate like parrot/string.h? 15:39
Actually, I may not need to move anything 15:40
16:10 Theory joined
wknight8111 purl seen chromatic 16:37
purl chromatic was last seen on #parrot 2 days and 9 hours ago, saying: Or maybe just code.google.com/soc/2008/mentor_step1.html [Mar 19 00:14:04 2008]
svnbotl r26510 | infinoid++ | trunk: 16:41
: Patch in RT #51976 seems very reasonable, and passes tests. Apply it.
: Thanks to Andrew Whitworth. wknight8111++
diff: parrotvm.org/svn/parrot/revision/?rev=26510
Infinoid "on first glance"? "it sounds?" I need a system to keep track of which metaphors are within lexical scope at any given time. 16:43
16:44 ambs joined
wknight8111 are you talking about my patches? 16:44
Infinoid no, I'm talking about my disorganized replies to your patches.
wknight8111 that's fine, i don't require organized replies 16:45
or any replies at all, for that matter
16:47 kj joined
Infinoid that function Parrot_char_digit_value() is pretty suspect... if -1 is a valid return value, then its return type shouldn't be UINTVAL 16:48
good find
purl it has been said that good find is hard to man...
Infinoid % man 'good find'
No manual entry for good find
wknight8111 I assume that they are trying to do something like return MAX_UINTVAL, just with shorthand
Infinoid -1 is returned if the character isn't in the range ('0'..'9') 16:50
the unsignedness of the result is just silly
wknight8111 I can go through and change the result to INTVAL or something else then, if that's a better idea
probably only need to change that file, the header, and a few other places 16:51
Infinoid I think so. could you?
INTVAL is right, I think
wknight8111 Okay, let me get started on it. I'll post a patch to the same ticket # for it
Infinoid wknight8111++ 16:52
ambs prepares to test strawberry perl 16:56
hmmms... yummy strawberry :) 16:57
erm, why am I talking on #parrot? wrong channel :-S
Infinoid heh. strawberry perl rules. 17:00
ambs I want to test some modules under windows 17:01
:)
Infinoid wknight8111: is there an option in tortoisesvn to always generate patches with pathnames relative to the parrot root? they would be a little easier to apply, that way
wknight8111 There might be. I didn't realize it was a problem. I'll work on it. 17:02
Infinoid its not a big deal, it just means I have to go find utils.c and cd into src/ before I can apply the patch 17:03
wknight8111 okay, i figured out what I was doing wrong. it will never happen again :) 17:07
Infinoid I'm going to reformat utilsmoveregtypecast.patch slightly so it passes t/codingstd/linelength.t 17:09
wknight8111 okay 17:15
Infinoid the code in jit_emit.h is there to reduce duplicated code; it is included and used in several source files
wknight8111 I'm searching diligently, and I can't find use of Parrot_char_digit_value outside of string_primitives.*
So, I think i'm going to just submit the patch with the .c and .h files changed
svnbotl r26511 | infinoid++ | trunk: 17:18
: Patch from RT #51984 doesn't appear to change any actual functionality, quiets
: some warnings on win32/msvc, and passes tests. Apply it.
: Minor modification: I split the line into 2, to pass t/codingstd/linelength.t.
: Thanks, Andrew Whitworth! wknight8111++
17:19 spez joined
Infinoid I don't see any actual usage of that function, either 17:19
ambs hmms, strawberry does not include a yacc. 17:24
Infinoid great, nothing to shave 17:33
17:36 Psyche^ joined
kj Anybody know how to express that somethign is "in the public domain"? (source code) 17:42
Like, "this source code is public domain" or whatever? How to express that properly?
wknight8111 I assume it would be something like "This source code has been released by the author into the public domain. Where this is not possible by law, the author grants license to use this file for any reason without any rights reserved, and with no warranty express or implied or fitness for a particular purpose" 17:44
kj sounds good :-) 17:45
wknight8111 of course, that may just be a mashup of memories from PD and GPL warnings
kj Are you ok if I use that text?
it's just for the PCT tutorial
wknight8111 I release that text into the public domain :)
kj ha ha
ambs dang 17:46
wknight8111 on Unix boxes, what parameters does the dup() function take? 17:49
on Win32 it takes a single int, and if that's standard behavior for it, I think I can fix a few warnings 17:50
ambs #include <unistd.h> 17:51
int
dup(int fildes);
int
dup2(int fildes, int fildes2);
Eevee yeah that
purl i guess yeah that is absolutely outrageous
ambs purl: forget yeah that
purl ambs: I forgot yeah that
wknight8111 thanks @ambs
17:56 askie joined 18:00 Psyche^ joined
kj PCT tutorial episode 6 is published! 18:01
davidfetter url? 18:04
kj parrotblog.org
18:08 lidi20 joined
kj www.bitwisemag.com/2/DLR-Build-Your-Own-Language 18:10
jhorwitz kj: very nice (PCT on parrotblog.org). i'm compiler-challenged, so i definitely need this. :) 18:13
kj cool! :-) good to hear.
I'll have about 3 more episodes, after which the complete language is implemented
cotto_work kj++ 18:19
jhorwitz i've been MIA the past week or two. did we hear what parrot OSCON talks were accepted? i know mod_parrot was. 18:20
kj w.r.t. the link DLR-build-your-own-language: again I notice that Parrot just ignored silently; there's only mention of JVM and CLR/DLR "new" support for dynamic languages. 18:36
wknight8111 Can somebody take a look at src/encodings/fixed_8.c:to_encoding()? 18:39
the function is declared PARROT_DOES_NOT_RETURN, and PARROT_CANNOT_RETURN_NULL, and has a return type of STRING * 18:40
either it does return a non-null string, or it doesnt return at all, and it can't be both 18:41
jhorwitz kj: based on the author's other articles, he seems very MS-centric
18:44 h0rn joined
cotto_work allison was at MS recently talking to some of the DLR guys 18:50
some of them know about Parrot 18:51
jhorwitz wknight8111: looks like that function is unimplemented for fixed8. this is a complete guess, but in that case the PARROT_DOES_NOT_RETURN might avoid some compiler warnings since nothing is actually returned, while the STRING * return type is needed for the ENCODING structure.
kj it just seems strange that if people start writing on dynamic languages, they don't know about what's out there
wknight8111 thanks jhorwitz, that's what I figured it was. So i guess the function definition will need to change when they implement it 18:52
kj but I can understand the focus on jvm and dlr
jhorwitz i'm hoping parrot has a strong showing at OSCON this year to help spread the word. 18:53
cotto_work maybe it'd be a good idea to submit your tutorials to slashdot once they're done, kj 18:56
that'd help raise awareness and show people the state of the project
kj i don't know if they're good enough for /. How does one get stuff there? 18:58
cotto_work slashdot.org/submit.pl 18:59
18:59 Psyche^ joined
Infinoid kj: use.perl.org would be a good place to put it, too 19:05
kj cotto_work: thanks for the link. That might be a good idea. Maybe it's good to wait till some people have read it and commented on it
to improve quality 19:06
Infinoid: yep: use Perl 6 to implement Your Favorite Language today!
:-)
19:12 jrockway joined
svnbotl r26512 | infinoid++ | trunk: 19:22
: Apply patch from RT#51982, thanks! wknight8111++
diff: parrotvm.org/svn/parrot/revision/?rev=26512
Infinoid wknight8111: the symbol "_dup" does not exist on linux 19:28
nopaste "Infinoid" at 75.5.242.149 pasted "RT#51988 build failure on x86-64 Linux" (17 lines) at nopaste.snit.ch/12543 19:29
wknight8111 it doesnt? All the stuff I saw (or thought I saw) online seemed to indicate it did 19:35
I mean, I won't argue with you about it, I must be misinformed
Infinoid nope, it's just dup() and dup2() 19:38
they don't really exist as symbols at all, at least not on my machine. they're inline functions in unistd.h that boil down to some assembly code that makes the system call and updates errno if necessary 19:40
wknight8111 Okay, then I guess we need to disregard that patch then. cl.exe warns about use of dup(), but a warning on one system is better then complete failure on another 19:41
wolverian a few remarks: parrot doesn't seem to configure cleanly from the git mirror, and doesn't find my local perldoc (when using my local perl) 19:42
Infinoid wolverian: can you paste what your Configure.pl invocation looks like, using nopaste? 19:43
purl, nopaste
purl it has been said that nopaste is at nopaste.snit.ch/ (ask TonyC for new channels) or rafb.net/paste or paste.husk.org/ or poundperl.pastebin.com/ or paste.scsys.co.uk/ or don't bother me while I'm eating
Infinoid wknight8111: what's the warning? I'm sure we can find a way to make both platforms happy :) 19:44
(and probably break Darwin in the process, tee hee)
wolverian Infinoid, do you want me to run it verboes? 19:45
s/oes/ose/
Infinoid I want to see how it's not configuring cleanly :) 19:46
wolverian well, sec. :)
it's trying to call svn to determine revision
Infinoid yeah, we keep having to fix that in various places
nopaste "wolverian" at 80.221.43.65 pasted "configure" (87 lines) at nopaste.snit.ch/12544 19:47
wolverian tries to find out why it doesn't detect perldoc 19:49
Infinoid well, does perldoc exist? (there's a funky setting called "versiononly" which will prevent perl from installing perldoc, h2xs, and some other tools) 19:50
wolverian yes, it does 19:51
purl stays quiet
wolverian furthermore, Configure.pl says "no", which seems to indicate (by perldoc.pm) that it is finding perldoc, but it's not getting any output from it
hmm.
Infinoid hmmm. there are some nice errors from lib/Parrot/Revision.pm, which hasn't been changed since kid51's branch merge in January 19:53
this is a normal git-svn checkout, right? 19:55
wolverian no, this is from repo.or.cz/w/parrot.git 19:56
which might be totally wrong, of course, since I don't know whose it is :)
huh. perldoc.pm seems to be trying to run '...//perldoc'
Infinoid the extra / won't hurt, but does the path look correct otherwise? 19:57
wknight8111 is kj still here?
kj ambsyup
ehm
whoes
wolverian no, it doesn't. I don't know what ... would mean, anyway
kj that should have been a yes :-) 19:58
wolverian I'm in ~/projects/parrot, perldoc is in ~/local/bin
wknight8111 kj, I really like your PCT tutorials, would you mind if I "stole" them and turned them into a book at Wikibooks?
I mean, I know they are PD, but I like to be curteous anyway
courteous*
kj wknight8111: no problem. 19:59
Infinoid wolverian: what's the output of: perl -e 'use Config; print($Config{scriptdir}, "\\n")'
wknight8111 excellent. I'll send you a link when I get something up and running
wolverian Infinoid, ".../"
I wonder how I managed that ...
kj great. I'm writing the last ones right now (ep.7; 8 is finished, 9 is jsut a conclusion) 20:00
wolverian the paths in perl -V look correct otherwise
Infinoid that's pretty bizarre. mine is "/usr/bin"
maybe there are multiple Config.pm's in your @INC and its choosing the wrong one? 20:01
wolverian okay
$Config{prefixexp} is /home/wolverian/local/bin/..
(note the ..)
and binexp, like scriptdir, is .../
scriptdirexp is correct though 20:02
Infinoid so... we could work around your bizarre and broken scriptdir by using scriptdirexp. trying to think if there's any downside to doing that 20:03
wolverian I'm pretty sure this shouldn't be happening in any case, i.e. perl should have configured itself correctly, regardless of my stupidity :) 20:04
Infinoid I'd really like to know how your system ended up in this state, and whether its a case which will come up again :)
I don't think "..." is a valid directory-alias on any platform. so I'm wondering if its some kind of internal placeholder 20:05
if so, its definitely something we need to recognise
wolverian any hints how to debug the perl source tree (after config,make,make install) to see what went wrong? 20:06
Eevee I think some versions of Windows accept ... to mean ..\\.. 20:07
not sure if XP/Vista still do
Infinoid wolverian: what platform are you running on? 20:08
wolverian Infinoid, linux, ubuntu 8.04
Infinoid self-built perl, or .deb package?
wolverian self, 5.10
Infinoid ok. I'm using a half-broken 5.10.0 gentoo ebuild :) 20:09
as for debugging it... its all about the values in Config.pm, which (judging from comments at the top) was generated by configpm
davidfetter you're repeating yourself
"half-broken" + "gentoo" ;)
Infinoid thank you for that, davidfetter :) 20:10
wolverian ;)
davidfetter hey, it used to be 90% broken ;)
half- is a big improvement
wolverian I think I built with userelocatableinc, btw
yes, I did 20:11
that's the only special flag besides prefix that I used
and prefix was ~/local
perhaps you can reproduce it with that?
Infinoid but remember, gentoo ebuilds aren't release quality until they're *fully* broken
users demand nothing less
davidfetter Infinoid, are you over 30 years old?
Infinoid I'll make a build and see what it looks like... will probably take a little while
davidfetter: not quite, but gimme a year 20:12
wolverian ah, it's explicitly mentioned in perldoc perldelta
davidfetter has noticed a marked paucity of post-30 people in gentoo
wolverian That means that, if the string ".../" is found at the start of any path, it’s substituted with the directory of $^X. So, the relocation can be configured on a per-
directory basis, although the default with "-Duserelocatableinc" is that everything is relocated. The initial install is done to the original configured prefix.
sorry for the horrendous paste.
so it _is_ a placeholder 20:13
Infinoid riight. so we're trying to use the bare value and it isn't working, because they are smarter than we are 20:14
$Config{scriptdirexp} is looking nicer and nicer
wolverian isn't it nice, though, that they let you peek at the internal value, though? :)
s:2nd/, though//
nopaste "Infinoid" at 75.5.242.149 pasted "scriptdir -> scriptdirexp" (13 lines) at nopaste.snit.ch/12546 20:15
Infinoid wolverian: does that reduce your breakage any?
wolverian Determining whether perldoc is installed...Use of uninitialized value in concatenation (.) or string at config/auto/perldoc.pm line 39. 20:16
.............................no.
even though this prints it fine: perl -MConfig -E'say $Config{scriptdirexp}' 20:17
Infinoid ok, please add scriptdirexp to the list of variables to steal, in config/init/defaults.pm 20:18
fortunately it looks like nothing else in parrot uses scriptdir
wolverian that gets past the perldoc check, yay 20:19
s,gets,gets Configure.pl,
next problem: make (dist)clean leaves behind these: languages/perl6/Makefile languages/perl6/src/utils/Makefile runtime/parrot/include/signal.pasm 20:20
I'm sure I should just submit these as bugs instead of complaining here, sigh 20:21
Infinoid sorry, I'm not ignoring you, I'm preparing this config tweak for commit 20:22
wolverian no worries 20:24
Infinoid might be interesting to go through %Config looking for other instances of ".../"... that perldelta entry looked an awful lot like it's gonna occur elsewhere 20:25
wolverian just don't want to stop discussion here with these issues
Infinoid try: make realclean
hmm, that doesn't appear to remove languages/perl6/Makefile either. oops 20:26
Infinoid crosses his fingers and hopes he didn't just break older versions of perl
nopaste "wolverian" at 80.221.43.65 pasted ".../ in Config.pm" (16 lines) at nopaste.snit.ch/12547
Infinoid thanks 20:28
wolverian you're welcome
Infinoid looks like scriptdir was the only path-related variable we were reading in config/init/defaults.pm, so we might get lucky and not care about the rest
wolverian should make succeed even though the svn revision check failed? 20:30
because, it doesn't. :)
"src/oo.c:25:27: error: pmc/pmc_class.h: No such file or directory"
svnbotl r26513 | infinoid++ | trunk:
: [config] perldoc check should use $Config{scriptdirexp}, not $Config{scriptdir}.
: The "scriptdir" value from p5 is sometimes interpreted in strange ways, but
: the "scriptdirexp" value is expanded at compile-time, so it's guaranteed to be
: sane.
Infinoid msg kid51 please take a look at r26513 for me... It Works For Me, but I'm hoping I didn't break older versions of perl, and I'm hoping we won't run into this issue anywhere else.
wolverian hmm. make -j1 seems to fare better 20:31
Infinoid stares at the lack of purl
we are not make -j-safe, and I would love to fix that
wolverian ah, that's it, then. thanks.
make fails building revision.c, calling git-svn 20:33
which obviously doesn't work if it's not a git-svn clone.. 20:34
Infinoid yep, that makes sense
is the problem that you don't have git-svn installed, or that your git checkout doesn't have git-svn markings?
wolverian the latter. I just cloned from the repo I linked to earlier, since it's by far faster than running git-svn. 20:35
(if I want history, anyway)
Infinoid trying it here...
wolverian that is, this: repo.or.cz/w/parrot.git 20:36
Infinoid hmm. I can't clone that 20:37
wolverian the actual git repo is: git://repo.or.cz/parrot.git
Infinoid oh, I'm a bonehead. right-o. 20:38
20:40 purl joined
Infinoid msg kid51 please take a look at r26513 for me... It Works For Me, but I'm hoping I didn't break older versions of perl, and I'm hoping we won't run into this issue anywhere else. 20:51
purl Message for kid51 stored.
Infinoid watches git-svn spin taking all the CPU 20:52
wolverian heh, yeah. I wonder what it's trying to do 20:54
there is no svn stuff in .git/config
Infinoid according to strace, it is running "git config -l <something>" over and over and over 20:55
git-svn spins forever once during Configure.pl, and again during tools/build/revision_c.pl
wolverian it doesn't actually run forever for me during Configure.pl 20:56
just for a while :)
30 seconds, perhaps 20:57
Infinoid ok, s/ <something>//
wolverian longer during revision_c.pl, perhaps forever.
Infinoid well, maybe I'm just impatient
wolverian nods
Infinoid apparently git-svn thinks something is going to change in the output of "git config -l", if it keeps trying
wolverian hmm 20:59
wolverian clones git
ouch, 30kB/s 21:00
this will take a while :)
I guess kernel.org needs a fatter pipe...
davidfetter wonders whether it needs a phat sack to go with the phat pipe 21:02
Infinoid I think the phat sack is what they use to pay for the phat pipe 21:04
but I'm not sure :)
wolverian this talk of sacks and pipes is making me uncomfortable, though I'm not quite sure why 21:05
wolverian twiddles thumbs
Infinoid discovers git-svn clone -r, and spares poor svn.perl.org 21:07
wolverian har
I'm sure it works with a git-svn clone .. or someone would have complained
Infinoid everyone who ever ran it without -r is probably still waiting and watching 21:08
wolverian haha
well, that's why I used the git mirror :]
oh, yay, git-svn is still a 4863 line perl script 21:09
yes, that's the exact linecount
Infinoid awesome
is it possible to convince the maintainer of git://repo.or.cz/parrot.git to insert some git-svn config values into their repo? 21:10
wolverian probably
I don't know him though; his e-mail is on the gitweb page
Infinoid say what you like about the awesomeness of raw git, but it *is* very useful to us to be able to obtain the svn revision, instead of a git hash 21:11
in this case, I guess the only way to do so is to look at the last line of the commit log
wolverian add a commithook on the svn server that adds the revision to REVISION? ;)
what's the git-svn invocation that parrot runs? 21:19
ack found it for me 21:24
I can't follow git-svn's logic, at least not without refactoring it into separate files and making some diagrams :) 21:29
it's clearly a bug in git-svn though 21:35
hm, it doesn't happen on an empty projet 21:36
s/projet/project/
i.e. git svn log --limit=1 doesn't spin
neither on git.git 21:37
I wonder if it's doing some kind of dumb loop and running git config once for each commit, or something 21:39
I'll let it run and see if it finishes
Infinoid could probably just parse the output of "git log -1" instead
wolverian looks like it 21:40
it'd be good to know what causes this behaviour and get it fixed in git-svn though 21:41
wolverian scratches his head and resumes the hunt for the red october
Infinoid that movie rocks.
wolverian yes :)
aha! it finished!
Infinoid yeah, git-svn could be fixed, but a more general solution for parrot would also help
wolverian (git svn log --limit=1 did, that is)
I don't see the looping behaviour on the git.git repo 21:42
what's different about parrot's repo?
okay, now that I let make spin a while on the revision.c stuff, I get this: compilers/imcc/main.c:243: error: expected expression before ‘)’ token 22:36
funnily enough, that line is: printf(" (r%d)", PARROT_REVISION); 22:37
:)
revision.c ended up with this (nopasting):
nopaste "wolverian" at 80.221.43.65 pasted "sad revision.c" (35 lines) at nopaste.snit.ch/12550 22:38
wolverian I suppose that's not a surprise though.. I'll just wait for the git log fix :)
22:43 skids joined
Infinoid working on that now 22:46
wolverian Infinoid++ 22:47
Infinoid running git-svn has a side effect of creating .git/svn/ even if it doesn't exist. which is a shame, because testing for the existence of .git/svn/ would be an easy way to choose whether to use raw git or not 22:50
hrm. "git log -1" works in both cases. might as well use it in both cases. 22:51
Determining Parrot's revision.......................................r26513.
\\o/ 22:52
does git run on non-unix platforms? do I have to worry about the portability of checking for the existence of a ".git" directory? 22:59
svnbotl r26514 | infinoid++ | trunk:
: [configure]
: * patch Parrot::Revision to use "git log -1" instead of "git-svn log --limit=1".
: This fixes the case where the user is building from a (non git-svn) git
: checkout from git://repo.or.cz/parrot.git - git-svn hangs for long periods of
: time (and eventually fails) when run on a non-git-svn checkout.
: * wolverian++ for mentioning it.
: * git-svn checkouts still work just fine... this solution covers both git cases.
diff: parrotvm.org/svn/parrot/revision/?rev=26514
wolverian it run-ish on windows with mingw nowadays, soon to be integrated into the mainline code
Infinoid did they change the name of the .git directory? 23:00
wolverian I don't think so
Infinoid I want something that always exists for any git checkout
wolverian that wouldn't make any sense
Infinoid suppose I could just skip the directory check and rely the return value of git, like the other code in that file does
wolverian let me test, I happen to be on vista right now 23:01
(games-- # come on, support linux already)
Infinoid thanks
as for that, there's always tuxgames
wolverian it's still .dir 23:05
er
.git
Infinoid great. does the patch work? (I can nopaste a copy, if it helps.) 23:06
wolverian still cloning.. sec 23:07
oh, wait
I suppose I should boot back to ubuntu to test it. :)
well, I guess I can test from vista too
wolverian is too tired
23:07 Psyche^ joined
wolverian of course, windows brings its own interesting features.. 23:08
Infinoid wolverian: tomorrow is fine, git-svn doesn't have many users :)
wolverian s/git-// # in the near future, hopefully ;)
meh, I don't have gcc on windows 23:09
does parrot build with mingw?
Infinoid yes
strawberry perl has a nice bundle that works fine for me
wolverian oh, good point 23:11
the paths aren't set up though 23:12
ah, there it is
eh, I don't know how to use this stuff 23:13
I'll test in ubuntu tomorrow :)
gotta sleep now
thanks and cheers 23:14
23:16 Limbic_Region joined 23:27 h0rn left 23:52 jrt4 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!