Parrot 1.9.0 "Blue-fronted Amazon" released! | parrot.org | Roadmap: icanhaz.com/parrotroadmap | Latest modified TT's: icanhaz.com/parrotbugs
Set by moderator on 16 December 2009.
poet what confuses me is why g++ is giving the error message as opposed to gcc 00:08
Coke PerlJam++ 00:21
poet: my mac is dead atm, sorry, cannot verify. 00:22
poet Coke: That's ok. I tried on the latest from SVN as well with the same results. I'll boot up a VM until I can figure out what the problem is 00:23
Coke hurm. is "assert.h" a parrot file or an OS one? 00:31
00:31 davidfetter joined
cotto Coke, it seems to be an os file. 00:34
Coke k 00:35
00:37 payload joined 01:04 bacek joined
dalek rrot: r43267 | coke++ | trunk/src/gc/malloc_trace.c:
This .h file doesn't exist.
 01:12
rrot: r43268 | coke++ | trunk/compilers/pirc/src/piremit.c:
Switch to <> include to simplify our dep checking.
01:28 zak_ joined 01:51 lucian joined 02:15 eternaleye joined 02:32 JimmyZ joined 02:35 bacek joined
dalek TT #1395 created by jimmy++: [patch]various minor change to parrot 02:51
TT #1396 created by jimmy++: [patch]changed two PMCs to use GET_ATTR syntax
rrot: r43269 | coke++ | trunk/tools/dev/checkdepend.pl:
Fix include dep-checking so it works regardless of directory.
 03:06
rrot: r43270 | coke++ | trunk/config/gen/makefiles/root.in:
Pass latest tools/dev/checkdepend.pl
TT #1397 created by jimmy++: [patch]changed ATTR n_regs_used[4] to use UINTVAL type in sub.pmc 03:07
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#31345), fulltest) at r43270 - Ubuntu 9.10 amd64 (g++ with --optimize) 04:07
had to run make -j corevm twice though
Coke mikehh: ORLY? 04:50
purl YA RLY.
Coke when you get a make -j run that fails to complete, I need to see the build log so I can figure out what dep is missing.
(and then fix checkdepend.pl to catch it.)
(and failures due to "resource not available" are not our problem. =-) 04:52
dalek rrot: r43271 | coke++ | trunk/config/gen/makefiles/root.in:
Don't bother trying to delete files that are no longer created.
 05:00
05:02 bacek joined
dalek rrot: r43272 | coke++ | trunk/config/gen/makefiles/root.in:
Remove historically amusing comment.
 05:16
rrot: r43273 | coke++ | trunk/compilers/pirc/src/bcgen.h:
'make headerizer' run.
07:20 jjore joined
cotto Coke, make headerizer is known to break the pirc build. Did you test that change? 08:30
08:35 iblechbot joined 08:50 eiro_ joined 09:21 bacek joined 09:26 fperrad joined 10:45 hejki joined
hejki ahoy there. i noticed there isn't ISLNK for stat(). any other ways around this? 10:45
11:03 bacek joined 11:04 szabgab_ joined 11:05 szabgab joined
bacek aloha 11:45
hejki ahoi. happen to have any ideas how-to figure out whether a dir is indeed a link? 11:47
11:54 masak joined
bacek hejki, hm... Os.stat? 11:57
12:13 ruoso joined
dalek rrot: r43274 | bacek++ | trunk/src/pmc/os.pmc:
[cage] Use INTERP instead of interp in Os PMC
 12:18
12:45 bluescreen joined
hejki bacek: i'm looking into it.. looks ok this far. i just have issues with my PIR "skills" :) 12:45
bacek hejki, check t/pmc/os.t for samples 12:47
hejki sure 12:48
i'm just wondering why i still keep getting same return values for links and dirs :P
(i'm returning [2] from the FixedPMCArray Os.stat returns)
the modes are correctly mirrored but somehow dir and link with same perms give same return code 12:49
bacek Os.stat is just wrapper around system's stat. 12:50
hejki but that gives separate retvals for links and dirs with similar permissions
mode from stat() should include both file type and perms
? 12:51
bacek yes, according to man page. 12:52
But anyway, S_ISLNK isn't exposed into PIR. 12:53
So, feel free to open trac ticket for it :)
12:56 payload joined
hejki hmm.. stat doesn't actually make a difference in mode-number between dirs and links :O 12:57
hejki wonders how the S_ISLNK is implemented 12:58
bacek system depended I think. 13:00
hejki i should be on a POSIX-compliant one
heh.. by a quick glimpse it seems that my original idea of implementing S_ISLNK into parrot seems to have loads of work to do.. i guess i'm better off filing a ticket :> 13:17
hmm.. there seems to be METHOD is_link(STRING *path) in pmc/file.pmc tho which uses S_ISLINK 13:25
s/LINK/LNK/;
13:25 cybertom joined
Coke cotto : nope. just ran it blindly! 13:32
(testing now) 13:33
13:39 cybertom left
Coke msg cotto: reverted. cotto++ 13:42
purl Message for cotto stored.
dalek rrot: r43275 | coke++ | trunk/compilers/pirc/src/bcgen.h:
Revert r43273 -
 13:55
14:03 JimmyZ joined
dalek rrot: r43276 | coke++ | trunk/config/gen/makefiles/nqp-rx.in:
nqp-rx has no tests, don't offer to run any.
 14:12
14:16 mikehh joined
Coke pbc_to_exe seems to have an undocumented -o option that doesn't work. 14:27
JimmyZ good evening, #parrot 14:33
Coke ~~
JimmyZ: I'll re-raise your commit bit question in PS tomorrow. 14:34
14:37 patspam joined
JimmyZ Coke, thanks 14:37
Coke bah. I tried to move nqp-rx's build into root.in as part of my makefile consolidation strategy. Get it 99% there... it depends on PCT, which is in a separate makefile. 14:47
dalek TT #1398 created by coke++: pbc_to_exe should be able to specify the output files. 14:51
hejki i can't compile Rakudo anymore as of yesterday. i'm getting: pastie.org/758753 also when i'm compiling Parrot i get a bunch of "ERROR: ld.so: object 'libfakeroot.so' from LD_PRELOAD cannot be preloaded: ignored." error messages. Parrot doesn't fail to compile tho. Any ideas/pointers? 14:55
masak I have this build problem too. it seems to have surfaced when I updated to the latest Parrot. 15:09
15:12 jan joined 15:14 bubaflub joined
Coke ... libfakeroot? 15:16
fakeroot doesn't appear anywhere in parrot's src.
hejki wot :D
Coke what does your LD_PRELOAD eval to?
Coke again finds a wish for parrot on git. 15:17
(but without git-svn or the unofficial mirror.)
bubaflub i'm with ya Coke, i'm more comfortable with git. esp. branching and working out-of-sync with others 15:18
Coke I am newly converted.
Right now, I just want to snapshot my work in progress so I can continue to break things. 15:19
bubaflub I had been using SVN for a few years but my current job used git
so i had to pick it up. it's been less than a year but i'm definitely converted.
i can't speak for hg or bzr but distributed is definitely the way to go if there are more than a handful of developers working on something 15:24
dalek rrot: r43277 | coke++ | branches/one_make:
Reduce the number of separate top level makefiles in the project.

  �groups.google.com/group/parrot-dev/...7ee0374b77		 15:34
rrot: r43278 | coke++ | branches/one_make (6 files):
Eliminate nqp-rx's Makefile, avoid unnecessary copying of PBCs.
TT #231 closed by coke++: make -j2 test fails 15:40
masak <Coke> fakeroot doesn't appear anywhere in parrot's src.
Coke: so, what does that mean for the Rakudo build error that hejki and I are currently experiencing? 15:41
hejki masak: which distro you're on? 15:42
Coke masak: ... that it's not parrot's fault?
hejki :D
Coke does your LD_PRELOAD include "fakeroot" ? 15:43
ah.
Coke googles for fakeroot.
masak what's LD_PRELOAD? it doesn't seem to be an environment variable.
hejki archlinux uses fakeroot for building stuff.. so i'm looking for possible similarities between me and masak :>
masak hejki: I'm on Darwin. 15:44
hejki ok
Coke darwin uses different varnames for LOADING than other osen.
man dyld, iirc, to find them. 15:45
but I have never heard of using fakeroot on darwin until just now. =-)
masak I could build Rakudo until just now. :)
hejki same 15:46
masak I find nothing in `man dyld` containing the word PRELOAD.
there's DYLD_LIBRARY_PATH, though.
oh, oh. I'm not getting the fakeroot errors. those are from hejki's Parrot build process only. 15:47
but I am getting the same Rakudo build error, and for code that built yesterday with an earlier Parrot version. 15:48
hejki (which might actually be caused by my distro (archlinux) using fakeroot for building)
Coke masak - did rakudo just update it's parrot dep revision? 15:50
*its
masak Coke: I'm not running against Rakudo's recommended Parrot revisions. I always run against the latest Parrot.
also known as "that might indeed be the reason". 15:51
if it helps, I could downgrade Parrot to the recommended version and rebuild.
Rakudo recommends Parrot r43076. I'm running Parrot r43275. 15:52
hejki i'm on 43276 :) 15:53
JimmyZ /me couldn't open groups.google.com/group/parrot-dev/...7ee0374b77 15:54
masak still doesn't explain why a newer Parrot makes some Rakudo .c file not compile anymore...
hejki ye
masak: in case you gain success with the preferred Parrot, lemme know about it so i can start testing out with the 15:55
S_ISLNK-macro*
i'm writing improper multiline
masak hejki: I will. 15:56
dalek TT #1343 closed by coke++: parrot's Test::More's is(PMC,Float), with precision
15:56 ash_ joined 15:57 Psyche^ joined
masak hejki: yes. that worked. 16:01
hejki nice 16:02
now i just need to do some SVN-hackery and see if my patches are still applicable :P 16:03
masak svn up -r43076 16:04
hejki ye 16:05
ah.. patches went all ok 16:08
hejki crosses fingers
let's see what happens for my rakudo now :)
Coke msg jimmyz I can open that fine here; GFWoC issue? 16:12
purl Message for jimmyz stored.
Coke msg jimmyz it was the 'makefile thoughts' thread on list.
purl Message for jimmyz stored.
Coke masak: if you could do a bisect to figure out the issue, that would help. 16:13
*the revision that caused the issue. 16:14
masak Coke: ooh, good idea.
purl masak: Good Idea: Tossing a penny into a fountain to make a wish. Bad Idea: Tossing your cousin Penny into a fountain to make a wish.
masak purl: you're a ceaseless fount of inanity.
purl masak: i'm not following you...
masak Coke: I'm kinda just leaving, but I'll get to it later today or tomorrow. 16:15
Coke hokay. I recommend infinoid's "svn-bisect" tool.
(from cpan)
masak googles it
got it.
Coke svn-bisect?
purl svn-bisect is interactive and modeled closely after git-bisect (and is written by infinoid). the other is intended to be run on autopilot.
Coke svn-bisect is also search.cpan.org/search?query=svn-bi...p;mode=all 16:16
purl okay, Coke.
masak Infinoid: search.cpan.org/dist/App-SVN-Bisect.../Bisect.pm -- I'm reading the synopsis and the documentation of the C<new> method. still not too clear on what to give as a value for $actions. 16:21
hm, the ACTION METHODS section below that mostly cleared it up... 16:22
still, there is some confusion in the Pod between Action, $action, and %actions. 16:23
hejki i'm trying to implement the stat(2) S_ISLNK-macro in parrot. i've added it in include/parrot/stat.h heikki.mehtanen.fi/parrot_patch/patch2.txt and in config/gen/platform/generic/stat.h heikki.mehtanen.fi/parrot_patch/patch1.txt i get no errors, but calling stat filename, 4 doesn't WAI. what am i missing here? 16:24
Coke masak; why read the docs of the module and not the command line util? 16:28
(the one just exists to support the other)
msg masak why read the docs of the module and not the command line util itself (for svn-bisect) (the one just exists to support the other) 16:29
purl Message for masak stored.
dalek rrot: r43279 | coke++ | trunk/PLATFORMS:
Fix typo in filename.
 17:12
Coke hejki: sorry, no clue here. 17:21
17:38 theory joined 17:44 davidfetter joined 18:07 joeri joined
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#31355), fulltest) at r43279 - Ubuntu 9.10 i386 (gcc with --optimize) 18:40
18:43 Zak joined 18:50 iblechbot joined 19:19 mikehh joined 19:29 bacek joined 19:38 mariano__ joined
cotto good morning 19:53
bacek cotto :) 19:57
19:57 mikehh joined 19:58 masak joined
bacek Ooookey... Current GC design is terribly awfully bad. 20:06
Coke don't mince words, bacek. What do you really think? 20:09
bacek Coke, I'm thinking about plugging Boehm GC. 20:10
And current design enforces usage of Fixed_Pool, manual separation of "constant" PMC, etc.
dalek rrot: r43280 | bacek++ | branches/boehm_gc:
Branch for pluggin in Boehm GC
 20:11
Coke bacek: good luck. =-) 20:14
bacek Coke, I don't need luck. I need time :)
cotto good time then 20:15
Coke rereads "recursive make considered harmful" and chuckles at the memory limits of current machines at the time. 20:39
(heck, many machine even have > 100MB!) 20:42
dalek rrot: r43281 | bacek++ | branches/boehm_gc (3 files):
Stub for detecting Boehm GC in Configure.pl
 20:43
ash_ Coke: but what if i want to build on my iPhone, it only has 256MB of ram, and i need part of that to keep my music player going /sarcasm 20:53
Tene h8 recursive make 21:02
Coke Tene: I'm killing it.
21:04 payload joined 21:06 payload joined 21:07 payload joined 21:15 mj41_ joined 21:20 payload joined 21:31 mj41_ joined
Coke are $%, $@, $<, etc. available in all the makes we're using? 21:34
I see we're using them in a few places already, so I'm guessing "yes"
GeJ Good morning everyone. 21:44
Coke anyone on atm that uses nmake to build parrot? 21:48
hurm. nevermind, I'm not ready.
21:53 mj41_ joined
bacek *incoming* 22:11
Ok, major GC problem is Pools. It should be other way around. GC use Pools for allocating objects. Not Pools use GC... 22:12
22:19 cognominal joined
dalek rrot: r43282 | bacek++ | branches/boehm_gc/config/auto/boehm.pm:
Set HAS_BOEHM_GC flag in auto::boehm.
 22:21
rrot: r43283 | bacek++ | branches/boehm_gc (3 files):
Almost blindly copy gc_inf into gc_boehm.
rrot: r43284 | bacek++ | branches/boehm_gc/src/gc/gc_boehm.c:
Actually build with Boehm GC.
 22:22
rrot: r43285 | bacek++ | branches/boehm_gc/src/gc/gc_private.h:
Add BOEHM_GC as available GC subsystem.
rrot: r43286 | bacek++ | branches/boehm_gc (2 files):
Add Parrot_gc_switch function to switch GC in "run-time"
rrot: r43287 | bacek++ | branches/boehm_gc/compilers/imcc/main.c:
Implement parsing --gc CLI option and switching GC.
rrot: r43288 | bacek++ | branches/boehm_gc/src/gc/gc_boehm.c:
Use GC_FREE instead of system's free in gc_boehm.
rrot: r43289 | bacek++ | branches/boehm_gc/src/gc/api.c:
Reinitialize pools in Parrot_gc_switch.
Tene bacek: is that branch "runs and passes some tests" or "omg experimental"? 22:26
bacek Tene, it crashes badly atm.
Tene Okay, thx 22:27
22:29 mj41_ joined
nopaste "bacek" at 122.110.43.111 pasted "Interesting. Boehm GC is slower with naive implementation..." (13 lines) at nopaste.snit.ch/19178 22:30
dalek rrot: r43290 | bacek++ | branches/boehm_gc/compilers/imcc (3 files):
Fix IMCC to use mem_sys_free
 22:38
rrot: r43291 | bacek++ | branches/boehm_gc/src/gc/gc_boehm.c:
Enable incremental GC
rrot: r43292 | bacek++ | branches/boehm_gc/src/gc/alloc_memory.c:
Use Boehm GC in mem_sys_* functions
22:40 mikehh joined 22:46 mj41_ joined 22:48 ttbot joined 22:51 solarion joined 23:01 bacek joined 23:04 cotto joined 23:21 kid51 joined 23:28 Zak joined
dalek tracwiki: v120 | fperrad++ | Languages 23:31
tracwiki: update Forth
tracwiki: trac.parrot.org/parrot/wiki/Languag...ction=diff
23:46 davidfetter joined
dalek rrot: r43293 | plobsing++ | branches/pmc_freeze_cleanup/src/pmc_freeze.c:
reuse id_list on thaw for finish_list, reducing garbage created during thaw
 23:59

Powered by Cro and the Raku® Programming Language.

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

Thank you!