Camelia, the Perl 6 bug

IRC log for #parrot, 2008-08-17

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:10 AndyA joined #parrot
00:15 apeiron joined #parrot
01:48 Whiteknight joined #parrot
01:48 Andy joined #parrot
01:57 dalek r30272 | rgrjr++ | trunk:
01:57 dalek : [typos] Neither pdd24_events.pod nor pdd25_concurrency.pod is in draft.
01:57 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=30272
01:58 zostay joined #parrot
02:04 Ademan joined #parrot
02:23 dalek r30273 | rgrjr++ | trunk:
02:23 dalek : [NEWS] Flesh this out a bit.
02:23 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=30273
02:42 Theory joined #parrot
02:45 buildbot joined #parrot
03:25 buildbot joined #parrot
03:29 dalek r30274 | jkeenan++ | opsrenum:
03:29 dalek : Rename sample files to conform to coding standards (no multiple-dot filenames) and change citations of those files inside 05-ops_renum_map_file.t.
03:29 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=30274
03:31 dalek r30275 | jkeenan++ | opsrenum:
03:31 dalek : Add coda to conform to coding standards.
03:31 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=30275
05:20 Theory joined #parrot
05:37 Psyche^ joined #parrot
07:05 Ademan joined #parrot
08:06 dalek r30276 | fperrad++ | trunk:
08:06 dalek : [Lua]
08:06 dalek : update NEWS for 0.7.0
08:06 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=30276
08:55 bacek joined #parrot
09:20 jkva_ joined #parrot
09:33 Whiteknight joined #parrot
09:51 Ademan joined #parrot
09:57 bacek joined #parrot
10:05 bacek_ joined #parrot
11:03 Zaba joined #parrot
11:23 masak joined #parrot
12:08 Casan joined #parrot
12:32 Infinoid joined #parrot
12:39 rurban joined #parrot
13:07 tetragon joined #parrot
13:42 Whiteknight joined #parrot
13:52 dalek r30277 | Whiteknight++ | gsoc_pdd09:
13:52 dalek : [gsoc_pdd09] update to trunk from r30235
13:52 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=30277
14:03 tetragon joined #parrot
14:10 kid51 joined #parrot
14:11 Debolaz joined #parrot
14:28 tetragon joined #parrot
14:56 dalek r30278 | Whiteknight++ | trunk:
14:56 dalek : [GC] Remove mention of GC_no_trace_volatile_roots from PDD09 and include/parrot/dod.h. The flag is not used anywhere in Parrot, and is not documented well enough to be implemented at a later time.
14:56 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=30278
14:59 Andy joined #parrot
15:02 jan joined #parrot
15:10 dalek r30279 | jkeenan++ | trunk:
15:10 dalek : Per http://rt.perl.org/rt3/Tic​ket/Display.html?id=57958, make sure that this config step leaves no 'test*' files behind in the top-level directory after Configure.pl has run.
15:10 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=30279
15:15 tetragon joined #parrot
15:37 dalek r30280 | jkeenan++ | trunk:
15:37 dalek : Per http://rt.perl.org/rt3/Tic​ket/Display.html?id=58036, make sure this config step cleans up after itself properly.
15:37 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=30280
15:41 kid51 Is there anyone on channel who could evaluate the C code in http://rt.perl.org/rt3/Ticket/Attachme​nt/459206/212356/auto.sizes.test3.txt -- specifically in proposed new probe file config/auto/sizes/test3_c.in ?
15:43 nopaste "kid51" at 68.237.14.234 pasted "Is this C code acceptable?" (19 lines) at http://nopaste.snit.ch/13819
15:44 s1n kid51: sure, but size_t isn't technically guaranteed to be any kind of integer type
15:46 s1n it's usually safe to assume that unless you're dealing with highly portable code (most libc implementations are unsigned ints iirc)
15:54 s1n also, your comment indicates you're trying to determine 64bit values? you might try shifting to find the sign bit
15:55 kid51 s1n  Thanks.  My aims are more modest here; I don't claim to know anything about sizes.  I'm just trying to fix some code that was leaving behind C test probes in the top-level directory after running 'perl Configure.pl'.
15:56 kid51 I.e., I'm hoping not to change the functionality that's there.  I just want it to be called in a cleaner, more standard way.
15:57 kid51 see the part of the patch that pertains to config/auto/sizes.pm.  I'll paste that as well.
15:58 nopaste "kid51" at 68.237.14.234 pasted "config/auto/sizes.pm: Cleaner way of getting hugefloatval" (37 lines) at http://nopaste.snit.ch/13820
15:59 kid51 In other words, this is a "do no harm" patch rather than a "change configuration process" patch.
16:10 Ademan joined #parrot
16:17 desertmax joined #parrot
16:29 jkva_ joined #parrot
16:38 rurban t/pmc/namespace_65.pir hangs for me
16:43 kid51 rurban:  Is that what's been reported in http://rt.perl.org/rt3/Tic​ket/Display.html?id=57668
16:56 smash joined #parrot
16:56 smash hello everyone
17:07 kid51 Since purl doesn't say it anymore, I'll say it:  Howdy smash you fine fellow!
17:08 rurban_ joined #parrot
17:12 tetragon Hrm... Configure.pl isn't finding my readline install and is instead tripping over editline
17:12 tetragon Time to find that patch of mine and try using it
17:13 tetragon Actually, that patch wouldn't fix this
17:13 rurban kid51: #57668 mentions a segfault. my test just hangs.
17:14 rurban $ ./parrot t/pmc/namespace_65.pir
17:14 rurban bar
17:14 * tetragon grumbles some more about OS X's editline
17:14 rurban 2
17:14 rurban <hang>
17:15 rurban I'd like that to be skipped as chromatic also voted
17:15 rurban (I'm on cygwin btw)
17:26 tetragon ld can't find my real readline install in /usr/local/lib (without modifying search paths) because it finds "readline" first in /usr/lib and terminates with an error over that unimplemented symbol in editline.
17:31 kid51 tetragon:  Add any findings to http://rt.perl.org/rt3/Tic​ket/Display.html?id=52212.
17:32 kid51 At the Parrot buildfest at YAPC, I tried to get people to try out that patch and give us the feedback that would be needed to commit it -- but we ran out of time on that one (though we got, IIRC, one of your other patches applies).
17:33 kid51 rurban:  Is your problem related to what Coke reported in http://rt.perl.org/rt3/Tic​ket/Display.html?id=57824 ?
17:34 rurban No. Coke's report also has segfault. Maybe my "hang" is a long gc, because it times out eventually, but not always.
17:35 kid51 ok.  Sounds like this is basis for new RT.
17:35 kid51 On my 2 OSes, I have not experienced the problem, IIRC.
17:36 tetragon kid51: The patch in #52212 wouldn't help due to changes in the flags since then.  It's now failing at linktime not runtime thanks to not having '-undefined dynamic_lookup' set.
17:36 rurban Or is it continues eating memory? nope. Just burning cpu. MEM=0.3% for the whole time.
17:36 tetragon And only five crashes (excluding the HCF test) for make test on my 64-bit build
17:36 kid51 tetragon:  Would that be a consequence of the changes we've made in config/init/hints/darwin.pm?
17:36 tetragon kid51: Quite likely.
17:38 tetragon kid51: The way around the problem now that doesn't involve breaking system symlinks is to check for readline without including /usr/lib and /usr/include in the search paths
17:38 tetragon (or to place everything else, such as /usr/local/lib,  first)
17:39 kid51 Well, patches welcome.  I don't claim to understand these issues well, so any explanation you can give in the RT would be welcome.
17:40 tetragon Hrm... -undefined dynamic_lookup is still mentioned in config/init/hints/darwin.pm, but it wasn't in any of the commands shown when I ran the readline step verbosely
17:40 kid51 You might want to configure with --configure_trace, then use Parrot::Configure::Trace to track the evolution of a particular value in Parrot::Configure object over the steps.
17:42 * kid51 would welcome a "cc for dummies"
17:42 tetragon kid51: Which cc?
17:43 tetragon kid51: Some of the IBM redbooks I worked with had comparison charts between a load of different cc implementations
17:43 kid51 Well, gcc as prime example.  But the things I don't understand about compilation I think lie at the next more general level.
17:44 kid51 I know there have been plenty of discussions about things like:  "this flag should go before that flag" or "this lib has to precede that one".  But I don't understand the context.
17:46 kid51 At the same time, I'm suspicious of things that unnecessarily complicate the code in our config system, since that makes it more unmaintainable over the long run (e.g., triggers).
17:46 particle1 joined #parrot
17:48 desertmax_ joined #parrot
17:54 tetragon Erm, according to what I'm seeing from Parrot::Configure::Trace, none of the values I'm seeing being set on line 42 of config/init/hints/darwin.pm are really being set
18:00 tetragon Erm, used the wrong value in my script for that
18:01 tetragon Looks like the value got changed at inter::progs
18:04 nopaste "tetragon" at 76.10.153.9 pasted "readline with some configure_trace output" (227 lines) at http://nopaste.snit.ch/13821
18:05 Limbic_Region joined #parrot
18:08 rurban interesting, my namespace_65 hang only appears with --optimized, with the debug build it works okay. Now this deserves a fresh ticket.
18:09 kid51 Have to leave now; will check back this evening.
18:13 NotFound rurban: not sure, but I think that is the test that is eating a lot of memory from some days ago.
18:14 rurban mem is 0.3%, cpu is 100%
18:14 rurban there is some optimisation bug
18:14 NotFound In 32 or in 64?
18:14 rurban 32
18:14 rurban cygwin
18:14 rurban jit
18:15 rurban I'll try with different runcore's now. But first I'll try to get an optimized libparrot.dll with symbols
18:15 NotFound Maybe is not the same test, I will check more carefully.
18:17 rurban No, it's the same test which segfaulted for coke and francois http://rt.perl.org/rt3/Tic​ket/Display.html?id=57668 http://rt.perl.org/rt3/Tic​ket/Display.html?id=57824
18:31 AndyA joined #parrot
18:39 rurban I summarized the upcoming cygwin parrot-0.7.0 at http://use.perl.org/~rurban/journal/37217
19:15 Auzon joined #parrot
19:50 davidfetter joined #parrot
20:07 Auzon joined #parrot
20:09 Auzon joined #parrot
20:14 dalek r30281 | chromatic++ | trunk:
20:14 dalek : [IMCC] Fixed compiler warnings in IMCC optimizer.
20:14 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=30281
20:17 dalek r30282 | chromatic++ | trunk:
20:17 dalek : [config] Fixed Solaris hint probes to respect parallel testing (RT #57900,
20:17 dalek : Andrew Johnson, Andy Dougherty).
20:17 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=30282
20:20 dalek r30283 | rgrjr++ | trunk:
20:20 dalek : Rename to match the new PMC class name (as required by
20:20 dalek : t/distro/test_file_coverage.t).  See RT#58046 for details.
20:20 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=30283
20:27 dalek r30284 | chromatic++ | trunk:
20:27 dalek : [PDD] Fixed PDD linewrapping at 78 characters, per codingstd test.
20:27 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=30284
20:37 dalek r30285 | rgrjr++ | trunk:
20:37 dalek :    + Two overlooked updates.
20:37 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=30285
20:51 Theory joined #parrot
21:04 Whiteknight joined #parrot
21:36 Theory joined #parrot
21:41 gmansi_ joined #parrot
21:48 gmansi joined #parrot
21:50 dalek r30286 | chromatic++ | trunk:
21:50 dalek : [PMC] Fixed co-recursion bug in NameSpace, when iterating through its contents.
21:50 dalek : The problem is that, when the key is a Key and not a String, sometimes there's
21:50 dalek : an infinite loop when the Key isn't a string, integer, or PMC key -- the
21:50 dalek : default path for the switch in key_string() calls the get_string() vtable entry
21:50 dalek : on the key, which calls key_string(), which....
21:50 dalek : The short-term solution is to call the parent PMC (Hash) implementation of
21:50 dalek : get_pmc_keyed() and return the results, if they're accurate.  This fixes the
21:50 dalek : crash and keeps Rakudo working.  Replacing the body of the vtable entry with
21:50 dalek : the SUPER() call breaks Rakudo.
21:50 dalek : The long-term solution is to fix the Key PMC, which is poorly designed and
21:50 dalek : poorly implemented.
21:50 dalek : See RT #57668 and RT #58040, for starters.
21:50 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=30286
21:55 dalek r30287 | chromatic++ | trunk:
21:55 dalek : [distro] Updated NEWS per my work for this release.
21:55 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=30287
21:56 gmansi joined #parrot
22:06 gmansi joined #parrot
23:02 teknomunk_ joined #parrot
23:11 kid51 joined #parrot
23:21 s1n joined #parrot
23:22 particle joined #parrot
23:27 dalek r30288 | jkeenan++ | trunk:
23:27 dalek : Shorten name of one file to meet coding standards.  Shorten names of two of its siblings for consistency.
23:27 dalek diff: http://www.parrotvm.org/svn​/parrot/revision?rev=30288
23:31 Zaba_ joined #parrot
23:34 tetragon kid51: inter::progs is clobbering the values set bin config/init/hints/darwin.pm with the ones on the command line for my linker flags.  This is also how the -L/usr/local/lib from perl 5 isn't making it through to the later tests
23:38 Whiteknight joined #parrot
23:50 kid51 tetragon:  I'm not surprised.  How should we proceed (in a way that keeps all variants of OS X happy)?
23:55 tetragon Not sure.  System perl users with no command line flags aren't affected, thanks to how integrate works
23:56 kid51 This makes me wonder whether *my* command-line flags are getting clobbered.
23:56 tetragon And I would need a 64-bit ppc to check for one of the conditions that I want to
23:57 tetragon Although, I may be able to do the check through -arch ppc64 flags
23:57 kid51 Refresh my memory.  When you used P::C::Trace, which value were you tracing?
23:57 tetragon linkflags
23:58 kid51 Let me try that on my iBook (10.4;ppc); haven't tried it in months.
23:59 tetragon I remember ppc being a bit picky about -undefined settings
23:59 tetragon (my builds)

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

Parrot | source cross referenced