Parrot 2.0.0 "Inevitable" released! | parrot.org | Priorities: merge branches, remove deprecations | Roadmap: icanhaz.com/parrotroadmap | Latest modified TT's: icanhaz.com/parrotbugs
Set by moderator on 9 February 2010.
Whiteknight chromatic: TT #1443 is sufficiently vague. 00:07
can't imagine I will get far with it without further info 00:08
jonathan: ping 00:09
dukeletoz == bizarro dukeleto? 00:10
rakudo doesnt even seem to build 00:18
fuggedaboudit 00:21
purl I fuggodaboudit
Whiteknight eh tony wes got a wiseguy
00:47 bacek joined
Austin Oi. 00:49
Can someone explain to me why Parrot/Test.pm always runs parrot by cd'ing to the binary dir? 00:50
chromatic No idea. 00:53
Austin Nor me. It's a problem for reuse, since passing CD=>some-other-dir apparently overrides the binary-dir location, and ./parrot is never found. 00:54
bacek o hai 00:57
Whiteknight you know Austin, if you found half as many solutions as you found problems, you would be our MVP 00:58
Austin :)
Whiteknight, every time I report a problem it means I've either found a solution or a workaround for the last one. 00:59
Since N > 2 in this case, I have found *more* than half as many.
Whiteknight okay, fine. Next time I see you, I'll give you the game ball
Austin This after you've spent the evening buggering? Keep your balls, sirrah. 01:00
Whiteknight urg, this rakudo error is already a pain in the ass
dalek rrot: r43947 | bacek++ | trunk (2 files):
Split parseflags into two functions. minimal version doesn't use any parrot subsystem and used to parse flags required for initializing interpreter
 01:07
TT #1436 closed by bacek++: Crash on parsing unknown command line arguments. 01:10
01:30 bacek joined 01:47 jsut joined 02:31 bacek joined 02:44 bacek joined, theory joined, mikehh joined, sri joined, cotto_work joined, jan joined, confound joined, TonyC joined, hicx174 joined, jhelwig joined, PacoLinux joined, eiro joined, wagle joined, NotFound joined, athomaso1 joined, Hunger joined, redbrain_ joined, treed_ joined, tewk joined, frodwith joined, spinclad joined, baest joined, elmex joined, dalek joined, jjore joined, Infinoid joined, preflex joined, integral joined, cotto joined, kvorg joined, szabgab joined, nopaste joined, Coke joined, Util joined, dukeleto joined, cxreg joined, Essobi joined, slavorgn joined, workbench joined, GeJ joined, rhr joined, leto joined, solarion joined, KatrinaTheLamia joined, ingy joined, Tene joined, ascent joined, hudnix joined, bacek_at_work joined
Austin Hmm.. it would be 'nice' if this segfault were persistent. 03:31
03:35 janus joined
nopaste "bacek" at 122.110.37.173 pasted "Wallpapering TT#1443" (13 lines) at nopaste.snit.ch/19613 03:40
dalek TT #1444 created by doughera++: Test failures in t/run/exit.t 03:55
kapo: 682822a | austin++ | (17 files):
Tweaked globals to prevent init problems.
 04:01
kapo: 4219dc0 | austin++ | t/ (9 files):
Added test cases for basic KRT0 loading function.
04:19 cognominal joined 04:49 cognominal joined 04:59 preflex joined
shockwave What is the exit op-code for quitting a script? I have tried 'quit' and 'exit', among others. 05:22
Austin What language? 05:25
purl language is horrifyingly verbose
Austin Ultimately, parrot has the exit opcode. If you're coding pir, just do exit <integer>. If you're doing nqp, try pir::exit(<expr>) 05:26
shockwave I'm doing PIR. I tried 'exit <int>' and it worked. 05:52
06:26 theory joined 06:39 kurahaupo joined
Austin Here comes my kiwi friend; 07:01
I'm alone in #parrot no more!
A short poem: Hi, Ku!
07:02 cognominal joined
dalek kapo: b00cfd6 | austin++ | (5 files):
Checkpointing before notest branch.
 07:39
08:19 bacek joined 08:21 patspam joined 08:51 iblechbot joined 08:59 cognominal joined
Austin haiku? 09:00
purl haiku is haiku-os.org/ for making an open-source BeOS. Or: Some gumballs are great / But as far as gumballs go / Tim's balls are the best. or arstechnica.com/open-source/news/20...f-beos.ars
dalek rrot: r43948 | mikehh++ | trunk/src/pmc/imageio.pmc:
fix codetest failure - line length
 09:51
rrot: r43949 | mikehh++ | trunk/src/main.c:
fix codetest failure - there should be at least one space between a C keyword and any subsequent open parenthesis
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#32190), fulltest) at r43949 - Ubuntu 9.10 amd64 (gcc with --optimize) 10:14
All tests PASS (pre/post-config, make corevm/make coretest, smoke (#32191), fulltest) at r43949 - Ubuntu 9.10 amd64 (g++ with --optimize) 10:47
11:36 patspam joined 11:38 patspam joined 11:53 woosley joined 12:04 joeri joined 12:10 kurahaupo joined 12:33 ruoso joined
dalek rrot: r43950 | gerd++ | trunk/config/gen/makefiles/root.in:
This update is to fix the make target "hello". I looked back until 0.8.2

will not recognized like it is used so I take pbc_to_exe.		 12:50
12:59 woosley1 joined, woosley1 left 13:09 cognominal joined 13:21 payload joined 13:54 Whiteknight joined
Whiteknight good morning #parrot 14:08
14:37 ruoso joined
dalek rrot: r43951 | whiteknight++ | branches/op_pmcs/src/pmc (2 files):
opcode can only be created from OpLib, not directly
 14:46
rrot: r43952 | whiteknight++ | branches/op_pmcs/src/pmc/opcode.pmc:
opcode can only be initialized once
15:08 cognominal joined
Whiteknight build appears to fail on x86-opensolaris 15:27
15:33 bacek joined 15:38 Psyche^ joined
nopaste "pmichaud" at 66.25.4.52 pasted "bnot opcode in parrot 2.0.0" (9 lines) at nopaste.snit.ch/19615 15:44
"pmichaud" at 66.25.4.52 pasted "bnot opcode in parrot trunk" (10 lines) at nopaste.snit.ch/19616 15:47
15:54 tetragon joined 15:57 theory joined 16:04 fperrad joined
dalek rrot: r43953 | whiteknight++ | trunk/src/ops/bit.ops:
fix the bnot_p_p opcode to preserve the same semantics as it had before vtable_massacre branch
 16:08
rrot: r43954 | jonathan++ | trunk/t/op/arithmetics.t:
[t] Test to cover bnot_p_p regression that whiteknight++ just fixed.
 16:24
16:27 lucian joined 16:33 fperrad_ joined 16:34 jan joined, ash_ joined 16:41 kid51 joined 16:45 payload joined
kid51 seen gerd? 16:50
purl gerd was last seen on #parrot 60 days, 20 hours, 16 minutes and 5 seconds ago, saying: Coke: the release in on the ftp-server I am still working on the website [Dec 15 20:33:50 2009]
16:56 AndyA joined 17:11 AndyA joined
mikehh kid51: ping 17:43
dalek kudo/master: abc9b2e | jonathan++ | docs/ROADMAP:
Move five completed, top-priority ROADMAP items to the completed section, now ng is master. :-)
 17:44
kid51 mikehh: pong 17:49
mikehh kid51: was under the impression that make coretest was essentially the same as the runcore tests from make fulltest, but it also now runs src and run tests 17:50
kid51 Hmm, I'll have to check that out. Am not as familiar with make coretest
It's documented as: Run the minimal 'core functionality' suite. 17:51
Coded as: # "core tests" -- test basic functionality but not ancillaries 17:52
coretest : corevm
$(PERL) t/harness $(EXTRA_TEST_ARGS) --core-tests
mikehh ah so, it used to be the same, bit it changed a while back - not exactly sure when - who generally looks after that?
kid51 It's a collective responsibility; there's no one lord and master 17:53
mikehh the thing is that the core tests from fulltest used to be the same but now are, for example: 17:55
testb : test_prep
purl test_prep is the 'make' target on which 'make test' immediately depends and is currently equivalent to 'make'
mikehh $(PERL) t/harness $(EXTRA_TEST_ARGS) -b $(RUNCORE_TEST_FILES)
kid51 see lines 61-78 of lib/Parrot/Harness/DefaultTests.pm
This may be a case where our terminology has gotten more confusing over the years. 17:56
We are perhaps overloading the term 'core'. 17:57
We have these things called runcores (which I do not really claim to understand very well). In make fulltest, we test each of these runcores in succession. 17:59
But then, I suspect, we also use the term 'core' in its more generic sense: the most essential thing to test
IIRC, about two years ago Allison created make corevm and make coretest to be the quickest possible way of testing changes to the most essential code in Parrot, i.e., something quicker and more to the point than 'make test' -- but not intended to be the definitive test suite. 18:00
mikehh kid51: the runcores are different ways of running the ops for example -S uses a giant switch statement and -r (not really a runcore) generates and runs the .pbc 18:01
kid51 Note how similar these two phrases are: run the tests which test the core of Parrot. run Parrot's runcore tests
mikehh: okay, but my point is that 'make coretest' is supposed to be an accompaniment to 'make corevm' 18:03
that, on the assumption that I'm recalling 2 years ago discussion correctly
mikehh I started running make corevm/make coretest as part of my test run after dukeleto/plobsing found that some of those tests failed while make test ran ok 18:04
kid51 Hmm. Well, we should try to understand how that can possibly come to pass, as it suggests a defect in the structure of our testing regimen. 18:05
mikehh and I remember when testing one of the calling conventions branches codetest passed but not test 18:06
kid51 Well, in those branches all bets are off!
But my assumption was that 'make coretest' would be a proper subset of 'make test'.
mikehh I think it was one of our first online hackathons - getting that working 18:07
kid51 In which case, anything that PASSed during 'make coretest' should PASS on 'make test' -- Unless something got added during 'make' that wasn't part of 'make corevm'
mikehh we had make corevm building but not make 18:08
kid51 Yes, I can easily see how, while working in a branch that touches many source code files, you could get 'make corevm' working. 18:09
and you would want a way of *demonstrating* that corevm was working, i.e., 'make coretest'.
Are there any test files currently FAILing in 'test' that PASS in 'coretest'? 18:10
(i.e., is this an issue that we must fix before tuesday?)
mikehh at the moment all tests PASS for me (at least at r43949) 18:11
kid51 and for me as well
dalek kudo/master: e778f84 | (Martin Berends)++ | docs/compiler_overview.pod:
[docs/compiler_overview.pod] any excuse for a commit to the new master branch formerly known as ng
 18:12
mikehh I think the problem was (somewhere around r43000) that a couple of the coretest tests had dependencies on files built in make but not corevm - so I test for that now 18:14
we sorted it out then, but I still test for that possibility - I think it was something due to extensions to ok or like 18:19
like or something similar doesn't work with corevm 18:20
kid51 You mean, the testing function 'like'?
mikehh that was pir based tests
and yes the testing function 'like' 18:21
kid51 mikehh: It's good that you test for that possibility. But I don't think the distinction between "tests that get run in make coretest" and "tests that get run in make test" ought to hinge on whether a particular testing function is available. 18:24
Mis dos centavos.
mikehh kid51: it is just that it came up that that one of the 'like' type tests depended on regex stuff built later so that make test passed but make corevm/make coretest had failures 18:29
18:29 kurahaupo1 joined
mikehh it was the same test, but the necessary functionality had been built yet in make corevm - it has been fixed, but we need to wartch out for that, which is why I test for it 18:31
kurahaupo1 leaving for @dayjobs 18:32
18:32 plobsing joined
dalek rrot: r43955 | plobsing++ | branches/tt362:
creating branch to address issue raised in tt3562.
 18:51
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#32197), fulltest) at r43954 - Ubuntu 9.10 amd64 (gcc with --optimize) 19:04
dalek kudo/master: 4afc080 | (Martin Berends)++ | docs/announce/20 (3 files):
[docs/announce] restore the Lisbon #23 Seoul #24 and Minneapolis #25 announcements, uniejo++ for notifying
 19:09
kudo/master: b96c6ef | (Solomon Foster)++ | src/Perl6/Grammar.pm:
Add <=> to the grammar.
 19:56
20:20 GeJ joined
dalek kudo/master: ac581a1 | (Solomon Foster)++ | src/core/Any-list.pm:
Add $by argument to Any.min and Any.max.
 20:30
20:45 bacek joined
dukeleto 'ello 20:47
21:10 KingOfKarlsruhe joined 21:12 theory joined 21:16 chromatic joined 21:30 payload joined 21:45 bacek joined
bacek aloha 21:48
dukeleto o hai 21:52
bacek dukeleto, can you benchmark boehm branch please? :) 21:57
chromatic Hey buddy, there's a line! 21:59
dalek rrot: r43956 | bacek++ | trunk/src/gc/alloc_resources.c:
Change aligned_mem function to behave exactly same as
 22:07
purl dalek: that doesn't look right
chromatic That looks like more of a "fix" to me. 22:10
#if 0 ... #endif isn't particularly clean, even with a comment.
That said, any bigger or more invasive fix may need to wait until after the release. 22:11
bacek chromatic, I'm not sure why aligned_mem/aligned_string_size have different semantic
compact_pool uses aligned_mem, allocate_string_storage uses aligned_string_size. 22:12
chromatic Me neither. The compacting code has always seemed baroque to me. 22:13
I understand why alignment might matter there, but not what happens or how it happens.
bacek I can rip-put commented code if this fix doesn't affect other parts.
chromatic, alloc_resources.c:543
chromatic Mostly I want to see a cleaner commit before we close the ticket. A workaround for now is fine, if we don't let the ugly linger. 22:14
bacek In nutshell:
1. compact_pool calculate required new size based on individual pools.
2. Use memory based on individual Buffers.
During second step it calls C<aligned_mem> which behave differently to C<aligned_string_size> (and use more memory) 22:15
Which cause memory overrun
chromatic Is this because of the "store a reference count in the (aligned) INTVAL immediately preceding the buffer to account for COW?" design? 22:16
bacek nope
chromatic Too bad; that's an easy and obvious place to make a mistake.
bacek Buffers (can be) aligned by 8 bytes
Strings always aligned by 4. 22:17
And this is what my commit changed. Now we always align Buffers same as Strings.
chromatic I can't think of why there should be a difference. 22:18
bacek me either
Thats why I left old code commented out :)
dalek kudo/master: 6d40788 | jonathan++ | build/PARROT_REVISION:
Bump to a Parrot with a fix for various segfaults we've been seeing, thanks to bacek++.
 22:19
bacek $dayjob time 22:20
See you
dalek TT #1443 closed by jonathan++: Segfaults possibly caused by pool compaction bug 22:34
22:55 payload joined
dalek rrot: r43957 | plobsing++ | branches/tt362 (4 files):
eliminate thaw_ptr from ImageIO API. This makes ImageIO easier to use from PIR.
 22:56
Austin chromatic: 8 bytes is often an align requirement for floating point, and may be an align requirement for 64-bit cpus (I don't know about this last). Do we ever do anything with Buffers other than use them for IO? 23:13
chromatic Not to my knowledge.
Austin Ok. 23:14
23:16 AndyA joined 23:24 masak joined
masak I'm getting a failed assertion when trying to build Parrot: 23:25
gist.github.com/304336
chromatic masak, is this a realclean build? 23:31
masak yes, I always realclean before I build. 23:33
I have it on automatic.
but mberends++ just gave me the tip of doing 'rm -rf' on the whole parrot directory, so I'm trying that now. 23:34
(he says he had to do that.)
chromatic That sounds like a bytecode change.
masak has such a change happened recently? 23:35
chromatic The most recent in PBC_COMPAT is from 31 January. 23:36
masak then I was probably bitten by that, yes. 23:37
it works now.
is there any reason to still be using the 'install-dev' target, or should I be switching to just 'install'? 23:38
chromatic I *think* they're the same now, but I don't know for certain. 23:39
They're synonyms in the current Makefile.
masak ok, sounds like it doesn't matter much. :)

Powered by Cro and the Raku® Programming Language.

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

Thank you!