Parrot 0.9.1 Released | parrot.org/ | 458 RTs left!
Set by moderator on 18 February 2009.
cotto You're right, although Parrot would have been broken since my NCI ATTR conversion if that code ran during make test. 00:02
jonathan ah, fail... 00:03
cotto (If there were a test failure, I would have caught it and converted the Bound_NCI PMC too.) 00:04
00:09 AndyA joined 00:15 kid51 joined
NotFound I'm not sure what a NCI method call is supposed to be. Just a NCI function call that takes a PMC as his first argument? 00:17
rg kid51 :) i have a question for you. is there already a ticket somewhere documenting the test problems when parrot is using a different compiler from perl? 00:19
Whiteknight cotto: Bound_NCI is used 00:26
I need to find where, but I know it is
I don't know what it is *
kid51 rg: Not sure I understand you. Parrot ought to use any C-compiler ... though we have vastly more experience with GCC. What do you mean by "different compiler from perl"? 00:28
NotFound Is used in object.pmc 00:29
rg there are a number of tests in t/steps/ that appear to test the compiler that they get from perl's configuration (in my case gcc).
Whiteknight yeah, and in default.pmc
rg if that compiler can't be found (because i'm using sun's cc for parrot) then those tests fail. 00:30
kid51 Ah!
This has nothing to do with perl (i.e., nothing to do with the Perl 5 you use to run Configure.pl). 00:31
It has everything to do with your OS and your C-compiler.
Can you paste output of Configure.pl (including any command-line options you use with it)? 00:32
NotFound But do we have some test were an NCI call is used as method on an object?
rg Configure.pl works just fine. but the tests don't seem to pick up those options.
when doing a make test i mean.
kid51 Alright, can you paste result of 'perl Configure.pl --test'? 00:33
That will run all pre- and post-configuration tests, including t/steps/*.t.
rg sure, just a second 00:34
moderator Parrot 0.9.1 Released | parrot.org/ | 451 RTs left! 00:34
nopaste "rg" at 62.216.218.102 pasted "perl Configure.pl --test output" (181 lines) at nopaste.snit.ch/15703 00:36
NotFound I inserted an 'abort()' in the object.pmc usage of Bound_NCI and all test pass. So this feature is at least seriously undertested 00:39
kid51 rg: I'm not surprised at those failures. They reflect a long-standing problem we've had with Sun/Solaris. 00:42
rg: If you look at config/init/hints/solaris.pm, you will see that it differs significantly from the hints files for all other actively maintained OSes. 00:43
rg are you sure it's just solaris? i can't think of another platform where gcc wouldn't be available
Whiteknight NotFound, it does look like a very complicated case where it gets invoked 00:44
NotFound Whiteknight: I'll try to use it in embed.t
kid51 rg: The hints file for Solaris uses callbacks: Parrot::Configure::Data::settrigger(), gettrigger() and deltrigger(). 00:45
Whiteknight NotFound: Yeah, that looks like what's needed. You have to define the Class in C through an embedding, and then write the method in C 00:46
kid51 Those P::C::Data methods were written by Andy Dougherty, who also wrote the relevant code in config/init/hints/solaris.pm. The latter is the only hints file that uses those callbacks.
Whiteknight Although the way it does invocation looks pretty arcane, I'm not happy about that 00:47
NotFound Whiteknight: I'll try first to create a class in pir and use add_method with an nci sub
kid51 rg: What the callbacks do is, in a subsequent config step, set a trigger that goes off in a yet later config step.
rg: Which means that the config steps are not, in a certain sense, self-contained. Which means that tests such as those in t/steps/, which assume that the config steps are (more or less) self-contained, won't work. 00:48
rg: IIRC, Andy D's rationale for that approach was to be able to try out different C-compilers, including gcc, on Solaris. 00:49
rg: But it means that Solaris, in effect, is configured differently by Parrot from all other OSes. 00:50
rg: Andy D advocates this use of callbacks during configuration. And since (a) he's the principal author of Perl 5's Configure (or so I was told) and (b) for a long time he was the only person developing on Solaris, his opinion has to be given weight. 00:51
rg: But I have never been very happy with that approach, because it means that the configuration steps can never be fully tested on that OS. 00:52
Now, some (perhaps, a lot of) people will say, "As long as Configure.pl completes, why bother to test the individual configuration steps at all?" 00:53
I don't share that opinion, needless to say.
But it does mean that we no longer include the pre- and post-configuration tests in the set of tests covered by 'make test'. 00:54
And that's where things are likely to remain until we purge all the Perl 5 from the Parrot configure/build process and scrap Configure.pl entirely. 00:55
... unless someone (hint, hint) actually wants to be Parrot's porter to Solaris, a role that is currently unfilled. 00:56
rg well the easiest "fix" would be not to run the offending tests on solaris
heh, i guess i was asking for that ;) however my main motivation for trying to run parrot on solaris was to help out with a 64bit big endian platform for rurban. 00:57
kid51 True. And if you'd like to submit a patch for that, I'll give it serious consideration. As I said, I'm not surprised that it was those three config steps that experienced the failures.
dalek rrot: r36950 | allison++ | trunk/config/gen/makefiles/dynoplibs.in:
[build] Convert Makefile for generating dynamic opcodes to a proper

with hardcoded paths in it. Partially resolves TT #338, as part of TT #238.
kid51 rg: For the purpose of helping out rurban, I don't think you need to worry about those tests.
If Configure.pl gives you a Makefile that you can then use to build parrot, then that's all you need. 00:59
In general, we do not yet have "porters" who have/take overall responsibility for making sure that Parrot works on a given platform/OS combination.
rg well my idea was to run smolder tests on solaris, and that doesn't look too good when you constantly have failing tests. 01:00
kid51 Hmm, does that mean that 'make smolder_tests' still runs the configuration tests? I guess so. 01:02
Speaking of smolder tests, we've got some failures on Linux ... which is not good.
kid51 reconfigures and rebuilds on Linux. 01:03
rg if make test does, i would guess make smoke does.
kid51 Oh, I guess I thought that we had restricted 'make smolder_test' to the set of tests run by 'make coretest'. 01:04
rg i think the linux failure are a new test in embed.t that notfound added some hours ago. i was waiting for my own smolder to confirm ;)
kid51 rg: So we have two alternatives:
(1) Patch to exclude those failing step test files on Solaris. 01:05
NotFound rg: what linux?
purl it has been said that linux is linux is linux by another name. or 80% of the world's top 500 super computers right now and the number one embedded system broadcast.oreilly.com/2008/10/how-l...evice.html
rg smolder is only adding --archive --send-to-smolder as args to harness
kid51 (2) Additional 'make' target to do smolder report with only the tests performed by 'make coretest'.
NotFound: See smolder.plusthree.com/app/public_pr...ails/18282 01:06
shorten kid51's url is at xrl.us/behbbg
kid51 ... which smolder run appears to have used lots of optimization.
... as did smolder.plusthree.com/app/public_pr...ils/18282, probably by the same tester. 01:08
shorten kid51's url is at xrl.us/behbbg
01:08 braceta joined
NotFound kid51: strange, linux i386 is where I'm working 01:08
purl working is a good approximation.
kid51 NotFound: I'm merely reporting Smolder tests, such as this one: smolder.plusthree.com/app/public_pr...ails/18285 01:09
shorten kid51's url is at xrl.us/behbbt
kid51 My own Smolder report just started running. It'll take ~ 10 minutes. 01:10
rg i don't know what happened for 18285. that's a lot of failures.
kid51 Perhaps r36949 or r36950. 01:12
rg he picked up 36949. looks like you also need at least 36950
kid51 Probably there were some errors in 36949 that allison corrected in r36950. She used the same log message in both commits.
kid51 is testing at 36950. 01:13
rg kid51: so do you think i should open a ticket about the solaris test failures? 01:14
kid51 If you open such a ticket, will you take it? 01:15
kid51 is seeing test failures on Linux at 36950.
rg as far as excluding the tests, yes. i'm not sure i'll have the knowlege or time for a proper fix.
kid51 Many failures in t/dynpmc/*.t. 01:16
Well, we can collaborate on it. I don't have Solaris available for testing, so we'd have to rely on you for that. But I could code review any patch you submit. 01:17
rg i find it interesting, that the smolder 18285 didn't fail embed.t like 18282
kid51 It would mostly be along the lines of 'skip all tests if $^O =~ /solaris/i' (pseudo-code) 01:18
smolder.plusthree.com/app/public_pr...ails/18286
shorten kid51's url is at xrl.us/behbcq
kid51 seen allison? 01:19
purl allison was last seen on #parrot 5 days, 21 hours, 20 minutes and 50 seconds ago, saying: Infinoid: cool [Feb 18 03:58:28 2009]
rg kid51: i'll try again to have a closer look at the tests and get back to you in a few days 01:21
kid51 Thanks: Feel free to email me.
oops 01:22
rg ok, my smolder is done aswell :) 01:23
i'm failing the same 27 tests, but not embed.t. so i guess 18282 was a fluke
01:26 jimmy joined
kid51 creates trac.parrot.org/parrot/ticket/368 for those failures. 01:27
rg: Yes, do get back to me on that. We definitely need Smolder reports on Solaris, and we shouldn't allow these config step test failures to pose an obstacle to that. 01:28
rg is wondering if he should run smolder tests more than once per day. 01:34
it has to be allison's change in r36949. r36948 is fine. 01:35
kid51 rg: No particular harm in that. Some smokers do it hourly, but that's probably excessive. 01:36
NotFound Looks like the BoundNCI must be created in the add_method vtable function of the class pmc to be useful. 01:37
rg well there is such a thing as too much information ;)
kid51 presumes that allison will spot TT 368 ... and goes to dinner.
01:39 Theory joined 01:46 dukeleto joined 02:13 Theory joined
Whiteknight NotFound, you find a way to test BoundNCI? 02:35
03:08 Theory_ joined 03:18 japhb joined
dalek rrot: r36951 | jkeenan++ | failed to fetch changeset:
Add new 'make' target 'smolder_coretest'. Want to generate a Smolder report,

smolder_coretest'.		 03:30
kid51 See smolder.plusthree.com/app/public_pr...ails/18290 for an example of 'make smolder_coretest'. 03:32
shorten kid51's url is at xrl.us/behbsa
03:54 janus joined 04:14 davidfetter joined 05:03 masak joined 05:20 jimmy joined 05:35 Tene joined 06:04 galf joined, galf left 06:57 uniejo joined
dalek rrot: r36952 | allison++ | trunk/MANIFEST.generated:
[install] A few more PMC resource files needed by Partcl.
 06:59
07:12 alvar joined
dalek rrot: r36953 | fperrad++ | trunk/t/configure/034-step.t:
[makefile] initialize value before tests, fix TT #369.
 08:07
08:08 iblechbot joined 08:14 Ademan joined
dalek kudo: 1fa37c0 | (Moritz Lenz)++ | src/ (2 files):
first shot at implementing Str.split in Perl
 08:19
kudo: 215ad17 | (Moritz Lenz)++ | (4 files):
add Any.split(stuff) and Str.split(Str)
shorten dalek's url is at xrl.us/behcez
shorten dalek's url is at xrl.us/behce3
dalek kudo: 1fe71b8 | (Moritz Lenz)++ | src/setting/ (2 files):
work around Rakudo bugs to get Str.split working
kudo: fa51bd7 | (Moritz Lenz)++ | src/ (2 files):
implement Str.comb in pure Perl 6
kudo: 92d94aa | (Moritz Lenz)++ | Test.pm:
[Test.pm] don't autothread over ok(), is()

was involved - which is an implementation detail and should be hidden. Now no test function should autothread at all, which is usually what the test write expects.
shorten dalek's url is at xrl.us/behce5
dalek's url is at xrl.us/behce7
shorten dalek's url is at xrl.us/behce9
dalek kudo: 108dfb5 | (Moritz Lenz)++ | src/builtins/any-str.pir:
Str.index() should work when the search string is "0"
shorten dalek's url is at xrl.us/behcfb
dalek kudo: 0e7dd5e | (Moritz Lenz)++ | (3 files):
Merge branch 'master' into str_split_in_setting
shorten dalek's url is at xrl.us/behcfd
dalek kudo: 592bd42 | (Moritz Lenz)++ | src/setting/Str.pm:
fix previous merge, and remove workaround
shorten dalek's url is at xrl.us/behcff
dalek kudo: 2cc539b | (Moritz Lenz)++ | src/setting/Str.pm:
much saner handling of limits in Str.split
shorten dalek's url is at xrl.us/behcfh
dalek kudo: e074bf4 | (Moritz Lenz)++ | src/setting/Str.pm:
fix Str.split for zero-width matches
shorten dalek's url is at xrl.us/behcfj
jimmy wow!!
karma Moritz Lenz
purl moritz lenz has karma of 56
jimmy karma (Moritz Lenz)
purl (moritz lenz) has neutral karma
08:23 Ademan_ joined
dukeleto how can I get permission to create tickets on trac.parrot.org ? 08:38
moritz dukeleto: get an account where you also give trac your email address 08:40
dukeleto: and confirm that email address
dukeleto moritz: sweet deal, I just got the confirmation email and now I am creating a ticket :) 08:41
moritz: thanks for the clarification though, i was a bit confuse
moritz dukeleto: it *is* confusing, and badly documented 08:42
dukeleto I am trying to write a parrot test using pasm_output_is and I am getting : # attempt to access code outside of current code segment
# current instr.: '(null)' pc 12 (/Users/leto/svn/parrot/t/op/arithmetics_12.pasm:6)
what does that particularly mean? 08:43
test is here: pastey.net/108882 08:44
just created trac.parrot.org/parrot/ticket/370 08:56
cotto dukeleto, you're the third person to mention being confused by that in the last 2 days. 09:03
dukeleto cotto: yeah, the error message should really say "go verify your email so you can create a ticket" or somesuch 09:04
cotto I'm raising the issue tomorrow at the weekly #parrotsketch meeting. Thanks for mentioning it. 09:06
dukeleto no problem, gladly someone was here to push me in the right direction 09:07
cotto Someone often is. 09:08
dalek rrot: r36954 | fperrad++ | trunk/MANIFEST.generated:
[MANIFEST] re-sort
 09:19
cotto I hope that's not the last re-sort.
09:25 bacek joined 09:31 skv joined 09:43 szabgab joined 09:53 alvar joined
dalek rrot: r36955 | cotto++ | trunk/src/packfile.c:
[codingstd] remove extra space after if
 09:54
rrot: r36956 | fperrad++ | trunk/config/gen/makefiles/dynpmc.in:
[build] fix linking, add missing libraries (crypto & gdbm)
 09:57
09:58 s1n joined
dalek rrot: r36957 | cotto++ | trunk/src/pmc:
[PMC] switch Iterator to ATTRs
 10:01
rrot: r36958 | cotto++ | trunk/src/pmc/iterator.pmc:
[PMC] use PMCNULL instead of NULL in Iterator
 10:13
10:42 s1n joined
mikehh r36956 appears to break parrot on linux i386 - cc: libcrypto.lib: No such file or directory 10:46
from cc -o md2.so md2.o -L/usr/local/lib -Wl,-rpath=/home/mhk/parrot/blib/lib -shared -O2 -g -L/usr/local/lib -fPIC -L/home/mhk/parrot/blib/lib -lparrot libcrypto.lib 10:47
make[1]: *** [md2.so] Error 1 10:48
make: *** [dynpmc.dummy] Error 2 10:49
moritz same here 10:53
cotto It's nice when not me breaks something. 10:54
dalek rrot: r36959 | fperrad++ | trunk/config/gen/makefiles/dynpmc.in:
[build] fix configuration logic, != is not available
 11:04
11:05 bacek_ joined
dalek rrot: r36960 | fperrad++ | trunk/config/gen/makefiles/root.in:
[install] remove win32 specific
 11:08
rrot: r36961 | fperrad++ | trunk/MANIFEST.generated:
[install] A few more PMC resource files needed by Lua & WMLScript.
 11:16
mikehh my gmail has gone AWOL 11:18
builds ok - testing now 11:27
still getting dynpmc failures - smolder.plusthree.com/app/public_pr...ails/18298 11:34
shorten mikehh's url is at xrl.us/behco6
mikehh probably from r36949/r36950 11:38
dalek a: 7b09a8c | (Francois Perrad)++ | config/makefiles/pmc.in:
build with installed tree
 11:45
shorten dalek's url is at xrl.us/behctv
11:48 jimmy joined
dalek lscript: 8907885 | (Francois Perrad)++ | config/makefiles/pmc.in:
build with installed tree
 11:49
shorten dalek's url is at xrl.us/behct3
dalek rrot: r36962 | NotFound++ | trunk:
[examples] add embed example
 11:59
rrot: r36963 | fperrad++ | trunk/tools/dev/mk_language_shell.pl:
[install] refactor targets 'clean'
 12:04
NotFound Can someone take a look at examples/embed and give me some feedback?
mikehh BTW 10 more dynpmc tests passed between r36953 and r36960 12:11
sorry 11 - r36953 failed 27 and r36960 failed 16 12:13
12:15 jimmy joined
NotFound #IF(win32 and cc!=gcc):LIB_CRYPTO := -lcrypto 12:29
67\t#ELSE:LIB_CRYPTO := libcrypto.lib
WTF?
12:33 rg1 joined
dalek rrot: r36964 | NotFound++ | trunk/src/extend.c:
[cage] add missed ASSERT_ARGS
 12:36
rrot: r36965 | NotFound++ | trunk/include/parrot/packfile.h:
[cage] update packfile.h headerizing
 12:39
12:45 kid51 joined 13:13 bacek joined 13:20 Whiteknight joined
dalek rrot: r36966 | NotFound++ | trunk:
[cage] Fix Parrot_set_executable_name declaration
 13:39
Infinoid rg1: Thanks for the patch. I did the counting in a different way that won't result in undef warnings, and I made the output format consistent with the other plugins 14:04
mikehh The failing dynpmc tests all fail with Class 'foo' not found
for MD5, SHA etc 14:05
Infinoid rg1: I had also changed some other things related to this module (detection of changeset fetch timeout), which screwed up the indentation and made me merge it manually. I'm going to throw the whole lot up on my git server, to help that in the future
14:09 gryphon joined
mikehh actually t/dynpmc/foo.t passes so I shouldn't have used that as an example 14:19
dalek rrot: r36967 | NotFound++ | trunk/t/oo/subclass.t:
[t] DO oo/subclass.t 'set the value of a non-constant key PMC'
 14:20
mikehh t/dynpmc/md2.t, t/dynpmc/md4.t, t/dynpmc/md5.t, t/dynpmc/ripemd160.t, 14:21
14:22 dalek joined
mikehh t/dynpmc/sha.t, t/dynpmc/sha1.t, t/dynpmc/sha256.t and t/dynpmc/sha512.t fail with Class 'MD2'. 'MD4' etc not found 14:25
rg infinoid: thanks. i hope i'll find that if i think of anything else i'd like changed ;) 14:30
octo moritz: Is `parrot-dev at lists.perl.org' the correct list address? Sure it's not `parrot-dev at lists.parrot.org'? 14:40
moritz octo: it's lists.parrot.org, sorry if I said that wrong previously
octo moritz: That explains why the mail wasn't delivered ;) 14:41
lists.perl.org / x3.develooper.com doesn't even have a MX record.. 14:42
moritz right, the perl lists all go to perl.org directly
davidfetter does gerd pokorra hang out here?
dalek kudo: 70d9081 | jnthn++ | src/setting/Pair.pm:
Use $.key rather than self.key etc in Pair.pm; improvement suggested by moritz++.
 14:47
shorten dalek's url is at xrl.us/behdau
NotFound "a Polish programmer has succeeded in installing a copy of Windows 3.1 on his Nokia N95" 14:49
Wrong channel, sorry
jonathan Can he try installing Parrot on it next? ;-)
Infinoid rg: git://squawk.glines.org/botnix-plugins 14:52
NotFound Some day I'll try to build parrot for Nintendo DS
Infinoid NotFound: I'd smoke that
davidfetter hands Infinoid a lighter 14:53
15:05 jan joined 15:25 Theory joined 15:30 cas joined
dalek rrot: r36968 | NotFound++ | trunk/examples/embed/lorito.c:
[examples] add --start option to lorito
 15:46
15:50 PacoLinux joined 16:00 clunker9 joined 16:30 DietCoke joined, DietCoke left, DietCoke joined
DietCoke any drupal wizards here? 16:30
dalek rrot: r36969 | NotFound++ | trunk/examples/embed/lorito.c:
[examples] add several runcore options to lorito
 16:31
16:35 Tene_ joined
dalek lscript: 6ee6800 | (Francois Perrad)++ | (22 files):
reorg tree source
 16:40
shorten dalek's url is at xrl.us/behdn2
dalek lscript: a98ce86 | (Francois Perrad)++ | (4 files):
update paths in Makefiles
shorten dalek's url is at xrl.us/behdn4
dalek lscript: 0feb8bb | (Francois Perrad)++ | .gitignore:
update .gitignore
shorten dalek's url is at xrl.us/behdn6
dalek lscript: a7e17a1 | (Francois Perrad)++ | config/makefiles/ (3 files):
re-run tools/dev/mk_language_shell.pl (simply expanded variables)
shorten dalek's url is at xrl.us/behdn8
16:41 riffraff joined
Coke I added the parrot calendar to the main site. 16:44
(don't need drupal wizards any more. =-)
NotFound Drop all wizards
Coke waits for suggested calendar events to rolll in. 16:50
16:54 jan joined
particle 1171/60 17:07
purl 19.5166666666667
Infinoid Coke: nice. the height attribute on that calendar iframe is a little small tho, it cuts off the bottom here 17:08
it might be showing some extra text here, because I haven't set up a local timezone so it's explaining about GMT
particle Infinoid: here too 17:09
coke++
Infinoid looks good now 17:10
dalek rrot: r36970 | NotFound++ | trunk (3 files):
[core] make explicit the assumption about non-empty signatures in src/call/ops.c
 17:14
particle probably can't make parrotsketch, again 17:15
moritz neither can I 17:16
particle coke, others: i want to propose a 0-day bug policy. all smoke failures must be fixed within 24 hours, at least until 1.0
thoughts?
purl i like cheese
particle purl: you smell like roquefort 17:17
purl particle: huh?
PerlJam particle: must be fixed or ... ?
particle commit reverted
+ public flogging by purl
PerlJam so ... we smoke after each commit? 17:18
or, how do you know which commit broke things? 17:19
who is responsible for finding and reverting the commit?
particle it's not a perfect system. we know between which two commits on the same smoke platform something broke
the community is responsible
i'd love for someone to take a leading role, though 17:20
PerlJam um ... it sounds good to me except for that "community is responsible" part :) 17:22
particle well, the community is responsible for releases, code review, and commits. why not checking smokers and reverting bad commits? 17:23
PerlJam because all of those are community initiated events. reverting a commit that breaks a particular platform is not. 17:24
Infinoid here we go, applying cosmic numerological significance to the ascii string "1.0" again 17:25
17:25 geof joined
Infinoid if we're going to do something like that, I'd vote for doing it forever and ever after, amen. 17:25
particle Infinoid: this is forever-and-ever 17:26
Infinoid In that case, I have no problem with it :)
particle trunk shouldn't break smokers from now on.
you want to fail tests? do it in a branch
we can smoke branches, too
it'd be nice to have an integration branch just for that purpose
Infinoid and in fact, we probably should do so more often
17:28 mikehh joined
Infinoid "integration branch"? e.g. blead? 17:29
particle sure
Infinoid interesting idea 17:30
particle develop in blead, or in private branches then merge to blead, merge to trunk when smokes all pass, release from trunk 17:31
Infinoid in this case, trunk is really just a long-running prerelease branch 17:32
particle trunk smoke failures are reverted
yep
makes it something rakudo could target, for example
PerlJam sounds good 17:33
Infinoid things will get out of sync if you revert them from trunk, though. You'd have to maintain some kind of "waiting patches" queue out of tree to make that work
or revert them from both trunk and devel
particle well, it shouldn't happen, as merge to trunk should happen after green smoke from blead 17:34
PerlJam If the commit could automatically be reverted and a ticket created that has the patch and a note that it causes failure and on which platform, that might be nice :)
particle but perhaps that means there's poor blead smoke platform coverage
Infinoid Sure, it won't happen, except when it does. :)
particle perljam: indeed. automation is key to making this process work
Infinoid Oh, I like the autoticket idea 17:35
PerlJam particle: with sufficient automation, you can ignore what I said earlier about the community being responsible.
Infinoid and autoassigned to the committer
Of course, if the act of reverting it breaks something else, your script will have to be a little smarter 17:37
PerlJam Infinoid: and a bot could carp on #parrot about it too (It's too easy to imagine an actual parrot squawking "Infinoid broke the build on OSX!" or somesuch :-)
particle :) 17:38
dalek rrot: r36971 | NotFound++ | trunk/t/op/arithmetics.t:
[t] add rounding Inf/NaN test, TT #37 dukeleto++
 17:39
17:41 Theory joined 17:42 kj joined
particle hrmm. this plan won't work until smokers report build/config failures, too. 17:47
Infinoid you'll have to retrain everyone to talk about "trunk r12345" or "blead r12345", not just r12345 17:48
pmichaud wouldn't blead and trunk be in the same repo?
Infinoid yes, but trunk will have a subset of the changes in blead 17:49
moritz this sounds like a big infrastructure to solve a problem that isn't actually all that big
pmichaud oh, I see .
moritz the problem seems to be more of a social one - people don't want to revert other people's commits for social reasons 17:51
particle well, it's also a way we can get a release-ready-anytime culture 17:53
but i'm not so sure that's the highest priority 17:54
we could have trunk = blead, and stable = branches/release
so occasionally, when trunk passes all smokes, merge to branches/release 17:55
that tends to discourage last-minute pre-release commits, which imo is a good thing
17:56 wknight8111 joined
PerlJam particle: maybe make that part automatic too. 17:56
smoke at regular intervals and if all pass, automatically merge to a release branch
(that may not be possible though)
particle yep, need a good size list of platforms to smoke on 17:57
i'd like to talk to amazon about using ec2
pmichaud I chatted with amazon about it already. 17:58
the guy I talked to said it probably wouldn't be of much benefit to us if we're looking for platform variety.
because ec2 is all about platform sameness.
particle no, it's not 18:01
that might have been a chat some time ago
they're x86/x64 (no sparc), but have bsd, suse, rhel, win, etc 18:02
pmichaud okay, then that's changed since I chatted, yes. And yes, this was 12-18 months ago.
particle cygwin, debian/ubuntu
yeah, they've come a *long* way since then. 18:03
pmichaud anyway, at the time I talked to the amazon guy he said it's so easy (and relatively cheap) to set up an ec2 account that we should just do that.
particle yes, but i'd prefer free to cheap :) 18:04
pmichaud we could certainly do it that way as proof-of-concept, and then if we decide we need bigger support go for free.
particle yep
pmichaud rather than wait for free access before we see if it even works.
NotFound Or even better, that amazon pay us ;)
PerlJam NotFound: it could happen. 18:05
NotFound Errrr... donate, be careful with PR X-)
18:17 allison joined 18:20 Psyche^ joined
dalek rrot: r36972 | fperrad++ | trunk/config/gen/makefiles/dynpmc.in:
[build] all digest PMC are grouped
 18:29
cotto Coke, do you still have a Drupal question? 18:32
18:34 chromatic joined
cotto Coke, nm 18:34
chromatic masak, is RT #62730 really still a problem? 18:39
18:39 rurban joined
chromatic cotto, did you figure out the NCI struct val thing? 18:39
cotto nope 18:40
The conversion is fine, but I don't have a solution to the workaround in TT #365. 18:41
18:43 particle joined
cotto Util++ 18:44
api sanity ftw
chromatic Hm, I remember at PDS storing the memory allocated to hold the JITted function in the NCI's struct val. Is that it? 18:45
cotto That may be what's happening. 18:48
If I initialize PMC_struct_val, Parrot dies trying to execute whatever's at that value. 18:49
chromatic That was to fix a memory leak, where we never freed that memory.
There's code in the NCI PMC that checks for a JITted flag.
Somewhere in invoke, I think.
cotto yes 18:50
dalek rrot: r36973 | fperrad++ | trunk/config/gen/makefiles (2 files):
[makefile] refactor targets 'clean'
 18:58
particle nopaste? 19:03
clunker3 pasta.test-smoke.org/ or paste.husk.org/ or nopaste.snit.ch:8001/ or rafb.net/paste or poundperl.pastebin.com/ or paste.scsys.co.uk/
purl nopaste is at nopaste.snit.ch/ (ask TonyC for new channels) or rafb.net/paste or poundperl.pastebin.com/ or paste.scsys.co.uk/ or App::Nopaste or tools/dev/nopaste.pl or at www.extpaste.com/ or paste.scsys.co.uk (for #catalyst, #dbix-class, #moose and others)
particle ...i thought we did away with encapsulation in the great war of 89 19:05
chromatic Get off my lawn, grandpa. 19:06
diakopter when building on VC++9sp1 (cl.exe v15.00.30729.01), cl doesn't like the space between -Fo (cc_o_out) and the filename... (it ignores the option) 19:13
chromatic Maybe Microsoft will allow spaces in filenames one of these years.
cotto aim high 19:14
octo Hahahahaha ;)
diakopter well, cl -c -Fo"dyn lexpad.obj" DWIM, but cl -c -Fo dynlexpad.obj ... doesn't work
NotFound Maybe -Fo means -Fo"" 19:15
dalek tracwiki: v116 | allison++ | ParrotRoadmap
tracwiki: trac.parrot.org/parrot/wiki/Parrot...ction=diff
shorten dalek's url is at xrl.us/beheez
dalek tracwiki: v22 | cotto++ | PMCUnionDeprecationTasklist 19:19
tracwiki: add reason for UnionVal deprecation
tracwiki: trac.parrot.org/parrot/wiki/PMCUni...ction=diff
shorten dalek's url is at xrl.us/behefb
dalek tracwiki: v117 | allison++ | ParrotRoadmap 19:32
tracwiki: trac.parrot.org/parrot/wiki/Parrot...ction=diff
shorten dalek's url is at xrl.us/behegz
dalek tracwiki: v118 | allison++ | ParrotRoadmap
tracwiki: trac.parrot.org/parrot/wiki/Parrot...ction=diff
shorten dalek's url is at xrl.us/beheg3
dalek l: fd374b9 | (Francois Perrad)++ | (2 files):
re-run tools/dev/mk_language_shell.pl
NotFound Talking about windows, can someone take a look at examples/embed and check if it can build on windows?
shorten dalek's url is at xrl.us/beheg7
19:35 barney joined
dalek tracwiki: v119 | allison++ | ParrotRoadmap 19:40
tracwiki: trac.parrot.org/parrot/wiki/Parrot...ction=diff
shorten dalek's url is at xrl.us/beheh2
dalek tracwiki: v120 | allison++ | ParrotRoadmap 19:44
tracwiki: trac.parrot.org/parrot/wiki/Parrot...ction=diff
shorten dalek's url is at xrl.us/beheig
dalek rrot: r36974 | cotto++ | trunk/DEPRECATED.pod:
[DEPRECATED] add Bound_NCI and Slice to deprecated PMC list
diakopter all the '$(CC_OUT) ' need replaced with '$(CC_OUT)' in config/gen/makefiles/dynpmc.in and config/gen/makefiles/dynoplibs.in $(CC_OUT) should already have a trailing space on the configurations that require it. at least, it does on some of them. 19:45
dalek tracwiki: v121 | allison++ | ParrotRoadmap 19:48
diakopter should I file an RT bug report for that?
dalek tracwiki: trac.parrot.org/parrot/wiki/Parrot...ction=diff
shorten dalek's url is at xrl.us/behei9
dalek tracwiki: v122 | allison++ | ParrotRoadmap
tracwiki: trac.parrot.org/parrot/wiki/Parrot...ction=diff
shorten dalek's url is at xrl.us/behejb
dalek tracwiki: v9 | allison++ | ParrotLongTermRoadmap
tracwiki: trac.parrot.org/parrot/wiki/Parrot...ction=diff
shorten dalek's url is at xrl.us/behejd
chromatic It should be fixable without a patch, diakopter.
diakopter ( :P not by me since I haven't signed a CLA and such) 19:49
chromatic Yeah, but I'm testing it now. 19:51
diakopter oh, cool.
dalek rrot: r36975 | chromatic++ | trunk/config/gen/makefiles (2 files):
[config] Removed unnecessary space after CC_OUT expansion in dynops and dynpmc
 19:56
diakopter chromatic: neat, thanks. 20:00
NotFound: I can try the examples/embed when this finishes building
NotFound diakopter: ok, thanks
chromatic Hm, t/src/embed.t #5 segfaults for me. 20:03
NotFound chromatic: platform? 20:04
chromatic x86 Linux 32 bit
NotFound Works fine for me 20:05
dalek tracwiki: v123 | allison++ | ParrotRoadmap
tracwiki: trac.parrot.org/parrot/wiki/Parrot...ction=diff
tracwiki: v124 | allison++ | ParrotRoadmap
shorten dalek's url is at xrl.us/behemw
dalek tracwiki: trac.parrot.org/parrot/wiki/Parrot...ction=diff
shorten dalek's url is at xrl.us/behemy
chromatic 0xb7de535a in Parrot_io_STDOUT (interp=0xb7cef1f3) at src/io/api.c:647
647 {
(gdb) bt
#0 0xb7de535a in Parrot_io_STDOUT (interp=0xb7cef1f3) at src/io/api.c:647
#1 0x0804881c in hello ()
#2 0x095ff5af in ?? ()
#3 0xb7e1dcb3 in Parrot_NCI_invoke (interp=0x958b008, pmc=0xbfe19834,
next=0x96503bc) at ./src/pmc/nci.pmc:320
diakopter chromatic: actually same with '$(LD_OUT) ' in at least those two files... 20:06
Infinoid nice corrupt interp pointer 20:07
NotFound chromatic: which runcore?
chromatic slow core (standard) 20:08
diakopter chromatic: wait, here's more 20:09
dalek tracwiki: v125 | allison++ | ParrotRoadmap 20:10
tracwiki: trac.parrot.org/parrot/wiki/Parrot...ction=diff
shorten dalek's url is at xrl.us/behem8
dalek rrot: r36976 | cotto++ | trunk/docs/project/release_manager_guide.pod:
[docs] update release manager guide to reflect recent numbering scheme decisions
cotto allison, trac's milestone numbers have been updated. 20:11
allison cotto: excellent, thanks!
20:11 Theory_ joined
diakopter chromatic: same problem (both LD_OUT and CC_OUT) in tools/dev/mk_language_shell.pl 20:12
chromatic Is this the right place to fix it, rather than the spot in Configure that sets the cc_out variable? 20:13
I don't doubt that it needs fixed. I'm just curious if it's a one-versus-many fix. 20:14
NotFound chromatic: I think a signature is wrong, let me check... 20:15
diakopter yes; each platform/configuration/hint should define the ones that (optionally) need a space as having a space, and the ones that cannot accept a space (ms vc15) as without the trailing space
chromatic I thought that might be the case. Thanks for confirming. 20:16
diakopter it's defined correctly for msvc (without the space)
yw
rurban NotFound: test embed on win32 since which revision? I have r36906 and don't want to update now 20:17
dalek rrot: r36977 | NotFound++ | trunk/t/src/embed.t:
[t] fix a nci signature in embed.t
 20:18
NotFound rurban: don't worry, no urgency
rurban I just bought 2GB ram for my laptop so I can run several vmware images also here 20:19
diakopter mistakenly bought a laptop with a bios editor that won't allow Intel VT to be activated. :| 20:20
won't make that error again.
rurban NotFound: I started fixing examples/c/test_main.c though. This is very similar 20:21
chromatic Yeah, don't buy laptops.
rurban chromatic: I'm 3 weeks on a trip and am forced to laptop 20:22
dalek rrot: r36978 | chromatic++ | trunk (3 files):
[config;dev] Removed more trailing spaces for $LD_OUT and $CC_OUT in Makefiles.

which allow it. Not all platforms (MSVC 9, for example) do. Thanks to
  diakopter for diagnosing this.
rurban diakopter++
diakopter: without IVT you cannot run 64-bit images? 20:23
Because that's my problem right now 20:24
diakopter yeah
rurban shit, thanks a lot!
diakopter ducks
rurban because I got a Intel Core2 with x86-64 cpu 20:25
but VirtualBox fails to initialize debian64, solaris64 and so on
diakopter your bios might let you turn it on. mine is Sony. :| 20:26
dalek tracwiki: v1 | cotto++ | ParrotVersionNumbers
tracwiki: initial version, mostly copy/paste from Allison's proposal on parrot-dev
tracwiki: trac.parrot.org/parrot/wiki/Parrot...ction=diff
shorten dalek's url is at xrl.us/beheq2
dalek rrot: r36979 | NotFound++ | trunk/examples/embed/Makefile:
[examples] make embed Makefile more portable
rurban I have Dell d820
Infinoid my thinkpad has an option for VT 20:27
diakopter the cmos address for the setting is known for some Sonys, but not for mine (yet). and I'm not about to spend the 10^n hours to [figure out how to] find/fix it.
rurban NotFound: $(CC) $(CCFLAGS) -c -I ~/parrot/include lorito.c sounds wrong 20:29
~/parrot
`bla` will also not work on cmd.exe
I think $(SHELL bla) will do it 20:30
But I'll test it first, after my reboot
rurban reboots to IVT 20:31
particle rm -f isn't portable either
NotFound: please see perldoc ExtUtils::Command 20:32
Hinrik OOJJJJ
oops
diakopter I think he's in prison? 20:33
NotFound particle: I try to avoid to depend on perl, the example is now intended to be used in a parrot development tree but that can change.
particle NotFound: parrot depends on perl 20:37
you can't build parrot portably without it 20:38
NotFound But the example is not building parrot, is linking with an already built parrot library
particle either depend on perl, or mark the example (and test) as your-platform-only 20:39
jonathan Does anyone remember or have a reference handy to how to make something "does array"? 20:40
So you can use it with :flat?
diakopter NotFound: $(LD) -o is not right.... you need $(LD_OUT)
particle notfound: or maybe 'fails on win32'
jonathan: pir, or c? 20:41
jonathan particle: pir
C is easy ;-)
NotFound diakopter: You mean parrot_config ld_out ? 20:42
jonathan I'm sure I remember someone hitting this recently-ish, but I forget who. 20:43
diakopter NotFound: yes. instead of '-o ' (no intervening space) 20:44
NotFound diakopter: I see, thanks
particle jonathan: i think it was cotto
diakopter I guess it might be a good idea to cache a bunch more of the config-(platform-)specific makefile values in parrot_config, such as "rm -rf" and the rest of ExtUtils::MM stuff 20:46
chromatic Hm, so that the Makefiles themselves don't have to have them. 20:47
That's a good idea.
NotFound We already have a lot of that 20:48
rm_f => '$(PERL) -MExtUtils::Command -e rm_f'
diakopter ooo you're building parrot.exe hotpatchable on msvc ;) 20:51
21:03 rurban joined
dalek rrot: r36980 | NotFound++ | trunk/examples/embed/Makefile:
[examples] make embed Makefile more portable
 21:03
rurban diakopter: my dell laptop Latitude D820 supports IVT. It's called VMM there. I need an admin to turn it on for me though. 21:06
octo According to docs/embed.pod, `Parrot_find_global_cur' will return `PMCNULL' if the symbol is not found. `Parrot_find_global_cur' internally calls `Parrot_find_global_n' which does the following: return PMC_IS_NULL(res) ? NULL : res;
Which is the intended behavior, `PMCNULL' (as documented) or `NULL' (as coded)?
GeJ Good morning everyone 21:07
octo Morning ;)
rurban ocot: I think the docs has to be fixed. But better ask in the parrot-dev list 21:08
s/ocot/octo/
octo rurban: Okay, thanks 21:09
NotFound octo: to be on the safe side, check that the returned value !PMC_IS_NULL instead of some value.
octo NotFound: Okay.. But I guess `PMCNULL' is NOT `((void *) 0)' so that exactly these problems are found, right..? 21:10
chromatic There's a Null PMC -- a singleton -- we use sometimes. 21:12
21:13 mikehh joined
octo While I'm changing this in docs/embed.pod, could one of you verify or correct the description of those flags, perhaps? <paste.debian.net/29027> 21:15
NotFound octo: I think the doc is incorrect anyway, because none of embed.h and extend.h does include global.h 21:19
chromatic octo, I'm not sure anyone understands those flags, unless we wake up leo or chip by saying their names. 21:20
octo NotFound: It worked without `<parrot/parrot.h>' up to the point where I needed `Parrot_find_global_cur', so just executing some PBC is possible without it..
NotFound And the EXAMPLES in that doc file uses #include parrot.h, wich is explicitly forbidden.
octo chromatic: Okay, thanks
NotFound: Oh, okay.. What else do I use then? 21:21
NotFound octo: nothing for a now. I'm working on it but takes time.
octo NotFound: Okay, since I basically want to find out if reentrancy is possible to the extent I need, I'll stick with parrot/parrot.h for now ;) 21:22
NotFound What I'm using in the example and test is to take use the namespaces directly.
octo You lost me.. How do you use the namespace directly? 21:24
NotFound octo: for testing and development, fine, but document and examplify incorrect usage, not fine.
octo: see examples/embed/lorito.c
octo NotFound: Already put this in front of the #include in docs/embed.pod: /* FIXME: Inclusion of <parrot/parrot.h> is actually forbidden. */
Will do
Uhm, there is no file called `lorito.c'. Nor is there a directory named `embed'.. 21:26
.. in the sources 21:27
NotFound octo: you need the current cvs release
I mean svn 21:28
octo I'm having the Git repository from repo.or.cz.. Last commit is Tue Feb 24 00:56:17 2009 +0000
NotFound Too old, we work fast ;) 21:29
octo Hehe, okay
NotFound Take a look at t/src/embed.t then
rurban seen rg 21:30
purl rg was last seen on purl 6 hours, 57 minutes and 59 seconds ago, saying: <private message>
octo That exists at least ;)
rurban notfound, particle: embed failing on win32 is too bold. only failing on msvc with cmd.exe. mingw and cygwin work fine as I assume 21:31
NotFound rurban: you mean the example or the test?
rurban makefile.vc would be the normal solution for this problem 21:32
NotFound They are not directly related
rurban NotFound: yes, the example with the new description
NotFound Ah, fine
I'm working on both at the same time, sometimes is confusing %-) 21:33
rurban or better: makefile.msvc
on perl5 we have win32/makefile.vc elsewhere we are more verbose
but using perl for helping embedders would be preferred I guess 21:35
NotFound rurban: I don't have msvc and my knowledge of it is a decade old, if someone wants to maintain it, fine.
21:35 gryphon joined
rurban I'll give it a try then 21:36
NotFound The long term solution IMO will be to use parrot to help parrot embedders, not perl5, but in the meantime... 21:37
rurban cygwin works fine for loriito 21:40
NotFound Nice :)
rurban now mcvs and mingw...
PATH=$PATH:/parrot_directory is wrong on an already installed parrot 21:42
I got the wrong include dir: parrot/embed.h: No such file or directory. 21:43
PATH=/parrot_directory:$PATH is better
NotFound rurban: that's the problem we talked last week about the installed path and includes 21:44
octo Amazing: As soon as you do it right, it starts working..
NotFound Ah, you mean working in dev tree while having other parrot instaled? 21:45
rurban yes, also parrot_config includedir instead of parrot_config prefix is much better 21:46
NotFound rurban: I think you are right
rurban parrot_config includedir => /usr/include
NotFound rurban: includedir gives me /usr/local/include, that's wrong working in the devel tree 21:47
rurban true, bad for testing
NotFound Anyway, this example is one day old, no need to make it perfect today;) 21:49
rurban one more catch: lorito$(EXE): lorito$(O) 21:51
NotFound Done 21:55
dalek rrot: r36981 | NotFound++ | trunk/examples/embed/Makefile:
[examples] some fixes in emebed/Makefile
 21:57
rurban I cannot find $(SHELL ..) or `` for nmake though. I guess we have the user to fill that in.
rurban maybe ron knows that for nmake 21:58
dalek rrot: r36982 | allison++ | trunk/docs/project/release_manager_guide.pod:
[doc] Add a note to release manager's guide about marking Trac milestones as completed.
 22:01
22:01 rob joined
particle typically it's cmd /c 22:03
... > file 22:04
then read into a var
Coke pmichaud: remember my comment about having to update our svn usage? based on a faulty assumption about version numbers. carry on. 22:08
nopaste "rurban" at 93.210.233.45 pasted "NotFound: Now I've got a problem: the msvc linker is strict." (10 lines) at nopaste.snit.ch/15711 22:11
octo What's the preferred method to get patches? Via email or via Trac?
rurban unresolved external symbol _Parrot_pbc_load and 3 more not exported 22:12
diakopter rurban: libparrot.lib needs created with its exported hints
rurban yes, the dllattributes are missing
chromatic octo, trac is better.
diakopter or, pass in the .def file /DEF:filename
NotFound rurban: What is the problem? Lack of decoration in extern header?
chromatic We can't miss them there.
octo chromatic: Okay 22:13
NotFound I mean extend
rurban we don't create a def file yet, with correctly attribute them, but only win32 strictly requires it.
I'll try to fix it...
dalek rrot: r36983 | rurban++ | trunk/examples/embed (2 files):
[examples] msvc fixes for examples/embed
 22:17
rurban particle: read into a var might be tricky. but we can create an incude file 22:18
particle err, sorry, yeah, include file.
rurban NotFound: wait a bit. maybe I'm too old with my msvc6 dir here 22:20
octo Sorry, but I guess I need your help again: The `str_key' argument to `Parrot_make_namespace_keyed_str' is documented as `NULL okay', but the argument is then passed on to `internal_ns_keyed_str' where a NULL pointer raises an exception. Unfortunatly, I have no idea what the intended behavior is.. :/ 22:21
22:23 Andy joined
NotFound octo: good catch 22:25
Infinoid octo: good catch indeed. I'd say remove _NULLOK from the Parrot_make_namespace_keyed_str and see if anything breaks :)
oh, that change will require "make headerizer" to update the ASSERT_ARGS definitions. 22:26
NotFound octo: Why were you calling with NULL, just for testing?
octo You'll have to change a bunch of functions then.. `Parrot_store_global_s' for instance..
Infinoid if those functions are not written to handle NULL pointers, then that's a good thing
octo NotFound: Didn't know how to get the global namespace, and `Parrot_store_global_s' said it's okay to pass NULL.. 22:27
So, in `Parrot_store_global_s' I've done this: if (str_key == NULL) { ns = CONTEXT(interp)->current_namespace; } else { previous action } Is that reasonable?
Infinoid is current_namespace the HLL root? 22:29
NotFound octo: note that that function is not in extend or embed usage, then it can be reasonable that they don't have a full check of his arguments.
octo Infinoid: I have no idea what `HLL' stands for, leave alone what `HLL root' *is* ;)
Infinoid octo: high level language (e.g. rakudo, partcl, etc)
rurban High Level Language
Infinoid So HLL root for rakudo would be the top level rakudo namespace 22:30
octo What I wanted to do was add a new subroutine (in form of a PMC) to a simple test script without any namespaces really..
So, yes, HLL root is probably what I wanted..
NotFound octo: you get the global namespace with Parrot_get_root_namespace, in embed.h
octo Maybe I should use `Parrot_get_ctx_HLL_namespace(interp)' in the NULL handling of `Parrot_store_global_s'? 22:31
Infinoid is parrot root == HLL root? (I'm not being sarcastic, I really have no idea)
particle no
NotFound Infinoid: no, is 'parrot' inside root
cotto It's nice to see the version number thread devolve into silliness. 22:32
diakopter rurban: web.archive.org/web/20041011160658/...rtlib.html
shorten diakopter's url is at xrl.us/behfdk
rurban NotFound: msvc6 works fine. my error (old build)
diakopter (maybe not related) 22:33
NotFound rurban: fine
rurban diakopter: thanks. I know a better tool: dllwrap
NotFound: It even works without the PATH= part 22:34
just calling lorito.exe fails when not finding libparrot.dll
NotFound octo: getting the HLL from context in a external function is risky, you get the HLL of the sub that is calling it, and that may not be the desired
Infinoid cotto: hey, at least I didn't suggest TAI64NA (accurate to the attosecond)
NotFound rurban: I don't know if there is a win32 equivalent of LD_RUN_PATH 22:35
rurban PATH=
syntax set PATH=N:\\usr\\src\\perl\\parrot\\parrot-msvc6;%PATH%
NotFound octo: to add a sub and find it easily from pir you need to put it in the 'parrot' namespace 22:36
rurban NotFound: for mingw we would need another special Makefile when using cmd.exe. maybe particle's trick will be better 22:37
octo NotFound: I'm trying to register that function right after the interpreter has been started up, not from an external frunction..
NotFound octo: external to parrot itself 22:38
octo NotFound: All I'm just calling `Parrot_store_global_s' from outside Parrot.. 22:40
NotFound octo: there is a test of that in t/src/embed.t , "Insert external sub in namespace", but I'm not sure about from what release number
octo: well, I don't used that function in the examples because is the wrong way, at least in the current state 22:42
octo NotFound: Ah, I think I see now what's done there.. I was always looking for functions like *store*..
NotFound Get the current release and look at lorito code, is cleaner than the tests 22:44
22:45 szabgab joined
octo I got it ;) 22:46
NotFound Uh, not, in lorito that feature is not used yet :-X 22:47
rurban btw: we still need the win32 manifest stuff for 1.0 who's doing that. there's even no todo ticket for this
octo I meant, I got it working, not I got the new revision.. Guess I'll solve the revision issue by getting some sleep and hoping that the Git repository is updated in the meantime..
NotFound octo: I hope so 22:48
octo: sorry if the current way is convoluted, but these are the first steps to make the embedding work without inappropiate workarounds 22:49
octo NotFound: No problem, I'm just totally new to Parrot and don't understand most of what I'm doing ;) So if I start asking dumb^Will-informed questions, please do point me to the appropriate documentation ;) 22:52
NotFound octo: the questions will help us to write that documentation ;) 22:53
octo Okay, I'm off to bed.. Good night (+ timezone correction), talk to you later.. 22:55
rurban oops, midnight. I leave also, bye 22:58
23:05 mikehh joined 23:25 Whiteknight joined 23:26 bacek_ joined
rurban does FREEZE_ASCII 1 dont work anymore? I get a src/io/api.c:231: failed assertion 'pmc' 23:27
dalek rrot: r36984 | coke++ | trunk/DEPRECATED.pod:
Fixup file to be more user-facing

  - standardize formatting slightly
  - add release numbers where some were skipped		 23:37
cotto rurban, I don't think it's ever worked. 23:40
rurban but a good idea though
cotto rurban++ if you can make it work 23:41
rurban naa, I think I have nbetter things todo right now.
cotto It would be nice.
rurban If I'm really desperate at freeze/thaw then maybe 23:42
Coke (if it doesn't work, let's rip it out.)
cotto I'll file a tt and see if it gets any takers.
rurban The most useful information would be if it ever worked, and would are the major points to watch for. 23:43
The failing assertion is just a missing pmc arg to Parrot_io_flush 23:44
NULL vs PMCNULL I guess
Nope. It's dying in a an exception: Unknown PMC type to thaw 0 23:47
p *io
dalek rrot: r36985 | NotFound++ | trunk (2 files):
[cage] drop NULLOK modifier in Parrot_make_namespace_keyed_str str_key arg
 23:49
cotto ticket filed 23:54
rurban At least the freeze shift functions ARE portable now. I checked that
dalek rrot: r36986 | rurban++ | trunk/src/pmc_freeze.c:
[cage] update TODOs
 23:57
23:58 Tene 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!