Please mentor for SoC | parrotcode.org/ | YAPC::NA talks deadline is Sat Mar 15 | tinyurl.com/2pmnlq
Set by moderator on 8 March 2008.
00:11 Limbic_Region joined 01:00 Pabellon joined 01:08 mdiep joined 01:19 Davemon joined 01:26 kid51 joined 01:41 petdance joined
svnbotl r26341 | petdance++ | trunk: 01:55
: localizing and consting
diff: parrotvm.org/svn/parrot/revision/?rev=26341
02:00 Theory joined
skids nopaste? 02:29
purl well, nopaste is at nopaste.snit.ch/ (ask TonyC for new channels) or rafb.net/paste or paste.husk.org/ or poundperl.pastebin.com/ or paste.scsys.co.uk/ or don't bother me while I'm eating
svnbotl r26342 | petdance++ | trunk: 02:32
: consting and annotating
diff: parrotvm.org/svn/parrot/revision/?rev=26342
nopaste "skids" at 71.233.204.100 pasted "parrot build warnings and failure on PA-RISC linux" (71 lines) at nopaste.snit.ch/12492
svnbotl r26343 | petdance++ | trunk: 03:03
: reflagged a PMC
r26344 | petdance++ | trunk:
: reflagged a PMC
diff: parrotvm.org/svn/parrot/revision/?rev=26344
03:11 jjore_ joined 03:36 rhr joined
jrt4 Anyone know offhand how to use gdb to scan for a byte pattern in a specified address range? Like if the heap for the parrot process ran from 08050000-0827d000, and I wanted to have gdb return the first byte it found in that range that equals 0xa17ad376 ... is there a command to do this sort of thing? 04:41
obviously the byte doesn't equal 0xa17ad376 ... I mean the address of the first byte where it finds the match :) 04:42
04:55 Theory joined
cotto FREEZE_ASCII in src/pmc_freeze.c seems to have bitrotted to the point where it breaks the build process if set to 1 05:01
Is this worth filing a bug report over?
05:22 Theory joined 06:34 tewk joined
svnbotl r26345 | duff++ | trunk: 06:40
: [rakudo] rough cut implementation of auto-{increment,decrement} for Perl6Str
diff: parrotvm.org/svn/parrot/revision/?rev=26345
PerlJam okay ... what's the way to add a ticket to RT? 06:46
and what's the practical difference between "new" and "open" (I just accidentally changed a ticket's status from new to open) 06:47
Tene purl: parrotbug 06:51
purl parrotbug is mailto:parrotbug@parrotcode.org or svn.perl.org/parrot/trunk/docs/submissions.pod
Tene That's how I do it.
PerlJam yeah, I just thought there was a webby interface too. 06:53
anyway, I need some sleep
PerlJam &
07:06 uniejo joined 07:31 allison joined 07:43 marmic joined 07:44 UltraDM joined 08:05 iblechbot joined 08:18 cognominal joined 08:29 AndyA joined 08:37 wknight8111 joined 08:51 kj joined
svnbotl r26346 | fperrad++ | trunk: 08:56
: [WMLScript]
: typo Lua -> WMLScript
diff: parrotvm.org/svn/parrot/revision/?rev=26346
09:37 slavorg joined 09:59 slightlyoff joined 10:42 ruoso joined 11:09 skv_____ joined 11:10 rblackwe joined 11:11 cognominal joined
Coke there is no webby interface for new RT, unless you use the command line client. 11:46
(which does sneaky web things on your behalf) 11:47
Coke sighs in the general direction of TPF. 11:48
lathos What now? 11:50
purl The chalice contains human blood cursed in Death's name. The hooded figure kneeling beside you passes it to you and you have to drink. LOSE 3 HIT POINTS.
lathos OK, I did not expect that.
12:09 wknight8111 joined 12:15 IllvilJa joined
Coke methinks purl has been playing a little too much D&D. 12:27
Coke summons ambs 12:30
svnbotl r26347 | duff++ | trunk: 12:39
: [rakudo] make Perl6Str increment() work on empty strings
diff: parrotvm.org/svn/parrot/revision/?rev=26347
12:59 kj joined 13:06 skids joined 13:35 Psyche^ joined 13:50 gryphon joined 13:51 petdance joined
lathos static.simon-cozens.org/memories-photos/3270.jpg # More crazy HDRity. 14:34
And the wrong challenge.
s/llenge/nnel/
14:34 davidfetter joined
Coke heh. 14:51
15:48 pfig joined, shamu joined, ewilhelm joined, Khisanth joined, arcady joined, silug joined, Alias_ joined, cout joined, purl joined, particle joined, TimToady_ joined, teknomunk_ joined, zev joined, cxreg joined, amoore joined, jjore joined, cotto_work joined, tewk joined, rblackwe joined, wknight8111 joined, skids joined, davidfetter joined, gryphon joined, diakopter joined, Ademan joined, Sartak joined, Crassworm joined, Tene joined, spinclad joined, jdv79_ joined, Coke joined, c9s_ joined, shorten joined, buildbot joined, cotto joined, Piper joined, TonyC joined, workbench joined
Juerd Piper is annoying. 15:49
PerlJam indeed.
Juerd Does it do more than logging and privmsg'ing? 15:50
i.e. is it safe to ignore it?
PerlJam moreso when combined with Tweety
hah ... I was wondering why I only see people talking to gumbybrain now and nothing from gb itself. I forgot I had ignored it. 15:51
ignore++
15:52 rhr joined
Juerd PerlJam: /ignore -replies in irssi :) 15:53
Then it also ignores lines beginning with the name 15:54
particle piper and tweety are safe to ignore 15:55
PerlJam Juerd: thanks. 15:56
16:21 jrockway joined 16:22 rdice joined 16:46 peepsalot joined 16:49 clunker joined, dngor joined, confound joined 17:02 Infinoid joined 17:05 mercutioviz joined 17:06 mercutioviz left
Infinoid anyone mind if I fix up t/codingstd/trailing_space.t to give line numbers, the way check_isxxx.t and c_cuddled_else.t do? (perl6str.pmc is failing all 3 of those, at the moment.) 17:08
17:11 avar joined
Coke no, please do 17:12
(it's simple enough to fix the entire file when you know it's bad, but having the line number can't hoit.)
17:14 BitPoet joined
particle infinoid: "/ $" is a good command in vim for fixing trailing spaces 17:14
particle wishes somebody would refactor the common code in t/codingstd and t/distro tests into a module
700*.75 17:16
purl 525
Infinoid that sounds like a good idea. they all have basically the same runloop 17:17
17:17 Ademan joined
Coke particle: "%s/ +$//" 17:17
particle sure, that'll work.
sometimes (rarely) trailing spaces are necessary 17:18
usually it files under examples/ or t/
Infinoid when are they necessary? (and should there be a special skip list for that?) 17:19
17:19 kj joined
Coke presumably they are special cased or they'd be failing by now? 17:19
Infinoid apparently so
Coke ah.. 17:21
;the test is only testing c files.
that's not right. =-)
Infinoid indeed 17:24
nopaste "Coke" at 72.228.52.192 pasted "This is probably better." (18 lines) at nopaste.snit.ch/12496
Infinoid how many failures do you get with that? :) 17:25
Coke 249 fies.
249 files.
Infinoid awesome. want to fix them up, or should I?
particle btw, i created a test recently based on codingstd tests, for a $work project... 17:26
Coke You're welcome to take the patch and fix up the failures. =-)
particle it also included the total number of files tested, which i like better.
failed 249 of 1178 files.
Infinoid hmm, nice
particle gives me a better level of confidence
simple scalar counter that's incremented 17:27
Infinoid makes sense. sounds like something that should go into the common runloop tho :)
svnbotl r26348 | infinoid++ | trunk:
: [t/codingstd] when trailing_space.t fails, it should report line numbers,
: like the other codingstd tests do.
diff: parrotvm.org/svn/parrot/revision/?rev=26348
particle precisely, all the better to refactor
Infinoid yep!
Coke particle: you're duplicating something the harness gives you because you avoided the harness! 17:28
particle huh?
ok, well, kinda
Coke ^_^
Infinoid the alternative is to have a separate ok() for each file, I guess 17:29
particle i think it's a Bad Idea to have 45000 codingstd tests
Coke Not if we don't run them as part of the standard test run. =-)
particle by "standard" you men dev tests? 17:30
Coke "make test"
purl "make test" is possessed!
particle it can be argued that codingstd tests should be run *only* for development testing 17:31
Coke in which case it doesn't matter how many of those tests there are.
particle "make test" is for the end user.
Coke not entirely, no.
(not as it stands today.)
particle right, not today.
today, i'ts a mess
by the end of `08, that'll be cleaned up 17:32
17:35 Daveman joined
Infinoid ok, I have a huge whitespace diff 17:35
Infinoid tests the heck out of it
Coke Infinoid++
particle of note are the stream files under examples 17:36
examples/shootout/*.pir_{input,output} 17:37
Coke just adding the perl type is probably an improvement.
particle ayep
Coke and we can add a list of skips, of course.
Infinoid should I abort the pir stuff for now, then? 17:38
particle well, i think it's safer to commit one filetype at a time 17:39
Infinoid can do, I'll start with .pl
particle probably more manageable too
Infinoid++
Infinoid note that the test never looks at examples/shootout/*.pir_{input,output}
only the .pir sources, not the data
particle sweet
does it look at .t files?
Infinoid probably not :) 17:40
particle good. should be quite safe
Infinoid are they returned by $DIST->get_perl_language_files() ?
Coke yes, the perl does look at .t files.
particle possibly .t files that have a perl #!
Coke right. it respects shebangs. 17:41
Infinoid well, nothing in t/ was reported by coke's version of trailing_spaces.t
particle ok
Coke could be those are already fine. =-)
Infinoid yeah, I was worrying about stuff in <<__END__ blocks 17:42
particle heredocs and end blocks are my concern, too
pir_output_is(<<CODE,<<SOME_OUTPUT_WITH_TRAILING_SPACES,'oops');
Infinoid exactly 17:43
just doing perl-language files now... test caught 32 files 17:48
17:51 Theory joined 17:56 Pabellon joined
Coke lathos: ping 17:58
I have concerns that we've had to invent something that unicode didn't handle. 18:06
ewilhelm what has become of the unified_testing branch? 18:22
Coke I believe it's pretty stale at this point. 18:23
I have a TODO note somewhere to ping the list about it.
18:27 barney joined
Coke barney: how goes? 18:29
particle ewilhelm: unified_testing branch is indeed stale. 18:35
if trunk head is merged there, i think it's now possible to get all languages using the new harness 18:36
Coke I suppose someone could try to merge changes in branch to it.
"jinx"
particle i need to customize the test harness for rakudo's spectests
i think my tuits are tied to the dollar
they're not worth much these days 18:37
18:37 Andy joined
barney I tried merging the 'unified_testing' branch some weeks ago, but got stuck somewhere 18:49
I think I also encountered some hardcoded pathes
ewilhelm seen PerlDreamer?
purl PerlDreamer was last seen on #parrot 150 days and 23 minutes ago, saying: you're right, it was tewk [Oct 15 11:26:53 2007]
barney is fighting with VPN 18:50
18:58 Psyche^ joined
davidfetter o/` i fought the VPN and // the VPN won o/` 18:58
ewilhelm hmm, looks like Colin got bogged down by the TAP::Harness API 19:01
that, and the somewhat ad-hoc "what to test" selection codes that seem to be embedded in the various harnesses 19:07
Coke it's got to go somewhere? neh? 19:11
ewilhelm yeah, see the svn.perl.org/parrot/branches/unifi...RANCH_TODO 19:13
shorten ewilhelm's url is at xrl.us/bhnf8
ewilhelm I was thinking normalizing that to $(t/harness --files) might be the best first step 19:15
it does look like a lot of what needs done is some moving of files and other stirring-up, which means the branch goes stale pretty quickly 19:16
Coke: the what-to-run decisions are a somewhat opaque (at least to me) mix of checking preferences and resources 19:18
Coke We're not tied to --files, btw. It was the initial stab at "how do we combine all these harnesses." 19:19
I don't know if there's a better idiom.
ewilhelm e.g. 'if -e "$Bin/../DEVELOPING";'
svn mkdir xt; svn mv t/codingstd xt/ 19:20
Coke ? 19:24
barney Couldn't we move away from 'harness runs all test scripts' to 'combine test reports from various subsystems/languages' ? 19:25
ewilhelm several pm authors have been using that for "extra tests"
barney: yeah, but first I would like to move toward "prove can run the tests" 19:26
so e.g. prove -j 9 allows them to run in parallel
Infinoid Coke: does the tcl testsuite pass for you? I'm having trouble getting through a languages-test to make sure I haven't broken anything
Coke Infinoid: have you committed any changes to tcl recently? 19:27
It was passing all but 2 tests last I checked (with some noise on another)
2 individual tests, that is.
ewilhelm: tcl jumps through hoops to allow you to run prove on them. 19:28
Infinoid no, though languages/tcl/tools/tcl_test.pl has a whitespace diff in my tree
barney So shouldn't we take a first step and require Test::Harness 3.0. So that gradual improvements and simplification can be achieved?
Coke Infinoid: that file is only used to generate the tests against the 8.5.0 spec tests.
ewilhelm I'm actually not seeing any sort of platform-based test selection here, is all the pick code in the harnesses just for "developing or not"? 19:29
Infinoid ok. well, looks like most tests are crashing on linux x86-64
(with or without my diff)
ewilhelm barney: yeah
Coke ewilhelm: yes. most of the platform specific skip stuff is *in* the test files.
ewilhelm hmm, well just moving them into xt/ would eliminate the need for the --files stuff then 19:30
Coke except for probably rakudo, which is special.
what is this xt idiom again?
ewilhelm "extra testing"
Coke oh. run everything in t by default?
ewilhelm you can use a longer directory name if you want
yeah, that's how e.g. M::B and EU::MM do it 19:31
and `prove -r t`
Coke so our "coretest" would be all that was left in t/, and then we could have xt/config, xt/otherthing ...
ewilhelm yep
barney and/or use $ENV{AUTHOR_TEST} 19:32
ewilhelm and you can of course have your own ./tools/runtests instead of prove
Coke and then how do you add back in some of the tests in xt? another harness?
ewilhelm prove t/ xt/config
Coke well, we already have harness. would runtests replace harness?
ewilhelm yeah, but using the App::Prove API
so it would get all of the same option processing as prove 19:33
Coke That all sounds vaguely reasonable.
ewilhelm svn.perl.org/parrot/branches/unifi...l/runtests 19:35
shorten ewilhelm's url is at xrl.us/bhng7
19:36 Ademan joined
ewilhelm ouch, that @ARGV handling isn't right 19:36
Infinoid ok. I don't seem to have broken anything new. any objections to my checking in trailing-whitespace fixes to 32 perl files? 19:37
ewilhelm I think that's where I need to hammer on the App::Prove API
Coke Infinoid: nope 19:38
Infinoid (if I have broken something, feel free to flame me and I'll fix it)
Coke human torch?
purl FLAME ON!
Coke fixes that emote fowl. 19:39
confound buk-awk!
ewilhelm Coke: example of hoops jumped through in TCL tests? 19:40
use Tcl::Test; #\\
__DATA__
purl __DATA__ is in perldata
ewilhelm ?
Infinoid svn: MERGE request failed on '/parrot/trunk' 19:41
svn: MERGE of '/parrot/trunk': 200 OK (svn.perl.org)
Coke half the tests are self hosted in tcl, half are perl (too hard to self host at the moment) the self hosted tests are run through perl first and a little module is used to sanitize the cur working dir before running.
so you can run the tcl tests from tcl, from languages, or, I think, from top level parrot build dir.
(despite the fact that there are hardcoded paths like "languages/tcl/foo" all through the code. 19:42
particle Infinoid: svn up puts me at r26349, with many changes to .pm files
Infinoid yeah, it went through, just failed to mark the local copy as such, looks like
Coke happens occasionally.
particle svn up should fix that
Infinoid I've seen it before. and yeah, svn up fixed it
so now I can look at breaking all the .pir files! :) 19:43
particle i'll rebuild and test
Infinoid particle: thanks, that'll be appreciated
hmm. before breaking .pir, I'll take a stab at unifying these test runloops 19:44
svnbotl r26349 | infinoid++ | trunk: 19:45
: [t/codingstd] check trailing whitespace in perl language files, not just C.
: [languages, lib, config] fix up 32 perl source files to pass trailing_space.t
diff: parrotvm.org/svn/parrot/revision/?rev=26349
Coke Infinoid: that would be appreciated. 19:50
Infinoid so what do I call the common infrastructure? 19:52
Parrot::Test::Runloop::LotsOfFiles?
(I'm really bad at naming things.)
particle Parrot::Test::Util
stick everything under there 19:53
Infinoid: all my failing tests look valid 20:03
Infinoid great, thanks 20:04
particle coding standard problems in perl6str.pmc, recently modified by PerlJam
Infinoid yep, that's what got me looking at trailing_space.t initially
(it didn't give me line numbers, the other two did)
particle scratch that itch! 20:05
Infinoid so I'm thinking about something like:
use Parrot::Test::Util::Runloop
name => 'No trailing spaces or tabs',
files => [ $DIST->get_c_language_files() ],
once_per_line => sub { !/.?[ \\t]+$/ };
I don't think it'll be exactly that, depending on whether $DIST is available at compile-time 20:06
particle yeah, you might need Parrot::Test::Util::Runloop->run({...}); 20:07
Infinoid that's more likely
ewilhelm $class->files(@list)->run_per 20:08
Infinoid I figure some things will be once_per_line, others might be once_per_file
ewilhelm err, $class->files(@list)->run_per_line(sub {...}) ?
Infinoid that looks nice too 20:09
ewilhelm or just $class->new(name => '...', files => \\@list)->run_per_line(sub {...})
for_lines()/for_files() ? 20:10
Coke: looks like the self-hosting tcl tests were working in the unified_testing branch with just a '#!tcl' 20:11
particle $class->new(...)->files(...)->slurp_scalar()->run() ?
svnbotl r26350 | bernhard++ | trunk: 20:12
: [codingstd]
: Uncuddle elses.
diff: parrotvm.org/svn/parrot/revision/?rev=26350
ewilhelm now it's starting to look like a mix of List::oo and File::Fu
ewilhelm is wearing lambda.pm colored glasses 20:13
Coke is reminded to get [apply] working in tcl. 20:14
ewilhelm: huh. 20:17
skids cognomial: see my nopaste from last night's log, and this... 20:20
mail-index.netbsd.org/tech-kern/200.../0001.html
cognominal: s/cognomial/cognominal/ 20:21
There being the problem on PA-RISC I I would've thunk ia64 too, but probably also IIRC alpha.
svnbotl r26351 | bernhard++ | trunk: 20:24
: #51662: [TODO] Remove 'classname' rule in imcc.y
: Avoid deprecated PIR syntax in languages/perl6
diff: parrotvm.org/svn/parrot/revision/?rev=26351
ewilhelm Coke: there is still a chdir issue -- they expect they're being run from languages/tcl 20:34
I suppose you could build that "always chdir to languages/foo" into the runtests code 20:35
svnbotl r26352 | particle++ | trunk: 20:37
: [docs] add DONORS file, to thank those who have helped make parrot what it is by donating funds
diff: parrotvm.org/svn/parrot/revision/?rev=26352
20:42 teknomunk joined
svnbotl r26353 | particle++ | trunk: 20:42
: fix svn metadata
diff: parrotvm.org/svn/parrot/revision/?rev=26353
r26354 | bernhard++ | trunk: 21:02
: #51662: [TODO] Remove 'classname' rule in imcc.y
: Avoid deprecated PIR in 'languages/jako'.
: - Diese und die folgenden Zeilen werden ignoriert --
: M jako/lib/Jako/Construct/Expression/Call.pm
: M jako/lib/Jako/Construct/Expression/Value/Identifier.pm
: M jako/lib/Jako/Construct/Statement/Assign.pm
: M jako/lib/Jako/Construct/Statement/Increment.pm 21:03
: M jako/lib/Jako/Construct/Statement/Decrement.pm
: M jako/lib/Jako/Construct/Statement/Concat.pm
: M jako/lib/Jako/Construct/Statement/Arithmetic.pm
: M jako/lib/Jako/Construct/Statement/New.pm
: M jako/lib/Jako/Construct/Declaration/Variable.pm
diff: parrotvm.org/svn/parrot/revision/?rev=26354
Coke thinks bernhard has a language issue. =-)
barney ME HAZ ERROR , is lolcode broken?
Coke ewilhelm: not every language would be happy with that.
barney: I don't think anyone checked it after the merge. =-) 21:04
davidfetter I CAN HAZ ERRORZ?
Infinoid OHNOES
ewilhelm Coke: that seems to be the main blocker in "unified" 21:05
I suppose a set of rules could be written such that the harness goes by the location of the test
particle each language should work completely and correctly from it's root directory
if they don't work that way, they're broken 21:06
ewilhelm so "always chdir to languages/foo" ?
diakopter ewilhelm: you can ask perlDreamer about his status/thoughts on that branch. Though he's not at the moment, he's often on freenode as the same nick.
particle yes, i say.
ewilhelm diakopter: I think his work stopped when I did 21:07
particle ewilhelm: we have an (unwritten) rule that "all languages must be self-contained"
barney for a language implementation you at least need to know a list of test scripts, from where to run it, and who executes the test scripts 21:11
where different scripts can have different executors 21:12
and as a bonus, a single script might have multiple executors, e.g. different implementations of the same language 21:14
svnbotl r26355 | bernhard++ | trunk: 21:18
: [parrot_compiler]
: Let the 'test' Makefile-target depend on 'all'.
diff: parrotvm.org/svn/parrot/revision/?rev=26355
nopaste "Infinoid" at 75.31.90.47 pasted "Proof of concept: test runloop consolidation for t/codingstd/ and t/distro/" (112 lines) at nopaste.snit.ch/12498 21:24
svnbotl r26356 | bernhard++ | trunk: 21:25
: #51662: [TODO] Remove 'classname' rule in imcc.y
: Avoid deprecated PIR syntax in languages/unlambda
diff: parrotvm.org/svn/parrot/revision/?rev=26356
Infinoid particle: was that the kind of thing you had in mind? 21:26
(see my nopaste please)
particle looks pretty good 21:28
i notice a small output difference with parens around line numbers as opposed to no parens 21:29
Infinoid and there's an extra "failed N files out of M total" line, too
particle but that's not important right now
Infinoid and the prefix is a little bit different
there is currently no consistency for this, for the various tests.
particle right. 21:30
i like the failed x of y
Infinoid c_cuddled_else.t currently outputs the line numbers with no parens, like my current output does
svnbotl r26357 | bernhard++ | trunk:
: #51662: [TODO] Remove 'classname' rule in imcc.y
: Avoid deprecated PIR syntay in languages/Zcode.
diff: parrotvm.org/svn/parrot/revision/?rev=26357
particle Runloop.pm should use $Test::Builder->level to set the caller depth
i'd rather see 'at t/codingstd/check_isxxx.t line x' 21:31
Infinoid oh, awesome. I was just going to complain about that.
particle++
hrm. the problem with testing tests is, you have to generate a lot of bogus code to get good coverage for the failure case 21:35
21:40 IllvilJa joined 21:59 buchet joined 22:00 Andy joined 23:12 jrt4 joined 23:43 skids joined
Coke particle,Infinoid: I'd rather see file:line 23:45
but don't care enough to bikeshed more than that. =-)
Infinoid nice thing is, you'll only have to change it once for everything 23:48
I'm finding myself leaving some tests unchanged, which are significantly different than the boilerplate (c_function_docs.t and c_header_guards.t for example) 23:50
23:52 nowhere_man joined
nowhere_man hi all 23:53
Infinoid hello 23:54
purl hola, Infinoid.
Infinoid allright. I'm out of time for the day, got 5 tests converted 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!