Parrot 2.6.0 | parrot.org Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | close 13 tickets, merge outstanding branches, help NotFound clean up PBC tests
Set by moderator on 17 August 2010.
mikehh bah - no it still had a failure in that test 00:00
00:00 Psyche^ joined
mikehh ok - so what do I do about that test 00:01
00:01 pgollucci joined
chromatic Sounds like a problem in ld.so, not the test. 00:02
Maybe run it through Valgrind?
mikehh it passed in make test, failed in testb in fulltest and fails prove 00:07
00:08 wagle joined
mikehh no apologies it passed testb, it failed testf 00:11
00:12 Administrator_ joined
mikehh chromatic: how do i do that 00:12
chromatic Do you have Valgrind installed?
mikehh I think so 00:13
yes and kcachegrind 00:14
chromatic If you have a single .pir file, you can run: 00:15
vg?
purl vg is probably volume group
chromatic vgp?
purl vgp is probably valgrind --suppressions=tools/dev/parrot.supp --num-callers=500 --leak-check=full --leak-resolution=high --show-reachable=yes ./parrot --leak-test $@ or tools/dev/vgp
chromatic tools/dev/vgp path/to/file.pir
cotto_work The profiling runcore, while giving lta line numbers, seems to deal fine with a simple tailcall example. 00:19
chromatic Let's fix the LTA line numbers, then move on.
nopaste "cotto_work" at 192.168.1.3 pasted "profiling tailcall test" (42 lines) at nopaste.snit.ch/22862 00:20
cotto_work (the line numbers don't seem to have anything to do with the tailcall) 00:21
certainly easy enough to add a failing test for
msg cotto add a failing profiling runcore test for nopaste.snit.ch/22862
purl Message for cotto stored.
nopaste "mikehh" at 192.168.1.3 pasted "vgp output" (556 lines) at nopaste.snit.ch/22863 00:26
chromatic ==19244== by 0x4F65481: Parrot_ParrotLibrary_destroy (parrotlibrary.pmc:69) looks suspicious. 00:27
mikehh yeah - but I don't think it actually failed 00:29
chromatic Seems like the source of your trouble though.
kid51 make fulltest: PASS Linux/i386 r48557
00:31 ruoso joined
chromatic All testb passes for me too. 00:32
mikehh shall I commit and you can look at what I have got
or put it in a branch 00:33
chromatic What kind of commit is it?
mikehh well the release stuff
chromatic Go ahead.
cotto_work same test failure here 00:34
00:36 davidfetter joined
cotto_work but nothing when running them manually with prove -b 00:37
mikehh ok r48558 - all tests pass except for that t/pmc/threads.t test 14 now in testf and if I try prove
but it seems to pass make corevm/make coretest, make test and now tesb and testr in fulltest 00:39
testb
Infinoid Any word on when feather is coming back online? I just finished up the github feed parsing work I started this morning, and can update dalek when I can get to it again. 00:40
00:46 Paul_the_Greek joined
Paul_the_Greek Does a PMC attributes block have an extra pointer in addition to all the data values? 00:47
particle hey Infinoid, hope you had a happy birthday last week.
chromatic I don't believe so, Paul_the_Greek. 00:48
00:48 plobsing joined
Paul_the_Greek Gak, then I really don't understand what is going on with the allocation of them. I have to wait until I get back home to sort it out. 00:49
00:51 dduncan joined 00:53 dduncan left
Infinoid particle: Thanks, I did. Hope you did too 00:56
particle indeed, thanks. 00:57
whiteknight Infinoid: how old are you now? 01:05
01:09 dalek joined
Infinoid has to remember what year it is 01:09
whiteknight: 32
cotto_work or 20 01:10
whiteknight 0x20
01:10 Util joined, pmichaud joined, dukeleto joined 01:11 PerlJam joined
Infinoid Ok, this dalek restart should get longest-prefixes for changed files working again on the github feeds. 01:12
01:12 dalek joined
cotto_work Where does juerd hang out 01:14
seen juerd 01:15
purl juerd was last seen on #perl 218 days, 13 hours, 35 minutes and 30 seconds ago, saying: tarski: See topic [Jan 11 11:40:22 2010]
Infinoid He's on #perl6 01:16
cotto_work Heh. I just looked for him and didn't see him, then he joined.
pmichaud is there a way from PIR for me to get some sort of measurement of used CPU time as a parrot program is running? 01:18
whiteknight I seem to remember such a beast 01:19
sorear cotto_work: that's because juerd IRCs from feather
pmichaud i.e., if I want to measure CPU usage instead of wall-clock time
sorear cotto_work: he can be found easily when he's not needed
cotto_work pmichaud: there's the C-level function Parrot_hires_get_time but no pir support for it 01:23
It sounds like something that'd be pretty trivial to use. It uses a per-processes cpu timer, if available 01:25
mikehh ok I don't know what is going on - I did a clean checkout of r48558 and it passes fulltest with g++ and with gcc 01:27
cotto_work If it'll make nqp faster, we can always add another experimental op.
mikehh I need to tag that as the release
pmichaud cotto_work: I might do that eventually; I was just wondering if anything was available immediately.
wall-clock time may be good enough for now. 01:28
mikehh I think I am losing it - how do you create the Release Tag?
pmichaud mikehh: svn copy, I think.
mikehh I am not thinking very straight at the moment
cotto_work it's in the release manager's guide
pmichaud item #7 01:29
I bet we could do a lot with an experimental 'profile' opcode. 01:33
cotto_work pmichaud: I'd appreciate a test case that demonstrates the profiling runcore failing at tailcall recursion if you find one. A quick pir test looked fine. 01:34
pmichaud: how would it work?
pmichaud cotto_work: (profiling runcore) that's part of the problem. simple cases seem to work, but when I run it on something from nqp I get nonsensical output. 01:35
and when I asked about it before, speculation was that it was tailcalls likely causing the issue.
is there documentation on the profiling runcore somewhere, or can I get a quick refresher on how to use it?
cotto_work a chunk of nqp code is a valid test case 01:36
docs/dev/profiling.pod is as much as exists
suggestions welcome
Coke I would love it if someone else created the docs.parrot.org update. =-) 01:38
cotto_work now that the permissions are fixed, I can do that
Coke ]o/ 01:39
(ow, cramp!)
pmichaud gist.github.com/533009 # what I get now when trying to use -Rprofiling with nqp.pbc 01:40
cotto_work what's par? 01:41
purl rumour has it par is the answer to 'why are these URI modules loaded but not explicitly used'
pmichaud symlink to the parrot executable
I can do it directly
dalek rrot: r48559 | mikehh++ | tags/RELEASE_2_7_0:
tagged release 2.7.0
 01:42
cotto_work Nice. I have a bug.
pmichaud gist.github.com/533018 # 2nd attempt, avoiding symlink 01:43
cotto_work same happens here
pmichaud (profile opcode) - I'm just thinking that if we had an opcode like "profile 'START'" it could record information in a file somewhere 01:45
01:45 bacek_in_sing joined
pmichaud so upon entering a subroutine, we do profile 'START' 01:45
and upon exit, we do profile 'EXIT'
and when profiling isn't enabled, the opcode is a noop
cotto_work and it'd just spit the message and a timestamp to somewhere useful?
pmichaud and the name or subid of the subroutine it was called from 01:46
and perhaps even the subid of its subroutine's caller
cotto_work do you have the tuits to file a bug on the profiling runcore? 01:47
pmichaud from what I just did? sure.
01:47 ash_ joined
cotto_work thanks 01:47
pmichaud trac.parrot.org/parrot/ticket/1745 01:49
cotto_work Next time, I hope you can take two steps before something explodes. 01:51
01:56 theory joined
dalek TT #1745 created by pmichaud++: profiling runcore fails with NQP 01:59
TT #1745: trac.parrot.org/parrot/ticket/1745
01:59 theory joined
mikehh oh damnit - I don't think I sorted out adding my ssh key 02:01
how can I soirt that out?
02:03 kid51 joined
cotto mikehh, what's your public key? 02:09
dalek p-rx: a528b79 | pmichaud++ | build/PARROT_REVISION:
Bump PARROT_REVISION to 2.7.0 release.
kudo: 67778a6 | pmichaud++ | build/PARROT_REVISION:
Bump PARROT_REVISION to Parrot 2.7.0 release, freeze until Rakudo release.
 02:11
cotto I don't see anything that looks like it's you
mikehh cotto, I am not exactly sure how to set that up for mikehh
cotto Do you have an ssh publick key? 02:12
mikehh on this system I am user mhb
cotto ~/.ssh/id_rsa.pub
mikehh ssh-dss AAAAB3NzaC1kc3MAAACBAMi4qq9hYjNYYDbZgBT5Ln63jSCjOP7KEWLFJVL6/MMSq7+mmVOIee+tqKBrWKxPjJYgKP3XHDjwLmMDLjnD30zYeD5nLRXSfAaDdFighsy5FRtz+w/4uDFpKPIVTjjjwnGSdnalsc8UVylAu2ZOKpr0Qe4ejdBwXGbihVi7qaMdAAAAFQCNCqF21BKLXTdLDCgKdVXy9BcUWwAAAIBrfeSErdult8YVYYwjDNRQPrAzIpryIaNezxv85X2w6PfhpC37zy1ZBZOwRC5KCfVVi5g4yVT9FyN8r/e71QcB4kbNV21YOc2hB1owKtcuKTJwGRJtdQVZqN4VFxm5iYb9MPAuzCRQ/ 02:13
wqcU+gy7p1vGD49aZE5UzqofSULvxErWQAAAIBYm0u0bKW8IGtGlSaG0M4Sb4U6pXHHn8WakEv4+5zxcqRr5u4DjQ6ZRtDZ7JmjMIayjphYeymB9T3+07r7TT5LVXZsElBZdyUqUJhQmXh/bhMbL0lGpsVmcktorkIeTW/zpnC9d9Q/4LWG2qjtQ3pEjkKyV8Jmta5HFkuCmkkmow== mhb@mhb-desktop
but that is for user mhb@mhb-desktop 02:14
cotto can you ssh into the file server?: ssh parrot@ftp-osl.osuosl.org 02:15
02:16 darbelo left
cotto mikehh, does that work? 02:18
mikehh cotto: it requests a password
maybe you can email me what to do at mike.hind@gmail.com 02:19
cotto That should work. Is that the username and hostname you're trying to log in from? 02:20
(i.e. what the computer thinks its hostname is)
try it again 02:22
I missed the first line
(Once I know what works, I can email you.) 02:23
I need to be afk for a while. I you can't get to the ftp site, post the tarball somewhere and I'll make sure it gets up. 02:27
pmichaud (possibly useful datapoint) In nqp-rx commits 4dc9c4b and 6e87137 all I did was modify the regex engine to check the debugging flag to avoid calling the !cursor_debug method if debugging was disabled. That resulted in a 9% performance in Rakudo compilation speed. 02:28
which tells me that method call overhead is a significant chunk of nqp-rx parsing speed.
(as opposed to being algorithmic in nqp-rx itself)
because all !cursor_debug was doing was to check the debug flag and return immediately if off. 02:29
s/performance/improvement/ 02:30
Coke thinks methods have always been expensive. 02:31
one of the reasons I avoided pct for so long. =-) 02:32
pmichaud sure
Coke (and pge. =-)
pmichaud but it also means that we either improve method calls substantially, or we find a way to avoid them.
and the latter is not very appetizing for long-term maintenance and extensibility.
Coke <nod> 02:33
pmichaud ...but now I'm also guessing there must have been something wrong with my measurements. I don't think the debug method should've been responsible for that big an improvement. 02:36
Maybe I accidentally grabbed a parrot change in the process (I don't think I did, but it's possible).
I do know that simply avoiding the debug method from protoregexes made a significant improvement.
02:42 janus joined 02:46 mikehh joined
dalek p-rx: 8b38129 | pmichaud++ | src/ (4 files):
Refactor cursor_debug output a bit to allow some simple profiling.
 02:55
02:57 japhb joined 02:59 tcurtis joined
Coke is there a nice shortcut for //= in nqp-rx? 03:02
(doubt it since it's = and not :) 03:04
dalek rrot: r48560 | gerd++ | trunk/examples/languages/squaak/doc/tutorial_episode_1.pod:
update to current situation
 03:05
Coke hurm; how to check to see if a dynamic var exists before using it? 03:08
pir::defined seems to barf on it.
cotto mikehh, how's the release going? It's getting pretty late (or early) in your neck of the woods, isn't it? 03:13
pmichaud Coke: no shortcut, sorry 03:14
and yes, it currently dies on a non-existent contextual. 03:15
mikehh yeah - about 04:15
pmichaud you could perhaps do pir::isnull(pir::find_dynamic_lex('$*XYZ'))
might need a __s on the find_dynamic_lex
pmichaud@plum:~/nqp-rx$ ./nqp 03:16
> pir::isnull(pir::find_dynamic_lex('$*ABC'))
1
cotto post here if you need a hand with something
pmichaud it won't find '$*ABC' in the current lexical block, but it'll find it in outer dynamic blocks
(find_dynamic_lex only looks in callers, not in the current block) 03:17
Coke urk. ok.
pmichaud you can always turn it into a sub that does what you want, though:
sorear pmichaud: p6eval lurks here now, btw. 03:18
nopaste "mikehh" at 192.168.1.3 pasted "uploaded the files to uploading.com" (7 lines) at nopaste.snit.ch/22864
pmichaud nqp: my $*ABC := 5; { pir::isnull(pir::find_dynamic_lex('$*ABC')); }
p6eval nqp: ( no output )
pmichaud nqp: my $*ABC := 5; { pir::say(pir::isnull(pir::find_dynamic_lex('$*ABC'))); } 03:19
p6eval nqp: OUTPUT«1␤»
pmichaud hmmmm
nqp: my $*ABC := 5; { pir::say(pir::isnull(pir::find_dynamic_lex__s('$*ABC'))); }
p6eval nqp: OUTPUT«rtype not set␤current instr.: 'parrot;PCT;HLLCompiler;panic' pc 146 (compilers/pct/src/PCT/HLLCompiler.pir:109)␤»
pmichaud huh.
mikehh cotto: if you could put those files up on the site
pmichaud nqp: my $*ABC := 5; { my $dummy; pir::say(pir::isnull(pir::find_dynamic_lex__s('$*ABC'))); } 03:20
p6eval nqp: OUTPUT«rtype not set␤current instr.: 'parrot;PCT;HLLCompiler;panic' pc 146 (compilers/pct/src/PCT/HLLCompiler.pir:109)␤»
pmichaud oh well.
cotto sure 03:21
Coke ok. I am now passing in a :namespace(@array), but it doesn't seem to effect the runtime. (I have [namespace current] working, I think. if I do a [namespace eval foo {puts [namespace current]}}, I still get the root NS. 03:22
whoops. :namespace<@NS> ... fixing...
no impact. 03:23
mikehh they should go to ftp/releases/devel/2.7.0
cotto yup
Coke (there's only 2 Past::Block.new()'s in my Actions, and I"ve updated both to respect the @*VAR. 03:24
(and it's definitely finding the @*VAR when it's set.
03:24 ppant joined
cotto worst. filehost. evar. 03:25
mikehh now to post the announcements etc.
cotto mikehh, can you post or nopaste the checksum here? 03:26
Coke feather.perl6.nl/~coke/foo.diff
namespace current must be broken. looks like the EVAL_ generated code contains the right directive. 03:30
sorear cotto: osuosl? or feather?
Coke pmichaud++
cotto neither
uploading.com in this case 03:31
mikehh, checksums?
46296ad1a60c5933fc229ec64f638f2abb66de23bb9b01b3e7c7b248c2b32df0 parrot-2.7.0.tar.bz2 03:33
51f206172b8b6699cc71e366feb111dc32bdd8ed5e3ecfd59ed898c1ee2de2a1 parrot-2.7.0.tar.gz
how do those look?
03:34 estrabd joined
Coke pmichaud: do you know if the :namespace() arg to the past block initialization is relative ? 03:42
ah... was adding in a top level "tcl" I didn't need to (because it's explicitly in the hll) 03:43
pmichaud++
pmichaud :namespace() is never relative, iirc. 03:44
afk briefly
cotto mikehh, are those checksums correct? 03:47
Coke has [namespace eval barf {puts [namespace current]}] working, just need to fix everything else I broke... 03:48
mikehh cotto, dunno, they were generated by the make release VERSION=2.7.0 03:49
cotto you can check them with sha256sum 03:50
Coke cotto: if they're created by the make target, they're fine, neh? 03:51
cotto I just want to be sure that they're the same files. 03:54
mikehh it checks out ok for me as in sha256sum --check parrot-2.7.0.tar.gz.sha256 and sha256sum --check parrot-2.7.0.tar.bz2.sha256 03:58
cotto wfm
mikehh 46296ad1a60c5933fc229ec64f638f2abb66de23bb9b01b3e7c7b248c2b32df0 parrot-2.7.0.tar.bz2 03:59
51f206172b8b6699cc71e366feb111dc32bdd8ed5e3ecfd59ed898c1ee2de2a1 parrot-2.7.0.tar.gz
cotto The files should propagate soon-ish.
mikehh they look the same to me 04:00
dalek rtcl-nqp: 5d277ca | coke++ | / (5 files):
First pass at [namespace eval]

But not much else works. Breaks some tests that use [namespace eval] because it's incomplete, so skip some of those tests.
  Tene++ && pmichaud++ for pointers on the $*PARTCL_COMPILER_NAMESPACE hack.		 04:18
website: mikehh++ | Parrot 2.7.0 "Australian King" Released! 04:27
website: www.parrot.org/news/parrot-2.7.0-au...g-released 04:28
pmichaud mikehh++ # excellent!
Coke mikehh++ !
mikehh Coke: I tried to get the Announcement uop on the site, but it is not positioned right
Coke: and can you update the docs on the site as well 04:30
Coke mikehh: docs.parrot.org ? hell no! =-) 04:31
I can try to fix www.parrot.org though. moment.
mikehh I've had it, anything else I do will really screw up, I'll make the other announcements after some sleep 04:32
Coke mikehh: edit the article, make sure you pick "publishing options -> sticky at top of lists"
mikehh Coke: ah ok, let me see 04:33
pmichaud gist.github.com/533441 # a rough call profile for parsing and compiling src/NQP/Actions.pm 04:37
mikehh How do you get it in the news page as well? 04:38
pmichaud looks like quote_escape wants some optimizing 04:39
Coke mikehh: you have to select News under Categories: News. 04:41
cotto mikehh++
pmichaud, is that just with wallclock timing? 04:42
pmichaud cotto: yes, since that's really what I have available atm.
the timings are a bit bloated by the cost of the profiling, as well.
mikehh must sleep, will do the rest later 04:44
pmichaud I'm going to take a break for now, though. Tomorrow I'll clean it up a bit more and then I'll try running this profiling on core.pm
which is a fairly good stress test 04:45
Coke processes some rakudo tickets to mix it up. 05:27
dalek TT #1174 closed by plobsing++: Disallow .local declarations in long-style call statement 05:54
TT #1174: trac.parrot.org/parrot/ticket/1174
06:23 uniejo joined
dalek rrot: r48561 | NotFound++ | trunk (4 files):
generate a pbc file for several Packfile PMCs tests and use it instead of one in t/native_pbc/, first step in TT #1712
 06:26
06:39 kurahaupo joined 07:06 rblackwe joined, Hunger- joined 07:07 preflex joined 07:14 pjcj joined 07:31 jhelwig joined 07:57 bacek_in_sing joined 08:11 kurahaupo1 joined 08:29 ash_ joined 08:46 ash_ joined
dalek parrot: bedb521 | leto++ | t/sql/plperl6.sql:
Add a test for PL/Perl6 with regexen
 08:53
09:16 masak joined
dzoe win 42 09:25
oops, sorry
09:29 lucian joined 09:37 theory joined
sorear phasers? 10:21
purl Phasers and futon turpedoes ready, Captain sorear.
sorear rakudosketch?
moritz phasers is the new rakudosketch 10:22
purl: phasers?
purl Phasers and futon turpedoes ready, Captain moritz.
sorear yeah, I was trying to get the timeslot 10:24
rather than having to convert CDT to UTC by hand
moritz should be 19H UTC 10:25
ah, I see you've already found that out :-)
masak purl: no, phasers is a weekly Perl 6 implementation update meeting on #phasers on freenode, every Tuesday at 19 UTC o'clock 10:29
purl okay, masak.
masak phasers?
purl phasers is a weekly Perl 6 implementation update meeting on #phasers on freenode, every Tuesday at 19 UTC o'clock
masak purl++
11:36 lucian joined 11:59 Paul_the_Greek joined
Paul_the_Greek Australian King! 11:59
mikehh++
12:13 pmurias joined 12:25 JimmyZ joined, Paul_the_Greek joined
Paul_the_Greek In the implementation of a PMC, what specifies that it is a subclass of the scalar PMC? 12:27
12:30 ruoso joined
NotFound Paul_the_Greek: extends scalar 12:33
Paul_the_Greek How is that specified in one of the built-in PMCs, in its .pmc file? 12:36
NotFound Paul_the_Greek: in the pmc file, in the pmclass declaration.
Paul_the_Greek I don't see it in BigInt or Complex. 12:37
NotFound pmclass Integer extends scalar provides integer provides scalar auto_attrs {
Paul_the_Greek Wait, it is in Integer. 12:38
Integer extends scalar but BigInt doesn't. 12:40
NotFound Yes, but don't ask me why, I don't know. 12:42
Paul_the_Greek And it means, for example, that BigInt doesn't handle the logical operators.
At least not at first glance. 12:43
I will ask this question again later today.
NotFound Maybe it has something to do with multiple dispatch. scalar provides it for most basic types, others must do by themselves. 12:44
12:46 whiteknight joined
NotFound mikehh++ 13:00
dalek rrot: r48562 | NotFound++ | trunk (5 files):
generate a pbc file for packfileannotations test, TT #1712
 13:24
13:26 khairul joined
Coke is there any way to use an array like ['tcl', 'namespace', 'barf'] as a way to find an HLL global in NQP? or do I have get the hll namespace and walk it? 13:27
(I have an array that represents the NS)
NotFound Coke: I think you can use only a key, not an array 13:29
Oh, NQP, sorry. I don't know. 13:30
Coke turns out I already have the object I'm trying to get, so I can avoid the lookup entirely. 13:40
dalek rrot: r48563 | NotFound++ | trunk/docs/embed.pod:
remove mentions of Parrot_find_global_s and Parrot_store_global_s from docs/embed.pod, TT #1660
TT #374 closed by NotFound++: Parrot_store_global_s: Use HLL root namespace if str_key is NULL. 13:46
TT #374: trac.parrot.org/parrot/ticket/374
13:46 bubaflub joined 13:50 bkuhn joined
dalek kapo: cb9b6b0 | whiteknight++ | src/Exceptions.nqp:
renumber exceptions since Parrot lost one
 14:07
14:09 hercynium joined 14:14 plobsing joined 14:23 ash_ joined 14:25 davidfetter joined 14:46 estrabd joined 14:57 allison joined 15:04 davidfetter joined 15:14 brianwisti joined 15:34 Andy joined 15:37 pyrimidine joined 15:55 tcurtis joined
Coke hurls failblog.org/2010/08/18/epic-fail-p...-name-fail for TimToady 16:13
brianwisti eq. 16:14
ew, that is. So disgusted that I couldn't even get two characters out properly.
jnthn Or is it just a clever bit of viral marketting?
Coke IT'S A WRAP! 16:16
brianwisti "Try our new pandemic burger!" 16:17
atrodo I like the 14lb burger for 2.20
brianwisti Yeah. Anyways.
Good price.
atrodo Could feed a family of 28 for that
s/for/with/
brianwisti Now I'm thinking of "A Modest Proposal" instead of code. 16:22
cotto docs.parrot.org is up 16:23
dated
dukeleto cotto++ 16:24
tcurtis interpreted that as STD.pm. 16:28
s/pm/pm6/ 16:29
Coke huh. it /feels/ like partcl is faster.
if I run 'git diff', long lines are chopped off, but git diff | more shows them. is this git, or my pager, or...? 16:30
(piping directly to $PAGER, I don't see the issue.) 16:31
dalek rtcl-nqp: 63f2bf6 | coke++ | / (5 files):
fix [namespace eval foo { proc ...}].
 16:34
atrodo when I run git diff, if it's more than a page, it pipes it to my pager(less) 16:37
so I'd guess, it's git diff trying to be magical
16:42 theory joined
cotto_work ~~ 16:45
pmichaud Coke: partcl may be faster due to the parsing improvements I made, fwiw. 16:57
(or it could be due to parrot improvements, also)
16:58 Paul_the_Greek joined
Paul_the_Greek purl,messages 16:59
Coke pmichaud++
pmichaud I'm making more today -- I found a nice place for improving parsing speed 17:00
it's not likely to be order-of-magnitude, but it should be a bit more help
Paul_the_Greek msg cotto Will you commit the patches in tickets 481 and 1207? 17:03
purl Message for cotto stored.
cotto_work updates his todo list 17:04
17:05 ash_ joined
Coke Paul_the_Greek: did anyone volunteer to be your mentor? 17:06
cotto_work We forgot to do that. 17:07
Paul_the_Greek Not yet, no.
Oh, unless it was done on the list.
Coke ugh. every time I think I made process on this namespace thing (and move on to the next chunk) I realize that the only reason it looked like it worked is because I had 2 different errors that were cancelling each out.
Paul_the_Greek Yup, I still need a mentor. 17:09
If the two errors always cancel, call it a feature!
cotto_work: Sent you a msg at cotto. 17:10
Is there a document for newbie comitters to read? 17:14
cotto_work I saw
docs/project/branching_guide.pod helps with branching 17:15
I'm not sure about a basic svn doc
Paul_the_Greek Oh, and a committer guide, too. 17:17
whiteknight committer guide is short: "you break the build, I break your face" :)
just kidding. I don't do any face-breaking
Paul_the_Greek A little face slapping will be okay. 17:19
I hear the beach calling. Be back later. 17:21
Coke hurm. I have a namespace eval foo {proc joe ...} and later, namespace eval foo {joe} - when trying to invoke joe, I cannot find it, despite the fact that joe is compiled into the right namesace, and just before I try to find "joe", I verify that my current namespace is in foo. I'm trying to use "get_global" to find the named sub in my current namespace. does this sound like it should work? 17:30
moritz are you in the same HLL both times? 17:31
Coke yup.
(always in 'tcl' hll)
ah... my current namespace fubar. 17:32
(because I'm using a trampoline that's only in one place)
17:37 bubaflub joined
cotto_work msg paul_the_greek TT #1207 needs a documented upgrade path before the patch can be applied. 17:37
purl Message for paul_the_greek stored.
dalek kudo: 7f39959 | pmichaud++ | docs/announce/2010.08:
Update release contributors.
 17:43
kudo: 72f5b8b | pmichaud++ | src/Perl6/Grammar.pm:
Refactor handling of $*QSIGIL in token variable to avoid 'return ()'.
kudo: e260f09 | pmichaud++ | build/Makefile.in:
Add 'make parsetrace' target for internal profiling/debugging.
pmichaud ♥ git. 17:45
Two people who have made my life far simpler and more fun: Larry Wall (Perl, patch, rn) and Linus Torvalds (Linux, git) 17:46
oops, wrong chan (but still useful :) 17:47
dalek kudo: 56c8464 | pmichaud++ | docs/ROADMAP:
Move "port to other VMs" to a priority 3 level -- I think priority 2 overstates it a bit.
 17:49
17:53 x3nU joined
x3nU hi 17:53
purl hi, x3nU.
18:09 p6eval joined, tadzik joined 18:19 davidfetter joined
dalek p-rx: d304887 | pmichaud++ | src/Regex/Cursor-builtins.pir:
Add debugging/profiling to <ww> builtin.
 18:30
p-rx: 1408049 | pmichaud++ | src/cheats/hll-grammar.pir:
Add debugging/profiling for <EXPR>.
p-rx: f49a206 | pmichaud++ | tools/analyze-parse:
Add analyze-parse, a simple Perl script for summarizing --parsetrace output.
18:34 tadzik joined
dalek parrot: ac58f42 | davidfetter++ | html/index.html:
Added download link.
 18:41
18:43 Andy joined
davidfetter say, Andy, what do i do to get vim 7.3 to see i'm writing perl6? is it a suffix-based thing, or...? 18:51
Andy you have a "use v6" at the top?
You may have to "# vi: ft=perl6"
in your file somewhere
davidfetter k
Andy It is not robust, as the perlbuzz article notes 18:52
davidfetter is there an agreed suffix for perl6 files that's not .pl ?
sorear Is .pm6 supported?
davidfetter 'sup stefan
Andy davidfetter: No, there is not a standard yet. 18:53
so I didn't try to cram one in
davidfetter lobbies for .p6
and .pm6
Andy github.com/petdance/vim-perl/issues#issue/30
18:55 Chandon joined
cotto_work msg paul_the_greek If you want to add the upgrade path yourself, look at docs/project/support_policy.pod under Deprecations and Deprecation Notifications. 19:00
purl Message for paul_the_greek stored.
19:03 kurahaupo joined 19:04 Chandon joined
Coke wonders why tcl allows 2 /or more/ semicolons as a NS separator. 19:20
cotto_work foo;;;;;;;;;;;;;;;;;;;bar;;;;;;;;;;;;;;;;;;;;;;;;baz
definitely more readable 19:21
Coke er, s/semi//
odd thinko.
Just annoying because it means I need split-with-regex, not split-on-chars. =-) 19:22
on the plus side, I love that they have a comparator called "ni" ! 19:23
cotto_work !
what does it mean? 19:25
Coke not in list.
partcl-nqp: set a [list 1 2 3 4] ; puts [expr 1 ni $a]; 19:26
p6eval partcl-nqp: OUTPUT«1␤»
Coke partcl-nqp: set a [list 1 2 3 4] ; puts [expr 1 in $a];
p6eval partcl-nqp: OUTPUT«1␤»
Coke looks like it's broken. =-)
cotto_work has a new favorite comparator
atrodo ni++ # perl has missed the target on this one 19:30
19:32 tcurtis joined
cotto_work hio tcurtis 19:33
tcurtis Hello, cotto_work.
19:58 davidfetter joined
dalek rrot-linear-algebra: 6dfdf6d | Whiteknight++ | / (2 files):
improve stringification of elements in NumMatrix2D to follow best-practices and be more compatible with the get_string vtables of other matrix types. Fix string tests for the type to match the new behavior
 20:07
rrot-linear-algebra: 21e2e9b | Whiteknight++ | / (2 files):
Add missing vtables to pmcmatrix2d, which helps avoid some segfaults when I added all those new tests. Update MatrixTest tests to be more general and not specific to NumMatrix2D. All tests for PMCMatrix2D and NumMatrix2D pass now
rrot-linear-algebra: 2836508 | Whiteknight++ | src/pmc/complexmatrix2d.pmc:
Add several arithmetic VTABLEs to ComplexMatrix2D, and add the elementary row operation methods too. Still fails some tests, but runs the suite to conclusion without segfaults and passes most tests which have been implemented (fails some tests which haven't been written yet)
rrot-linear-algebra: 2c6fd2b | Whiteknight++ | / (3 files):
ComplexMatrix2D and NumericMatrix2D both 'does numericmatrix' to show their similarities. Also, the NumMatrix2D tests should inherit from NumericMatrixTest, which adds several more tests for arithmetic routine. Changing this exposes some broken tests, but it's more honest
rrot-linear-algebra: 752cf19 | Whiteknight++ | t/pmc/nummatrix2d.t:
remove TODO'd tests which don't even have the correct names
rrot-linear-algebra: ae283cf | Whiteknight++ | t/harness:
some upgrades to the harness to give a comprehensive report of all tests which fail in each file. This is something that's been missing and is a welcomed addition
20:21 kurahaupo_mobi joined 20:22 fperrad joined 20:54 theory joined 20:57 Paul_the_Greek joined
Paul_the_Greek Can someone tell me what vtable.h is generated from? 20:57
sorear vtable.tbl?
cotto_work yup, that 20:58
purl rumour has it yup, that is the one i installed
Paul_the_Greek Checking ...
purl it has been said that Checking ... is it complicated to get to work with google and yahoo ?
cotto_work forget checking ... 20:59
purl cotto_work: I forgot checking
Paul_the_Greek Yes, it could be that, sorear. I can't see vtable.h online, so I can't tell. 21:00
cotto: Question about the upgrade path here, or send you a message? 21:02
cotto_work here works 21:03
Paul_the_Greek Since find_lex throwing an error was deprecated in 2.4, why would we need an upgrade path now? 21:04
cotto_work A user might be relying on that behaviour. We want to have a well-documented upgrade path for any features that are removed so that people know what needs to change to get their code to work with the latest Parrot. 21:06
s/latest/latest stable/
Paul_the_Greek But wouldn't that upgrade path have been specified when the error-throwing was deprecated in 2.4?
Oh wait ...
purl wait ... is that one of his other silly conventions? placing a 'G' in front of any account that he thinks is global?
Paul_the_Greek Maybe you are referring to the flag that we are deprecating now. 21:07
cotto_work Nope. The deprecation notice just means "this feature will be removed after the next stable release".
Paul_the_Greek But the upgrade path isn't described until after it's actually removed? Isn't that too late? 21:08
cotto_work Not really. People can continue to depend on 2.6.0 while we work on 2.7.0 and 2.8.0 and only upgrade at the next stable release. 21:09
Users are free to track trunk or the latest dev release if they want, but we work under the assumption that most users will track the stable releases. 21:10
Paul_the_Greek Hold on ... 21:11
cotto_work holds on
brianwisti I track stable for my own work, and just keep abreast of what's going on in dev.
21:13 whiteknight joined
Paul_the_Greek Okay, so we can wait until I return home to add the upgrade path, or you can do it for me. 21:14
It's up to you, since you have done so much.
cotto_work I'd prefer that you do it, if only to increase the number of people who have added upgrade path documentation. 21:16
Paul_the_Greek Okay, will do. I'll add a note to the ticket that it's waiting, if only to remind myself.
cotto_work Paul_the_Greek++ 21:17
21:17 kurahaupo joined
Paul_the_Greek The upgrade path won't be clever. The user of find_lex will have to check the return value and throw an exception. 21:17
cotto_work That's fine, as long as it's reasonably sand and well-documented. 21:18
Paul_the_Greek Okay, ticket flagged. I'll jump on it as soon as I return home. 21:19
Thanks again, cotto.
cotto_work np and thanks for your work so far 21:20
Paul_the_Greek My pleasure. 21:21
The link to dep.pod on the Support Policy page is still broken.
21:23 bubaflub joined
dalek p-rx: aa0e12b | pmichaud++ | src/cheats/hll-grammar.pir:
Add debugging/profiling to quote_EXPR.
 21:24
p-rx: e0d2bec | pmichaud++ | src/PAST/Compiler-Regex.pir:
Optimize quantified+ratcheted character lists a bit, such as \\h* and \\v+ .
p-rx: a56f707 | pmichaud++ | src/stage0/ (4 files):
Update bootstrap files.
Paul_the_Greek Hang on! I can edit the deprecations page from here.
cotto_work Yes, you can.
Just don't get sand into your keyboard. ;) 21:25
Paul_the_Greek Well then, I shall do that now and you'll be able to commit the patch. I'll message you when it's ready.
cotto_work sounds great
Paul_the_Greek Brushing away sand ...
21:29 lucian_ joined
Paul_the_Greek cotto? 21:37
purl cotto is Christoph Otto <mailto:christoph@mksig.org> or a cooked salami or The Decider or the Lorito Coordinator
cotto_work me? 21:38
purl cotto_work is always at work or cotto's employed alter ego
Paul_the_Greek Right, so I edited the wiki page. But now it describes an upgrade path to something that is not in the current release.
cotto_work That's fine. It describes an upgrade path to something that'll be in the 2.9.0 release. 21:39
Paul_the_Greek Okay then, we're all set. I'll annotate the ticket yet again to say that I added the upgrade path.
cotto_work Actually, the change should go into ParrotDeprecations and ParrotDeprecationsFor2.9 21:40
There are some examples you can copy/paste from as templates.
whiteknight would be much nicer if track wiki supported template transclusion 21:41
whiteknight pines for MediaWiki
cotto_work That other deprecation page should go away or be modified to avoid confusing more people.
and a simple toc
(without stupid hacks) 21:42
Paul_the_Greek Oops, dinner. I'll leave this for now and visit it again tomorrow. You'll be hearing from me.
cotto_work That's what they all say.
Paul_the_Greek I'll call you ...
dalek tracwiki: v12 | Paul C. Anagnostopoulos++ | Deprecation 21:45
tracwiki: trac.parrot.org/parrot/wiki/Depreca...ction=diff
tracwiki: v13 | Paul C. Anagnostopoulos++ | Deprecation
tracwiki: trac.parrot.org/parrot/wiki/Depreca...ction=diff
cognominal how come perl6.c takes so long to compile on a 2Go box? 21:49
davidfetter maybe it's part of the slow food movement 21:52
sorear 2Go? 21:55
giga-octet?
cognominal yup
davidfetter any relation to 2pac? 21:57
cognominal oops, I meant 2GB 21:58
davidfetter suddenly this makes *way* more sense 22:03
cotto_work thought it was some kind of trendy netbook 22:04
cognominal no, that's just a clueless French talking proken English.
*broken
davidfetter parle francais comme une vache espagnole :P 22:07
22:08 bubaflub joined
davidfetter can't even figure out to get c-cedille on his macbookpro 22:09
NotFound davidfetter: muuuu 22:13
davidfetter mdr
cognominal :) 22:14
Alt-C C
davidfetter 's first language was french, but that was a *long* time ago
comme ça? 22:15
cognominal I have an azerty keypbard but I prefer to use it as a international english keyboard. It is easier for Perl and C. 22:16
NotFound Maybe I should do the same with my spanish keyboards. 22:17
It's a vaste of space having a key for the Ñ 22:18
cognominal you will do it mañana... 22:19
davidfetter oh, and who thought it was a *good* idea to put an "enter" key right between the right apple and the left-arrow? :P 22:21
NotFound ...en España.
davidfetter notes that modern macbookpros do not have this silliness
sorear a ñ key sounds like a waste of space even if you *are* using Spanish... 22:28
22:29 tcurtis joined
davidfetter Feliz Año Nuevo is *very* different if the ~ isn't there 22:30
NotFound sorear: it is, but when it was suggested years ago to supress it, people took it like an insult to the country. 22:31
sorear davidfetter: it doesn't need a whole *key* though, that sounds like a job for shift-something 22:32
NotFound davidfetter: we don't need a key for a acute, we also don't need it for n tilde.
davidfetter fair enough
22:45 theory joined 23:07 hercynium joined 23:29 theory joined 23:31 ruoso joined, aloha joined 23:32 bacek joined 23:34 dngor_ joined 23:55 bubaflub 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!