Parrot 2.6.0 | parrot.org Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | fix 'make html' (talk to Coke), update tutorial (talk to tcurtis)
Set by moderator on 21 July 2010.
00:31 kid51 joined 00:33 s1n joined 00:51 ingy joined 00:53 snarkyboojum joined 00:57 ingy joined 01:17 rurban_ joined
darbelo NotFound: ping 01:48
01:54 cognominal joined
cotto It's nice to see someone obliterating whitespace instead of just removing or deleting it. 02:04
darbelo Next time, It'll be anihilation. 02:07
02:09 snarkyboojum joined
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#35089), fulltest) at r48178 - Ubuntu 10.04 amd64 (g++ with --optimize) 02:26
02:46 tcurtis joined 02:47 janus joined
ttbot Parrot trunk/ r48179 MSWin32-x86-multi-thread make error tt.taptinder.org/file/cmdout/359555.txt ( tt.taptinder.org//buildstatus/pr-Pa.../rp-trunk/ ) 03:04
03:04 Coke joined
dalek rrot: r48179 | chromatic++ | trunk/config (2 files):
[config] Allow spaces in build directory.

to build Parrot. pbc_to_exe needs additional logic to escape spaces in paths, but that's a bigger challenge for later. Tests pass for me on a checkout without spaces too.		 03:06
sorear Do we have CJK in build directories working yet?
mikehh rakudo (b46a3b3) builds on parrot r48178 - make test PASS, spectest_smolder -> #35090 (pugs r31823) PASS - Ubuntu 10.04 amd64 (g++ with --optimize) 03:21
t/spec/S04-statements/for.rakudo - TODO passed: 49
partcl-nqp builds on parrot r48178 - make test PASS - Ubuntu 10.04 amd64 (g++ with --optimize) 03:38
winxed builkds on parrot r48179 - make test/test1/test2 PASS, some examples ok 03:44
that should be r14178 - haven't tried r48179 yet 03:45
03:59 LoganLK joined
mikehh plumage ok on parrot r48178 - tests ok - Ubuntu 10.04 amd64 (g++ with --optimize) 04:01
plumage - tree-optimization builds, tests, install ok
04:39 Austin joined 04:56 jsut joined 05:00 plobsing joined 05:05 snarkyboojum joined 06:13 uniejo joined 06:21 galf joined 06:25 galf left 07:24 baest joined
dalek kudo: 807748a | (Fitz Elliott)++ | src/core/operators.pm:
force xx op to numify rhs except Whatevers
 07:25
07:47 snarkyboojum joined 08:30 fperrad joined 08:35 AndyA joined 09:17 rurban_ joined 09:30 lucian joined
dalek kudo: 4884806 | moritz++ | src/Perl6/Actions.pm:
turn s/// into a call
 09:56
kudo: ed98e0f | moritz++ | src/Perl6/Actions.pm:
add :node to substitutions, to get line numbers in error messages
09:57 AndyA joined 10:04 jan joined
dalek kudo: 9808d7c | moritz++ | (4 files):
remove old Substitution hack
 10:19
10:32 macroron joined
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#35101), fulltest) at r48179 - Ubuntu 10.04 amd64 (gcc) 11:33
nopaste "mikehh" at 192.168.1.3 pasted "still major problems with space in directory name (t parrot)" (88 lines) at nopaste.snit.ch/22302 11:48
11:51 whiteknight joined
whiteknight goodmorning, #parrot 11:54
moritz good morning mr. withworth!
mikehh hi whiteknight 11:58
12:03 bluescreen joined
whiteknight hello moritz, mikehh 12:05
how are you fine gentlemen doing todaY?
moritz well, it's Monday. 12:06
any further questions? :-)
actually, not too bad. Already got a patch into rakudo today, s/// style substitution now works in any context
whiteknight yes, I saw that. very nice 12:07
12:07 lucian_ joined
whiteknight some people read the newspaper in the morning, I read the commit logs :) 12:08
moritz :-) 12:09
Coke if we're using NQP's expr parser, should we get parens for free? 12:45
... hey, do we still have an evalbot in here? 12:46
moritz Coke: want one? 12:51
Coke moritz: yes please. (mainly for partcl-nqp for me) 12:52
12:55 p6eval joined
moritz nqp: say(1) 12:55
p6eval nqp: OUTPUTĀ«1ā¤Ā»
atrodo oooh, shiny
Coke moritz: would it be possible to add partcl-nqp there? 13:11
moritz Coke: yes, I just have to set it up first 13:12
13:35 particle1 joined 13:46 bluescreen joined, bubaflub joined 13:48 bubaflub left 14:16 whiteknight joined 14:23 Andy joined 14:28 preflex joined 14:56 bubaflub joined 15:13 jsut_ joined 15:16 bubaflub joined
Coke man, #perl6 is going to get chaaaaaaaatty. 15:32
atrodo freenode, right?
Coke ja 15:33
atrodo joins to watch the chatter
16:06 eternaleye joined
cxreg hey, if i modify src/nci/extra_thunks.nci, it generates src/extra_nci_thunks.c, but does not modify src/nci/extra_thunks.c 16:07
am i doing it wrong?
the latter appears to be the important one
Coke: due to R* you mean? 16:08
that's thursday
in addition, src/extra_nci_thunks.c is only local, but src/nci/extra_thunks.c is tracked by SVN 16:12
and the diff between them is not just my changes
although a lot of it is just whether CURRENT_CONTEXT(interp); is const 16:13
16:19 theory joined
dalek rtcl-nqp: 69a50b8 | Coke++ | (3 files):
make [expr] support parens
 16:31
Coke cxreg: yes, due to R*. it's already more chatty today.
Coke susses the paren issue.
Coke was expecting some magical circumfix rul einherited from the expr grammar, but that wasn't it. =-)
dalek nxed: r581 | NotFound++ | trunk/examples/fly.winxed:
add a bridge and use real time in example fly
 16:38
16:39 Austin_Hastings joined
cxreg nobody has a hint on generating a proper version of extra_thunks.c? 16:54
mikehh cxreg: I think plobsing did most of the work there 16:57
cxreg it looks like maybe the generating program is just outdated 16:59
but i'm not sure
./parrot_nci_thunk_gen
darbelo It could be. We have a checked-in version of the output for bootstraping purposes. 17:00
cxreg darbelo: right, so the checked-in one is the one that's actually used, but the makefile writes to a different file
and the resulting files are different
darbelo Can you nopaste a diff? 17:01
cxreg sure
nopaste "cxreg" at 192.168.1.3 pasted "src/extra_nci_thunks.c vs src/nci/extra_thunks.c" (2330 lines) at nopaste.snit.ch/22308 17:02
cxreg the most interesting and significant diff is the function name at the bottom 17:03
NotFound darbelo: pong 17:04
darbelo Yeah.
cxreg while we're working on it, I'd like to add some thunks for postgres' SPI 17:05
darbelo cxreg: src/nci/extra_thunks.nci is the file to do that, I think. 17:06
cxreg right, that's how i found out
i added to that file, ran make, and got no results
migrated the generated changes over to src/nci/extra_thunks.c, and then it worked
Coke I thought one of those files was for embedding, and one was for parrot. 17:07
darbelo You can pass a -o /path/to/outfile option to nci_thunk_gen
cxreg darbelo: except that the contents are wrong, including the function name
anyway, i think it's just that this is outdated 17:08
cxreg looks at the src
mikehh nearly all the diffs are comst or not
darbelo NotFound: Do you think the bytebuffer PMC is up to replacing the various _BYTE and _BYTES accessors in the string API?
mikehh const
cxreg mikehh: yes, except for the last hunk 17:09
darbelo A few of them are unused, and others look mis-used in places, if the bytebuffer PMC can cover the same needs for external uses, I can probably refactor them away. 17:11
NotFound darbelo: What accessors? The ones in the charsets and encodings?
darbelo Yep.
NotFound darbelo: If they aren't used is speed critical parts, I think BB can do the job. 17:12
darbelo Excellent. 17:13
NotFound++ 17:14
NotFound Maybe we need to add a sort of substr to BB 17:16
darbelo Probably, that would be a good replacement to ENCODING_GET_BYTES(), but I wouldn't really hurry, since it's unused iside parrot :) 17:18
17:19 rurban_ joined
NotFound We can just wait to see if someone ask for it. 17:19
cotto_work ~~ 17:22
cxreg ./parrot_nci_thunk_gen --core --loader-name=Parrot_nci_load_extra_thunks --no-warn-dups --output=src/nci/extra_thunks.c <src/nci/extra_thunks.nci 17:23
seems to work
maybe we just need to add that to the makefile
17:26 jsut joined
darbelo We store the makefile templates in config/gen/makefiles/ and great at applying patches form out users :) 17:30
cxreg yeah, i'm working on it :)
darbelo cxreg++
cxreg what's the best form for patches?
email?
purl email is for old people
cxreg lol
darbelo Attaching them to a Trac ticket or just nopasting them here. 17:31
cxreg oke
17:32 tcurtis joined
nopaste "cxreg" at 192.168.1.3 pasted "nci makefile patch" (13 lines) at nopaste.snit.ch/22309 17:36
cxreg that turned out to be as simple as I hoped
the rebuilding and testing took a while, lol
darbelo cxreg: Looks good. Any name you want to be credited by in the commit message? 17:43
cxreg Dave Olszewski <cxreg@pobox.com> 17:46
thx
dalek rrot: r48180 | darbelo++ | trunk (2 files):
Modify the makefile to keep src/nci/extra_thunks.c properly updated.
 17:58
cxreg woot thx 18:02
plparrot will be actually useful soon 18:03
dukeleto++ for all his work on it so far
cotto_work cxreg is Dave Olszewski <cxreg@pobox.com>
18:14 patspam joined
dukeleto cxreg++ for hacking on PL/Parrot and improving Parrot on the way 18:17
18:50 ruoso joined 19:07 bubaflub joined 19:16 dalek joined
Coke whee, only one more feature needed before I can run init.tcl (again) with no errors. (of course, it won't DO anything, but at least it will complete) 19:30
19:53 dalek joined
tcurtis github.com/ekiru/nqp-rx/commit/3df1...bbb2c60a1c Tree::Optimizer is sufficiently working to rewrite my example constant-fold optimization with it. :) 20:17
20:27 ash_ joined 20:33 lucian joined 20:40 mokurai joined
cotto_work I don't entirely understand what that means, but it looks shiny. tcurtis++ 20:44
darbelo I think it's a dog food advertisment. 20:48
cotto_work I love dogfood.
I do, however, note that the change results in more code instead of less. 20:49
tcurtis cotto_work: it means fewer monkey-patched methods onto compiler classes for writing optimizations with my GSoC. :) And eventually shiny new features.
cotto_work so better encapsulation? 20:50
sorear seen Schwern 20:57
purl Schwern was last seen on #perl 8 days, 14 hours, 16 minutes and 18 seconds ago, saying: Which is actually the easy way to do it [Jul 18 06:41:05 2010]
tcurtis cotto_work: That's all it allows for now, yes. 21:00
cotto_work encapsulation is worthwhile
21:01 whiteknight joined
whiteknight Tene: ping 21:11
Tene whiteknight: pong
tcurtis cotto_work I don't think it's a significant increase in the amount of code compared to doing the same thing without Tree::Optimizer, though. Just 3 lines really(load Tree/Optimizer.pbc, create a Tree::Optimize, run it after registering passes). The encapsulation is possible without it, but it would be a bit more fragile in that you have to handle threading the result of each optimization to the next yourself and handle the ordering yourself. 21:12
whiteknight Tene: that :nsentry change you made to NQP-RX, is that in the repo now?
Tene whiteknight: I was hoping to get confirmation from you or Austin_away that it worked before trying to find the docs to import it into the parrot repo.
So, no, it's not in the parrot repo yet.
whiteknight Tene: oh, okay. I'll have to figure out how to test it then 21:13
darbelo whiteknight: grab a nqp-rx checkout from github and install it. 21:15
Tene you don't even need to install it.
whiteknight darbelo: will it install over the parrot-nqp that comes with parrot?
Tene you can just compile it and use the 'nqp' binary in the local dir 21:16
darbelo It'll be available as nqp, instead of parrot-nqp, IIRC.
Tene I did my testing with just ./nqp
whiteknight gotcha 21:18
21:44 snarkyboojum joined
dalek nxed: r582 | NotFound++ | trunk/examples/fly.winxed:
reset timer before exiting and add alternate point f view in example fly
 21:47
22:23 snarkyboojum joined 22:26 AndyA joined 22:29 AndyA joined 22:34 jan joined 23:09 hudnix joined 23:10 snarkyboojum joined 23:26 theory joined 23:27 snarkyboojum joined 23:32 AndyA_ joined 23:40 ruoso joined 23:47 mikehh joined 23:57 Psyche^ joined 23:59 snarkyboojum joined

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!