Parrot 2.10.1 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Long live Git! github.com/parrot/parrot | git clone git://github.com/parrot/parrot.git
Set by moderator on 18 November 2010.
allison it didn't used to, we used to separate "make install" from "make install-dev" 00:00
packages for Linux distributions still make that distinction
(because it's a standard distinctions for all packaging systems)
another vector to think about is who is offering to support the code 00:01
00:01 dmalcolm left
allison if the parrot developers aren't willing to support it, it shouldn't be in the base tarball 00:01
TypeNameHere_ Excuse me for being a n00b around here, but what exactly is the point of parrot? What is the benefit of using this over separate VMs for different laguages?
allison TypeNameHere_: it's difficult to answer that question the way you've asked it, perhaps you could be more specific? 00:03
dukeleto TypeNameHere_: one VM codebase can be used to run many many dynamic languages
TypeNameHere_: and then languages can share resources, like libraries and other bits
cotto TypeNameHere_, it depends on who's asking. There are different answers for HLL users, HLL developers and library developers
dukeleto TypeNameHere_: because things boil down to a bytecode that they all understand
TypeNameHere_: HLL = High Level Language, i.e. a language running on Parrot VM 00:04
allison cotto: for example, we ship some copies of external software in /ext, but even there, we are prepared to answer questions, accept bug reports, and solve problems for the version we ship
TypeNameHere_ dukeleto: alright, thank you.
plobsing TypeNameHere_: there's a lower barrier to entry to get a minimal language up and running, less work to get a decent library, and less work to get it embedded in places you never thought of (eg: postgres)
dukeleto TypeNameHere_: do you have an interest to use Parrot for something?
TypeNameHere_ dukeleto: not directly; I have an interest in perl 6 00:05
dukeleto: although, it sounds cool enough to start seeing how I can help.(starting at www.parrot.org/dev) 00:08
cotto allison, thanks 00:16
00:28 whiteknight joined
dukeleto TypeNameHere_: sounds good. If you use Rakudo Perl 6, your interest in understand Parrot will grow :) 00:29
s/in understand/in understanding/
TypeNameHere_ dukeleto: actually, I only use 5 so far; I want to adopt the new standards for 6 and, conveniently, I have been very strongly considering going for python as well 00:34
00:34 mikehh left
dalek rrot: 3b80d03 | jkeenan++ | / (2 files):
[configure] Applying patch for TT #1862:

gcc -fvisibility=hidden support.		 00:35
dukeleto I would like feedback on our brand spanking new Google Code-In 2010 Homepage: www.google-melange.com/gci/org/home...oundations 00:36
00:40 jsut joined 00:41 kid51 joined
dalek TT #1861 closed by jkeenan++: ipv6 config probe too noisy 00:44
TT #1861: trac.parrot.org/parrot/ticket/1861
00:45 jsut_ left
nopaste "kid51" at 192.168.1.3 pasted "generational_gc branch: build failure at commit 86871e50abf (11 Nov)" (8 lines) at nopaste.snit.ch/26102 00:46
kid51 That was on a Linux/i386 box with ample resources. 00:47
bacek_at_work kid51, erm. There is no "failure" in nopaste
whiteknight dukeleto: page looks nice. Very functional 00:51
dukeleto whiteknight: danke
whiteknight: i went through about 20 revision
s
whiteknight kid51: I don't think the generational_gc branch is buildable now. The important part there is bacek's new GC design in the comments 00:52
dukeleto just made the tinyurl icanhaz.com/gci2010 for our page 00:54
dalek rrot/embed_api: c4ccd13 | Whiteknight++ | / (10 files):
first prototype of a new mechanism for setting the configuration hash as a PMC
 00:55
rrot/embed_api: 1a72bd7 | Whiteknight++ | / (3 files):
miniparrot builds again and does not segfault. It fails in other ways
rrot/embed_api: 65bf81a | Whiteknight++ | / (5 files):
actually run the code, don't just look at it. Miniparrot runs now. parrot executable fails (but I suspect the problem is easy to fix)
rrot/embed_api: 291f4c3 | Whiteknight++ | src/embed/pmc.c:
add pmc.c for bluescreen++
rrot/embed_api: 5c2c3d4 | Whiteknight++ | / (3 files):
fix library paths. The build now progresses through all the parts that require the parrot executable and we now need to fix all the ancillary utility programs to also work
 00:56
cotto We seem to have 1995K worth of functions in libparrot.so. It'd be interesting to survey where they come from. 00:57
whiteknight 1995K worth of functions? 00:59
cotto yup 01:00
dalek TT #1861 reopened by coke++: ipv6 config probe too noisy
TT #1861: trac.parrot.org/parrot/ticket/1861
TT #1861 closed by coke++: ipv6 config probe too noisy
TT #1861: trac.parrot.org/parrot/ticket/1861
nopaste "cotto" at 192.168.1.3 pasted "lots and lots of functions" (2 lines) at nopaste.snit.ch/26103
cotto the stripped library is 2.2M 01:02
kid51 bacek: I should have made it clearer: The build hung indefinitely at that point. Tried it twice.
So, as whiteknight said, it's not currently buildable
bacek_at_work kid51, ok. How much memory do you have?
(And it was buildable when I stop working on it)
kid51 This is not one of my personal boxes .... but it's much faster than any of mine: dual core and all that 01:03
It hung for at least 5 minutes before I killed it
whiteknight cotto: did the dynop_mapping branch land?
kid51 Will try to file more complete report next week
nwellnhof The generational_gc build also hangs for me when building ops2c
Linux/i386 in a VM with 768MB RAM 01:04
cotto whiteknight, it landed quite a while back.
01:05 pmichaud left, kid51 is now known as kid51_at_dinner, Util left, tadzik left
whiteknight oh, nice 01:06
cotto++ on the blog post 01:10
dukeleto link? 01:11
cotto reparrot.blogspot.com/2010/11/what-...ranch.html 01:12
dukeleto how do we make planet.parrotcode.org/ -> planet.parrot.org ?
can someone setup the CNAME?
everything else on parrocode is deprecated
i meant planet.parrot.org > planet.parrotcode.org 01:13
cotto whiteknight, thanks
bacek_at_work dukeleto, trac.parrot.org/parrot/ticket/1413 01:14
dukeleto has evidently missed like 20 blog posts in the last few weeks. Good lord.
bacek_at_work kid51_at_dinner, it hangs because I hardcoded very-very-very-tiny GC threshold. For testing purposes. 01:15
01:22 dngor_ joined
whiteknight "PARROT VM: Could not load or run bytecodePARROT VM: Catastrophic error. Cannot recover" 01:22
so there's a great new error
gaze upon the results of the new embed_api work in awe 01:23
cotto I like it. 01:24
needs a newline though
01:24 dngor left
whiteknight Don't criticize my art! 01:25
I haven't implemented error messages yet. So the second message "Catastrophic error" comes from the error-handler not being able to read the error message 01:30
eventually I'm going to have to do that 01:31
01:32 lucian left 01:33 mikehh joined
whiteknight I need to figure out how to do it 01:33
The problem is that at the point of passing out a raw error message, we're at the end of the rope with an unhandled exception. I suspect that creating a new STRING at that point may be fraught with danger 01:34
but trying to work with C strings is even more stupid 01:35
dalek rrot: 14257c7 | mikehh++ | config/auto/warnings.pm:
fix codetest failure - trailing spaces
cotto C strings are so much fun. 01:38
01:44 tadzik joined 01:47 pmichaud joined 01:48 Util joined 01:58 kid51_at_dinner left
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#1287) fulltest) at 14257c7 - Kubuntu 10.10 amd64 (g++-4.5 with --optimize) 02:12
02:12 tadzik left 02:13 dngor_ is now known as dngor 02:14 Util left, pmichaud left
cotto dukeleto, ping 02:17
plobsing, ping 02:21
plobsing cotto: pong
cotto plobsing, do you have any plans for pirc's makefile infrastructure?
plobsing not as of yet. I'm currently trying to find the git-fu to eliminate empty branch merges after git-filter --subdirectory-filter 02:22
cotto stackoverflow.com/questions/359424/...repository <-?
plobsing after reasonably translating all of the history, I'll worry about making it stand alone. unless you have some bright ideas
dalek rrot/embed_api: 46b7a9d | Whiteknight++ | / (5 files):
upconvert pbc_to_exe to use the new API. it builds, but segfaults building parrot_nci_thunk_gen
rrot/embed_api: b62a636 | Whiteknight++ | tools/dev/pbc_to_exe.pir:
fix the runcore so it's a string instead of an integer
rrot/embed_api: b574648 | Whiteknight++ | tools/dev/pbc_to_exe.pir:
use the correct size for the bytecode array, this gets us past the error we were seeing and allows pbc_to_exe to run correctly after bootstrapping. However, now we get an unhandled error in parrot_nci_thunk_gen
 02:23
cotto plobsing, no bright ideas from me
plobsing cotto: that describes what I've done. but now I have a bunch of empty branches merging back at various points in history
cotto perhaps you could add parts of the generated makefiles as a starting point
plobsing that sounds like it could work. thanks. 02:24
cotto it had its own makefile before we moved to a monolithic Makefile with includes 02:27
02:30 bluescreen joined
cotto whiteknight, embed_api doesn't build for me 02:33
is that expected?
bluescreen are you using gcc? 02:34
cotto sure
bluescreen there is a minor fix in api.c
cotto that's the file that doesn't compile 02:35
bluescreen around line 159, the last ; from EMBED_API_CALLIN(interp_pmc, interp)
has to be removed
dukeleto cotto: pong
bluescreen stupid gcc thinks ";" is a line of code
dukeleto all this embedding API work is very awesome
cotto dukeleto, I was going to ask about the git-fu behind moving pirc to a separate repo, but plobsing seems to be on it already 02:36
bluescreen, where's the stray ;?
dukeleto cotto: awesome. laziness works again!
cotto I love that virtue.
bluescreen line 159
02:36 rurban_ joined
bluescreen where is no paste 02:37
cotto plobsing, ok
aloha, nopaste?
aloha cotto: nopaste is is nopaste.snit.ch (works with the script in $_PARROT/tools/dev/nopaste.pl)
dukeleto plobsing: what git-fu are you doing? 02:38
plobsing dukeleto: I've done a filter-branch --subdirectory-filter to extract only the history that pertains to pirc. unfortunatly, it doesn't eliminate useless branch merge commits. 02:39
02:39 rurban left
nopaste "bluescreen" at 192.168.1.3 pasted "api.c diff" (13 lines) at nopaste.snit.ch/26106 02:39
02:39 rurban_ is now known as rurban
plobsing so I've got a ton of empty branches getting merged back in the history. I'd like the history to be clean before I push it 02:39
dukeleto: any idea how to delete these automatically (or even manually?) 02:40
02:42 whiteknight left
cotto bluescreen, it looks like there are a handful of other stray semicolons too. 02:42
bluescreen cotto, getting other errors? 02:43
or warnings?
cotto errors, though after fixing those the build explodes in some other places
bluescreen pbc_to_exe? 02:44
cotto runtime/parrot/library/PGE/Perl6Grammar.pir actually
er, compilers/pge/PGE/builtins.pg
bluescreen what do you get? 02:45
cotto segfault
I'm running realclean to make sure there's no leftovers interfering with the build. 02:46
same thing 02:47
bluescreen I'm going to checkout github repo again and give it a try
dukeleto plobsing: you can use filter-branch again to remove the commits, but it sounds like your original filter-branch probably had a bug in it 02:51
plobsing: you can tell filter-branch to ignore merge commits that don't have a diff, or something like that
mikehh rakudo (90cb844) - build on parrot (471-g14257c7) - make test PASS, make spectest_smolder[(#1288) roast (2c952c4)] FAIL - Kubuntu 10.10 amd64 (g++-4.5 with --optimize) 02:53
26,262 ok, 8 failed, 607 todo, 1,858 skipped and 1 unexpectedly succeeded
02:53 tadzik joined
dalek rrot/embed_api: 7836edb | cotto++ | src/embed/ (2 files):
remove some stray semicolons that break the gcc build
 02:53
02:56 Util joined
plobsing dukeleto: really? I read in the doc that the --prune-empty command (which is supposed to do something to that effect) only works on commits with 1 parent 02:56
02:57 pmichaud joined
dukeleto plobsing: you will have to actually write a commit filter 03:01
03:01 hercynium left
bluescreen cotto: you were right there were bunch other of those errors 03:01
cotto bluescreen, fixed
bacek_at_work, how current is docs/memory_internals.pod? 03:03
dukeleto plobsing: github.com/leto/Util/blob/master/b...fix_gej.sh
plobsing: it's an env-filter, actually 03:04
plobsing: time git filter-branch -f --env-filter '. /home/leto/fix_gej.sh' -- --all
plobsing: that is the command i used
plobsing dukeleto: I'm thinking of doing a --parent-filter 03:05
dukeleto plobsing: sure, TIMTOWDI
plobsing if a merge commit has no diff with one of it's parents, delete the other parent
is there some way to dry-run these things? 03:06
dukeleto plobsing: i think so 03:12
plobsing: git help filter-branch
cotto does anyone know what happened to src/gc/alloc_register.c? It's referenced in src/gc/api.c but doesn't exist. 03:17
plobsing alloc_register? what does gc have to do with register allocation? 03:18
cotto 1 - nothing apparently. 2 - I think it was allocating space for registers, not allocating their use 03:20
03:21 bluescreen left 03:23 nwellnhof left
cotto anyone know what "interact with some of the architecture of Parrot" in src/gc/api.c means? 03:29
git blames whiteknight
plobsing cotto: probably got moved into contexts then 03:30
or pcc
cotto plobsing, either way, it sounds like I can safely remove the reference to it from the docs 03:32
GeJ dukeleto: I'm sorry, but I do not approve. What makes you think that I need to be fixed?
oh wait...
dukeleto GeJ: just showing plobsing how i used filter-branch. Just an example. 03:39
cotto GeJ, we don
't fix what's not broken. ;) 03:40
bacek_at_work cotto, docs/memory_internals.pod require some cleanup. With fire. 03:50
cotto bacek_at_work, why is src/gc/variable_size_pool.c empty? Do we need it?
bacek_at_work cotto, no idea. Did I add it? 03:51
cotto no. You just happen to know the gc code better than most.
bacek_at_work just kill it then
probably my mistake anyway
cotto svm 03:52
wfm
. o 0 (at least I got one letter right)
bacek_at_work, why do we keep src/gc/malloc.c and src/gc/malloc_trace.c around if they're unused?
btw, variable_size_pool.h has a struct, so it might have been your plan to move some functions into the .c file 03:53
bacek_at_work erm. Looks like you're right. 03:55
GCI task?
afk # another (4th for today) meeting 03:56
cotto if I don't have the tuits, sure. I wouldn't mind an excuse to mess with some gc code though. 03:57
wheee. We have a gci pull request. 04:09
04:13 rurban left
dukeleto cotto: link? 04:13
GeJ github.com/parrot/parrot/pull/1 04:15
cotto what GeJ said
GeJ Looks like Andy's already on it.
dukeleto I commented as well :) 04:16
cotto no moss on us
If those are part of our api, won't they need a deprecation cycle? 04:17
dukeleto Also, won't anything that uses them not compile now?
Ah, i see. Only 2 other files used them.
04:31 cognominal left
dalek rrot: 6821e89 | petdance++ | / (4 files):
annotations and markers on arguments
 04:34
plobsing I feel as though I should be deleting the parrot release tags as well. they don't really have meaning for pirc which was never released. 04:41
cotto plobsing, sure 04:45
dalek rrot: bee67f0 | petdance++ | src/packdump.c:
removed unused argument from static function
 04:53
nopaste "plobsing" at 192.168.1.3 pasted "[git parent-filter] delete useless parents" (36 lines) at nopaste.snit.ch/26108 04:54
plobsing github.com/plobsing/parrot-pirc \\o/ 04:55
(I didn't know if/how to make parrot/parrot-pirc) 04:56
cotto it'll be tricky to use without a proper Makefile, but whether you bother is entirely up to you. 04:58
plobsing++
plobsing I may or may not put more work into this at some point. the important thing is the history is extracted and preserved 04:59
cotto agreed 05:00
are you comfortable with someone ripping out pirc at this point? 05:01
actually, it doesn't matter. I can do it in a branch and merge when you are.
plobsing heh, I'm halfway through that already
>:) 05:02
cotto then I won't bother
plobsing++
That's certainly the fun part.
05:16 rurban joined
dalek rrot: f4cabb1 | plobsing++ | co (54 files):
delete pirc

Per discussion on #parrotsketch (2010-11-23) this will be removed as it is unlikely to serve our purposes in its current form.
It has been split out into it's own project currently hosted at
  github.com/plobsing/parrot-pirc. Any further developments should occur
there.		 05:17
rrot: 829c859 | plobsing++ | MANIFEST:
mk_manifest_and_skip
cotto I need to write a policy document on the proper use of "its" and "it's". 05:22
plobsing++ 05:23
plobsing yes, my grammar is horrid. that's how you can tell I'm a native english speaker. 05:24
cotto it does seem to happen with native speakers more often 05:25
plobsing now how do I get dalek to monitor the new pirc repo? 05:27
cotto do you have a commit bit for perl6/mu on github? 05:29
the config file is misc/dalek-something.json
plobsing ENOPERM 05:30
cotto I can do it.
done 05:33
plobsing cotto++
cotto nice use for github's web editing interface 05:34
dalek rrot: 85ae767 | plobsing++ | NEWS:
add news entry re: pirc removal
 05:41
cotto now the question is whether to close all of pirc's tickets outright or to make an attempt to migrate them to github
plobsing do we have a list of all of them?
cotto they should have the pirc tag
plobsing how do I search tickets by tag? 05:44
cotto trac.parrot.org/parrot/report/30
It's a bit weak. I think you have to make a report. 05:45
plobsing I found the "custom query" button
it appears to do what I want
cotto so it does 05:47
plobsing it appears pirc has 16 open tickets
cotto I'll reject all the tickets so they won't be confused with tickets that have been fixed. 05:48
plobsing I'd prefer to hold off until we import them 05:49
I have no idea how to do that
cotto plobsing, you could make a query show everything about a ticket for trac for export and use github's api to import them. 05:55
or use this: github.com/davglass/trac2issues 05:58
Hm. That has to be run on the trac server. 05:59
plobsing it's a start 06:01
cotto I appreciate that you're trying to be fair to pirc. You're more patient than I am. 06:03
plobsing trac doesn't appear to have a web api 06:12
I'll look into this further tomorrow 06:13
bacek_at_work plobsing, why /plobsing/parrot-pirc not /parrot/pirc? 06:30
you should able to create new repo in /parrot
cotto bacek_at_work, only a few people have admin bits now 06:31
bacek_at_work ah, ok. Then ask one of them to create new repo. 06:32
cotto dukeleto, ping 06:38
NotFound Do you want to create parrot/pirc on github? 06:49
cotto NotFound, that'd be ideal 06:53
do you have a commit bit for that?
06:53 bacek joined
NotFound Yeah. Please tell me the description. 06:53
cotto ohai bacek 06:54
aloha, clock?????
aloha, clock
aloha, msg aloha aloha
aloha cotto: OK. I'll deliver the message.
cotto aloha, clock?
aloha cotto: cotto: LAX: Tue, 22:54 PST / CHI: Wed, 00:54 CST / NYC: Wed, 01:54 EST / UTC: Wed, 06:54 UTC / LON: Wed, 06:54 GMT / BER: Wed, 07:54 CET / TOK: Wed, 15:54 JST / SYD: Wed, 17:54 EST
NotFound Created: github.com/parrot/pirc, empty for a now, someone must initialize it. Description is: "Experimental PIR compiler" 06:59
cotto NotFound, thanks 07:00
sorear NotFound: That was a Spanish question, not a Perl 6 one 07:11
NotFound: there's some very handwavey stuff in the Perl 6 spec about language-specific concepts of a character that may cause .chars and .codepoints to be different 07:12
NotFound sorear: I've not looked at that yet. 07:31
sorear NotFound: I'm just asking about the principle of least suprise 07:36
to someone who has not read the Perl 6 spec, are there any contexts where Perl should treat "ch" as one letter? 07:37
NotFound Anyaway, composing characters is not frequent in Spanish. Composing keys generate individual codepoints. We even have a key for n tilde, for maximum laziness. 07:38
sorear: ah, ch. No, even the academics finally aknoledged that ch is a c and a h.
An ll is two l. It's a wonderful world. 07:39
It always amazed me how language academics kept confusing characters and phonemes for so long time.
Even now, they say that 'ñ' is a character, they are unable to see n + ~ 07:40
There were public protests some years ago when was proposed to drop the 'ñ' from keyboards. The apocalype will come if people must do more than one keystroke to get a ñ. 07:42
The same people that never had any troule pressing ' + a to get á. 07:44
But at least, ch and ll are no more in dictionaries. We've got some progress. 07:45
sorear "á".chars == 1 falls out naturally of NFG 07:46
I wonder if NFG might be all we really need
NotFound For Spanish, I think yes. For languages with much more complex character sets, I dunno. 07:47
sorear Can you name a language that I should study for this next? 07:48
NotFound Maybe Klingon? ;)
Don't know, I've read a lot about that things but never remeber the details. 07:49
sorear Perl 6 only cares about those languages that Unicode cares about
07:50 fperrad joined
NotFound sorear: unicode has Klingon... well, some unicode implementations has it in optional planes, that is. 07:50
sorear could only think of Spanish and German as languages with interesting character concepts, and moritz has already said basically what you did about ss
NotFound sorear: I think there are some problems with romanic languages written with cirillic characters...maybe Croatian. 07:51
cotto bacek_at_work, what's gc_ms2_magical_mark? 08:09
08:21 theory left 08:38 plobsing left
dalek rrot: fa56f62 | fperrad++ | tools/install/smoke.pl:
PIRC is gone
 08:47
sorear has anyone been troubled by the github url butcherer? 08:48
cotto sorear, what do you mena? 08:49
*mean
moritz you mean that it's not the full SHA1 in the URL?
08:50 lucian joined
sorear moritz: yes 08:50
I put in an ad-hoc pass to remove the last 30 characters of URLs matching /github.*commit\\/[0-9a-f]{40}4/
s:2nd'4'$' 08:51
cotto I like that. It removes some visual noise.
sorear++
moritz likes it too
szbalint yeah, collisions are rare anyway, but much better for humans this way. 11:23
mikehh got a wierd test failure here - t/pmc/fixedstringarray.t is failing in corevm/coretest but not in test 11:39
1) it failed after running test 40 - looks like a segfault (test 41 the next test is in sub test_gc 11:40
2) extracted the test_gc sub - ran ok 11:41
3) extracted the tests - sub equality_tests, added them , ran ok 11:42
4) prove t/pmc/fixedarray.t still fails in the same place
so running the tests (corevm/coretest) fails after test 40 - extracting tests 34 - 44, runs ok 11:48
BTW this is on Kubuntu 10.10 amd64 (g++-4.5) - (RELEASE_2_10_1-477-gfa56f62) 11:52
12:07 cognominal joined 12:18 bluescreen joined 12:37 fperrad_ joined 12:40 fperrad left, fperrad_ is now known as fperrad 12:59 bluescreen left 13:04 mikehh left 13:14 bluescreen joined, bluescreen left 13:18 rurban_ joined 13:20 Patterner left 13:21 rurban left, rurban_ is now known as rurban 13:22 whiteknight joined
whiteknight got my first pull request from a GCI student!! 13:23
good morning, #parrot
brb, firefox is in ur computer, eatin all ur RAMs 13:26
13:26 whiteknight left 13:32 darbelo joined 13:35 Psyche^ joined, Psyche^ is now known as Patterner 13:41 mikehh joined
dalek rrot/datatypes_func_renames: 45e28a4 | (Matt Rajca)++ | / (5 files):
Rename public functions in src/datatypes.c
 13:44
rrot/datatypes_func_renames: 1309405 | (Matt Rajca)++ | / (2 files):
Ran headerizer
mikehh get that test failing in Ubuntu amd64 as well, just corevm/coretest and with g++, but passes gcc and with --optimize (t/pmc/fixedarray.t) 13:52
mikehh got to go out for a while, bbl 13:53
13:54 mikehh left 13:55 bluescreen joined
Coke .. why move pirc to a separate repository? is someone going to work on it? 13:56
(why not just kill it with fire) 13:57
dalek TT #1869 created by whiteknight++: Move Parrot::Embed into it's own repo 14:01
TT #1869: trac.parrot.org/parrot/ticket/1869
TT #1870 created by whiteknight++: Move OpenGL Bindings to their own project/repo
TT #1870: trac.parrot.org/parrot/ticket/1870
Coke aloha: msg cotto - there's a lot of residual "pirc" cruft left in the repo. 14:03
aloha Coke: OK. I'll deliver the message.
14:04 whiteknight joined
Coke mk_manifest_and_skip.pl doesn't seem to care that compilers/pirc no longer exists... 14:06
darbelo whiteknight: ping 14:13
whiteknight darbelo: pong
darbelo whiteknight: Did you do the opengl move? 14:14
whiteknight darbelo: no, I've only suggested it
darbelo Ah, ok. I just remembered something that might be an issue. 14:15
whiteknight what is it?
#1870, to track the issue
darbelo That code uses NCI extensively, with some very hairy signatures, so it hooks into configure to get the thunks it needs compiled into parrot. I remember japhb mentining it couldn't move out wihtout proper run-time NCI support. 14:17
dalek TT #1629 closed by whiteknight++: find_codepoint experimental op
TT #1629: trac.parrot.org/parrot/ticket/1629
whiteknight darbelo: I think the gsoc_nci branch merged. We have full libffi support now 14:18
opengl probably needs an update, but it shouldn't be too big a hassle
darbelo Last I ckecked it, it worked fine (to the level my onboard graphics could exercise it) 14:19
You can use it from winxed too. 14:20
whiteknight yes
dalek rrot: 2b58997 | Whiteknight++ | / (2 files):
find_codepoint is no longer experimental, it is supported. Remove the entry from DEPRECATED.pod and add a short blurb to NEWS
 14:22
rrot: 1d07f22 | Whiteknight++ | / (82 files):
merge in trunk, fix conflict in NEWS
Coke sees the answer to his question in #Ps logs: plobsing has too much spare time. Fair enough. ;) 14:25
14:26 plobsing joined
Coke ugh. feels like there's more slow NQP stuff getting built. 14:26
slow-to-build, that is.
14:29 bluescreen left
plobsing darbelo: (re: opengl) the thunks it requires are no longer baked into parrot, but are in stead generated as a dynlib by nci_thunk_gen 14:32
whiteknight plobsing: and even that should be able to go away with the libffi stuff merged, right?
plobsing no
whiteknight why not?
plobsing libffi is not available everywhere 14:33
whiteknight ah, okay. But where libffi is available, those don't need to be built
?
plobsing possibly not.
whiteknight awesome
plobsing libffi is decidedly less efficient than aot or jit thunks
whiteknight ah, gotcha
plobsing but if you're worried about thunk efficiency, you're probably doing it wrong
whiteknight maybe next step on the road to NCI world domination is an LLVM-backed thunk generator 14:34
plobsing sure. why not? start a project. it doesn't need to be internal 14:35
my libjit thunk generator works fine as an add-on library (modulo users not knowing how to install libraries on their own)
darbelo The problem I was referring to is that those thunks are generated at parrot build-time, they'll have to be moved to opengl build-time, at which time the compiler used to build parrot may or may not be present. 14:37
plobsing thats the same problem any library with C components faces.
we have libraries with ops and pmcs. they don't seem to be bothered too much.
14:38 dmalcolm joined
darbelo No, but currently opengl 'Just works' by riding on top of configure, we have to make sure it continues to efter moving out. 14:38
There's piles of arcane knowledge in the configure step, I'm just trying to make sure it doesn't get lost in the move. 14:39
14:40 bluescreen joined
plobsing There's piles of arcane knowledge in that configure step, that's why we're trying to remove it. 14:40
well, part of why at least 14:41
whiteknight plobsing: where is your libjit thunk generator? 14:42
14:42 plobsing_ joined
whiteknight plobsing_: where is your libjit thunk generator? 14:43
plobsing_ github
github.com/plobsing/parrot-libjit-fb
dalek rrot: d3046c1 | plobsing++ | NEWS:
update pirc removal news item to use official repo
 14:47
rrot: dab3652 | plobsing++ | / (3 files):
Merge branch 'master' of github.com:parrot/parrot
whiteknight ah, okay. That's what I was wondering. a dynpmc does seem like the perfect vehicle for this
14:47 plobsing left
whiteknight Parrot's NCI frame builder system is all PMC-based internally? You can register any dynpmc as a framebuilder? 14:48
plobsing_ yes
pluggable everything 14:49
whiteknight It's been a while since I really looked at that subsystem. I really need to give it another glance over
plobsing_ it's still a little hairy inside, but I feel like we've made some improvements
whiteknight Where does the libjitframebuilder PMC actually get registered as the system thunk builder? 14:50
VTABLE_set_pmc_keyed_int(interp, interp->iglobals, IGLOBALS_NCI_FB_CB, callback);? 14:51
plobsing_ yes, there are *two* iglobals that need to be set
callback and userdata 14:52
whiteknight and userdata is an instance of libjitframebuilder?
plobsing_ callback is a PMC on which get_pointer returns a function of a particular signature. 14:53
userdata is any pmc you like. it gets passed to the function
whiteknight hmmm... this is all very cool looking
plobsing_ in this case, yes, it is a libjitframebuilder 14:54
whiteknight so the system gets the function pointer out of IGLOBALS_NCI_FB_CB, and then calls that function with the data from IGLOBALS_NCI_FB_UD and the signature string
I assume these thunks are getting cached somewhere, so we aren't constantly regenerating them? 14:55
plobsing_ yes
whiteknight in short I would say "yes, we've made a shitload of progress in this subsystem" 14:56
and you are in no small way responsible for it
plobsing_ src/nci/api.c:build_call_func takes care of the cacheing
whiteknight: I did most of that back in the spring. the subsystem's been mostly dormant since then. gsoc_nci simply uses the existing hooks. 14:57
whiteknight as one small suggestion, I think we should encapsulate the IGLOBALS garbage behind a function like Parrot_nci_register_frame_builder()
but i'm in the "make better interfaces for all this shit" mode 14:58
plobsing_ that might be nice. overkill possibly, but nice nonetheless. my next move for the subsystem is to complete the encapsulation of frame builders from the signature strings 14:59
darbelo 'One signature to bind them all'? 15:00
whiteknight so where does the libffi stuff come in, does it have a PMC type somewhere?
plobsing_ whiteknight: it piggy-backs on a ManagedStruct PMC. ManagedStruct is a poor-man's PMC class.
whiteknight ah yeah, I'm looking at the src/nci/libffi.c file now 15:01
plobsing_ oops, make that UnManagedStruct. potato potato. 15:02
whiteknight potato, unpotato
plobsing_ internet is not a good medium for accents
whiteknight internet isn't so bad, it's the keyboards that are lacking 15:03
15:05 Andy joined
dalek rrot/datatypes_func_renames: 6821e89 | petdance++ | / (4 files):
annotations and markers on arguments
 15:08
rrot/datatypes_func_renames: bee67f0 | petdance++ | src/packdump.c:
removed unused argument from static function
rrot/datatypes_func_renames: f4cabb1 | plobsing++ | co (54 files):
delete pirc

Per discussion on #parrotsketch (2010-11-23) this will be removed as it is unlikely to serve our purposes in its current form.
It has been split out into it's own project currently hosted at
  github.com/plobsing/parrot-pirc. Any further developments should occur
there.
rrot/datatypes_func_renames: 829c859 | plobsing++ | MANIFEST:
mk_manifest_and_skip
rrot/datatypes_func_renames: 85ae767 | plobsing++ | NEWS:
add news entry re: pirc removal
rrot/datatypes_func_renames: fa56f62 | fperrad++ | tools/install/smoke.pl:
PIRC is gone
rrot/datatypes_func_renames: 2b58997 | Whiteknight++ | / (2 files):
find_codepoint is no longer experimental, it is supported. Remove the entry from DEPRECATED.pod and add a short blurb to NEWS
rrot/datatypes_func_renames: 1d07f22 | Whiteknight++ | / (82 files):
merge in trunk, fix conflict in NEWS
rrot/datatypes_func_renames: a68cfef | Whiteknight++ | / (63 files):
Merge branch 'master' into datatypes_func_renames
plobsing_ I feel like I've seen this before
darbelo plobsing_: git merges replay the commits 15:09
You have seen this before.
plobsing_ sure merges replay commits, but does dalek have to? 15:10
whiteknight oh shit, i didn't realize all those commits were going to come in here
darbelo He reports commits. Those are commits. He doesn't ask questions.
He just pulls the trigger. 15:11
whiteknight if git is going to generate more IRC traffic, maybe we need a second chatroom for it
#parrotfeed or something like that
ha, has anybody ever seen the commit feed for english wikipedia?
darbelo Unrelated thought. plobsing_, where do you see IMCC fitting in the bright shiny 'no implicit PIR compiler ' world? 15:12
whiteknight darbelo: somewhere in a trash dumpster
:)
plobsing_ imcc or more likely pirc might get resurected as dynlibs 15:13
whiteknight They are going to be stand-alone front-ends for libparrot
and dynlibs too
you will load_language "PIR" to load it in 15:14
plobsing_ unless bacek beats us them to it. never try to race the coding robot.
darbelo Let me put it in end-user terms. I downloaded parrot, I typed make install and have a 'parrot' executable. Can I type 'parrot some.pir' and have it work? 15:15
If so, then there's still an implicit PIR compiler in the parrot executable (let's call it libpirc). 15:17
whiteknight darbelo: step back. The better question is "will I have a parrot executable"
?
think of java and javac. The java.exe file does nothing except run compiled bytecode
plobsing_ darbelo: what most people expect from parrot.exe actually belongs in parrot-pir 15:18
whiteknight javac compiles the java language into bytecode but does not run it
Coke you crazy kids. I remember when parrot had an explicit assembler and an explicit ... runner.
and then we absorbed imcc.
whiteknight Coke: yeah, it's never too late to correct the mistake
Coke and now we're heading back the other way!
whiteknight I don't think we are going to have multiple executables like that. That's not really what we want 15:19
plobsing_ Coke: yes, but now IMCC is the greater of the evil options
darbelo That's the clarification I want. We'll have libparrot, a library that runs bytecode, and then a small constellation of language fakecutables, with no implicit 'parrot language'
Coke Yup. not saying you're going in the wrong direction. It's just amusing.
whiteknight but libparrot is just a library, and we can link it into any front-end that we want
Coke I think not having a specific, supported, minimal language is a mistake. 15:20
whiteknight there will be a parrot-pir frontend (probably based on IMCC or PIRC initially, but eventually will be a fakecutable for PIRATE)
Coke (I don't think it has to be PIR in the form of IMCC)
whiteknight Coke: we do ship NQP-RX in with our releases
we could easily include another front-end there too
darbelo starts chewing his own boot straps.
whiteknight The difference is that libparrot is our main "product". Anything else we have is just an addon to that 15:21
atrodo I agree with Coke, having a shipped language will help people that want to dive into developing/playing with parrot 15:22
Coke nqp-rx is fine, if even-slower-than-pir-and-can't-do-everything-imcc-does. 15:23
darbelo Hm. If we want PIRATE to fully bootstrap we'll need a stable bytecode format. 15:24
whiteknight Coke: NQP is just one example of an alternate front-end that we ship with our releases
There will be more. Winxed is an example of a low-level language with most of the capabilities of IMCC and good speed
Coke www.theregister.co.uk/2010/11/23/da...u_updates/ 15:25
whiteknight I don't know if we'll ever be able to include that in with the repo or not
plobsing_ whiteknight: why not? it's not like we own nqp. we ship it anyways. 15:26
whiteknight plobsing_: It's NotFound's decision whether we incorporate it or not. I don't presume to know what he would want
plobsing_ afk # meeting
whiteknight legally it's under a free license and we could just take it and tell NotFound to suck an egg. Realistically we're not going to do that without the author's permission 15:27
NotFound winxed is (C) me, not pafo. If you want it on the parrot repo, I don't have any problem. Just ask. 15:28
And if you want to transfer ownership to pafo, I can agree provided that I retain some design control. 15:29
whiteknight NotFound: we're not there yet, but when somebody wants an alternative to PIR, I'm comfortable suggesting Winxed
NotFound: what license does it use? I don't see a LICENSE file or anything in your repo 15:31
ah nevermind, I see it on googlecode: Artistic/GPL 15:32
NotFound whiteknight: yeah, I'm always lazy about including legalese and such in the repos.
whiteknight Basically, it wouldn't be a transfer of ownership. You always own it. under the artistic license anybody can use it, copy it, modify it, so long as we always give you proper attribution
but again, we're not there yet 15:33
NotFound Transferig copyright mainly means that if the need arise to sue someone, other people will pay the legal costs ;) 15:34
whiteknight :) 15:35
well then you're looking at the wrong people, because PaFo ain't gots no money 15:36
I'm going to merge that datatypes_func_renames branch. Incoming. 15:39
dalek rrot: 6993dc3 | Whiteknight++ | NEWS:
mention the purpose of this branch merge in NEWS
 15:40
rrot: 2f5ab6e | Whiteknight++ | NEWS:
Merge branch 'master' of github.com:parrot/parrot
rrot: a3a874d | Coke++ | / (15 files):
remove more pirc references.
rrot: 31eeb30 | Coke++ | docs/ (2 files):
Update/Remove some stale references to languages/
whiteknight I love looking at the network graph for parrot github.com/parrot/parrot/network 15:57
I don't always understand why there are so many lines in there, but it's still cool to see how much we are acheiving
dukeleto 'ello 16:08
whiteknight hello dukeleto
we have our first successfull GCI project. I just merged it
several of the other tasks are listed as ActionNeeded because of the short timeouts, but I can't seem to take any action on them. So they look kind of stuck 16:09
16:10 bluescreen left
Coke we have 35 build warnings in a gcc 4.2.1 build without --cage 16:17
nopaste "coke" at 192.168.1.3 pasted "non cage warnings" (35 lines) at nopaste.snit.ch/26119 16:20
Coke that's a non-helpful IP address. ;) 16:21
whiteknight ah, many of those should be easy to resolve
Coke: can you post that to the parrot-dev list? Getting all those cleaned up by 2.11 would be nice. by 3.0 should be a necessity
dukeleto whiteknight: awesome to hear that we have a GCI pull request! 16:23
Coke how about I open a ticket instead? 16:24
hurm. actually there's a wiki page...
16:24 theory joined
whiteknight Coke: however you want to advertise it is cool. Getting warnings fixed is something we definitely should be doing regularly 16:25
16:26 bluescreen joined
dalek tracwiki: v13 | coke++ | BuildWarnings 16:29
tracwiki: trac.parrot.org/parrot/wiki/BuildWa...ction=diff
Coke aloha, karma Coke 16:41
aloha Coke: Coke has karma of 21.
Coke aloha, karma coke
aloha Coke: coke has karma of 21.
atrodo That's either a really good coincidence or a really smart aloha 16:44
whiteknight I think aloha does everything lower-case 16:48
It definitely hasn't been updated with all the karma from the purl days however
16:56 plobsing joined
dalek rrot: 60c49d4 | Whiteknight++ | DEPRECATED.pod:
add note to DEPRECATED.pod about #1800
 16:56
16:58 fperrad_ joined 17:00 fperrad left, fperrad_ is now known as fperrad 17:05 hercynium joined
NotFound whiteknight: ping 17:09
cotto ~~ 17:20
whiteknight pong 17:25
NotFound whiteknight: about find_codepoint, we should now move it from experimental.ops to string.ops, isn't it? 17:26
whiteknight NotFound: yes. You want to do that?
NotFound Yes, I'm testeing now.
whiteknight awesome
dukeleto Is it just me, or has Parrot development sped up quite a bit in the last week? 17:36
dukeleto can barely keep up
NotFound dukeleto: about myself, I was a bit inactive in previous weeks, and also getting used to git. Now I'm a bit better. 17:38
dukeleto NotFound: great to hear that git it not holding you back. Let me know if you have any questions, or suggestions to improve the Git docs. 17:40
NotFound dukeleto: you can add "Plase don't delete our repos" X-) 17:41
whiteknight NotFound: 1) See "Delete" button. 2) Don't push it. 3) Repeat. 17:42
dukeleto NotFound: ;)
NotFound whiteknight: that reminds me the joke: Programmer dead in the shower. The shampoo said: "wash, rinse, repeat" 17:44
whiteknight :) 17:46
dalek rrot: 3cfeb72 | NotFound++ | / (7 files):
move find_codepoint from experimental.ops to string.ops TT #1629
TT #1871 created by moritz++: Rakudo randomly segfaults (probably GC related) 17:51
TT #1871: trac.parrot.org/parrot/ticket/1871
cotto_work ~~ 18:06
coke++ for taking care of the pirc leftovers] 18:16
dukeleto yay, random segfaults! 18:17
whiteknight I wouldn't have it any other way 18:21
pirc sure got shuffled out quickly. Has anybody talked to kjs lately? 18:22
He should definitely receive an email, if he hasn't already
Coke git branch -a | ack pirc 18:26
my branch there can be removed.
whiteknight Coke: done 18:28
Coke is there a way to get the equivalent of svn's --stop-on-copy? 18:29
(a google for that is less than helpful) 18:30
googling for "git log stop-on-copy" is more helpful. 18:33
18:35 plobsing left
whiteknight Coke: git log --no-merge? 18:43
cotto_work whiteknight: emailing kj is a good idea. emailing him before removing it might have been a better one. 18:45
darbelo aloha: seen kj 18:46
aloha darbelo: kj was last seen in #parrot 25 days 5 hours ago joining the channel.
Util Coke: 1. The "Browse Source" section of Trac should support it, when it start working again. 18:47
2. From versioncontrol.wikia.com/wiki/List_..._on_branch :
"If you had the foresight to tag the start of your branch, say with <branch>-start, you're in luck: git-log <branch>-start.. 18:48
If you don't have a <branch>-start tag, I'm stumped."
3. www.spinics.net/lists/git/msg101285.html has an excellent explanation of "--stop-on-copy" semi-equivalence
END_VAGUENESS
18:50 dngor left
whiteknight Coke: I have some other ideas too, what exactly are you wanting to see? 18:50
cotto_work any volunteers to mail kj?
whiteknight I'm doing it right now 18:51
cotto_work whiteknight: thanks
Util From msg101311 : In general, you can't say "My branch starts _here_" unless you use the root commit(s) as the starting point, or you apply "extra" knowledge (you know from which other branch this branch forked). 18:54
dukeleto Yes, which you can get at programmatically with "git merge-base" 18:56
dukeleto drops some git knowledge bombs
whiteknight BAM!
cotto_work boom
dukeleto Run "git help merge-base" for some amazing ASCII art. Not kidding.
18:58 dngor joined
dukeleto whiteknight: i am changing some of your tasks to ask for "SVG" instead of "a vector graphics format". Many students are getting confused by that. 19:01
whiteknight okay, gotcha
I think I only requested one logo task
cotto_work: what's the story with the opmap_aware_pmcs branch? 19:09
dukeleto whiteknight: i like your tasks. Keep 'em comin'
whiteknight: i am approving a bunch now 19:10
whiteknight yay! I'm being helpful
cotto_work whiteknight: bacek says it's using the wrong approach. I partially understand what he's saying but I need to figure out exactly what he means and what approach he's suggesting.
atm it's stuck in debugging. it may or may not be worth fixing.
whiteknight okay. 19:11
I'm just digging through this network graph looking for some old straggling branch heads
what do you need to know in order to determine whether it's worth fixing? 19:12
cotto_work whiteknight: I need draft a plan and go back and forth with bacek a bit 19:13
aloha: clock? 19:14
aloha cotto_work: cotto_work: LAX: Wed, 11:14 PST / CHI: Wed, 13:14 CST / NYC: Wed, 14:14 EST / UTC: Wed, 19:14 UTC / LON: Wed, 19:14 GMT / BER: Wed, 20:14 CET / TOK: Thu, 04:14 JST / SYD: Thu, 06:14 EST
whiteknight okay. I don't want to rush you, I'm just trying to make sure we don't forget important branches from our huge collection of them
cotto_work There's no risk of that branch being forgotten. We can't get PIRATE back to working without it (or something like it). 19:15
whiteknight ah, gotcha. Is that branch the only outstanding requirement before PIRATE can move ahead? 19:17
cotto_work no. PIRATE also needs gsoc_past_optimizations
whiteknight okay then 19:18
the dominos are falling like a house of cards. Checkmate!
I may try to update gsoc_instrument to HEAD 19:19
dukeleto whiteknight: i think bacek already did that
whiteknight: as of a day or two ago
darbelo gsoc_past_optimizations is a remnant. The code moved to it's own repo on github. 19:21
The version in parrot's repo is stale.
Coke whiteknight: I'm wanting to see exactly what --stop-on-copy on a branch would have shown me: commits that were made to that branch since the branch point.
whiteknight Coke: the branch on your local machine, or the branch on github? 19:22
darbelo: I was told the exact opposite: the code on github was never made to work, and that I should use the branch from svn to work on it
Coke whiteknight: "does it matter" ? 19:23
whiteknight oh wait, wrong repo. I am talking about gsoc_instrument, not gsoc_past_optimizations
Coke: yes. One command includes "origin" in it, the other command doesn't
Coke ... isn't that just part of the name of the branch?
whiteknight Coke: git log --name-status master..<CURRENT BRANCH NAME> 19:24
shows you all commits in the specified branch but not yet in master
Coke whiteknight++! 19:25
darbelo whiteknight: the code on github is newer. tcurtiss mentioned a bug that never got fixed there, but the code in svn is older.
whiteknight Coke: actually the --name-status option isn't needed 19:26
the --name-status gives you svn-like status codes 19:27
Coke congratulates dukeleto on his TPF grant.\\ 19:28
cotto_work dukeleto got $$? Nice.
Coke not until he finishes the work, no. ;) 19:29
tadzik paste.perldancer.org/1JnW1c7qSp7zK -- any ideas on this?
whiteknight Coke: link to the grant? 19:30
dukeleto Coke: got a link?
Coke news.perlfoundation.org/2010/11/201...sults.html 19:31
whiteknight tadzik: weird. Where is that? master? HEAD?
Coke (is the congrats. has link to actual grant posting)
tadzik whiteknight: what Rakudo uses 19:32
19:32 nwellnhof joined
tadzik oh wait, now it works 19:32
whiteknight hmm
tadzik after some realclean
dukeleto Coke: whoa, i didn't even know that that grant was still being considered. I forgot that they auto-renew them
tadzik not funny that it's not the first time I see it
Coke surprise!
whiteknight dukeleto: kind of funny considering I'm completely rewriting all the embedding stuff 19:33
dukeleto whiteknight: yeah, quite funny.
whiteknight I've got an idea: howabout I do a bunch of work, and you get all the money? :)
Coke dukeleto: I did make a comment when I voted that Hoped you still had tuits for it. Apologies, I assumed that you had seen the posting.
dukeleto whiteknight: seriously, that grant was written and submitted long before you started hacking on this.
whiteknight: we can work something out, I am sure. 19:34
whiteknight dukeleto: Oh, I know. I'm just playing. I don't care about money
(don't tell my wife I said that)
dukeleto whiteknight: i can change my proposal so that it is in line with what you are doing
whiteknight: adding tests won't change because of your work, but the docs part of it will need to be modified 19:35
whiteknight dukeleto: if you can tackle the extend stuff (which is, IMO, a bigger project), that should be plenty
there's a reason why I'm doing embedding stuff first and saving extending stuff till later
Coke It's a little late to change the proposal. ;) 19:36
Once a grant manager is assigned you can chat with them about plans. 19:37
dukeleto Coke: i think the grant can be salvaged in it's current state, but it might need some tweaks, which, from what I understand, is allowed. 19:39
whiteknight dukeleto: are you even going to have the tuits to do what the grant needs to you to? You're a pretty busy guy 19:40
less busy now that the git migration is done, but still very much in demand
dukeleto whiteknight: good question. The grant is about increasing test coverage mostly. I know I can do that. The part of the grant about documentation needs to be tweaked. 19:41
whiteknight: The first inch-stone will be updating docs/embed.pod to contain the "Lexicals", "Type Signatures" and "Constants" sections. This should take one to two weeks. 19:43
whiteknight: is that still relevant after you merge your embed_api changes?
Now that the git migration is over, I can do a lot more and I am more efficient at doing it.
whiteknight the first stage of the new embed API will be very slim. There are only enough functions right now to do exactly what our in-repo utilities need to function 19:44
parrot.exe, pbc_to_exe, pbc_merge, pbc_disassemble, etc
That work is almost complete and could be ready for serious review as early as this weekend if I bust my hump a little
dukeleto whiteknight: but will the docs in docs/embed.pod be totally worthless or are they still relevant/close ? 19:45
whiteknight dukeleto: I have no idea, I haven't looked at them at all
dukeleto whiteknight: you are changing internals. Does the user-facing part of the embed API change much? 19:46
whiteknight dukeleto: I am only changing the user-facing parts
I'm not modifying internals at all
I'm adding a new API "layer" of functions to wrap existing functionality with improved error handling and stuff
dukeleto whiteknight: ok. then my grant-related docs changing is still relevant. cool. 19:48
nwellnhof bacek, ping
whiteknight nwellnhof: what is the status of the string_checks branch? 19:50
nwellnhof whiteknight, I consider it mergeable but I think it didn't get any testing. 19:51
I fixed the change that broke Rakudo's IO::Socket.
whiteknight nwellnhof: if testing is all it needs, I can run some tests right now for you 19:52
dukeleto nwellnhof: what change was that?
nwellnhof dukeleto, I originally changed the 'recv' method of the Socket PMC to return binary strings instead of ASCII. 19:53
It's the right thing to do, but Rakudo depends on the wrong behavior.
dukeleto nwellnhof: i have been adding many tests to Socket and SockAddr PMCs. Did you add any tests for that? 19:54
Coke needs a parrot-threat-level.
whiteknight Coke: Orange
nwellnhof dukeleot, tests for what exactly? 19:55
dukeleto nwellnhof: some of my tests are still in the tt1798_ipv6 branch, if you need an example
nwellnhof: tests that Socket does what you think it does
Coke whiteknight: mmm. that's probably the level I would have picked. if I can get all my partcl issues resolved, it'll drop down to yellow.
whiteknight when the level gets to red, we have to do full-body pat-downs before entering #parrotsketch, for safety 19:56
nwellnhof dukeleto: The string_checks branch doesn't really add any new features, so I don't know what Socket tests I could add. I added a lot of Unicode tests, though. 19:57
whiteknight nwellnhof: smoke report coming soon 19:58
anything else in particular you want to see tested?
nwellnhof whiteknight, "without ICU" is pretty important.
whiteknight I don't think I have ICU on this machine 19:59
nwellnhof nevermind then
dukeleto votes for a +1 to Parrot Threat Level "Plaid" 20:01
whiteknight slightly less harrowing than Parrot Threat Level "Tartan" 20:03
but much more severe than Parrot Threat Level "Polkadots" 20:04
nwellnhof: smolder.parrot.org/app/projects/rep...tails/1307 Result: FAIL 20:13
dukeleto: the prepare_2_10_0 branch, it looks merged. Can we delete it from github? 20:14
moritz it can 20:18
whiteknight BALEETED 20:21
moritz whiteknight++ 20:22
nwellnhof whiteknight, thanks for the smoke report.
dukeleto nwellnhof: tests are for when existing features/bugs change, not just new features. Thanks for the unicode tests, though! 20:23
whiteknight tests are for everything, all the time
nwellnhof dukeleto: Most of the time I don't change existing features, and for many of my bug fixes I'd need tests written in C. 20:29
whiteknight yay! another student forked parrot and has a commit in the works 20:30
dukeleto nwellnhof: Tests in C are very valuable.
whiteknight this GCI thing is making me very very happy
dukeleto nwellnhof: there are many examples that you can just copy and tweak
whiteknight: glad to hear it. It seems to be getting young smart people hacking on Parrot, and that is pretty cool. 20:31
dukeleto feels all warm and fuzzy
whiteknight yes, watching people we have never heard of before forking parrot and preparing patches out of nowhere is awesoem
AND awesome
cotto_work whiteknight: It's pretty great. 20:32
whiteknight and this program lasts for a few months, I think
NotFound Looks like the move to github is being helpful, too. 20:33
whiteknight yes, it is enabling this kind of workflow 20:34
At the end of this summer, we might be able to close that damned #443 task 20:36
and let us never speak of it again
NotFound "this"?
20:37 hudnix left
NotFound Are you in the southern hemisphere? ;) 20:37
cotto_work whiteknight: do you have a list of the functions that need to be renamed for #443? 20:39
Coke cotto_work: all of them! HAHAHAHAHAH
cotto_work Coke, good idea. I'll start a branch. They'll all be renamed in the form "Parot_<function name>_<subsystem>". 20:40
whiteknight cotto_work: I compiled a list of files yesterday, and turned every single file in the list into a GCI task 20:42
1 was merged this morning. One is in the works now
and it's Parrot_<subsystem>_<functioname>
cotto_work I'm just concerned about knowing when we can close the ticket.
whiteknight WE CLOSE IT WHEN I SAY WE CLOSE IT 20:43
Nah, when GCI is over we look over the list of files not converted and see what to do then
I suspect all the low-hanging fruit will be picked by that point 20:44
cotto_work ok. melange should at least tell us what did and didn't get done.
whiteknight yeah, I hope so
Maybe I'll copy the list to someplace else
dalek rrot/string_checks: 1e06c9f | nwellnhof++ | src/string/encoding/shared.c:
[str] Fix assertion in Unicode chr
nwellnhof whiteknight++ 20:45
whiteknight I added the list to the ticket itself 20:48
nwellnhof whiteknight: Can you test string_checks with the latest commit? 20:49
whiteknight will do. Building and testing now 20:50
NotFound make cover isn't working well, it doesn't report a lot of files. 20:52
whiteknight nwellnhof: smolder.parrot.org/app/projects/rep...tails/1309 all tests pass 20:56
nwellnhof: ++
nwellnhof++
20:58 hudnix joined
NotFound "Current convention is to create branch names of the form username/foo" I don't see any branch that follows that convention. 21:04
whiteknight yeah, that's sort of a lie
and by "sort of", I mean it's a lie
nwellnhof most of the branches are still from SVN 21:05
GeJ Bonjour everyone. 21:08
whiteknight hello GeJ
GeJ Heya whiteknight. How's life treating you these days? 21:10
whiteknight GeJ: Things are going well. I'm trying to stay busy and succeeding
21:18 rurban_ joined 21:21 darbelo left, rurban left, rurban_ is now known as rurban, darbelo joined, darbelo left 21:25 bluescreen left
dukeleto NotFound: yeah, i wanted that to be the case, but it didn't happen 21:25
NotFound: you can change that to be: "of the form ttXXXX_foo" where XXXX is the TT, if one exists. 21:26
NotFound: "otherwise, pick a branch name that will be meaningful to others"
21:26 lucian left
NotFound dukeleto: I dislike to document conventions that no one is going to follow. 21:26
That last one is sensible ;) 21:27
21:27 lucian joined, whiteknight left
dukeleto NotFound: I have been following the ttXXXX_foo convention, and so does kid51 21:28
NotFound: people can immediately see which ticket a branch pertains to, and i think that is useful
NotFound: it is optional, not mandatory
NotFound Fine, then.
cotto_work If I work on a branch for a ticket, I use that convention. 21:29
dalek rrot/string_checks: 741efbd | nwellnhof++ | src/string/encoding/utf16.c:
[str] Add missing ASSERT_ARGS
rrot/string_checks: 319feed | nwellnhof++ | src/string/encoding/u (3 files):
[str] Fix exception types in string scan functions
cotto_work It's hard to make something like that mandatory. 21:30
21:35 perlite_ joined
dukeleto The wording "Current convention" should be changed to "It is recommended but not mandatory" 21:37
NotFound The "but not mandatory" is just noise. 21:38
dukeleto NotFound: +1
21:39 perlite left, perlite_ is now known as perlite 21:46 fperrad left 21:58 lucian left
dalek rrot/mappedbytearray: e006c1d | NotFound++ | / (2 files):
skeleton initial implementation
 21:58
cotto_work should that be "bytemapped" ? 22:01
NotFound Naming is open to discussion. I followed the pattern of xxxxyyyyyArray PMC as a start. 22:02
dalek rrot/mappedbytearray: b585195 | NotFound++ | src/pmc/mappedbytearray.pmc:
a few checks for mmap availability
 22:04
NotFound Maybe this branch should be kept private for a now, but it helps me familiarizing with git branching. 22:05
dukeleto NotFound: i think it is good for you to familiarize yourself with git branching. 22:08
NotFound Is bad to have a metacommiter not used to git, but that wasn't my decision ;) 22:09
22:28 Andy left
dukeleto aloha: coverage? 22:31
aloha dukeleto: coverage is cv.perl6.cz or tapir2.ro.vutbr.cz/cover/cover-results/
22:31 Andy joined
dalek rrot/mappedbytearray: b999b3b | NotFound++ | src/pmc/mappedbytearray.pmc:
implmented minimal functionality, read only, linux only
 23:10
23:28 jan left 23:32 jan joined 23:33 hercynium left 23:57 whiteknight 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!