Parrot 3.3.0 released | parrot.org | Log: irclog.perlgeek.de/parrot/today
Set by moderator on 15 May 2011.
NotFound_b BTW I doubt people uncomfortable with C will be able to guess the right signatures and structures. 00:00
whiteknight that's another good point. Our "language" for NCI thunks is terrible 00:02
dukeleto It is quite a sour-tasting DSL. 00:04
whiteknight right. We do want to get rid of it eventually. 00:07
figuring out what to replace it with is the big question
NotFound_b What is that language? Did we have some updated documentation? 00:09
cotto_work It sounds like the conclusion is that we do want to add nci sigs (and generated thunks) to replace the old 't' versions. 00:14
NotFound_b cotto_work: big +1 00:15
pmichaud I'm hand-crafting a list 00:17
cotto_work pmichaud: thank you
pmichaud it's based on what existed in src/nci/extra_thunks.c in 2011.04 00:18
cotto_work pmichaud: once that's in master, should we consider the release unblocked or is there anything else?
pmichaud I think unblocked. I'd like to be able to actually test mysql on zavolaj before the release goes out... but I'm running out of time (also considering other things I have to get done today)
cotto_work pmichaud: ok. I'll let gerd know. 00:20
pmichaud: shouldn't ncifunc be merged? 00:22
pmichaud yes, if we're comfortable with merging. passes make test on my box, including the new ncifunc tests
cotto_work I'll give it a whirl'
pmichaud let me make sure I don't have any local commits
I do not. 00:23
merge away.
dalek rrot: 1f73a3a | pmichaud++ | src/nci_test.c:
Add some functions for testing char* parameters in NCI.
rrot: 6691e38 | pmichaud++ | runtime/parrot/library/NCI/Utils.pir:
[NCI] Add "ncifunc" as a replacement / approach to "dlfancy".
rrot: 3b55d79 | pmichaud++ | / (3 files):
Merge branch 'master', remote-tracking branch 'origin' into ncifunc
cotto_work Is anything needed other than make test?
dalek rrot: c1315ce | pmichaud++ | t/library/nciutils.t:
[NCI]: Add tests for 'ncifunc' in NCI::Utils.
rrot: 0366834 | plobsing++ | runtime/parrot/library/NCI/Utils.pir:
properly handle null arguments
pmichaud oops
rrot: 6e069f0 | plobsing++ | runtime/parrot/library/NCI/Utils.pir:
properly handle null string returns
rrot: 2fef4a1 | plobsing++ | src/nci/extra_thunks. (2 files):
add thunk for mysql_real_connect
rrot: f820150 | pmichaud++ | t/library/nciutils.t:
[nci]: Add test to NCI::Utils for functions returning a NULL pointer.
pmichaud looks like I just did, by accident
cotto_work I'll test master then.
pmichaud great
git diff ncifunc master shows no differences on my box, so I think it's a clean merge 00:24
cotto_work good sign
00:26 kid51 left
dalek rrot: fd81720 | cotto++ | MANIFEST:
manifix
 00:28
nxed: r1005 | NotFound++ | trunk/winxedst1.winxed:
declare some more functions with using static, for a little speed gain in stage

undeclared identifiers are globals in hash initializers		 00:29
cotto_work looks good on my machien 00:31
dalek rrot: cc9a715 | cotto++ | / (2 files):
minor codingstd fixes; pod and trailing space
 00:35
pmichaud gist.github.com/975660 # a hand-edited list of signatures we should have in the release 00:41
note that it's hand-edited, so I could've easily made a typo
but the instructions on how I started are there, so someone else can follow and reproduce
afk
mikehh hey cotto, you beat me to those, I commited but hadn't pushed yet :-{ 00:51
00:51 bubaflub left
mikehh how do you revert without re-cloning? 00:51
something to do with HEAD 00:53
plobsing reset HEAD 00:55
dukeleto mikehh: you want to blow away all changes? 00:58
mikehh: git reset --hard HEAD
mikehh: the --hard part updates your working copy. Without that, you will only reset your staging area
mikehh dukeleto: yeah, that's the one
dalek sella: 90b8b62 | Whiteknight++ | / (3 files):
More tweaks and fixes to the REPL. improve commands. Add an ability to add a common code preamble to each line.
 01:01
sella: 232fe78 | Whiteknight++ | src/winxed/Repl.winxed:
Add in some basic error recovery. Cleanup more code
01:13 woosley joined
dalek rrot: 8869229 | Whiteknight++ | docs/project/release_manager_guide.pod:
soh_cah_toa volunteers for September.
 01:14
whiteknight fulltest takes forever 01:20
if it were faster, I would do it more often 01:21
cotto mikehh, all part of my evil plan 01:22
sorear whiteknight: you should set up a remote server to asynchronously fulltest and send you the results after committing
whiteknight sorear: I'm stuck on the first step: set up a remote server 01:23
mikehh well TapTinder seems happy 01:26
whiteknight fulltest clean on my machine with clang 01:27
I'm going to run again with g++, since that's the most picky 01:28
maybe icc if I have time
sorear What's jitterbug in the parrot context? 01:29
whiteknight also, I can't parallelize fulltest, because the stupid threading test goes all stupid on me 01:30
cotto I added make not_stupid_tests to get around that. 01:32
01:33 lucian_ left 01:36 whiteknight left
dukeleto sorear: jitterbug is a language-agnostic continuous integration system for git 01:47
sorear: i was running the parrot test suite with it, but parrot found some bugs in it, so I am fixing those :) 01:49
sorear: here is a working jitterbug instance as an example: jitterbug.sgn.cornell.edu:8090/ 01:50
01:51 kid51 joined
NotFound_b dukeleto: I have the gcc farm account ready. 01:51
dukeleto NotFound_b: cool! 01:53
kid51 darwin/ppc: master: commit f8201502: make test: PASS
dukeleto NotFound_b: we don't really have a framework yet. If you log into some of the gcc machines, you will see that I have smokers setup to pull down the latest commits and submit reports to smolder 01:54
which is down.
cotto I also had good results with a debian ppc vm.
dukeleto which sucks.
NotFound_b: gcc64 is openbsd. I can't get parrot master to compile on it. 01:55
NotFound_b: if you want something to cut your teeth on :)
cotto dukeleto, is there anything on M0 I can do? I don't want to start implementing ops in the interp until I can run them. 02:00
mikehh All tests PASS (pre/post-config, make corevm/make coretest, make world/make test, fulltest) at 3_3_0-296-g8869229 02:03
cotto If you want bytecode, t/m0/m0_bytecode_loading.t will work for generating everything except the ops.
mikehh Ubuntu 11.04 i386 (g++ --optimize)
pmichaud 5h to release
cotto soh_cah_toa++ for release managering 02:04
NotFound_b dukeleto: how do you clone parrot into a machine in the farm? My first attempt is painfully slow. 02:09
It's better now in gcc64, the machine I was using must be busy. 02:14
dukeleto NotFound_b: how are you cloning? 02:16
NotFound_b: also, if you look for ~leto/git/parrot, you can just copy that, if it exists. Also, every machine's HD is NFS mounted at /N/home/foo, iirc
NotFound_b Public access git from github
git://github.com/parrot/parrot.git 02:17
kid51 Uh, oh. I have a failure on darwin/ppc:
dukeleto NotFound_b: parrot.git is just big.
kid51 t/library/nciutils.t
dukeleto NotFound_b: copy my version from ~leto/git/parrot
NotFound_b: i have it on man gcc boxes. gcc64 for sure
s/man/many/
NotFound_b dukeleto: yeah, but 6 KiB/s is lower than I expected
dukeleto NotFound_b: it usually is pretty fast. Github may be overloaded currently 02:18
NotFound_b Is faster now in gcc64, was testing in gcc10
cotto NotFound++ 02:20
nopaste "kid51" at 192.168.1.3 pasted "t/library/nciutils.t: darwin/ppc: cc9a7154802: FAIL" (16 lines) at nopaste.snit.ch/45998
NotFound_b Is nice to have access to such a bunch of machines, I'm tempted to start writing skynet.winxed 02:21
dukeleto NotFound_b: some gcc machines are not on the fast interwebs. Most are, though. But some are hosted outside of the main datacenter.
NotFound_b: indeed :)
NotFound_b: so many machines, so little time
cotto kid51, trying to duplicate on my ppc vm 02:22
qemulator is buggy 02:25
kid51 cotto: This is a strange one, since I had make test PASS on previous commit
But, then again, instability in 'make test' results has been the main theme on this platform over past two weeks. 02:26
cotto I think make reconfig on this vm takes longer than make does on my laptop. 02:27
kid51 Yes, that FAIL has to go onto the heap of unstable tests. I just got a pass when running 'make library_tests' 02:33
cotto kid51, ok 02:34
I doubt I'll be able to run the test before the release.
kid51 linux/i386: make test: PASS on: gcc build; gcc build --optimize; all g++ build; all g++ build --optimize
cotto I'm working on updating the nci thunk sigs in master now. 02:38
NotFound_b dukeleto: looks like it's failing in sparc specific hand-writen machine code. 02:43
kid51 make relcheck PASS 02:46
02:51 kid51 left
cotto that test looks good on ppc 03:05
dalek rrot: 9e53f16 | cotto++ | src/nci/extra_thunks.nci:
add extra nci thunk sigs

These are to replace the ones that were removed with the 't' deprecation and removal. pmichaud++ for sorting through these		 03:06
cotto that was not an exciting commit 03:08
pmichaud, can you verify that everything you need is in master? 03:09
pmichaud cotto: let's see if I can at least get the mysql example working 03:12
It'll take about 20 mins to set things up, maybe
cotto pmichaud, thanks 03:16
plobsing cotto: you might want to run 'make bootstrap-nci' for your last change to take effect 03:26
NotFound_b dukeleto: I think I've found a fix for the sparc problem.
cotto plobsing, thanks. I knew there was something I was missing.
done 03:28
dalek rrot: 20a159d | cotto++ | src/nci/extra_thunks.c:
rebootstrap nci
pmichaud I got a little farther: gist.github.com/975894 03:37
would be nice if that error told me what the illegal type was :-| 03:41
cotto pmichaud, are we still missing a signature?
lta
pmichaud I don't know yet.
cotto ok
pmichaud updates the error message 03:43
I'm guessing that perhaps the type numbers have changed
Illegal type (-70) in initializer for struct 03:44
yup 03:45
NotFound_b Some sparc guru available here?
plobsing pmichaud: illegal type means a type code (integer) was specified that does not fall within the range of defined types 03:47
pmichaud yes
I'm just not sure what the new type codes should be
the value is coming from zavolaj: 03:48
given $!of { 03:49
when Str { $typeid = -70 }
when Int { $typeid = -92 }
when Num { $typeid = -83 }
default { die "Unknown type"; }
}
I don't know where the -70, -92, -83 come from.
plobsing yikes. why are you using hardcoded integer constants?
pmichaud what else would we use?
plobsing use the symbols defined in datatypes.pasm
pmichaud from Perl 6?
not yet possible.
plobsing winxed seems to handle the issue just fine
pmichaud yeah, well, winxed has a lot fewer constraints than we do. 03:50
NotFound_b Yes, but in a hackish way.
I just read the .pasm file looking for macro const declarations
plobsing pmichaud: am I to assume that those type codes correspond to "auto-converting cstring", int, and float respectively? 03:51
NotFound_b: yes, the way we expose constants is flawed and generally leads to inconvenience of access and an increased risk in breaking PBC compatibility
we need someone architecty to sit down and figure out how to do it right. until then, we need to use what we've got. 03:52
pmichaud in 3.3.0 they were apparently DATATYPE_STRUCT (-70), DATATYPE_LONGLONG (-92), and DATATYPE_INT8 (-83)
that seems very not right.
NotFound_b However, the hack has proved to be better (for me) than keep track by hand of changes in constants. 03:53
pmichaud oh, looks like the values changed from 3.0.0 to 3.3.0
in 3.0.0 they were DATATYPE_INT, DATATYPE_CSTR, and DATATYPE_DOUBLE 03:54
NotFound_b constants should never change. They are constants, aren't they? ;)
pmichaud "constants aren't, variables won't"
so looks like zavolaj is using the 3.0.0 constants -- or at least the version I have is doing so.
for now I'll hard code the constants. I'll update things to try to read from datatypes.pasm a bit later 03:55
almost made it to the end: gist.github.com/975923 03:56
zavolaj is having trouble dealing with null returns... but I'm guessing that has to be a zavolaj issue 03:57
dalek rrot: 328122b | pmichaud++ | src/pmc/unmanagedstruct.pmc:
[nci]: Update the "Illegal type" message of UnManagedStruct to at least tell us what type was illegal.
 04:08
pmichaud cotto / others: I'm able to get a fair distance through the mysqlclient.p6 example, and unix-fork.p6 works just fine for me as well. 04:11
bacek_at_work ~~
pmichaud There's almost no way I can test every library that might exist before the release (or even within a day or so), so I think we'll have to accept what we have now and hope it works out
bacek_at_work
.oO( It took less effort to make GMS works after merge/unmerge before 3.0 than single NCI "t" signature )
 04:12
pmichaud oops, I broke a test 04:13
nopaste "NotFound" at 192.168.1.3 pasted "patch for sparc64, being tested on gcc farm" (30 lines) at nopaste.snit.ch/46005 04:14
NotFound_b Can someone with some knowledge of sparc take a look at that? 04:15
dalek rrot: b88be2a | pmichaud++ | t/pmc/unmanagedstruct.t:
[nci] Update UnManagedStruct tests for new illegal type message change in 328122b.
 04:16
pmichaud afk # sleep 04:27
cotto 'night 04:29
dalek rrot: 59be8ac | NotFound++ | src/gc/system.c:
use inline assembler in gcc/sparc64 instead of hand-coded
 04:56
dukeleto NotFound++ # utilizing the compile farm 04:57
NotFound_b dukeleto: it's magical, I've fixed a low level sparc problem and I don't have any idea about sparcs. 04:59
Of course, google helped a lot.
Well, don't know if it's fully fixed, make test is still running. 05:02
cotto NotFound_b, wfm
dukeleto cotto: to answer your question, you could help me by creating a binary representation of hello.m0 to test against 05:14
NotFound_b: great to see you putting the farm to use
NotFound_b Is more interesting than farmvile 05:16
cotto dukeleto, wfm 05:18
NotFound_b, a lot of things do 05:19
NotFound_b Don't know for sure, never used it. Don't even have a facebook account.
05:20 mtk left 05:21 theory left 05:26 mtk joined
cotto dukeleto, what's in hello.m0b now? 05:31
looks like it was generated with the assembler and isn't quite valid as is. 05:33
05:39 dafrito left
dukeleto cotto: hello.m0b is generated partially now 05:40
cotto: and I need what we actually think it should look like to test against
cotto: i am feeling under the weather and heading to bed. 05:41
dukeleto -> sleep()
cotto 'night 05:44
NotFound_b make test finised succesfully :) Going to sleep. 05:50
05:50 NotFound_b left
cotto goes to sleep 05:51
dalek rdinal: ec69f91 | (Daehyub Kim)++ | / (2 files):
add Float.pir to setup.pir as build target
 05:55
rdinal: 4f96cac | tadzik++ | / (4 files):
Merge pull request #9 from lateau/lib191

new methods
06:26 fperrad joined, jsut joined 06:31 jsut_ left
dalek rdinal: df0b3aa | tadzik++ | src/classes/ (3 files):
Dynamagically load OS. Fixes t/file/dir.t
 06:47
06:48 dod joined 07:27 mj41 joined
moritz gerd just talked to me over phone - his network connectivity broke down, so the parrot release will be delayed 07:36
chances are it get fixed later on today 07:37
so he suggests we wait until 1800 CEST (1600 UTC), and if he doesn't show up by that time, somebody else should take over
dalek a: 3e9ab2e | fperrad++ | lua/lib/luamath.pir:
refactor without Math/Rand
 07:41
07:49 rurban_ joined 07:50 UltraDM joined 07:51 rurban left, rurban_ is now known as rurban
dalek lscript: a19c3e5 | fperrad++ | wmlscript/library/wmlslang.pir:
refactor without Math/Rand
 07:57
a: 7edb8bf | fperrad++ | lua/lib/luamath.pir:
random: fix range
 08:07
08:28 contingencyplan left 08:31 gerd joined
gerd I am sorry, my network connection was broken. I would like to start now the checkout for release 3.4.0. 08:35
Would you please stop commits, the release update is running. 08:38
make fulltest is running 08:46
mikehh gerd: anything you want me to check on Ubuntu 11.04 i386/amd64 08:52
gerd mikehh: not I think running the "make fulltest" no my Fedora is enough 08:54
It is still runing
mikehh just finished fulltest on Ubuntu 11.04 i386 with g++ 08:55
with perl 5.14.0 08:56
gerd now it is finished on my mashine, successful 08:59
dalek rrot: f74aff9 | (Gerd Pokorra)++ | / (12 files):
update to version 3.4.0
 09:01
gerd "git push" is done
mikehh gerd++ 09:05
gerd I tagged the release 09:06
"make release_check" is running from the tarball sources 09:07
release_check is okay; the tarfiles are uploaded to ftp.parrot.org 09:18
moderator Parrot 3.4.0 released | parrot.org | Log: irclog.perlgeek.de/parrot/today 09:24
09:48 woosley left 09:50 SHODAN joined
bacek gerd++ # yet another release done 09:56
moritz gerd: did you forget to push the tags_ 09:57
s/_/?/
gerd: after git pull, the output from git describe --tags is still a RELEASE_3_3_0-something
dalek website: gerd++ | Parrot 3.4.0 "Pacific Parakeet" Released! 10:03
website: www.parrot.org/news/2011/Parrot-3.4.0
mikehh moritz: I got RELEASE_3_4_0 10:08
moritz huh, git tag shows it too 10:09
ah, my fault 10:10
my source tree was behind, not sure why it didn't merge
mikehh moritz: I think the tag was pushed a bit later 10:11
moritz bumps rakudo's PARROT_REVISION 10:12
10:34 ambs joined
gerd I am leaving for lunch, announce email is sent, a nice today to the developers, bye 10:34
10:35 gerd left 11:02 Psyche^ joined 11:07 Patterner left, Psyche^ is now known as Patterner 11:16 woosley joined 11:37 woosley left 12:10 whiteknight joined
whiteknight good morning, #parrot 12:11
tadzik good morning whiteknight 12:12
whiteknight hello tadzik
12:16 bluescreen joined 12:20 bluescreen left 12:29 lucian joined
whiteknight gerd++ on the release 12:32
Coke_ . o O ('ttttttttttttthpt' thunk.) 12:33
12:35 bluescreen joined 12:36 mtk left 12:40 mtk joined
dalek sella/gh-pages: e4b8ece | Whiteknight++ | libraries/memoize.md:
add descriptions of classes, fix formatting, add code examples
 13:05
13:07 JimmyZ joined
dalek sella/gh-pages: 568a03a | Whiteknight++ | / (3 files):
memoize is now stable. Update webpage to reflect that
 13:15
sella: 60f76ae | Whiteknight++ | s (13 files):
Memoize is now stable
13:20 mtk left 13:21 mtk joined, lucian left 13:25 bubaflub joined 13:35 contingencyplan joined
dalek nxed: r1006 | NotFound++ | trunk/winxedst1.winxed:
remove a redundant initialization
 14:09
14:12 alester left
dalek nxed: r1007 | NotFound++ | trunk/winxedst1.winxed:
fix and avoid cloning in findclasskey method chain
 14:19
rkdown: 5a3bcf8 | fperrad++ | / (2 files):
refactor without Math/Rand
 14:24
14:29 bluescreen left 14:43 UltraDM left, bluescreen joined
Coke_ drinks coffee. 14:44
tadzik feels like tea 14:48
Coke_ wonders if the parrot frappr group still exists. 14:49
(or if frappr was even the app name.)
(with geo-tagged users.)
15:01 SHODAN left 15:03 spinclad left
Coke_ wonders if he still has a system on which he can fight SHODAN. 15:16
15:23 alester joined
dalek rrot: 2b840ec | Whiteknight++ | runtime/parrot/library/distutils.pir:
Change exe_pbc build rules in distutils to use the specified bin name for the output .exe file instead of a mangled form of the .pbc file name. I can't find any other projects using this feature, but it seems to work well on my machine without breaking anything
 15:39
rrot: d852087 | Whiteknight++ | / (56 files):
Merge branch 'master' of github.com:parrot/parrot
rrot: 4a968e3 | fperrad++ | runtime/parrot/library/ (2 files):
[osutils] refactor without Math/Rand
 15:48
15:49 rurban_ joined 15:51 rurban left, rurban_ is now known as rurban 15:53 theory joined
dukeleto ~~ 15:53
15:55 mj41 left 15:59 particle left, particle joined 16:07 dod left
ttbot Parrot 2b840ec0 MSWin32-x86-multi-thread make error tt.taptinder.org/cmdinfo/9470 16:08
16:17 bluescreen left, bluescreen joined
Coke_ would be nice if "msg coke" would send me an email if aloha thought I was "offline" 16:18
16:18 particle left 16:22 lucian joined 16:27 particle joined 16:38 particle left 16:43 M_o_C joined 16:44 JimmyZ left 16:49 darbelo joined 16:53 mtk left, dodathome joined 16:54 mtk joined 17:02 mtk left, mtk joined 17:07 particle1 joined 17:12 hercynium joined 17:13 particle1 is now known as particle
cotto_work ~~ 17:20
gerd++ 17:24
nice to have another uneventful release 17:31
17:34 theory left
dalek TT #2113 created by whiteknight++: FileHandle .puts and .print 17:44
TT #2113: trac.parrot.org/parrot/ticket/2113
17:44 darbelo left
NotFound whiteknight: ping 17:46
whiteknight pong
cotto_work: I take it back. We do have a smooth, automated release system. It's name is G.E.R.D.
all we need now is to retroactively apply an acronym 17:47
NotFound About TT#2133: I think that taking a PMC argument has the advantage that derived classes can specialize
whiteknight NotFound: that's fine. We can have a PMC argument if we want. I just don't think we need both of those methods together 17:48
tcurtis ~~ 17:49
whiteknight It's absurd to have two methods, doing basically the same thing, calling the same API function, etc 17:51
NotFound This is Parrot version 3.4.0-devel built for sparc64-openbsd. (gcc compile farm)++ 17:53
tadzik any outside-us gsocers around? 17:58
s/us/US/
17:58 M_o_C left 17:59 hercynium left
whiteknight tcurtis: I added a simple tokenization library to Rosella 18:01
tcurtis: it does simple tokenization based on character classes, but it's better than nothing
cotto_work bellard.org/jslinux/
whiteknight it can correctly tokenize a string like "var foo = self.whatever(1, 2, 3);" 18:02
atrodo cotto_work> i saw that earlier. Will wonders never cease
cotto_work atrodo: clearly they will not 18:03
atrodo cotto_work> gives hope for M0 on JS
cotto_work once it's fully specified 18:06
whiteknight looks forward to JS-on-Parrot-on-JS-on-Lolcode 18:07
lolcode, of course, runs on turtles all the way down
plobsing those poor turtles 18:08
cotto_work speaking of which 18:11
allison: ping
whiteknight what part of that reminded you of allison? The lolcode or the turtles? 18:22
allison cotto_work:pong
allison dreams of lolturtles 18:24
cotto_work allison: privmsg 18:25
lucian tadzik: me 18:34
18:38 darbelo joined
tadzik lucian: might if I privmsg? 18:39
lucian tadzik: not at all 18:40
dalek nxed: r1008 | NotFound++ | trunk/winxedst1.winxed:
refactor handling of constants and drop some dead code
 18:47
18:50 darbelo left 18:53 RobertLJ joined
cotto_work pmichaud: does Rakudo want the Select PMC? 18:55
er, dynpmc
whiteknight NotFound: ping 18:56
NotFound whiteknight: pong
whiteknight NotFound: is it possible to share const declarations between files? 18:57
or do they have to be defined in each file
NotFound whiteknight: there is a way that I implemented by a plobsing suggestion: generating a .pasm file for including.
whiteknight NotFound: and that works fine with the parser? 18:58
NotFound You need to use $include_const 18:59
whiteknight is there an example of that anywhere?
NotFound A bit ugly, but there is no native include yet.
whiteknight is Native include on the roadmap?
NotFound --target=include to generate the pasm 19:00
Examples of $include_const usage, there are several in examples. 19:01
For parrot native include files, but is the same.
19:01 RobertLJ left
NotFound Native include is on the roadmap, yes, and the improvements in scope search are cleaning the path. 19:02
The idea is not to do a text editor-alike include, but allowing only declarations. 19:03
19:04 darbelo joined
whiteknight I'm going to add that --target=include thing to the distutils library. That's useful 19:25
19:34 ShaneC joined 19:36 darbelo left, darbelo joined, particle left 19:48 RobertLJ joined, particle joined 19:54 darbelo left, kid51 joined 19:55 particle left 20:00 PacoLinux left 20:01 particle joined 20:02 dodathome left
cotto_work #ps in 21 20:09
20:13 darbelo joined 20:17 soh_cah_toa joined
cotto_work whiteknight: ping 20:22
darbelo tcurtis: ping 20:26
whiteknight pong (short) 20:27
I'm packing up to go in ~3 min
pmichaud cotto_work: several Rakudo users want select() capabilities
so if it doesn't show up in Parrot, we may package it in Rakudo or Star
dalek sella/gh-pages: a0e4965 | Whiteknight++ | libraries/winxed.md:
+preliminary docs about Winxed.Repl
 20:29
whiteknight incoming
sella/gh-pages: 5b7ff94 | Whiteknight++ | libraries/winxed.md:
+list of new classes for Winxed library
sella: b54f9cc | Whiteknight++ | VERSION:
bump VERSION for memoize
sella: 960d0ba | Whiteknight++ | / (9 files):
break the repl logic into a series of delegate classes, which can be plugged.
sella: f2571a9 | Whiteknight++ | / (4 files):
add infrastructure to build and install utility/driver programs, such as the repl.
sella: 81b83bf | Whiteknight++ | src/nqp/ (25 files):
remove old NQP source files. They are all sufficiently out of date that they aren't worth keeping. They are all unchanged since the version_1 tag, so if we need to revisit them we can look there
sella: 8e76cde | Whiteknight++ | src/test/ (2 files):
Remove some TODO notes from Test.Asserter and add docs. Test.Builder takes a handle to use for output, defaulting to stdout
sella: 2319820 | Whiteknight++ | src/unstable/path/ (4 files):
Small refactors to Path, and documentation for that library
sella: c52b2cb | Whiteknight++ | s (4 files):
Winxed has an ability to generate a .pasm file with constants. Add pasm_winxed entries to the distutils library for handling those cases.
sella: 31a36df | Whiteknight++ | s (3 files):
Fix build problem with inheritance. Tokenizer needs to be included in the build before CClass. Cleanup/rename CClassTokenizer -> Tokenizer.CClass
sella: d2a5857 | Whiteknight++ | s (4 files):
Add a token class, and use that in the tokenizer instead of returning raw strings
whiteknight blah, we really need to set a lower limit on dalek
cotto_work back in 2m 20:30
20:31 PacoLinux joined
NotFound #ps in -2 ? 20:32
20:38 whiteknight left 20:40 darbelo left 20:42 darbelo joined
cotto_work pmichaud: I didn't know that Rakudo users wanted Select. I'll make sure it gets done soon (i.e. next couple days). 20:42
NotFound All test pass in gcc40 2.6.26-2-powerpc64 20:44
pmichaud cotto_work: thanks 20:48
cotto_work pmichaud: sure. Is the first time you've mentioned that Rakudo wants Select?
pmichaud maybe in so many words 20:52
I asked about the Select PMC a couple of days ago, but didn't necessarily say "Rakudo wants this"
looks like I asked about it yesterday; and tadzik/moritz inquired about it on May 6 20:53
tadzik I think I, personally, created the mess :)
cotto_work ok. In the future, let me know when Rakudo wants something like that.
pmichaud sure 20:54
and my statement above was supposed to be "maybe *not* in so many words"
(can't type these days)
tadzik ISTR tewk himself wrote select for he wanted async IO in Rakudo. When I wanted to implement some of that, I saw that Select is not in Parrot core
cotto_work that's how I read it
21:06 fperrad left
cotto_work dukeleto: ping 21:07
dukeleto cotto_work: pong 21:13
cotto_work dukeleto: where are the test results on jitterbug?
dukeleto cotto_work: borked
cotto_work dukeleto: ok
dukeleto cotto_work: github.com/franckcuny/jitterbug/issues/70 explains
cotto_work dukeleto: gotcha 21:14
dukeleto cotto_work: the test results were not useful, because previous builds ruined the cached git repo, and then it was reporting build errors when there weren't any
21:14 kid51 left
cotto_work better nothing than bad data 21:14
dukeleto cotto_work: i haven't had a single second to look into it since i noticed it was broken
21:14 hercynium joined
tcurtis darbelo: pong 21:15
darbelo tcurtis: Hey. Sorry I disappeared on you last week. 21:16
I should be more reachable this time arround.
Do you have time for a quick status update? 21:17
cotto_work dukeleto: will jitterbug eventually do everything we currently do with smolder? 21:18
(or try to do)
tcurtis darbelo: it's fine. sure.
darbelo: I've read the first four chapters of "Practical Translators for LR(k) Languages". I need to finish that. 21:22
I've got my development environment setup (I'm using a Parrot+winxed+plumage+anything-else specifically for working on GSoC). 21:25
21:26 soh_cah_toa left
tcurtis I haven't created a github repository for the project yet (partially because I've yet to name it). 21:26
I'll do that sometime in the coming week.
dalek rrot: 65b742b | NotFound++ | frontend/parrot/main.c:
cast isxdigit argument to unsigned char
 21:28
tcurtis And I'll play around with designing the object representation for grammars.
I think I'll write a grammar pretty-printer in order to test out using the grammar classes, as well as a few simple example grammars. 21:30
21:30 mj41 joined, GeJ joined 21:31 soh_cah_toa joined, eternaleye_ joined
tcurtis darbelo: One of the design decisions I'm uncertain about there is how to represent tokens. 21:32
NotFound All test pass in NetBSD 5.1 amd64 gcc 4.1.3
21:32 TiMBuS|Away joined 21:33 simcop2387_ joined 21:34 jjore_ joined, jtpalmer_ joined 21:35 autark_ joined
dukeleto cotto_work: smolder and jitterbug are orthogonal 21:35
cotto_work: smolder *accepts* test reports from random places, with a web interface. Jitterbug is a dedicated smoker with a web interface.
cotto_work: so, no. 21:36
darbelo tcurtis: Do you have any prior art in mind?
cotto_work dukeleto: ok.
dukeleto cotto_work: jitterbug could easily submit to smolder, tho. If smolder weren't down, that is.
21:36 AzureSto_ joined, ambs_ joined, hercynium left, mtk left, alester left, Patterner left, ambs left, preflex left, autark left, TiMBuS left, simcop2387 left, jtpalmer left, GeJ_ left, eternaleye left, jjore left, AzureStone left, TiMBuS|Away is now known as TiMBuS, simcop2387_ is now known as simcop2387, ambs_ is now known as ambs
cotto_work I'm a little sad to keep smolder, but having something working trumps most other concerns. 21:36
21:37 ambs left
dukeleto cotto_work: yeah. sad but true. 21:37
cotto_work until we start having excess tuits, that is
21:37 alester joined
cotto_work looks forward to the Great Coming Tuit Flood 21:37
21:37 preflex joined, ShaneC left
mikehh cotto_work: I keep meaning to do some work on smolder, but other things keep intervening 21:38
21:38 mtk joined
tcurtis darbelo: looking into that currently. 21:38
darbelo The way I think about it, a token isn't really an object, but more of a role or an interface. Which probably makes representing them kind of awkward. 21:40
21:41 davidfetter joined 21:42 estrabd left, estrabd joined
tcurtis ANTLR looks like it uses a single token class with an integer tag to denote its type (although custom classes can be used). 21:47
darbelo tcurtis: If you were to think really hard about it. What are the most basic things that the parser needs to know about a token? 21:48
If you were walking on the street and found totally opaque token from an unknow source. What questions would you ask it? 21:50
tcurtis What kind of token it is. Maybe its location in the input file for debugging purposes. Actions will want to know the actual substring of the input that composes it.
Although one could also delegate that portion into the lexer. 21:51
In which case, the Actions would possibly need to know some value that the lexer had given it.
aloha, msg whiteknight Since you were talking about perhaps writing a lexing library for Rosella, could we talk sometime about the interface it would present? 21:53
aloha tcurtis: OK. I'll deliver the message.
tcurtis msg whiteknight I'm particularly interested in your thoughts on token representation.
aloha OK. I'll deliver the message.
21:54 bluescreen left
bacek_at_work ~~ 21:55
tcurtis darbelo: I don't like the ANTLR approach. It looks like it makes it hard to make decisions about the semantic meaning of the token at the level of the lexer.
And it would require some degree of coupling between the lexer and the parser.
21:56 Patterner joined, hercynium joined
tcurtis One design I'm considering is to have one special token class for things like '(' or such where you really just care about the character value. And match everything else by isa. 21:58
With a class that provides built-in support for things like source location information that tokens can subclass to easily provide that suppport once the parsers start understanding it. 22:01
22:08 hercynium left
darbelo tcurtis: It sounds good to me, be carful about overdesigning stuff into the tokens. 22:09
tcurtis darbelo: could we set up a regular time to meet each week? 22:14
22:18 bubaflub left
darbelo tcurtis: We should. I can manage something lis this time on tuesdays/fridays 22:19
where this time is the vincinity of 19:00 in UTC-3 22:20
22:00 UTC
I don't know how that translates into your timezone.
tcurtis 17:00 for me, which is quite convenient for me on Tuesdays. 22:21
cotto_work www.worldtimebuddy.com/ 22:22
darbelo Tuesdays it is. Which makes it quite close to "after #ps", also convenient.
cotto_work silly name, but I really like the ui
tcurtis darbelo: great. Do you have anything else you'd like to talk about today? 22:24
22:25 jevin left
darbelo I had some questions, but I still need to read up before I can properly articulate them. So, not really. 22:27
msg me when you set up your project's repo.
aloha OK. I'll deliver the message.
22:28 darbelo is now known as me, me is now known as darbelo
tcurtis darbelo: will do. msg, email, or ping me if you figure out how to articulate your questions. 22:28
darbelo Will do. 22:29
22:30 whiteknight joined 22:42 bubaflub joined 22:45 xenoterracide joined 22:48 RobertLJ left, mj41 left
cotto_work whiteknight: ping 22:53
whiteknight pong
cotto_work whiteknight: what kind of plans do you have for profiling?
whiteknight cotto_work: surprisingly, I don't have any plans right now. I haven't dug into the code enough yet 22:54
cotto_work whiteknight: ok. I'm thinking about how we can most effectively split up the work. I don't have a good answer.
dukeleto mmmm, profiling 22:55
cotto_work profiles dukeleto 22:56
dukeleto just skyped with soh_cah_toa++ (a.k.a The New Debugger Guy)
cotto_work woot.
what news?
soh_cah_toa :)
cotto_work is he done yet? 22:57
dukeleto cotto_work: hah. Not yet. But he told me that he wants to be involved with the parrot community for longer than just GSoC, which is good to hear 22:58
cotto_work soh_cah_toa: how soon do you think you'll be able to start hacking full steam ahead?
soh_cah_toa: at that point, we should start regular contact of some kind
dukeleto: very
whiteknight turtis: ping
dukeleto cotto_work: i told him to pester you as much as he wanted. Blame me :)
soh_cah_toa cotto_work: right, after i send this email to parrot-dev. i'm going to blog about my plan/question for breakpoints 22:59
cotto_work soh_cah_toa: awesome
dukeleto soh_cah_toa: matt.might.net/articles/by-example-...ing-style/ CPS + Javascript 23:00
soh_cah_toa dukeleto: ah, that's right. thanks
dukeleto soh_cah_toa: github.com/parrot/parrot/blob/RELE...debugger.t 23:01
whiteknight cotto_work: cgaertner is currently MIA
23:01 alester left
dukeleto soh_cah_toa: ok, I have completed my homework :) 23:01
cotto_work whiteknight: ok. I was hoping that'd changed.
whiteknight nope, no change
soh_cah_toa dukeleto: yes, thanks
whiteknight cotto_work: if we don't hear from him, there's a real chance he could be dropped 23:02
so, I'm sort of hoping that doesn't happen
cotto_work whiteknight: do you know if he knows what's expected of him? 23:03
whiteknight cotto_work: I don't know anything, except that he has completely disappeared
I've sent him several emails
dukeleto whiteknight: i just sent him a github message as a hail mary
whiteknight I searched the internets for a phone number, but no dice 23:04
cotto_work whiteknight: ok. I'd hate for us to drop him for lack of contact, but it's slowly getting to that point.
dukeleto cotto_work: total lack of existence for a month does not bode well for completing GSoC successfully
whiteknight cotto_work: yeah, that's my thought exactly. This bonding period really is a time to get set up and familiarized with code. Missing this opportunity is a huge drawback 23:05
it is a bummer to think about, I was really looking forward to this project 23:07
PerlJam isn't he one of the ones with an odd schedule?
dukeleto PerlJam: what do you remember?
whiteknight PerlJam: yeah, maybe he's in that timezone that's a month and a half ahead of us, and he's still asleep 23:08
dukeleto PerlJam: regardless of him still taking classes, he could take 1 minute to email us that he is still alive
23:08 mtk left
whiteknight you want to know what real happiness is? I have a single alias that updates, builds, and installs Parrot, Plumage, and Winxed 23:09
dukeleto from my experience, he either got a job and doesn't want to tell us, or something bad happened that I would care not to theorize about.
I hope he is just ignoring us, for his sake. 23:10
cotto_work When that's the best option, we're not in a happy place.
PerlJam probably. I was just looking through my logs and such and he wasn't one of the ones with some weird schedule difference.
soh_cah_toa maybe someone could try contacting google about it. i had to enter my phone number and mailing address. maybe they can give you his or contact him for you 23:11
whiteknight already did 23:12
soh_cah_toa ouch...
cotto_work I don't think there's much more to be done until either he contacts us or GSoC coding starts and we need to say that he's disqualified. 23:14
PerlJam bummer 23:16
I wonder if there's any precedent for "slot redirection" when something like this happens so early in the game? 23:17
whiteknight nope 23:19
23:20 darbelo_ joined, darbelo left, darbelo_ is now known as darbelo
PerlJam btw, I just discovered that one of the students I hired for the summer is enough of a language geek that he might fit in around here. I'm trying to get him involved with parrot/perl6. 23:22
cotto_work PerlJam: do it now
darbelo PerlJam: You hired him. You can *force* him to get involved :) 23:23
PerlJam heh
If I could make parrot/perl6 work dovetail nicely with what we do at $work, I would force him to get involved.
:-)
But right now he's going to be working on a mobile app
darbelo "And by signing there you've just agreed to devote no less than 70% of your free time to helping Parrot." 23:24
23:24 bubaflub left
darbelo PerlJam: I'm working on mobile stuff as well. Sadly I haven't been able to sell anyone on using parrot for anything yet. 23:25
whiteknight msg tcurtis my prototype strings library is here: github.com/Whiteknight/Rosella/tre...ble/string tokenizer/Token.winxed is a very basic token implementation
aloha OK. I'll deliver the message.
23:29 lucian left 23:32 rurban_ joined 23:35 rurban left, rurban_ is now known as rurban
dalek sella: ddd9834 | Whiteknight++ | src/unstable/string/String.winxed:
use predefined cclass constants instead of magic literals
 23:36
sella: d39f45f | Whiteknight++ | src/unstable/string/ (3 files):
fix tokenization and removing debugging nonsense.
whiteknight cotto_work: what is the current state of the profiling runcore. That was your baby at one point, right? 23:37
cotto_work whiteknight: yes. Last I checked it ran fine. It has some tests in t/profiling/profiling.t but produces nonsensical output in some cases. 23:38
whiteknight okay, what exactly does it profile? Parrot Subs?
cotto_work let me dig up the case
ops
whiteknight okay, so not Subs? 23:39
cotto_work it profiles subs too
whiteknight: pmichaud++'s use case was profiling ops2c on src/dynoplibs/math.ops, so it might need reduction before a proper test case can be distilled 23:41
whiteknight that's a case where it produced nonsensical output? 23:45
I think what I'm most trying to do is figure out why what we have isn't what we need 23:48
then when I know what the difference is, we can chart a course from here to there
23:50 rurban_ joined
cotto_work whiteknight: in his message to parrot-dev on 1/31, he says that it seems unlikely that osp2c is spending 83/102 seconds in 'slurp' 23:50
if you want to try it, run ./parrot -Rprofiling ./ops2c.pbc --dynamic src/dynoplibs/math.ops; perl tools/dev/pprof2cg.pl parrot.pprof.X 23:52
23:52 rurban left, rurban_ is now known as rurban
cotto_work I also have plans to get the profiling runcore outputting kcachegrind output directly 23:53
23:57 darbelo left

Powered by Cro and the Raku® Programming Language.

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

Thank you!