Parrot 1.1.0 Released | parrot.org/ | 332 RTs left
Set by moderator on 21 April 2009.
00:09 AndyA joined
dalek rrot: r38277 | jkeenan++ | branches/install_tools/config/gen/makefiles/root.in:
Add 3 test files for install tools to make manifest_tests.
 00:33
01:01 wayland_ joined 01:06 LylePerl joined
LylePerl msg Infinoid I've got the steps and my output in this ticket rt.perl.org/rt3//Public/Bug/Display...l?id=65006 01:09
purl Message for infinoid stored.
01:11 he left 02:19 Debolaz joined 02:41 janus joined 02:53 japhb joined 02:55 eternaleye joined 03:17 contingencyplan joined 03:51 cspencer joined
cotto seenNotFound 04:01
seen NotFound
purl NotFound was last seen on #parrot 6 hours, 16 minutes and 34 seconds ago, saying: Last time was some time ago ;)
cotto NotFound, ping
04:07 Andy joined 04:09 tetragon joined
dalek rrot: r38278 | Infinoid++ | trunk/include/parrot/runcore_api.h:
[cage] Fix up a mismatched header guard comment.
 04:14
04:19 amoc joined
cotto chromatic++ #header cleanup 04:25
msg NotFound Can you take a look at trac.parrot.org/parrot/ticket/563#comment:7 ? 04:27
purl Message for notfound stored.
04:50 eternaleye joined
dalek rrot: r38279 | Infinoid++ | trunk (4 files):
[netbsd] Apply patch from he++ in TT #574.

  * Use math.h defines for PARROT_FLOATVAL_INF_POSITIVE and friends
This patch may be a starting point for further work in this direction, see doughera++'s comment on the ticket.		 05:13
05:26 flh joined
dalek rrot: r38280 | allison++ | trunk/docs/book/ch02_getting_started.pod:
[book] Some example updates, and a chapter title change.
 05:29
NotFound cotto: pong 05:33
cotto I msg'd you 05:37
It looks like the commit that kid51 mentioned was part of something that never got finished. 05:38
NotFound Yes, is unfinished. The idea is to check that the values for max and min are compilable, and to have a way to calculate appropriate values as fallback. 05:40
wayland_ Mark Lentczner? 05:41
purl?
purl yes, wayland_?
wayland_ purl, Mark Lentczner?
purl wayland_: bugger all, i dunno
cotto NotFound, ok. You should probably file a tt and put a comment in there so it doesn't get removed. 05:42
I'll add a comment to tt #563. 05:43
Thanks
NotFound I'll take a look at it later, now must go to my dayjob. 05:44
05:49 gaurav joined
Coke must go to sleep. 05:51
cotto seen particle 05:54
purl particle was last seen on #parrot 7 hours, 24 minutes and 24 seconds ago, saying: yep :D
06:08 uniejo joined
Tene hmm... I got as far as a nasty recursive loop leading t0 a segfault 06:10
nopaste "tene" at 166.70.38.237 pasted "weird recursive segfault" (128 lines) at nopaste.snit.ch/16388 06:11
cotto without looking, I wonder if it involves keys 06:15
too bad
clearly the problem is that you don't have enough ram 06:16
repro? 06:18
Tene working on migrating rakudo to a .HLL
I can push in a branch...
cotto likes playing along at home 06:19
Tene huh, 'git push' is taking a long time. 06:21
oh, there it goes.
the 'hll' branch
I'd love help. :) 06:22
cotto should 'git checkout hll' work? 06:26
in rakudo's root after a git pull
Tene, ^ 06:29
Tene cotto: try git fetch origin hll 06:31
if you didn't get the hll branch with a normal fetch 06:32
then git checkout origin/hll
you should have it already, though
git branch -r
should show you
cotto does that happen during the build? 06:35
Tene Yes. 06:36
just 'make' should fail for you.
it segfaults when trying to invoke perl6_s1.pbc 06:37
cotto there it is 06:38
06:43 allison joined
cotto wishes gdb had an easy way to view the beginning of a backtrace. 06:46
TonyC like C< bt 5 > ? or not what you're after 06:48
cotto except the opposite 06:51
hey! it works with negative numbers!
06:51 eternaleye joined
cotto TonyC++ 06:51
apparently processing 145000 stack frames takes a while 06:52
06:53 iblechbot joined 06:56 rblasch joined 07:38 AndyA joined
dalek rrot: r38281 | allison++ | trunk/docs/book/ch03_pir.pod:
[book] Beginning to rework the PIR chapter into a quick intro section

sections.		 07:52
purl sections are made of MEAT
08:02 masak joined 08:05 HG` joined 08:09 amoc joined
janus 1.1.0 isn't on the ftp in contrary to what the release announcement says 08:14
cotto istr some issues around the ftp upload 08:15
janus as long it's a known issue someone probably will take care of it :) 08:17
cotto yes, although I'm making a mental note to get that straightened before I make the 1.4 release. 08:20
allison I just sent a reminder about it 08:27
we know exactly what it is, the admins switched the FTP staging server over to ssh keys instead of passwords
eiro bye all 09:09
09:09 eiro left 09:21 GeJ joined 09:29 fperrad joined
dalek rrot: r38282 | cotto++ | trunk/src/pmc/resizablebooleanarray.pmc:
[PMC] use a properly-sized string when freezing RBA, fixing an OpenBSD segfault
 09:38
09:50 rblasch joined
bacek good evening 10:25
12:38 rg1 joined 12:39 LylePerl joined 12:40 helozjisky joined 12:53 Andy joined, GabrielVieira joined 13:09 gryphon joined
Coke wonders if the admins let anyone know about the switch before it happened. 13:42
(ftp/ssh)
moritz: ping. 13:44
moritz Coke: pong
Coke rt.perl.org/rt3//Public/Bug/Display...l?id=64276 seems to work now.
(it's also listed here: rakudo.org/status)
moritz Coke: ok, I'll make sure it's tested, and then closed 13:46
Coke apparently has a half baked ICU installed via fink, and ponders just killing fink entirely. 13:51
14:00 nicomen joined 14:06 iblechbot joined 14:09 Andy joined 14:23 helozjisky joined
particle coke: they sure didn't let me know. 14:24
fperrad: i assume you use a set of scripts to create those parrot-win32 packages... where are they committed? 14:30
fperrad particle, not exactly, see 14:32
parrotwin32.wiki.sourceforge.net/LaRecetteDuChef
pmichaud fperrad++ # just saw the binaries on sourceforge -- neat!
particle i wonder how much of that recipe can be generalized enough to help porters on other platforms 14:33
seems quite doable with some light abstraction 14:34
fperrad the recipe for language part is not up to date 14:38
I split it in 2 parts, some languages have targets 'installable', 'win32-inno-installer', but not all
particle ah, i see. 14:40
Andy whooo, GCC 4.4.0 just released gcc.gnu.org/gcc-4.4/changes.html 14:43
14:45 Theory joined
dalek tpfwiki: will@coleda.com | Parrot 14:48
tpfwiki: www.perlfoundation.org/parrot/index.cgi?parrot
tpfwiki: will@coleda.com | Parrot 14:51
tpfwiki: www.perlfoundation.org/parrot/index.cgi?parrot
fperrad particle, currently, I'm not enable to upload the tarball 1.1.0 on ftp-osl.osuosl.org
I upload it on SourceForge
sourceforge.net/project/downloading...a=10183653
shorten fperrad's url is at xrl.us/bep36g
fperrad I sent a rsa public key, I waiting for an answer.
particle fperrad: i see that, thanks for following up. allison also poked the admins via email. they're in my time zone, but probably don't get up as early as i do. i expect an answer in <2 hours, otherwise i'll poke again. 14:53
dalek tpfwiki: will@coleda.com | Parrot 14:54
tpfwiki: www.perlfoundation.org/parrot/index.cgi?parrot
Coke msg kj : can we delete www.perlfoundation.org/parrot/index..._proposals 14:55
purl Message for kj stored.
shorten Coke's url is at xrl.us/bep36z
dalek tpfwiki: will@coleda.com | Parrot 14:57
tpfwiki: www.perlfoundation.org/parrot/index.cgi?parrot
Coke msg kid51 I think we can delete www.perlfoundation.org/parrot/index...ke_reports - that's all written against smoke, which is dead. If you agree, please delete the page. 14:58
purl Message for kid51 stored.
shorten Coke's url is at xrl.us/bfeof
dalek tpfwiki: will@coleda.com | Parrot 15:00
tpfwiki: www.perlfoundation.org/parrot/index.cgi?parrot
Coke jonathan: ping 15:02
dalek tpfwiki: will@coleda.com | Parrot 15:03
tpfwiki: www.perlfoundation.org/parrot/index.cgi?parrot
Coke msg jonathan www.perlfoundation.org/parrot/index...erbs_notes is obsolete now, isn't it?
purl Message for jonathan stored.
shorten Coke's url is at xrl.us/2ors
15:05 darbelo joined
dalek tpfwiki: will@coleda.com | Parrot 15:06
tpfwiki: www.perlfoundation.org/parrot/index.cgi?parrot
particle coke++ # wiki cleanup/migration 15:07
Coke particle: since you're the only one with a picture, if you could migrate www.perlfoundation.org/parrot/index...ot_porters or delete it, that'd be helpful. 15:08
shorten Coke's url is at xrl.us/fn5dg
particle who put my photo up there? sheesh, that's a mugshot for pete's sake. 15:09
dalek tpfwiki: will@coleda.com | Parrot
tpfwiki: www.perlfoundation.org/parrot/index.cgi?parrot
particle nukes that page
Coke It's your gravatar, no doubt. =-)
particle yeah, i'll need a better one, it seems
pmichaud See also github.com/particle :-) 15:10
particle dang, what happend to my other photo...
PerlJam particle: you have a non-mugshot photo? :) 15:11
particle yes. one. somewhere....
jonathan Coke: yes
PerlJam (the one on github looks like you cropped the plaque with numbers on it :-)
wayland_ particle: I had a photo like that. It took me ages with gimp to get it right :) 15:12
dalek tpfwiki: will@coleda.com | Parrot
tpfwiki: www.perlfoundation.org/parrot/index.cgi?parrot
wayland_ Hey, don't nuke the page before I see it :)
particle aha! it's on dopplr.
jonathan -> .pm group 15:13
particle waves to beer-loving perl mongers 15:15
Infinoid the beer-hating ones are out of luck, I guess :)
particle i've got another gesture for them.
Infinoid Kinky.
particle that's how i lol. 15:16
wayland_ well, I don't get the wave, and I reject the other action too :) 15:21
particle anyone give ubuntu 9.04 a go yet? 15:24
pmichaud particle: I'm trying kubuntu 9.04 right now. 15:25
(installing it)
so far so good. I tried the beta version about two weeks ago but had some issues.
I'm hoping the release version will be clean enough that I can upgrade to it.
(and perhaps even get back to 64-bit) 15:26
when I tried 9.04-beta a couple of weeks ago, it did unlock some of my new laptop's hardware features, so I'm hopeful the non-beta will work out.
particle heh, that'd be nice 15:28
pmichaud we'll know soon enough. The biggest piece I'm missing thus far is that konsole has lost a few features in KDE 4 15:29
so I might switch to a different terminal program
rg used the 9.04 beta to debug the jit register problem
pmichaud downloads ubuntu 9.04 to perhaps give it a try. 15:30
15:31 HG` joined
nicomen has been using 9.04 for months 15:36
wayland_ waits for Eterm to support Unicode :)
Infinoid you'll be waiting for a long time, judging from how long I've been waiting for the same thing. 15:41
(or rather, how long ago I got tired of waiting and switched to something else)
wayland_ Oh, I'm using gnome-terminal 15:42
But I'm still watching so I can switch back :)
I think it basically means that Eterm has to be rewritten to use the E17 libraries, and they have to have TTF support 15:43
Then unicode should hopefully "just happen" :)
Infinoid mlterm has its own quirks, but it's been good enough for me for some time now
wayland_ I hate gnome-term because it steals the alt-key 15:44
but I haven't gotten annoyed enough to find something else yet :) 15:45
pmichaud if konsole doesn't work out for me I'm thinking of trying xfce4-terminal ... it looked reasonable at first attempt
wayland_ Mainly because I'm more worried about xorg not supporting multiple graphics cards properly any more
pmichaud I didn't like any of the other terminal programs I saw
Infinoid they all have quirks 15:46
pmichaud konsole has worked out very well for me over time, but they got rid of the ability to set the window size from the command line
so you're stuck with windows that always open to whatever size the last one you closed was 15:47
wayland_ What, of its own window, or one you're spawning?
pmichaud (same for window geometry)
for one I'm spawning
wayland_ yuk
pmichaud apparently the person in charge of konsole thinks that forcing the window to the size of the last close is a "feature" 15:48
Infinoid wayland_: edit -> keyboard shortcuts -> disable all menu access keys
that should free up your alt key.
PerlJam I like being able to switch tabs with Alt-#
Infinoid apparently the person in charge of konsole has never heard of people like me, who typically have 30+ terminal windows open, of all different sizes
I like having alt unhindered so I can switch windows in irssi with it. 15:49
pmichaud yes... on the kde bugs list, his reaction has been "why would you ever want to set geometry from the command line?"
PerlJam Infinoid: re konsole ... which is odd because I would expect people like you to be the most common 15:50
wayland_ Infinoid: Many thanks. I've looked in preferences, but stupidly ignored the other because this wasn't a shortcut in my mind :)
Infinoid wayland_: I had to google for it... but it annoyed the heck out of me too
pmichaud anyway, I'm off to lunch. bbl
wayland_ o/
PerlJam Infinoid: I just don't use tabbed windows when I'm running irssi 15:51
wayland_ I like alt because alt+b = backward-word in bash (and alt+f is forward word)
PerlJam (and I get rid of the menu)
rg infinoid: it's still sending esc-<key> instead of high-bit characters like xterm :(
PerlJam wayland_: I just use vi keybindings :)
rg you wouldn't know by chance how to change that aswell? ;) 15:52
Infinoid no, I don't know. esc-<key> is what I want it to send, the xterm stuff causes other problems for me
15:52 rdice joined
darbelo xterm can do both, actually "<ctrl>+click and select "Meta sends escape". 15:53
wayland_ I've never managed to get away from nano, so I'll stick with the Emacs keybindings for bash :)
Infinoid So many standards to choose from :) 15:54
darbelo I've found xterm to be "an improvement on most of it's succesors" 15:55
rg darbelo: i've dropped xterm because it doesn't have tabs or multiple windows. 15:56
wayland_ rg: by that, do you mean multiple windows in the one process? 15:57
darbelo I use multple xterms :)
rg yes, in one process of course.
Infinoid is used to spawning multiple terminal processes too
tabs look nice for organization... but unfortunately my common usage is too random-access for those things
darbelo Also, you can't look at two tabs at the same time. 16:00
Infinoid usually just has 4 terminal windows per virtual desktop 16:01
16:03 iblechbot joined
wayland_ I wonder what a RTL terminal would be like :) 16:03
Infinoid that would be beautiful, but probably murder to set up 16:05
wayland_ Not that I can read more than a few words of any RTL language anyway... :) 16:07
Infinoid hmm, my patch stack is empty. Time to find something new to break.
wayland_ I've got one per virtual desktop, but my penance is 800x600 until I or someone fix the xorg multi-video card problem 16:11
Infinoid: Does that mean you want people to send you more patches? 16:12
16:15 knewt joined
Infinoid by all means, more patches never hurt :) 16:15
wayland_ Well, except it's not ready yet :) 16:18
It tries to make an RPM of Parrot that Rakudo will happily build on top of 16:19
But it requires that the spec file run "make install-dev" instead of "make install"
I suspect that that's going to make people unhappy, even though Rakudo won't go without it :)
16:19 whoppix joined
wayland_ And so I'd like to at least wait until Rakudo builds on top of it before trying to get people to accept it :) 16:20
moritz wayland_: on most linux distributions you want to biuld a separate -dev package
Infinoid I'm not sure what the current thinking on that is... perhaps the "install" target is for running previously built HLLs and apps, and the "install-dev" target is for building HLLs and apps against it 16:21
that being the case, can you build a parrot-dev with the extra bits?
wayland_ moritz: Oh, it does that too
moritz so just require the parrot-dev package for rakudo?
Infinoid or parrot-devel, I guess, since this is rpm
16:21 flh joined
moritz ah right 16:21
moritz is too much into Debian
wayland_ (yes, devel instead of dev, but I know what you mean :) )
I was under the impression that install-dev was only for if you were developing parrot, but maybe I'm wrong 16:22
moritz no, also if you are developing languages
purl okay, moritz.
wayland_ no? 16:23
Infinoid hmm, this looks relevant
Building a language depends on a series of Parrot build tools,
installed in /usr/lib/parrot/<version/tools. These tools will generally
not be included in the default "parrot" package on most systems, but
will require a "parrot-dev" package to be installed before they can be
moritz if you want to develop parrot, you use svn anway ;-)
Infinoid built.
-- pdd30_install draft
wayland_ (sorry, trying to get a response out of purl)
moritz ah ;-)
purl: no, also if you?
purl bugger all, i dunno, moritz
wayland_ Ok, I'll send in my patch then
16:26 GabrielVieira left
wayland_ Nah, can't find it at the moment. There's one other patch that I'm waiting on (kid51 has requested comments on it), and then I'll be able to get back to looking at this properly 16:26
dalek rrot: r38283 | Infinoid++ | trunk/docs/pdds/draft/pdd30_install.pod:
[pdds] Minor clarification to PDD30 draft. I think it means *languages* will require the parrot-dev package, but right now it reads like the parrot-dev tools themselves require parrot-dev to be built.
 16:32
16:45 bsdz joined
bsdz msg pmichaud i've a trac with a patch for pct to add complex number support at trac.parrot.org/parrot/ticket/466 . when i first filed it, it was a bug but does have a patch now. can you have a look? 16:48
purl Message for pmichaud stored.
16:49 jhorwitz joined
Infinoid Hmm. My /usr/lib/libpcre.so is a linker script that aliases to /lib64/libpcre.so.0. For whatever reason, Parrot's pcre test can't open it, and the test fails. 16:59
particle that shouldn't stop you from debugging it :) 17:08
is the .c file around, and .cco? 17:09
...for that configure test.
Infinoid it's a testsuite test, t/library/pcre.t, not a configure test 17:10
to debug it, I need to figure out where we're calling dlopen() or whatever. Working on that part.
17:11 contingencyplan joined
particle oh 17:13
wayland_ I had a patch that shows you the path that it searched when trying to load a library
But it hasn't been committed yet
It's in Trac, and it works for me
Do you need it?
Infinoid nah, I know what it tried to open, from the strace output 17:15
My issue is that dlopen() doesn't seem to be able to handle GNU ld scripts
it just opens it, says "hey, this isn't elf" and barfs
wayland_ Racist! It needs to accept Dwarves too! :) 17:16
Coke moritz: ping
moritz Coke: pong
Coke: I've updated the rakudo status page 17:17
Coke rt.perl.org/rt3/Ticket/Display.html?id=58646 looks closable.
(presuming the patch was complete. =-)
hurm. only about half a dozen more remaining. 17:18
61744 seems old. 17:21
jhorwitz Infinoid: yeah, unfortunately linker scripts are only used by the linker. dlopen is expected to know what it's opening. we may need to address that when building NCI extensions. :-P
Infinoid an NCI extension is exactly what this is 17:26
I've created TT #578 for that issue. Would love some suggestions, if anyone has some. 17:27
I can think of a couple of possibilities, neither are very appetising 17:29
We could try to parse the linker script ourselves, which sounds like a neverending bug spigot. Or try libpcre.so.0 before libpcre.so and try /lib before /usr/lib on linux, which will require special-casing for all the other extensions, too, and will need updating again for the next time the distro boys play ugly games with script locations.
jhorwitz we will probably need to use the linker itself to do our dirty work 17:31
Infinoid hmm. You mean, like, build/link a stub .so file that dynlinks to the real library, and then dlopen() that?
I suppose it would fix the problem, but I wish there was a simpler way 17:32
17:32 ruoso joined
jhorwitz i'm thinking more along the lines of using the linker to find the path ahead of time and somehow inserting that into the dlopen....missing some details of course... :) 17:34
Infinoid that would work.
jhorwitz something that would happen during a "make" phase
Infinoid wonders if there's some parse_ld_scripts() function in -ldl 17:35
of course, resolving it at make time doesn't help packaged parrots
jhorwitz very true
Infinoid hmm. if there's any platform specific tools for manipulating ld scripts, they're probably in libbfd 17:39
what a gloriously nasty bit of code that library is 17:40
17:59 AndyA joined 18:06 rblasch joined 18:15 rakudohudson joined
dalek rrot: r38284 | rblasch++ | trunk/config/auto/format.pm:
[config] Fixed transposed "long long int" ivmin/ivmax.
 18:15
Tene jhorwitz: if I'm having mod_parrot issues tonight, think you'll be available for me to harass? 18:16
jhorwitz Tene: i should be around tonight after 8 18:17
Tene jhorwitz: thanks. :)
jhorwitz (EDT)
btw, i fixed some issues yesterday, bringing it up to speed w/ parrot trunk.
Tene ah, so I might not have problems. nice.
18:29 ilia joined 18:36 contingencyplan joined
Coke ponders a talk. 18:37
Coke doubts it. 18:38
18:39 jan joined
Coke finds www.tutorialspoint.com/parrot/ 18:40
... the f?
PerlJam Was the talk "crazy parrot stuff I found on the internet"? 18:41
Tene Parrot - PERL Virtual Machine 18:42
Infinoid and they use old snapshots from perl.org. and they explain things in pasm, not pir, and never actually get around to anything perlish 18:44
NotFound Looks like mk_language_shell.pl needs both a build tree and an installed parrot to work :? 18:55
Can't open perl script "/usr/local/lib/parrot/1.1.0-devel/tools/dev/gen_makefile.pl" 18:56
wayland_ Did you do "make install", or "make install-dev"? :) 18:57
NotFound None of then, that is the point 18:58
wayland_ Oh, ok
wayland_ keeps wearing his p6 shoes, and being glad that someone gets Parrot :)
NotFound And on ecmascript build out of a parrot tree with an installed parrot make test fails because it does not find modules. So it looks like we don't have a clear plan about paths 19:00
wayland_ NotFound: That's probably true. I've been working on building an RPM of Parrot that Rakudo can happily live on, and had to fiddle with the paths 19:03
Quite a bit
Coke ecmascript might be broken WRT pathing. 19:04
(I know many languages were patched for that pre-1.0)
NotFound I was trying to build a new language to check what it does, but given that it does not work... 19:05
my $build_tool = $config{libdir} . $config{versiondir} . '/tools/dev/gen_makefile.pl'; 19:09
No wonder it does not work without an installed parrot
$ ../../parrot_config libdir 19:11
/usr/local/lib
darbelo cotto, ping 19:12
19:29 megazooch joined 19:34 barney joined 19:37 megazooch joined 19:42 gryphon joined 19:58 rblasch1 joined
fperrad particle, tarball uploaded 20:07
particle \\o/
Infinoid fperrad: I've updated the URL redirects for release/current and release/developer 20:08
fperrad++
fperrad 2 days late 20:09
Infinoid not your fault :)
particle not our fault, though. could you send an email off?
20:09 rblasch joined
Infinoid is looking at his post to parrot-dev, Subject: Parrot 1.1.0 "Half-moon Conure" Released! 20:09
Coke particle: I had someone rant to me that http downloads would be nice. 20:10
Infinoid What do they think this is, the 21st century? :) 20:11
20:11 megazooch joined
fperrad Github avoids this kind of problem (see Rakudo) 20:14
particle we can get http downloads. i'll enter a support ticket. 20:16
Infinoid nice, automatic snapshots 20:17
20:20 megazooch joined
Coke IWBNI if we had a trac plugin that did that, yes. 20:23
but our tarballs aren't quite a snapshot of the repo. 20:24
Infinoid how about nightly snapshots (fsvo "night") which are only posted if they receive smolder's blessing? 20:25
particle how's the weather there in the future?
Infinoid it's gonna snow again. 20:26
wayland_ LOL (but not very loud :) ) 20:30
allison_afk Coke: actually, we already have http downloads via ftp.osuosl.org/pub/parrot/ 20:38
allison ftp.parrot.org is just an alias 20:39
they don't have the ftp.parrot.org alias set up, but wouldn't be difficult
Coke allison: the http loads redirect to an ftp url.
(and apparently we don't link to the actual http downloads from the downloads page.) 20:40
cotto darbelo, pong
allison Coke: for me it stays ftp.osuosl.org/pub/parrot/releases/...1.0.tar.gz all the way down
shorten allison's url is at xrl.us/bep53b
Coke If it's just a matter of changing the downloads page, awesome.
Infinoid I think you can just change the url redirects 20:41
Coke hurm.
allison Coke: oh, you mean our redirects on www.parrot.org?
our redirects do go to the ftp page
Coke www.parrot.org/release/supported seems to go .... right. 20:42
allison if people prefer the http link we can put that in instead
particle allison, i just mailed support about it
Coke both is fine, as long as they're labeled. 20:43
20:43 bsdz joined
darbelo cotto: got my last mail? 20:48
cotto darbelo, yup
You'll have a shiny new branch in the very near future.
particle logs into melange while darbelo and cotto are nearby
20:49 fperrad left
cotto Also, I haven't seen any packaged binary libraries of decnumber, but you're right that we should use them if they exist. 20:49
particle cotto: i thought there might be a way to set up an ml in melange for gsoc projects, but i don't see one 20:50
i'll talk to dukeleto about getting that done 20:51
cotto Thanks.
Coke particle: is the plan for gsoc access to give them access and have them work in a branch?
darbelo cotto: That's kind of the problem, the 'release' they ship is a .zip with the .h and .c files.
particle coke: yes 20:52
cotto darbelo, if you can find some distro that packages a library, then we should do library detection and fall back on the included code. Otherwise, we can just use the included code.
darbelo If we include the code, why bother with the installed one? 20:53
particle first thing i'd like to see is the decnumber tests imported to parrot, and a harness written to run them against current parrot 20:54
*to parrot repo
Coke particle: I would tend to prefer the approach partcl and rakudo take with spectests. 20:56
but whatever's easiest wins for now.
particle but they're not stored in a repo anywhere
cotto darbelo, to use what a user has installed in case they know something we don't.
particle *public repo 20:57
cotto I suspect the whole question is moot and that you won't find any distros that package decnumber separately.
particle we certainly can ask ibm, though.
cotto particle, there's no obstacle to including the files in Parrot as far as Parrot's licensing, right? (decnumber is ICU-licensed) 21:00
particle i believe that is correct. 21:02
allison's more of an expert than am i, but that's what my previous research determined.
cotto darbelo, you have your branch 21:03
particle has darbelo submitted a cla?
darbelo Not yet. I'll scan it today.
particle fab.
dalek rrot: r38285 | cotto++ | branches/gsoc09_decnumber:
creating a branch for darbelo++ to work on decnumber-based PMCs for his GSoC 2009 project
 21:04
particle ftp.parrot.org/releases/devel/1.1.0...1.0.tar.gz
looks like all we need to do is change the links on the website
allison cotto: we can't include ICU licensed files in Parrot, that's why we don't ship Parrot with ICU itself included 21:10
particle crap, then i was wrong.
cotto ditto
particle so, we must package independently? 21:11
...or rely on existing packages?
cotto ...which I haven't been able to find so far.
Infinoid if we make it obvious enough in the documentation, the distro guys should pick it up
cotto Unfortunately, decnumber doesn't even come with a makefile. It's just a bunch of .c and .h files with some examples. 21:12
allison either rely on existing packages, or package separately, or provide instructions on downloading and unpacking the source into the parrot tree at a specified location where the Parrot compilation process knows to look for it
but, we have to have a sane fall-back strategy (parrot has to work fine without it, though fine to disable some features) 21:13
darbelo I'm liking the last option. Less work :)
I was going to follow the current Big* PMCs behaviour for that: 21:14
allison darbelo: well, existing packages are the least work, and generally preferred, but sounds like not possible in this case
particle allison: know any open-source friendly folks at ibm willing to package that up? 21:19
21:20 flh joined
allison well, to do us any good, they'd have to package it for a dozen different operating systems and architectures. Seems unlikely. 21:20
the "unpack this into X directory" option seems like the best we've got ATM 21:21
21:22 Whiteknight joined
bsdz just listening in. if it's icu there are a load of binaries at icu-project.org icu-project.org/download/4.0.html#ICU4C 21:22
moritz Infinoid: it seems that dalek stopped to report in #perl6
cotto or make download_and_unpack_decnumber_to_directory_x ?
darbelo I think I could trhow together a OpenBSD port for it, and maybe a .deb package, if I put some effort into it. But that leaves out a whole lotta platforms. 21:23
Coke do we have to compile it ourselves?
cotto bsdz, decnumber is ICU-licensed, but isn't part of ICU.
Coke if so, then it seems reasonable to put the .zip file somewhere and just download it as part of the build.
cotto Coke, yes and yes
Infinoid moritz: when? 21:24
bsdz cotto@ ah okay. decnumber is a math library. wonder if there's an alternative at netlib
Coke (on parrot.org, or perhaps in a top level svn dir, e.g. "resources/", (next to branches/ etc.)
moritz Infinoid: my /lastlog dalek doesn't show any occurence...
it did report something yesterday, though 21:25
allison Coke: that requires a network connection for the build, which we can't do
Infinoid looks like the last rakudo updates from dalek in here was about 30 hours ago
pmichaud there have been lots of rakudo updates since then. :-)
Infinoid when in doubt, SIGTERM. 21:26
21:26 dalek joined
Infinoid now, commit something :) 21:27
particle SIGDOUBT.
Infinoid the atom feed looks fine
particle mmm. hungry.
cotto allison, what would you recommend for getting decnumber into parrot? Could there be a make or configure target/option, or does the download have to be manual? 21:30
I really want to avoid requiring a manual download. 21:31
allison cotto: I don't see any way around it
cotto: it's an optional download 21:32
so not too painful
cotto allison, as in optional manual download or optional make target? 21:33
allison cotto: both
darbelo "Until your distro starts packaging it uncompress decnumber into directory X if you want big numbers" 21:34
allison what would be really ideal is to make the whole thing a separate parrot extension
darbelo: yes, though I don't know that any distros will package it. It's not very general purpose
darbelo: it's also been around a *long* time, so you'd expect them to package it by now if they were going to 21:35
ingy hi allison
allison hi ingy 21:36
ingy ltns
21:36 cognominal joined
Whiteknight that's what we do with other libraries, so it seems reasonable to do it for decnumber too 21:36
allison Whiteknight: but so far, we always keep them in the Parrot repository.
darbelo allison: They never really had anything that depended on it, so no pressure to package. 21:37
allison like postgres
Whiteknight do we keep ICU in the repo?
allison Whiteknight: no, but it is packaged by most distros
purl okay, allison.
bsdz i'm confused, why can't you distribute a package with an icu license? 21:38
allison ingy: indeed, where are you?
bsdz: we can't include it in Parrot
bsdz: that is, we can't distribute it under the Artistic License 2.0
ingy allison: chamonix france w/ nothingmuch. been here since feb 21:39
bsdz allison: even if we follow what they say and provide copyright info?
allison bsdz: we can tell people to download it
ingy allison: you be at yapc?
Whiteknight darbelo and cotto: do either of you know where the best place to get decnumber for Ubuntu is?
allison bsdz: but then part of parrot is under a completely different license, and we're holding the users to two licenses
Whiteknight yeah, allison is right about that. Can't mix licenses. I've run into that problem in the past with other projects and it never turns out well 21:40
bsdz allison: ah okay. it's to maintain a single license. would this also cause problems with icu and libjit (lgpl)
cotto Whiteknight, the issue is that most distros (including Ubuntu) don't package it. You'll need to download the zip file from speleotrove.com/decimal/ .
darbelo Whiteknight: speleotrove.com/decimal/#decNumber is 'where to get decNumber' there is no 'for Ubuntu' to my knowledge.
allison bsdz: yes, icu and libjit have the same problem
bsdz: even LLVM, for that matter 21:41
bsdz: we can have people download and install the packages as prerequisites, but can't include them in Parrot
bsdz allison: got you. thanks. btw how does that work with fperrads win32 distro? we okay there? 21:42
Whiteknight cotto and darbelo: Thanks! I guess I need to have that downloaded if I am going to backup mentor this projects
21:42 donaldh joined
Whiteknight bsdz: It should work with the win32 distro. You can get around mixing licenses by creating an aggregate and clearly marking which parts have which license 21:42
bsdz Whiteknight: cool. thanks :) 21:43
21:43 rob joined
Whiteknight The problem is that we say the Parrot repo is under the Artistic 2.0 license, so we can't put things under other licenses into it 21:44
but we can create an aggregate distribution that contains clearly distinct things
Infinoid (and the parrot tarballs)
bsdz ah of course. it's tied to the repo
cotto Whiteknight, I image that an early part of the project will involve adding a make target that downloads decnumber so that it's easier for other people to play with.
Whiteknight cotto: I wonder how hard it would be to try to get some distros to start packaging it? 21:45
particle allison: non-parrot.org packagers can include them freely, though
Whiteknight particle: even parrot.org packagers should be able to do it, so long as we kept it out of the repo and were clear about the individual licenses used 21:46
probably more complication then we would like, but possible
allison cotto: only if it's optional, we can't depend on having a network connection
cotto: and for that matter, it has to be possible to build even with decnumber without a network connection
particle: ideally, they would be packaged separately 21:47
particle yeah, well, that just sucks. 21:48
allison Whiteknight: an aggregate distribution is just that, an aggregate.
Whiteknight allison: yes, I'm simplifying
allison Whiteknight: most open source licenses (particularly GPL, which is the most restrictive), have some language about "aggregation" 21:49
cotto allison, my thinking is that make or configure would look in a certain place for the decNumber-icu-361.zip (or whatever) and use it if it's there. There could also be a make target that downloads it to where make expects it to be to make life easier for users.
allison it basically comes down to a question of whether something is a "derived work"
Whiteknight if we include libdecnumber in a package and don't modify it's internals, we're using it as a simple aggregate and not as a derivative 21:50
it's the same way that e.g. Wikipedia is GFDL but can include CC-BY-SA images without conflict
allison I'm not sure how distributions like Debian feel about multiple licenses in a single package
in general they want you to be *very* clear about it
gravity It depends on the package 21:51
moritz and it depends on linking
bacek good morning people of past
gravity Yes, that's basically it. It needs to be clearly worked out.
Whiteknight yeah, that's a different question entirely. I'm not saying we should pursue maintaining our own complicated packages, but it is *possible* in the most general sense
moritz static linking between different licenses can cause much trouble
allison so, if we compiled decnumber into Parrot for the Parrot package, we'd have to explicitly say "X directory is under the ICU license"
which is effectively just as bad as including it in the Parrot tarball 21:52
Infinoid bacek: good afternoon, future person
bacek Infinoid: :)
particle i'm not from past, i'm from pir.
21:52 bobke joined
cotto allison, what about compiling decnumber to a shared library? 21:52
Infinoid is from post
allison that's why I was suggesting making the whole Parrot library a separate package, instead of including it in the Parrot tarball or packages
bsdz strawman: maintain a separate googlecode site for each license and compiled packages as prereqs for parrot 21:53
allison we could even license that extension under ICU, so there's no difference
particle yes, we could even package icu and decnumber together
allison particle: icu is already packaged, so no need
bacek from parse. Trying to parse PMC with PCT.. 21:54
allison bsdz: doesn't gain us much, since they're already available for download
cotto: I mean make the whole SOC project a separate package
cotto: A dynamic PMC 21:55
21:55 nopaste joined
particle allison: soc as separate package seems cleanest to me 21:55
allison cotto: install runs separately, so no licensing difficulty
particle keep license icu instead of artistic2
allison cotto: it can have its own package
bsdz allison: yes, true - guess was thinking more about compiled versions. but tbh i like it when my build script just downloads the right package or uses one locally. much like CPAN shell does
allison cotto: and hey, we'd be providing easy access to the decnumber library, which apparently no one else has done 21:56
CPAN is a packaging system
cotto: Oh, and if the dynamic PMC is ICU licensed, then decnumber can just be checked into its repository 21:58
22:02 donaldh left
cotto Hmmm. I was hoping that it could be included with Parrot somehow, but distributing it as a dynpmc would solve a number of problems. 22:04
darbelo It also has to be developed in a separate repo, right? 22:06
cotto but it'd also mean that distributors would be less likely to build packages for it
allison darbello: yes
cotto: well, I'll do the debian and ubuntu packages
particle it can be developed in a parrot branch, but never merged to trunk 22:07
allison we need to start packaging separate modules anyway
moritz particle: then I don't see the use for a branch
particle but probably better to start in a separate repo from the beginning
moritz: me neither
allison particle: aye, separate from the start is better
cotto yes, so it looks like I jumped the gun in creating a branch 22:08
particle that's okay, branches are cheap
allison particle: the temptation is to make it depend too closely on the parrot build directory (as we learned from languages)
particle yes indeed.
let's get a permissive repo set up somewhere, so folks can help with the packaging effort darbelo requires but didn't know he'd have to have 22:09
wayland_ Do you want me to send in my patches for RPM specfiling?
allison wayland: I thought you already did? 22:10
particle patches++
wayland_ Oh, wait, or are we doing libparrot as a separate thing?
The Mandriva specfile does that already, I think
No, I'll have to leave the specfile patch until I'm more awake, I think 22:11
Anyway, I'm not sure it's something you need for the current project
allison wayland: it will need a specfile (one for just that library) 22:12
wayland_ What does libdecnumber do for us?
Well, like I said, the Mandriva one is already split up into sections, and has one for libparrot
darbelo wayland_: decimal arithmetic on big numbers.
22:12 Fayland_logger joined
wayland_ ok, thanks 22:12
allison wayland: it's a separate package, not part of the parrot package 22:13
bsdz is it big numbers and/or no loss of precision?
cotto bsdz, it's bcd-like with arbitrary precision 22:14
bsdz ah cool. i like that. sounds like python's decimal package
cotto as long as you want less than a billion digits 22:15
moritz I don't have that much money anway ;-)
bsdz no one seems to nowadays.. 22:16
darbelo moritz: "a billion digits", he didn't say on which sise of the decimal dot. :)
1.2 * 10^-999999999 22:17
purl 18446744072709551629
bsdz is that a small paltry billion or a decently sized british billion..
hold on that's bit more 22:18
wayland_ Ah, so that was only a milliard, then :)
cotto darbelo, are you ok with the dynpmc approach or do you want to try to figure something else out? 22:21
Whiteknight i cant even figure out how to get libdecnumber built on my system. Good luck darbelo! 22:23
darbelo I haven't looked at dynpmcs in depth yet, but they sound better that "packaging denumber for a gazzilion platforms", so I'm OK with it. 22:27
cotto darbelo, ok. I'll kill the branch. I wonder if the repo for the dynpmcs could be hosted on parrot.org 22:28
particle, ^?
allison cotto: it's pretty easy to set one up 22:29
though, given the nature of the project, would google code make sense?
allison settles down in a big empy apartment to take a look at partcl 22:30
dalek rrot: r38286 | cotto++ | branches/gsoc09_decnumber:
the decnumber PMCs will be implemented as out-of-tree dynpmcs, so a branch here is unnecessary
 22:32
cotto Is the New BSD license compatible with the ICU license? That looks like the closest that Google Code offers. 22:41
darbelo Do the have 'Other' as an option? 22:42
cotto nope 22:43
darbelo It looks motre like the MIT License to me. Let me check something... 22:46
Infinoid it's a modified MIT license, plus the ncurses advertising clause 22:48
darbelo Then it's MIT, I guess. 22:50
22:50 Limbic_Region joined
allison either BSD or MIT will do 22:54
(they're largely equivalent)
BSD has the advertising clause, which the stock MIT doesn't, so slightly closer
22:55 zostay joined
allison (but they're both so bare it's not much difference) 22:55
cotto Do we have an equivalent of mk_language_shell.pl (or create_language.pl) for dynpmcs, or should we just use that tool and strip out the language stuff? 22:59
darbelo src/dynpmc/README.pod claims tools/dev/gen_class.pl is what we want. 23:01
23:06 Ademan joined
particle i believe that's the new hotness for dynpmcs, yes. 23:13
cotto It looks like it's meant for general PMCs. 23:14
The readme also assumes that the dynpmc will be in-tree 23:15
Infinoid hmm, apart from an extra tag on the pmclass line, I don't see much difference between dynpmc and pmc source files. 23:16
cotto I might have to hack on something to make out-of-tree dynpmcs easier to start, once darbelo and I figure out what's missing.
Infinoid, not much, other than not having a static id number. The trick is that they also need their own makefile, tests and possibly Configure.pl. 23:18
Infinoid ah
cotto actually, they can just use parrot_config 23:19
parrot_config++
I'd still like parrot to have make uninstall. allison, would a patch that adds that be ok to apply? 23:22
allison cotto: mk_language_shell.pl will also do the framework for a dynpmc (though it's intended for language dynpmcs, it's a start) 23:23
cotto: uninstall is really not a safe thing for a makefile to be doing, packaging tools are good at it, so we'll leave it to them 23:24
cotto allison, my reasoning is that it's nice to make install from svn, but I also want it to be reversible.
Infinoid if cotto really really wants one, perhaps "uninstall" should print a big fat warning and tell you to really_uninstall if you really want to break your system.
allison cotto: if you're doing testing and want to make sure you catch all files, just install into a temporary directory, like /home/me/temporaryparrot 23:25
cotto allison, what's potentially unsafe?
that'd work
allison deleting files with root access 23:26
really bad news, makes sysadmins cringe
Infinoid I've had a couple of uninstalls run away on me. It's a bad scene.
cotto I can buy that
darbelo tangent: Since I'll be working out of the parrot repo, do you still need the CLA?
cotto darbelo, it wouldn't hurt, since you may still want to contribute patches to Parrot. 23:27
allison darbelo: if you want to contribute the code to the Parrot foundation, then you can send it in, but not required
cotto but it won't be necessary for gsoc, since you can submit bugs and patches through other channels 23:28
darbelo ok, then should the date in page 3 be in dd/mm/aaaa or mm/dd/aaaa format?
allison darballo: I tend to go for the unambiguous 23 Apr 09 format 23:29
darbelo: or something like that
Infinoid iso8601++ 23:30
allison darbalo: but it doesn't matter 23:31
wayland_ Is ISO8601 YYYY/MM/DD ? 23:32
cotto YYY/D/MMM
Infinoid ISO8601 -s YYYY-MM-DD
is, too
wayland_ ISO8601++ :)
cotto: Stop being evil :) 23:33
Infinoid You could probably just fill in "1240529616"
cotto it's unambitouos 23:34
wayland_ What, Unix timestamp?
cotto ly a bad idea
s/ouo/uou/
Infinoid yeah, unix timestamps are a great way to piss people off. What other value they have, I dunno.
cotto s/bit/big/
wayland_ They save a few CPU cycles (like we can't afford them :) ) 23:35
Infinoid they take a lot more CPU cycles for humans to read
wayland_ Well, yes. I meant they save some but should be converted for external use :)
Although with Unix timestamps, conversion may involve repentance :) 23:36
Infinoid On the other hand, they're also timezone-independent 23:37
wayland_ That's ++ 23:38
Infinoid Ok, time for me to stop suggesting incredibly bad ideas and get some work done. Back later :)
wayland_ breakfast. AFK! :)
23:45 TonyC joined
cotto darbelo, are you starting a project on Google Code? 23:54
darbelo Nope, I thought you were. 23:55
cotto Since it's your project, you should probably start it. ;) 23:56
darbelo ok. Does parrot-decnumber-dynpmcs sound like a good name for it? 23:59

Powered by Cro and the Raku® Programming Language.

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

Thank you!