Parrot 1.0 Released | parrot.org | Thanks to everyone who helped get 1.0 out the door!
Set by moderator on 18 March 2009.
rg well test #13 isn't failing. it's sinh and tanh that are causing problems under specific circumstances 00:00
nopaste "rg" at 93.104.42.222 pasted "FreeBSD jit failure" (66 lines) at nopaste.snit.ch/15962 00:03
rg if you want to keep TT#38 specific to the atan failure, we could also reopen RT#36086. 00:05
nopaste "darbelo" at 201.252.61.125 pasted "OpenBSD jit failure" (68 lines) at nopaste.snit.ch/15963
rg that looks quite identical to what i'm seeing, so it could be a *bsd problem. 00:06
kid51 We don't want to open or re-open RT tickets.
rg we also don't want to migrate tickets. there's quite a lot of information in that rt already. something's got to give. 00:08
kid51 rg: Could I ask you to open a new TT with your paste output, specifying that the current problems are with sinh, tanh and pow under jit? 00:09
chromatic What's wrong with the existing sinh/arc TT?
kid51 chromatic: Didn't know about that? Which TT? 00:10
rg there's yet another one?
00:10 Whiteknight joined
chromatic Sure, that's why there's a TODO in one of those tests for JIT and some of those math opcodes. 00:11
atan2, for example
TT #38
rg c: that's the one we're discussing and kid51 wants to keep this specific to atan2 00:12
kid51 Well, as I read TT#38, it was focused on atan2, which now appears to be passing.
Am I misreading it?
chromatic I believe that's the one marked as unexpectedly succeeding.
kid51 Yes.
cotto asked if TT38 could be closed, as test #13 was now passing. 00:13
chromatic Hm. I guess these are different enough for a separate ticket.
kid51 I said, "Let's get confirmation on other platforms."
chromatic If the atan2 test succeeds on OpenBSD but these don't, that's different.
rg ok then i'll open a new ticket
kid51 Until this discussion, I wasn't aware of problems with *other* tests in this file under jit.
chromatic For some reason I thought these failures included atan2. 00:14
00:14 davidfetter joined
kid51 See post here of :51 last hour. 00:15
I don't claim to understand the deep issues here; I'm just trying to manage the tickets.
rg i was hoping someone could help me debug the cause, but right now i don't really have the time to dig in. 00:16
darbelo There is one caveat that i forgot here. I'm running a 4.5-beta which hasn't been released yet. 00:20
acording to the changelog it has "Vastly improved C99 support in libm" 00:21
rg darbelo: since i have the same problem, it's probably parrot ;) 00:22
also, you can try and run the same command without -j and it should pass. 00:23
darbelo Yeah, well I dont hava a 4.4 or earlier release to test, so i can't be sure tha it isn't a OpenBSD issue that got fixed.
janus darbelo: good stuff, i didn't find the time yet to investigate which math related things can be fixed now 00:32
Coke_afk quit 00:41
00:48 darbelo left 00:49 allison joined 01:09 eternaleye joined
kid51 Alright, I just closed TT 38. 01:24
If you want to report errors in t/op/trans.t under JIT for other math functions, please do so in function-specific tickets.
E.g., a ticket focused on 'sinh failing under JIT'. 01:25
rg what's different between (sinh|tanh|pow) and e.g. atan2?
i really don't want to open 3 tickets for the same problem 01:26
kid51 They're failing on >1 platform; atan2 no longer is.
Will there be one solution to all 3 failures? If so, then one TT will suffice.
rg i guess if i could tell i could fix it
the symptoms are the same, though 01:27
kid51 But if you think they will have discrete solutions, then I recommend individual tickets.
But, hey, I'm not the only one with a commit bit here.
rg i much rather write tickets with a patch ;P 01:28
uh. sinh fails, cosh works?! 01:36
dalek rrot: r37726 | jkeenan++ | trunk/MANIFEST:
Update MANIFEST to include files cited by Andy Dougherty in TT #496, plus one
 01:39
01:58 Theory joined
dalek rrot: r37727 | jkeenan++ | trunk (2 files):
Try to bring into conformance with coding standards; errors cited by Coke in TT 496.
 01:59
rrot: r37728 | jkeenan++ | trunk/src/io (2 files):
Make file conform to C macro args coding standard.
 02:02
rrot: r37729 | jkeenan++ | trunk/src/pmc/socket.pmc:
C++-style comments are forbidden.
 02:05
Tene pmichaud: ping 02:09
rg kid51: created TT#501 02:14
Tene pmichaud: nm
dalek rrot: r37730 | jkeenan++ | trunk/src/pmc/socket.pmc:
Correct error in C coda.
 02:15
rg can someone put a comment in RT#36086 without reopening it? 02:16
02:17 eternaleye joined
dalek rrot: r37731 | jkeenan++ | trunk/src/io/socket_win32.c:
Make file conform to line length coding standard.
 02:19
kid51 Damn. Making a file pass the coding test causes 'make' to fail. 02:20
dalek rrot: r37732 | jkeenan++ | trunk/src/pmc/sockaddr.pmc:
Make file conform to C parentheses coding standards.
 02:22
kid51 rg: Yes, I believe you can do that. Just check that status is 'resolved' when you're done. 02:25
Thanks.
dalek rrot: r37733 | jkeenan++ | trunk/src/io/socket_unix.c:
Reverting last commit, as solution to coding standard problem caused build to break.
rrot: r37734 | jkeenan++ | trunk/src/io/socket_win32.c:
Reverting a commit done for coding standard that broke build in similar file.
 02:29
nopaste "bacek" at 211.29.157.151 pasted "Proper socket macro fix for kid51" (106 lines) at nopaste.snit.ch/15964
02:32 msmatsko joined
kid51 bacek: thanks. How would I apply a patch created via git? 02:34
bacek_ patch -p1 < foo.patch 02:35
02:35 darbelo joined
rg kid51: since i don't think i could resolve the ticket, i don't think i can keep it resolved. 02:35
or rejected in that case 02:36
kid51 Well, Andy was correct to say let's move the discussion to TT, so 'resolved' in this case doesn't mean 'solved'.
Or am I completely misunderstanding you?
bacek_ Looks like that worked. Thanks. 02:37
rg i think you are. my point is that rt will reopen the ticket if i comment on it and i don't have permission to change the status, be it resolved or rejected.
kid51 repeats make test to make sure nothing else broke
rg (andy chose rejected)
kid51 looks at RT 36086
Enter your comment. I'll deal with the status (though perhaps in the morning). 02:38
Thanks.
rg ok
kid51 (A ticket that was opened in June 2005 -- and still kickin'!) 02:40
rg hmm looks like it actually stayed on rejected. 02:43
dalek rrot: r37735 | jkeenan++ | trunk/src/io (2 files):
Applying patch pasted by bacek++ to bring file into conformance with coding standard without breaking build.
 02:45
kid51 rg: So, we're cool. 02:46
rg yes, thank you. 02:47
kid51 404 unresolved RTs -- let's see if we can get below 400 within 24 hours! 02:48
dalek rrot: r37736 | jkeenan++ | trunk/t/op/trans.t:
Point the todo message to TT 501.
 02:49
rg you're todo-ing the wrong test 02:50
and i believe it only affects *bsd
kid51 Can you provide patch? 02:52
rg hmm let me see if i can find an example for a platform specific todo 02:53
btw RT #31170 looks closeable (just a token ticket anyway and implemented today if i was following the discussion correctly) 02:54
nopaste "RG" at 93.104.42.222 pasted "t/op/trans.t patch" (54 lines) at nopaste.snit.ch/15965 03:01
03:01 Andy joined
dalek rrot: r37737 | allison++ | trunk/compilers/pct/src/PAST/Compiler.pir:
[pct] Add a signature for 'bxor'.
 03:08
rg kid51: patch at nopaste.snit.ch/15965 03:29
rg is off to bed
kid51 too
dalek rrot: r37738 | jkeenan++ | trunk/t/op/trans.t:
Applying patch pasted by rg++ correcting TODO labels, etc.
 03:32
03:34 szabgab joined
dalek rrot: r37739 | tene++ | trunk (2 files):
Add CONTROL_LEAVE exception type for rakudo.
 03:35
03:37 mikehh joined 03:47 janus joined 03:48 pinkyandthebrain joined 03:49 pinkyandthebrain left 04:01 Andy joined 04:33 tetragon joined 04:38 pusky joined 04:41 pusky left 04:50 s1n joined
dalek rrot: r37740 | Infinoid++ | trunk/examples/io/httpd.pir:
[examples] httpd.pir should use socket methods (rather than ops) wherever possible.
 05:09
Tene purl: msg jonathan &foo.leave and such works now in the leave_exceptions branch. pmichaud wants to discuss it with you. 05:10
purl Message for jonathan stored.
05:17 pusky joined 05:41 masak joined
Tene purl: msg jonathan if you get &?BLOCK and &?ROUTINE working, most of leave.t (except the stuff using indirect method syntax) will pass 05:48
purl Message for jonathan stored.
06:03 Santtu joined 06:09 Geekneeus joined
masak there is wonderful generated documentation for Parrot Compiler Tools that I can view locally after doing 'make html'. however, when I go to www.parrotcode.org/docs/compiler_tools.html I get 'This document is a DRAFT' and nothing more. 06:09
why?
Geekneeus make html is my next step :D 06:15
masak I wish someone would fix that. it doesn't feel congruent with a 1.0 release to have a h1 heading screaming DRAFT. especially since nice documentation obviously exists if you dig a little. 06:20
06:40 eternaleye joined 06:43 chromatic joined 07:00 uniejo joined 07:14 eternaleye joined
dalek rrot: r37741 | cotto++ | trunk (3 files):
[h] move inclusions of execinfo and dlfnc headers into a template, fixing make cover on Linux x86
 07:40
cotto Cool. We got an LWN article. 07:50
lwn.net/Articles/325364/
Is the guy who wrote that here? 07:51
++ to him 07:53
moritz bah, that's subscription only
cotto lwn.net/SubscriberLink/325364/41db46bcd5d83e11/ 07:54
chromatic cotto, there's a C++ comment in r37741.
cotto chromatic++, not anymore 07:56
dalek rrot: r37742 | cotto++ | trunk/config/gen/config_h/feature_h.in:
[h] fix C++ comment, clarify
moritz cotto: aye, very nice article 08:01
08:10 PacoLinux joined 08:14 eternaleye joined
cotto seen wknight8111 08:31
purl wknight8111 was last seen on #parrot 8 days, 13 hours, 18 minutes and 25 seconds ago, saying: is too much of a wuss to quit his job [Mar 17 19:11:41 2009]
cotto night
08:59 AndyA joined 09:05 alvar joined 09:37 donaldh joined
dalek rkdown: 40b00e8 | (Francois Perrad)++ | src/ (3 files):
whole refactor codeblock
 09:39
shorten dalek's url is at xrl.us/bemnws
dalek rkdown: 2366993 | (Francois Perrad)++ | src/parser/ (2 files):
remove action RawLine
shorten dalek's url is at xrl.us/bemnwu
09:47 wayland76 joined
wayland76 gmake[1]: *** [TGE/Parser.pir] Segmentation fault 09:49
That's what I'm getting trying to compile Parrot from svn
10:34 Tene joined 10:37 gaz joined 10:51 mikehh joined 11:02 ruoso joined 11:04 Infinoid joined
dalek rrot: r37743 | fperrad++ | trunk (2 files):
[chm] compute an index

format output like HTML Help Workshop 4.1		 11:05
11:20 bacek joined 11:21 donaldh joined 11:22 wayland76 joined
bacek good evening 11:59
karma bacek
purl bacek has karma of 102
bacek ok, it's about right time :)
Can I apply for parrot commit-bit? ( I know about sending CLA, etc) 12:00
12:00 DietCoke joined
dalek rrot: r37744 | fperrad++ | trunk (3 files):
[chm] the book is now a regular section
 12:00
wayland76 It looks like commit bits require that you ask at the right time of day :) 12:03
bacek wayland76: we just far-far away from other world :) 12:06
wayland76 We are, rather, and it's my bedtime, so I'll abandon you too :) 12:07
12:15 cognominal joined 12:35 rg1 joined 12:49 pusky left 12:51 Andy joined 13:10 gryphon joined 13:19 ruoso joined
moritz bacek: commit bits are usually discussed on #parrotsketch 13:34
Coke who wants commit bits? 13:43
if for parrot, you'll need a CLA before anything could happen anyway.
13:44 donaldh left
Coke cla? 13:52
purl i heard cla was Contributor License Agreement or www.perlfoundation.org/contributor_..._agreement or www.parrot.org/foundation/legal
14:20 Theory joined
diakopter pmichaud: does a potential rakudo contributor need to submit a TPF CLA and/or a PF CLA? 14:51
(or both)
sorry, dumb question, obviously a PF CLA wouldn't be needed 14:52
diakopter wakes up a bit more
14:55 ruoso joined
pmichaud rakudo still requires TPF CLAs 14:58
15:01 szabgab joined
dalek kudo: 7d9cd97 | pmichaud++ | docs/spectest-progress.csv:
spectest-progress.csv update: 332 files, 7424 passing, 0 failing
 15:11
shorten dalek's url is at xrl.us/bemoqk
particle heh, there are gsoc students interested in porting parrot to plan9 15:17
...but there's no perl5 there. :(
well, there's a sorta-perl5 there 15:19
15:20 donaldh joined
Coke ooh, someone has submitted a portfile for osx that might fix our install problem. 15:24
s/fix/workaround/
Infinoid someone++ 15:26
moritz particle: did we get an application for it? or just an idea? 15:27
Coke secret is apparently: install_name_tool - change dynamic shared library install names
jdfrens++ 15:28
doing a build right now... (and it's /just/ a portfile update, no patch required)
particle moritz: plan9 folks got an idea, or maybe an app 15:31
15:33 gravity joined
Coke woot. initial patch doesn't work, but I can make it work. 15:33
particle i told them we can co-mentor, if it looks possible (perl support on plan9 is... not great) 15:34
simontwo (but that is probably a feature, since perl is Unix-ish and Plan9 is anti-Unix) 15:41
NotFound Windows is also anti-unix and has perl and parrot ;) 15:43
purl okay, NotFound.
NotFound purl: forget Windows
purl NotFound: I forgot windows
simontwo NotFound, I don't think Windows is actually anti-Unix... just look at the remnants of Unix in places like system32/drivers/etc and more obscure places... I think Windows is just confused. Vista is the glossy prostitute version. 15:44
purl, Windows?
purl i don't know, simontwo
NotFound simontwo: I think that things are here for the benefit of the posix subsystem
Or mayne just because that was the easier way to adopt the bsc tcp stack 15:45
s/bsc/bsd 15:46
And Vista, I'm happy because I don't have the need to use it at work nor at home 15:47
Coke that's a new one:
compilers/imcc/imcc.l:1339: warning: variable 'buffer' might be clobbered by 'longjmp' or 'vfork'
simontwo NotFound, ditto. I never touched a vista box yet. well, I did touch one that was turned off. 15:49
Infinoid on a dare?
simontwo more like an accident 15:57
16:07 darbelo joined
Coke WOOT 16:09
'sudo port install parrot'
'which parrot' -> /opt/local/bin/parrot 16:10
'parrot -v' -> Missing program name
parrot -[acEGhprtvVwy.] [-d [FLAGS]] [-D [FLAGS]][-O [level]] [-o FILE] <file>
16:14 bkuhn joined
dalek rrot: r37745 | coke++ | trunk/ports/macports/Portfile:
Get a working 1.0.0 Portfile!
 16:15
rrot: r37746 | coke++ | trunk/ports/macports/Portfile:
Revert some unintentional changes to the portfile.
 16:22
Coke reruns the port install to make sure those changes didn't hork anything.
yay, it still works. 16:26
16:33 davidfetter joined
dalek rrot: r37747 | fperrad++ | trunk/tools/dev/mk_language_shell.pl:
[languages] add the option --gen-parrot and its stuff
 16:39
Coke trac's dead. 16:41
davidfetter /requiescat in pace/ 16:43
mikehh can't access trac either 16:53
particle moritz: gsoc.cat-v.org/ideas/ (parrot on plan 9) 16:54
Coke I like the "first we need perl 5.8.4 ported" 16:57
I'm sure THAT'll be easy. =-)
davidfetter heh 17:00
"first, catch a rabbit."
17:13 NordQ joined
particle what's the tpf gsoc irc channel? 17:15
i can't believe i forgot.... 17:16
rg #soc-help on irc.perl.org. 17:19
particle thanks! 17:20
Coke hopefully the 1.0.0 macport will be live in their system by tomorrow. 17:25
dalek rrot: r37748 | pmichaud++ | trunk/compilers/pge/PGE/Exp.pir:
[pge]: Allow routines generated by PGE to honor other pirflags (e.g., from POST)
 17:28
particle wow, cool, coke! 17:29
Coke meh. I'm just the glue. 17:30
pmichaud glue can be very important.
17:30 frzntoz joined
Coke ^_~ 17:30
pmichaud without glue, everything tends to fall apart.
17:31 donaldh left
chromatic That would have solved Yeats's problem. 17:31
pmichaud afk # lunch 17:33
dalek kudo: 0f6354d | pmichaud++ | build/PARROT_REVISION:
Bump PARROT_REVISION to take advantage of PGE changes in r37748.
 17:34
davidfetter bon appetit
shorten dalek's url is at xrl.us/bempcc
17:38 rindolf joined
rindolf Hi all. 17:38
davidfetter hi
rindolf Does anyone know what is chromatic 's up-to-date email address? It's not on www.wgz.org/~chromatic/ or www.modernperlbooks.com/mt/2009/03/...ntals.html and his oreilly.com address may be out-of-date.
shorten rindolf's url is at xrl.us/bempcv
rindolf Hi davidfetter 17:39
moritz rindolf: chromatic at wzg iirc
chromatic yep
rindolf moritz: thanks.
moritz chromatic: you would know :-)
rindolf chromatic: thanks.
chromatic My previous oreilly.com address forwards to my real account, but it's full of press releases and other uselessness, so I'm about to filter it away.
rindolf chromatic: OK. 17:41
Coke chromatic?
msg purl chromatic?
purl Message for purl stored.
Coke odd. 17:42
purl, chromatic?
purl chromatic is <req>a lot of fun. For years I'd try to play stuff like `peter and the wolf', and then I'd be frustrated because it would use some note I didn't have. or the author of jellybean or mailto:chromatic@wgz.org or wgz.org/chromatic/ or the winner of the not-a-contest perl-bugathon. or best reached via email. or the guy who hit me in the eye.
Coke ah, there you are.
nice tmbg ref.
Coke finds a perlcast on his ipod.
Infinoid jonathan, pmichaud: So, I replaced socket ops with socket methods as much as possible in the httpd.pir example (r37740). Thus a bunch of ops can be removed, see the upcoming nopaste.
nopaste "infinoid" at 75.140.33.106 pasted "Remove unneeded socket ops from io.ops." (67 lines) at nopaste.snit.ch/15969
Coke Are the opcodes part of the spec or no? 17:43
Infinoid No idea, hence the question I was just about to ask :)
moritz Coke: they were, but are gone now
Infinoid That seems to require running opsrenumber.pl. Are there any reasons (e.g. deprecation or whatever) not to go ahead and do that?
Coke NOOO
Infinoid Coke: Can you be more specific? 17:44
chromatic I don't even remember the bugathon. Maybe I should get a trophy.
... unless it was that time I broke English.pm.
Coke please don't run opsrenumber.pl
instead, you can probably just remove the offending ops from ops.num
Infinoid Can I patch opsrenumber.pl to do that?
rindolf chromatic: OK, I sent you the email I wanted to. 17:45
Coke Infinoid: to do what?
Infinoid To remove the offending ops from ops.num
If there's a reason to preserve these things, I can see a use for a tool that does so
Coke I would rather we decide what the policy is first before we write tools.
Infinoid Well, this is why I asked the question 17:46
Sounds like I should ask #parrot-dev about it 17:47
err, without the #
Coke there's a trac ticket referring to a build test that fails without doing the renumbering, also.
My /guess/ is that pre 1.0, it made sense to renumber whenever we added/subtracted ops to keep things in order. 17:48
My guess is that post 1.0, it makes sense to not gratuitously renumber if we can get away with it.
PerlJam Coke: sounds like someone should lock ops.num so that it doesn't accidentally get changed. 17:50
Infinoid I can sit on this patch until I can get rid of the remaining "socket" and "sockaddr" ops, which involves adding equivalent methods to the Socket and SockAddr PMCs
So it's not a big deal. 17:51
17:58 frzntoz joined 18:16 barney joined
Coke Working on the macport is one of the few chance I get to code in tcl! 18:19
chromatic I don't think it matters if we have holes in the ops file.
We can remove ops without renumbering.
Coke I could see doing a partial renumber before a stable release to squash any gaps that occurred post-release. 18:20
post-last-stable-release, that is.
holy curd, batman. RT now at 396 tickets. 18:21
NotFound Coke: Who won the prize for breaking the 400 barrier? 18:28
Coke NotFound: no clue; happened while I wasn't looking.
now we'll have to get down to 300.
NotFound We need an automatic photo-finish system 18:29
Coke someone with icc installed could probably help us close 2 tickets.
chromatic That changes the outcome by observing it!
Coke can someone with install privs update the installed parrot on feather? 18:30
darbelo Hmm, more OpenBSD build problems. Could someone lend me a hand? 18:32
Coke possibly. 18:33
390 tickets. guess I had a stale copy. 18:35
rg notfound: that was probably kid51 last night
Infinoid just closed out his pdd13pbc tickets and created TT #504 18:36
Infinoid <-- doing his part to close out RT :)
Coke 388 18:37
rg darbelo: what's the problem? 18:38
purl SIR NO PROBLEM SIR!
Infinoid Coke: I'll update feather to parrot 1.0.0
will "make install" cleanly overwrite the previous stuff? Or do I have to go hunt down and remove that stuff first? 18:39
NotFound I'm thinking about doing the same for all debugger "whishlist" tickets 18:40
nopaste "darbelo" at 201.252.61.125 pasted "OpenBSD build failure" (25 lines) at nopaste.snit.ch/15970 18:41
darbelo That wasn't happening yesterday
I noticed cotto made some related changes inside config/gen/platform/ in r37741 18:43
rg that looks like a missing include
18:44 davidfetter joined
darbelo so right now I'm building r37740 18:44
Coke takes/steals 20 RT tickets.
NotFound darbelo: I think I catch it. Wait a minute... 18:48
dalek rrot: r37749 | NotFound++ | trunk/config/gen/platform/generic/dl.c:
[core] include dlfcn.h in generic/dl.c
 18:51
moderator Parrot 1.0 Released | parrot.org | 385 RTs left! 18:54
Coke I'm taking some of the short, obvious keepers from RT and transitioning them to trac. 18:54
pmichaud Coke++ 18:55
18:55 bsdz joined
Coke I just mention it so folks don't start transitioning /everything/ =-) 18:55
Infinoid RT #50212 has no end in sight, and don't have tuits for #51718. Those are the only two I have left.
Coke if you don't have tuits, give up the ticket, i suppose. 18:56
Infinoid but kid51 so graciously gave it to me :P
bsdz hi, anyone know of any plans to ditch the json compiler? i want to use it in another HLL. 18:57
Coke no plans regarding json either way. 18:58
if someone is actually USING it, I guess we'll keep it. =-)
if you use it, I ask if you can help write more tests for it. 18:59
bsdz ok, i'll use it and will submit some tests specific to what i'm doing :)
Coke fair enough. =-)
there are two rt's for that right now; I'll convert them to a single trac for you to hang patches off of. 19:00
Infinoid I need to move the installed parrot 0.9.1 out of the way to get a clean 1.0.0 build on feather. Does anyone care? 19:02
PerlJam Infinoid: no way. (I probably installed 0.9.1 in fact)
Infinoid No moving or no caring? 19:03
PerlJam no caring
Infinoid kthx
bsdz thanks Coke. i'll find them in trac later
Coke, there seems to be two JSON.pbcs. One in library that looks like it creates JSONs, the other in compilers that can parse them. Is that right? 19:12
Coke yes. compilers/json is for ``compling'' JSON into a PMC. 19:13
JSON.pbc is for outputing JSON given an existing PMC.
IIRC>
bsdz am i right in thinking that compiling JSONs is not a library thing 19:14
Coke I don't understand the question. 19:15
Infinoid Coke: parrot updated on feather
Coke Infinoid++
Infinoid the old version might still work as parrot-0.9.1
Coke hey, whoever is fixing old networking code; can you fix compilers/json/postalcodes.pir ? 19:17
bsdz i'm wondering how to load_bytecode the JSON pbc in my pir. If I do load_bytecode "JSON.pbc" i'll only be able to write JSONs. If i want to read JSONs I'll need to manually copy compilers/json/JSON.pbc to my oath an include it from there. I'm guessing? 19:18
Coke can't just include it from the installed path? 19:19
bsdz not sure. both have the same name 19:20
PerlJam Infinoid: Now we just need to get the parrot installation to actually work so that others can use it :)
Coke shouldn't matter.
NotFound $ find . -name "JSON.pbc" 19:21
./runtime/parrot/library/Config/JSON.pbc
./runtime/parrot/library/JSON.pbc
./compilers/json/JSON.pbc
19:21 pinkyandthebrain joined
NotFound ETOOMUCHJSON 19:21
Coke or perhaps... it's not enough. =-)
bsdz perhaps would be better if they were renamed to JSONReader and JSONWriter or like
darbelo okay, r3774 works. Building r377
sorry, r37740 works. Building r37741 to confirm. 19:22
rg notfound: something also need to #include "parrot/has_header.h" for dl.c to work 19:25
19:26 pinkyandthebrain left
Coke bsdz; that's not a terrible idea. 19:26
would probably require a deprecation cycle. open a ticket for milestone 1.5 ?
rg my guess would be config/gen/platform.pm needs to put it into platform.c
bsdz i can add a ticket to trac. should i also write a patch though?
rg or maybe just put it in dl.c 19:27
Infinoid bsdz: patches never hurt 19:28
bsdz okay leave it to me :)
darbelo Confirmed r37741 did it. RTLD_LAZY is defined in dlfcn.h which isn't included anymore. 19:29
rg why didn't putting the include back fix the build? no include for has_header.h has been removed 19:36
ah bad dependencies. it didn't rebuild platform.c 19:39
darbelo But shouldn't the inclusion in trunk/config/gen/config_h/feature_h.in be enough? 19:40
rg because it's built by configure. 19:41
rg needs to learn more about parrot ;)
darbelo: fixed in r37749. the feature that would include dlfcn apparently isn't built. 19:42
rg also wonders what's the connection between PARROT_HAS_BACKTRACE and dlfcn.h 19:43
chromatic Apparently dlfcn.h somehow defines DL_info, which the backtracing needs. 19:46
Coke msg particle: adding callgrind output to parrot would make an excellent SOC project.
purl Message for particle stored.
NotFound The backtrace use some features, if available, to show more detailed info 19:47
Coke wouldn't it BNI parrot_config had an option to dump everything?
Tene Coke: --dump ? 19:48
That gives me 363 items
is that missing something?
rg chromatic: ic. but dynamic loading needs it too, so putting it back in dl.c is the right thing to do. notfound++ :) 19:49
Coke Tene: WIBNI if parrot_config had a --help option? =-)
Tene: there, now it does. whee. 19:56
Tene Coke++
darbelo Is there any particular parrot-person that should be contacted for SoC stuff? 19:57
Tene darbelo: particle AKA Jerry Gay, iirc
dalek rrot: r37750 | coke++ | trunk/tools/util/parrot-config.pir:
[parrot_config] - use 'iter' opcode; add usage.
rg but you didn't add if key == '--help' goto usage ;P 19:59
Coke rg: done. 20:01
particle is the best person for GSOC parroty stuff, yes
I am probably a distant third. 20:02
20:04 eternaleye joined
dalek rrot: r37751 | coke++ | trunk/tools/util/parrot-config.pir:
Add the --help option I was complaining didn't exist.
 20:04
kudo: 00cd1fd | pmichaud++ | src/ (2 files):
Add perl versions of trim() to setting (RT #64096).
 20:07
shorten dalek's url is at xrl.us/bempwu
dalek kudo: a0c6e3d | pmichaud++ | src/ (2 files):
Add Perl 6 version of p5chop and p5chomp to setting (RT #64092).
shorten dalek's url is at xrl.us/bempww
dalek kudo: 3bbb1c4 | pmichaud++ | src/ (2 files):
Add Perl 6 versions of cis() and rand() to setting (RT #64108).
shorten dalek's url is at xrl.us/bempwy
dalek kudo: 8af4574 | pmichaud++ | src/setting/Any-num.pm:
Clean up rand() and cis() from RT #64108 patch.
kudo: 5c07c7b | pmichaud++ | src/setting/Any-str.pm:
Use \\x0a instead of \\o12 (I find it easier to deal with hex).

newlines instead of using \\n.
shorten dalek's url is at xrl.us/bempw2
shorten dalek's url is at xrl.us/bempw4
rg coke++ (being consistent)++
20:08 fperrad joined
Coke macport commit in. should be live in an hour or so. 20:12
Infinoid *cue frankenstein theme music* 20:13
dalek kudo: c3e5408 | pmichaud++ | t/spectest.data:
Update t/spectest.data with p5chop and p5chomp tests.
 20:24
shorten dalek's url is at xrl.us/bempzk 20:25
20:37 Theory joined 20:47 eternaleye joined 20:53 eternaleye joined 20:56 frzntoz left 20:58 bsdz joined
bsdz hi, anyone know how I can get parrot's builtin load_library search path? also is there a way to get parrot_config to dump all settings to the console? 20:59
pmichaud bsdz: parrot_config --dump
bsdz thanks pmichaud. is there a pod for parrot_config? 21:00
pmichaud not that I know of.
might look at the parrot_config source.
Tene Coke added a usage statement and a --help to parrot_config today 21:01
bsdz that's really handy. i knew it had everything i wanted to know but didn't know how to get it ;-) 21:02
Coke I had the same problem. =-) 21:04
21:04 eternaleye joined
bsdz hmm, will it have the load_library search path there? :-/ 21:05
pmichaud I don't think it's in the config, no.
I could be wrong about that.
I don't remember seeing it there.
bsdz must admit nothing obviously looks like it. 21:06
hmm, I guess I should just assume it looks in current dir then under runtime/parrot/library. 21:07
Coke still no macport! 21:10
SYNC DAMN YOU!
fperrad in PIR, interpinfo .INTERPINFO_RUNTIME_PREFIX gives the root of search path 21:11
bsdz thanks fperrad :) 21:12
:q
oops i guess that gives the game away - i'm using vi . 21:13
pmichaud I figured it was some weird smiley. 21:14
bsdz i hadn't thought of that - could be 21:16
fperrad today I work on language with an installed parrot, see r37747. 21:21
a Rakudo patch is available on parrot.privatepaste.com/500z2XLLsI
not completed, because @libparrot_shared@
I think we need a new variable @inst_libparrot_shared@
pmichaud I don't want to use the := form of makefile rules unless someone explains to me what they are. 21:23
21:23 GeJ joined
Infinoid := interpolates immediately, = defers (and possibly recurses) 21:25
pmichaud is that true for all make tools?
NotFound No
gnu make specific
pmichaud so what happens on a platform that isn't using gnu make? Or is that now a requirement for Parrot? 21:26
Infinoid I hope not.
Coke I believe it's setup so that we use := if we can, and = if we can't.
pmichaud Parrot is setup that way?
Rakudo isn't.
NotFound Will not be simpler to never use it?
fperrad with gmake :
Coke parrot, yes.
fperrad ':=' is simply expanded variable
'=' is recursely expanded variable 21:27
if you are not using gmake, configure replaces all ':=' by '='
Infinoid Oh, right. Configure.PL does a s/// to strip the colons in the case where the build architecture can't handle it
pmichaud I'm not using Parrot's Configure.PL
Coke NotFound: Yes, it would be simpler to never use it.
Infinoid simpler++
pmichaud Rakudo doesn't want to go through gmake-specific items.
Coke I'm not sure adding the complexity to the build process is a net win.
NotFound fperrad: For some reason other than have more things to test and that can fail in some platforms?
pmichaud not unless there's a big win, which someone would have to explain to me. I never understood why Parrot switched to using the := . 21:28
not to mention the mk_language_shell generator.
chromatic I don't think it's a requirement. Certainly a few porters bring it up every time a GNUism sneaks in.
Infinoid pmichaud: I don't think there is a big win. := has a lot of cultural inertia in the linux world, that's all I know.
pmichaud okay. Rakudo votes against the := in Makefile rules. 21:29
NotFound If we must take care that it always can be replaces by = , there is no reason to use it.
Just replace it, and magically we test that it can be replaced X-) 21:30
pmichaud I'm also not a fan of splitting Rakudo's Makefile into multiple Makefiles
I can be convinced on that one, though.
(er, "more easily" convinced)
NotFound Recursive make is evil, some people says 21:31
pmichaud I agree -- I like that Rakudo doesn't depend on recursive make.
Infinoid Recursive make is hell when you want a working -j
NotFound recursive make is just not make at all
Infinoid That's very zen. 21:32
NotFound Euro zent X-)
Tene recursive make is bad news... 21:37
Coke I would be happy to get rid of the various makefiles in parrot. 21:46
not personally, but to have them gotten rid of. =-)
I tried at one point to start putting compilers/ into the top level makefile, e.g.
WOOT.
sudo port -d selfupdate && port list parrot 21:47
parrot @1.0.0 lang/parrot
Coke announces to list. 21:51
Infinoid Coke++ 21:52
fperrad pmichaud, I notice your wishes 21:53
one makefile is easy if I use dynpmc.pl & dynoplibs.pl, but there are deprecated
pmichaud if the new tools require separate makefiles, I think they're misdesigned. 21:54
Coke ISTR the makefiles to replace those were to be generated; you could always include them. (or we could update the tools to generate makefile snippets for inclusion at config time.) 21:55
pmichaud I don't see the separate makefiles in the diff
Coke but allison was driving that.
GeJ Good morning everyone
fperrad in fact, there are no new tools, just makefiles 21:56
pmichaud can I see what the makefiles look like? 21:57
21:57 TiMBuS joined
Coke (ah. then I suppose you could just incorporate those makefiles into your makefile) 21:57
pmichaud (the templates are what I'm looking for)
nm, I'm assuming it's the templates that are in mk_language_shell 22:03
fperrad new patch at parrot.privatepaste.com/470oM5NPht
now with new files
git diff doesn't include new file 22:04
pmichaud what are the #IF(...): lines?
are those gmake specific also?
fperrad no gmake, parrot tools specific 22:05
pmichaud so my makefiles have to be processed through parrot to work?
I guess I should be clear on this: I fundamentally disagree with the direction that parrot's build system has taken over the past 6 weeks. I don't want Rakudo's build system to have to follow that direction. 22:07
and I'd prefer that Rakudo's build system not depend on a bunch of parrot-specific tools.
(e.g., to generate Makefiles)
fperrad I try to work with tools supplied (and installed) by Parrot. 22:12
I think it's important that most languages over parrot use same tools.
pmichaud Rakudo was using those tools, until six weeks ago when others started changing the tools.
indeed, the tools were based on Rakudo.
(e.g., mk_language_shell) 22:15
afk for a bit # errands
22:25 eternaleye joined 22:28 eternaleye joined
rg coke: you usually have one Makefile per directory. parrot is working against make by building stuff from the main makefile 22:28
chromatic One Makefile per directory is hateful. 22:30
rg why? i believe that's how make was designed to work 22:31
chromatic Make was designed to handle dependencies.
If you split up your project's dependencies to where Make can't see them (across process boundaries), Make can't handle dependencies.
Tene rg: that's very much not how make was designed to work. 22:33
miller.emu.id.au/pmiller/books/rmch/ is a standard document on this 22:34
chromatic Make's utter lack of code inclusion (really, it's worse than JavaScript!) makes recursive Makefiles tempting. 22:35
rg ok, let me put it another way. make has a lot of features that help you if your files are in one directory
you're right that its primary design is to manage dependencies
davidfetter what's the down side of recursive makefiles? 22:36
chromatic You lose dependency information. 22:37
davidfetter k
rg i guess it shows its age with parallel builds
chromatic Not just that, but you have to execute *every* recursive Makefile even if none of their targets need rebuilding, because you don't have any information about them. 22:38
confound rg: tene's link is well worth reading 22:39
davidfetter how big a PITA would adding some kind of code inclusion mechanism be? 22:40
are we talking SoC project? get 5 geniuses on it as their full-time job for a year? 22:41
confound it has a code inclusion mechanism, it just sucks
chromatic Sort of like it has a "Did my dependencies get modified?" mechanism? (It *barely* does.) 22:42
Given that Make is the poster example of why backwards compatibility is wrong, I'm not sure it'll ever get fixed.
rg schedules some time to read the paper 22:43
darbelo It also is so widespread it will never go away either. 22:44
confound and then everyone will keep using it because it's everywhere 22:45
Infinoid But projects slowly drift away from using recursive makefiles, which is good enough 22:46
chromatic Yeah, you can't ever fix it because it has 12 (!) users!
darbelo It's everywhere, and every "where" has its very own set of incompatible extensions. Whee! 22:47
Infinoid When in doubt, copy the linux kernel Makefiles. Buildroot did. 22:48
davidfetter oops. mischan 22:50
chromatic, how about if it has 12! users?
moritz davidfetter: I had the same idea ;-) 22:51
but it seems a bit high a number ;-)
pmichaud Rakudo has managed to avoid recursive makefiles, I see no reason for us to drift towards a bad design only to have to figure out how to extricate ourselves from it later.
davidfetter yeah, well, there's lots of people learning to code from an early age :) 22:53
moritz like, 1/10th of the entire population? ;-) 22:54
davidfetter who knows? 22:55
NotFound who make?
purl make is awful really
23:01 fperrad left 23:02 NordQ joined 23:27 bacek_ joined 23:32 contingencyplan joined 23:47 pinkyandthebrain joined 23:59 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!