Parrot 2.3.0 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | Priority : apply deprecations, merge branches, finish exceptions_refactor branch | GSoC students, please read trac.parrot.org/parrot/wiki/GSoCersStartHere
Set by moderator on 17 May 2010.
Coke ... one I think is a waste of time in most cases. Don't succumb to the peer pressure! 00:01
00:01 kid51 joined
Coke certainly for dev releases. 00:01
(seriously, if it's stressing you out, don't worry about it) 00:03
cotto_work when in doubt, "Pirates" 00:06
"release without an interesting name"
Whiteknight I've got a good name lined up now. It's a name of a type of parrot, though I don't generally think all releases need to be named after parrots 00:07
bacek_at_work Do we have list of all releases codenames? 00:08
darbelo "The Nameless Bird" would probably work as a release name. 00:09
Whiteknight docs/project/parrothist.pod
bacek_at_work there is not such file... 00:10
darbelo docs/parrothist.pod
bacek_at_work docs/parrothist.pod
cotto_work docs/parrothist.pod 00:11
darbelo Hmm. 2.1.1 is missing there.
bacek_at_work darbelo, blame Coke. He was release manager :) 00:12
00:26 Myhrlin joined
kid51 Smolder crapping out again 00:34
Whiteknight smolder is rediculously slow most of the time. Maybe time is coming where we need to consider setting up our own server 00:41
00:44 dmagnus_ joined, baest_ joined 00:45 elmex_ joined, jjore joined
kid51 Someone made an adjustment to the timeout value about a month ago, and it has worked fairly well since then. 00:46
In fact, just now was my first 500 error since that change.
In any event: make test PASS darwin/ppc
00:46 NotFound joined
kid51 make fulltest PASS linux/i386 00:46
00:46 simcop2387 joined, dalek joined 00:48 particle joined 00:52 abqar joined
darbelo OpenBSD amd64 is passing fulltest regularly this days. Figured I'd mention it. 00:55
Whiteknight nice 01:02
it's almost as if openbsd is a first class citizen in the world of operating systems 01:03
cotto_work That's what happens when you throw commit bits at people running openbsd. 01:13
dalek rrot: r46754 | coke++ | trunk/docs/parrothist.pod:
Add in missing release (bacek++ for noticing)
 01:18
bacek_at_work Coke, it was darbelo
01:24 hercynium joined
cotto_work time for home 01:32
and tomorrow, migrating my ubuntu box to btrfs 01:33
01:34 plobsing joined 01:37 bakkdoor joined
cotto ohai 01:47
Whiteknight ohey
bacek_at_work no way 01:50
plobsing Is there a way to tell parrot "Hey, give me a PMC, but I don't know what kind I want yet"? 01:56
Whiteknight ... no? 01:57
plobsing: where, in the PIR level or in C?
plobsing C
Whiteknight plobsing: I suppose it's possible, but far more work than it's worth
plobsing I'm trying to think of a way to do freeze/thaw as a single pass over a PMC 01:58
Whiteknight once you set the specific type you have to check all the flags, the whole initialization process
plobsing which flags? custom_mark, custom_destroy, etc? Don't those only get set by vtable init? 01:59
Whiteknight singleton, constant, whether to even run vtable init 02:01
check out Parrot_pmc_new to get an idea for all the crap it does 02:02
anyway, I'm out for the night. see y'alls tomorrow
plobsing my use is more like Parrot_pmc_new_noinit. but it appears that even it does some init. misleading name. 02:03
02:04 bluescreen joined
cotto It's already a long name. Parrot_pmc_new_lessinit would just be too much. 02:04
plobsing Parrot_pmc_new_someinit_somenotinit 02:05
bacek_at_work Parrot_pmc_partially_init_something 02:06
plobsing so why do classes get to allocate their own PMC headers? is it so they can tack stuff on to the end of the struct by allocating something slightly larger? 02:09
02:15 khairul joined 02:19 JimmyZ joined
kid51 Ah, my Smolder finally got thru 02:24
kid51 must sleep
purl $kid51->sleep(8 * 3600);
nopaste "tcurtis" at 192.168.1.3 pasted "Why am I getting this error trying to compile this?" (65 lines) at nopaste.snit.ch/20582 02:34
khairul tcurtis: maybe its 'elsif' instead of 'else if"? im not familiar with nqp though 02:39
tcurtis thanks, khairul++. You are indeed correct. 02:40
cotto good morning khairul 02:42
khairul hiya cotto, good evening to u
give me a moment before i send a privmsg. nature calls. 02:43
02:43 gbacon_ joined
cotto no rush 02:43
02:46 gpw joined 02:47 janus joined
tcurtis I'm considering sending a message to parrot-dev linking to my blog posts. Not getting much feedback with my current post-and-wait-for-comments-via-email-or-irc strategy. 02:56
sorear tcurtis: Is it possible to comment on blog posts at all? 02:58
Nobody has done it
plobsing: Why don't you want the libjit framebuilder to come with Parrot? 02:59
tcurtis sorear: Possible, yes: www.parrot.org/content/past-pattern...comment-56 03:03
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#33889), fulltest) at r46754 - Ubuntu 10.04 amd64 (g++)
sorear tcurtis: As far as I can tell, there are no comments on that post
Do I need a commit bit? 03:04
plobsing sorear: I tried. I felt there was a barrier to getting it accepted as most ppl don't have libjit (and MacOSX can't get libjit :-() 03:05
sorear MacOSX *can't* get libjit?
plobsing AFAIK it won't compile or somesuch 03:06
sorear Also, we're targetting toasters now bubaflub++. They don't have libjit.
sorear doesn't have libjit
plobsing I would have chosen a different library at the onset had I known that.
sorear also doesn't have postgresql
sorear feels that the absense of a library on some systems should not be the deciding factor in whether to standardize the binding 03:07
plobsing plus *requiring* libjit/llvm/etc for core feels bloaty
ash++ will be making one based on libffi that hopefully works in more places
pluggable everything ftw 03:08
sorear plobsing: I don't think it should be required 03:10
but, if you have it, I think you shouldn't have to chase random github extensions to make it work
plobsing fair enough. If there is a desire to get it in core, I'm not *against* it. I'm against me doing the work for that. 03:12
sorear: also, it is distutils-compliant, so likely no chasing down required once I figure out how to register my project. 03:20
plumage?
purl somebody said plumage was the future Parrot module ecosystem. It will include tools to search metadata, handle dependencies, install modules, and so forth. The repository is at gitorious.org/parrot-plumage/parrot-plumage and the design docs are at trac.parrot.org/parrot/wiki/ModuleEcosystem
cotto 24 makes me lol 03:36
sorear 24 in what list? 03:37
cotto the tv show
purl the tv show is, like, way more interesting
03:42 LoganLK joined 04:14 TonyC joined 04:15 nopaste joined
tcurtis How am I getting a "too few positional arguments: 2 passed, 3 (or more) expected" error in a function after everything I do in that function? 04:15
plobsing return = call perhaps? 04:24
tcurtis ".return ($P107)" immediately follows the say I inserted at the end of the function, and the output from the say is happening. 04:31
plobsing tcurtis: can you show us the code
what are you returning to? 04:32
tcurtis I'll paste the code and the backtrace. 04:33
nopaste "tcurtis" at 192.168.1.3 pasted "grammar, actions, backtrace, and the bug-causing code." (230 lines) at nopaste.snit.ch/20583 04:36
tcurtis backtrace shows it as being called from Regex::Cursor::!reduce 04:39
plobsing is this go on parrot? awesome!
tcurtis Hopefully eventually. Of course, I'll start coding my GSoC soon, so I'll probably not work on it much until that's over with. But I like writing(more accurately, starting to write) compilers. So I tend to do that when I'm bored. 04:42
bacek_at_work tcurtis, you are using old way of declaring Grammar with modern NQP 04:43
sorear you could write a custom language for tree manipulations
bacek_at_work #= <key> form is deprecated.
sorear {*} is also deprecated
purl okay, sorear.
plobsing {*}?
purl {*} is part of the non-slurp alternation. or deprecated
plobsing I understood about half of those words 04:44
bacek_at_work If you are using it you have to declare actions with optional $key. E.g. method foo($/, $key?) { ... }
04:44 aukjan joined
sorear we need to track down all the old PGE documentation and kill it with fire 04:45
bacek_at_work sorear, can you rewrite squaak tutorial? 04:46
tcurtis Is there any up-to-date docs or example code for PCT?
sorear no. But I've never read it, so it must not be important.
bacek_at_work sorear, it is important. It was best "PCT Getting Started Guide" for last 2 years. 04:48
cotto defintely 04:50
We should move it to the wiki. 04:51
nm. It's on wikibooks
plobsing are there any good docs on POST? I've been giving more thought to the "make IMCC optimizer understand/work with POST" idea from sorear++
cotto wiki++
I don't know how you'd do that. I don't think imcc uses post internally. 05:43
bacek_at_work plobsing, don't make "IMCC optimizer"... It's kinda useless. 05:45
arnsholt Up to date docs for PCT would be very nice
05:46 uniejo joined
plobsing I think there is value in some parts of IMCC. I'd like to lift those parts out. The optimizer has potential. 05:46
cotto: I know IMCC doesn't use POST internally. I'm wondering if there is much difference in information level between say the life analysis blocks and whatever POST uses to represent blocks 05:48
06:00 aukjan joined
JimmyZ AFAIK, IMCC was planning to be replace by pirc or PIR? 06:02
plobsing afaict pirc is stalled. 06:04
cotto yes but there's been no movement there
plobsing also IMCC optimizer is not totally tied to the huge mess that is the parser 06:05
cotto better to work with what we have now than waiting for a steady state
plobsing my strategy is to gradually make things better. Making the optimizer available to HLL developers is a step in that direction (I think)
sorear plobsing: post is very low level... it can be roundtripped to PIR if you're willing to lose includes
plobsing is compilers/pct/src/POST/ it? I thought there would be more to it. 06:07
cotto It's basically pir. There's not much to it. 06:08
plobsing not much to pir? 'wc -l compilers/imcc/imcc.y' would seem to disagree :p 06:09
sorear half of PIR needs to die 06:12
after the next stable release, let's rip out all the imcc features I don't use. 06:13
plobsing sorear: such as? if you can make a case, we might deprecate it.
sorear .meth_call
plobsing already deprecated
sorear .(all the PCC stuff except tailcall and return)
sorear goes off to get a list of PIR features 06:14
06:15 grom joined
sorear user-defined macros 06:15
plobsing ones that take args or constants too? 06:16
sorear mostly the first
grom i am trying to figure out how to use the Lua frontend. I have checked it out with git but not sure what todo from there
sorear getting rid of constants too + .include would be awesome, but we'd need to find another way to handle Parrot magic numbers
probably a special literal syntax for them 06:17
plobsing you mean like .MY_MAGIC_NUMBER ?
sorear I mean like Trace.Instructions 06:18
what is this setline/setfile stuff in imcc.l? I've never used it
what's :uniquereg? 06:19
plobsing deprecated that's what
it was used to tell the register allocater not to mess with a register 06:20
sorear I don't like the fact that you have to specify encodings on strings; they should inherit it from the PIR
the .const mechanism bothers me but I haven't fully figured out how
plobsing how are you supposed to get utf16 strings in an ascii encoded file then?
NotFound sorear: is more complex than that. We should inherit from PIR the encoding for the source, but we must be able to select the one for the result. 06:21
sorear NotFound: source? result? I'm talking literals here 06:22
grom am I suppose to run parrot setup.pir? 06:23
sorear yes
NotFound sorear: unicode:utf8:"\\x{AB}" ---> Here I don't care about the source encoding, but I want a utf8 string in the pbc
nopaste "JimmyZ" at 192.168.1.3 pasted "JimmyZ got many faild tests on rakudo perl 6 on Windows XP" (47 lines) at nopaste.snit.ch/20586
plobsing sorear: wrt .include files with parrot magic numbers, I'd try to convince the powers that be to add some form of "use x :from<PASM>"
sorear NotFound: That seems a rather corner case. Why can't you just use some kind of recode_s_s_s op? 06:24
argh, too much C in imcc.y
NotFound sorear: because I want it in the pbc, not recoding at runtime. 06:25
cotto you were expecting maybe perl?
sorear well, it would have been nice if the BNF part started earlier than 45% 06:26
setline and setfile should be merged with .annotate 06:27
NotFound sorear: BTW that corner cases are the majority of the population of the world.
sorear the majority of the population of the world wants to encode UTF-16 text in the middle of an ASCII file? 06:29
I have never seen a *text editor* that can handle that
nevermind a compiler
if you want Unicode in strings, encode the source in a UTF; that's the way HLLs work
or are we trying to optimize, because pirops are ascii? 06:30
plobsing PIR is not an HLL. it is a convenient ASM. perhaps it should be made less convenient so people see the difference 06:31
NotFound sorear: read my example again. There isn't any ut16 on it.
sorear oh, I was confusing you with plobsing, oops 06:32
NotFound The current implementation of pir is wrong because mixes source charset/encoding with the ones for the string result. But the ability to emit the strings as desired is very useful.
I wrote some days ago a new unescape function intended to that. 06:33
Parrot_str_unescape_string 06:34
That decouplig will avoid the need to write mixed-encoding pir source while allowing to emit the kind of strings wanted. 06:37
plobsing would that eliminate the problems JimmyZ experiences with his /tmp paths? 06:38
06:39 eternaleye joined
NotFound plobsing: probably not, the env vars get/setting and file opening functions may need some love. 06:40
06:49 eternaleye joined 06:59 iblechbot joined 07:09 fperrad joined 07:11 gpw joined 07:12 fperrad_ joined
nopaste "plobsing" at 192.168.1.3 pasted "HLL PASM constant import" (23 lines) at nopaste.snit.ch/20587 07:23
plobsing sorear: in stead of dropping down into PIR, why not do something like nopaste.snit.ch/20587 in your HLL of choice? 07:24
07:42 grom358 joined
nopaste "grom358" at 192.168.1.3 pasted "What is wrong with this?" (18 lines) at nopaste.snit.ch/20588 07:43
bacek_at_work grom358, .SOLAR_MASS 07:44
(note dot)
grom358 bacek: hey? its giving me an error on line 6 07:46
plobsing pmc 07:50
not pcm
grom358 oh.. lol.. thanks
fperrad seen jhorwitz 07:58
purl jhorwitz was last seen on #parrot 251 days, 15 hours, 23 minutes and 19 seconds ago, saying: Tene: not at all. i should have the tuits this week to work on your compile issue [Sep 8 16:35:38 2009]
Tene Looks like I scared him away. 08:00
sorear plobsing: Missing the point. I'm suggesting things that could be ripped out of IMCC at the next deprecation cycle; I wouldn't miss them and probably no-one else would 08:02
Less code means less maintaince 08:03
plobsing replace them with fpmcs perhaps? 08:04
08:04 viklund joined
plobsing we need something to contain that information 08:05
sorear I was going to propose magic literals 08:06
but fpmcs might work - what are they?
plobsing frozen pmcs
eg: config.fpmc
they're quasi-pbc
sorear This would make hand-coding pir more annoying.
I'm not sure I mind, though 08:07
plobsing I guess I'm just not seeing how removing macro constants and adding "magic literals" removes syntax
sorear it would also allow removing .include 08:09
and all the multi-file abstraction layer
cotto looks like the parallel build is a bit sketch 08:12
y
sorear is enlightened by the current discussion and will submit a reduced IMCC pruning ticket tomorrow. 08:13
"things that sorear, cotto, plobsing, and NotFound agree are unneeded" 08:14
plobsing was there any point on which all of us agreed? 08:15
sorear nobody spoke out to save the NCI directives (.set_arg etc)
nobody spoke out to save block-macros
plobsing those aren't nci directives. and .nci_call is already deprecated
sorear nobody spoke out to save variable source encoding in strings
sorry, PCC directives 08:16
plobsing those are actually used by a(x, y, z) under the hood
sorear nobody spoke out to save setline and setfile
plobsing just try ripping them out
sorear Then they should be made permanently deprecated. 08:17
plobsing you'd need to redesign PCC (again)
sorear Why?
plobsing because that's how PCC works IIUC 08:18
sorear PBC format is permanently deprecated
It hasn't caused any problems
plobsing there's 2 kinds of deprecated I suppose - "we might keep it" and "kill kill kill!" 08:19
sorear Removing stuff from IMCC is great. Removing stuff from PDD19 is almost as good
Code we don't need to support has a low maintaince cost, too
plobsing That I can agree with.
08:28 clinton joined 08:41 TiMBuS joined
dalek rrot: r46755 | cotto++ | branches/ops_pct/compilers/opsc/ops2c.nqp:
[opsc] add some short options for convenience
 08:42
bacek ~~ 08:57
dalek rrot: r46756 | fperrad++ | trunk/runtime/parrot/library/HTTP/Message.pir:
[LWP] add a default Content-Type in form-data
 09:15
09:18 bakkdoor joined 09:25 grom358 joined
grom358 how can I call a C function from parrot? or does parrot have sqrt built-in? 09:26
nvm.. I see it is in the opcodes documentation 09:27
moritz most simple math is built-in
09:30 snarkyboojum joined
grom358 how can I print out a num with so many decimal places (porting this line from Lua: io.write( string.format("%0.9f",energy(bodies, nbody)), "\\n") 09:39
moritz parrot supports sprintf or printf or so
bacek grom358, there sprintf opcode
$N0 = 'energy'(bodies, nbody); $S0 = sprintf "%0.9f", $N0; say $S0 09:41
(With ';' representing newline)
Yay, it will not work...
grom358 yeah.. it says it wants PMC for last parameter 09:43
bacek $N0 = 'energy'(bodies, nbody); $P0 = new ['ResizablePMCArray']; push $P0, $N0; $S0 = sprintf "%0.9f", $P0; say $S0
something like this 09:44
grom358, why do you want to use pir, anyway? 09:45
moritz nqp is so much nicer to write :-)
bacek
.oO( github.com/fperrad/lua is implementation of Lua )
grom358 I tried that lua implementation earlier and got compile errors 09:46
bacek moritz, nqp doesn't support :multi (yet :)
grom358, strange. It should work...
moritz grom358: did you try it with newest parrot? or with the last release?
grom358 against 2.3.0 IIRC. It whatever yum install on fedora 12 had 09:47
moritz grom358: then you likely need an older version of lua too
bacek grom358, "master" lua will not work with release parrot.
moritz grom358: but if you plan to do some development with parrot, it's worth tracking the newest SVN
grom358 i have 5.1 lua
bacek parrot-lua then :) 09:48
grom358 damn I should of come here earlier.. cause atm I was just tryign to compare lua and luajit with parrot vm
bacek luajit will be faster... 09:49
msg fperrad What do you think about tagging Lua (and other your HLLs) with matching parrot releases? 09:50
purl Message for fperrad stored.
bacek msg plobsing What about idea of deprecating PASM->PBC totally? 09:53
purl Message for plobsing stored.
grom358 .const num DAYS_PER_YEAR = 365.24 ; $N0 = 2.96460137564761618e-03 * DAYS_PER_YEAR ; print_num($N0) is giving me 0.00.. huh? 09:54
bacek What is "print_num"? 09:56
grom358 print_num is that wrapped up code you gave me before for out to 9 decimal places 09:57
bacek meh... IMCC... Constant folding is broken... 09:58
grom358, good catch. Care to create ticket in trac?
grom358 sure can.. i'll just make a smaller test sample 09:59
nopaste "bacek" at 192.168.1.3 pasted "Smaller example for grom358++" (23 lines) at nopaste.snit.ch/20590 10:01
grom358 bacek: :)
bacek grom358, you can attach output from "./parrot -t1 t.pir" to clearly show that constant folding is broken...
grom358, and assign ticket to plobsing. He will "enjoy" it :) 10:05
grom358 is it on core component? 10:08
bacek "imcc"
purl, imcc? 10:09
purl imcc is the c of parrot or the problem or the intermediate code compiler or will be murdered soon
moritz imcc is also "fun" to hack on
purl okay, moritz.
bacek purl, good girl
purl :)
grom358 I can't assign tickets. trac.parrot.org/parrot/ticket/1647 10:10
bacek grom358, fixed. Thanks! 10:13
moritz (the assignment is fixed, not the bug - right? :-)
bacek moritz, boring! 10:14
:)
bacek do want to resurrect and implement PIR compiler in PCT... 10:15
10:16 davidfetter joined
dalek TT #1647 created by grom++: Constant folding on num type 10:24
TT #1647: trac.parrot.org/parrot/ticket/1647
10:24 kid51 joined 11:02 KatrinaTheLamia joined 11:10 Patterner joined
grom358 thanks for all the help.. finally finished implementing nbody in pir 11:17
is something like, $N0 = $N0 + $N1 slower then $N2 = $N0 + $N1 ? 11:19
bacek grom358, 1 microsecond slower 11:20
may be even less
moritz nothing compared to a GC run
grom358 kk.. i was just looking at areas can try to speed this up.. my pir version runs slower then perl 5 version 11:21
any switches I should be using.. I am just doing $ time parrot nbody.pir 1000000 11:24
bacek grom358, can you nopaste full code? 11:35
nopaste "grom358" at 192.168.1.3 pasted "n-body benchmark" (300 lines) at nopaste.snit.ch/20591 11:36
grom358 i can see major thing could change already. Get rid of using Hash 11:44
i just did that cause I was translating from the Lua version
11:47 Mokurai joined
bacek grom358, what is your perl5 version? 11:47
(There is few redundant access to hash values in pir version... But it shouldn't be 5x times slower) 11:51
grom358, what is output of "parrot_config optimize" on your box? 11:54
11:55 bluescreen joined
grom358 i don't seem to have parrot_config 12:00
the perl5 version is the one of language shootout page
12:00 JimmyZ joined
grom358 my attempt at pir version runs at 48seconds compared to 38seconds on 1000000 iterations on my test box 12:02
lua version runs in 10 seconds 12:03
bacek grom358, do you have svn checkout of parrot? 12:08
grom358 wow.. luajit runs it in 0.5 seconds 12:09
bacek grom358, can you check this one? trac.parrot.org/parrot/browser/trun.../nbody.pir
luajit is very-very good.
I did mention it :) 12:10
12:11 grom358 joined
bacek grom358, on my box parrot is about 4 times faster than perl5 on nbody benchmark... 39.9 vs 10.9 seconds. 12:15
grom358, wow, exetel.com.au :) 12:16
12:19 grom358 joined 12:22 Mokurai1 joined 12:24 bluescreen joined
bacek grom358, mate, are you around? 12:24
12:30 grom358 joined
Coke sorear: you can't just rip stuff out based on 2 or 3 people's limited usage in #parrot. =-) 12:34
bacek Coke, 3 out of 5 is majority of customers! 12:36
12:36 iblechbot joined, khairul joined
Coke bacek: That may be tragically correct. 12:36
12:37 Whiteknight joined
bacek Coke, s/may be/is/... 12:38
Whiteknight good morning, #parrot 12:39
bacek Whiteknight, "morning" 12:40
12:40 Mokurai1 joined
Coke bacek: in any case, I don't think we can get rid of setline/setfile. 12:41
(unless I'm missing some newer way to do that.)
bacek Coke, do you know how many "pir directives" (staring with .) do we have?
12:41 grom358 joined
bacek More than 30... 12:42
E.g. ".begin_yield". Who is using it??? 12:46
Why we can use just ".yield()"???
And so on.
Coke bacek: certainly some of those can go, yes. 12:50
but defaulting to "everything except" is a bad model when we're nearing 3.0; Find the explicit cruft and remove it.
bacek Coke, I think broad deprecation notice will do
like "Everything is deprecated: pir syntax, pcc, younameit" 12:51
Coke also, user defined macros are nice for hand-rolled PIR. I'd hate to see that go, as it would involve a HUGE rewrite of partcl.
bacek And use only opt-in model.
Coke bacek: I think that's bad. Decide what you /want/ first before you throw out the old stuff. Then we can have a migration plan.
bacek I want world domination. 12:52
And pony.
Coke bacek: given our existing deprecation policy, switching it around per element is just confusing. and to change PIR is going to be VERY invasive. 12:53
bacek Actually - lets nominate 3-4 HLLs and focus on their requirements.
Coke better to decide what we want first rather than say "we don't know what we want, but it ain't this."
bacek E.g. "rakudo", "partcl-nqp" and "lua"
Coke bacek: ok. partcl says "Stop changing stuff so I have to rewrite my damn compiler every 3 months from scratch."
focusing on game changers like lorito is probably better cost vs. effort. 12:54
bacek I _do_ like lorito.
Coke (every 3 months and from scratch are slight exaggerations. But partcl is STILL broken from whatever the last few releases have done.)
perhaps I'll be in a better mood once partcl is fixed. 12:55
Coke needs koohii
bacek That's why I put some effort into ops_pct branch...
bacek have very-very bad idea... 12:57
Let's bring "partcl" and "lua" back into nest.
Just to make sure, that parrot development will not brake them 12:58
moritz rejected.
bacek moritz, why?
moritz because it's a very bad idea 12:59
as you said yourself
bacek but currently parrot is disconnected from HLLs development...
And we are trying to do something with very lagged feedback from HLLs 13:00
moritz tries to give Rakudo feedback 13:01
bacek So, having some king of "nominated HLLs" should help with it.
13:02 atrodo joined
bacek moritz, rakudo has more devs than parrot... 13:02
And (personally) I always judge my development based on rakudo needs. 13:03
moritz bacek: I don't think so (more devs)
13:04 Mokurai1 left
moritz we have 2 that grok the code base, and about 2 or 3 more that don't, but still contribute 13:04
bacek moritz, "git log HEAD~1000|grep Author|cut -f2 -d' '|sort|uniq|wc -l" 13:05
13:06 Mokurai joined
moritz bacek: that's flawed for several reasons 13:06
bacek it is
moritz bacek: one of them is that for rakudo we try to push patches with the correct author 13:07
which git allows, and svn doesn't
bacek But it's much more that 1 for Lua and 1 for Partcl
dalek kudo: 31598e9 | (Solomon Foster)++ | src/core/Cool-str.pm:
Add beginnings of support for :nth argument to .match.
 13:11
kudo: da6b23e | (Solomon Foster)++ | t/spectest.data:
Turn on new counted-match.t.
Coke having them in the nest isn't going to help. it didn't help before. =-) 13:13
(hell, we're lucky we get a fulltest. =-) 13:15
Whiteknight John W. Emerson?
Everybody: Please update NEWS 13:16
I'm scouring the SVN logs now, but I don't guarantee I'll catch everything worth reporting. 13:17
bacek Whiteknight, "We are totally optimal now"...
Whiteknight ah, that's easy 13:18
Coke at least a few HLLs are using a parrot-friendly "perl Configure.pl && make test", or perhaps the distutils from fperrad; shoudl be pretty easy to test on a regular basis. partcl is only there as a reference at this point, so I don't actively maintain it, I just fix it when I noticed it's broken. Hard to fault parrot for not taking at least that much of an interest.
bacek Coke, just because partcl is very disconnected from parrot... In terms of development/testing/younameit 13:19
Whiteknight GZipHandle PMC, is written in PIR? 13:28
dalek rrot: r46757 | fperrad++ | trunk/runtime/parrot/library/LWP.pir:
[LWP] fix
 13:40
Coke Whiteknight: PMCs are written in C
(you can override certain things in PIR, but assume C) 13:41
bacek: the main problem I have is that parrot isn't really stable. Every time I pick something to target, I'm already hopelessly out of date. 13:42
This is, generally, a good thing because it means things are getting done.
Practically, it sucks.
(it's been 9 years and I'm still an early adopter. =-) 13:43
bacek: so, what do you suggest? 13:44
And, honestly, I'd focus more on lua (which, SFAIK, is complete), or rakudo (which has a much larger (that is, >1) audience than either partcl flavor) 13:45
13:45 ruoso joined
Coke IWBNI if we had an HLLTinder. 13:46
moritz rakudo has a smolder report page 13:47
Coke (which would have to respect build/PARROT_VERSION
rakudo smolder? 13:48
moritz rakudo smolder is smolder.plusthree.com/app/public_pr...reports/18
Coke moritz: rt.perl.org/rt3//Public/Bug/Display...l?id=63100
moritz Coke: thanks, will close 13:49
Whiteknight I've posted an update to NEWS with some of the things I found in the SVN logs. Needs more
I especially am having a hard time keeping track of all the things fperrad was adding, so notes about those would be nice 13:50
Coke msg him. 13:51
purl Message for him stored.
Coke msg purl
purl Sorry, I've never seen pur before.
Coke Whiteknight: when you say "outside of trunk", do you mean "outside of build dir" ? 13:52
Coke hates our release announcements. 13:53
bulleted lists of things that don't mean anything to the casual reader. Me ponders how to fix this in future releases. 13:54
Whiteknight Coke: that's what the svn logs say. I copy blindly
Coke Yah. that's doesn't make a good release announcement.
moritz Coke: maybe something like Rakudo does for release announcements? 13:55
Coke not blaming you; it's pretty much been done the same way since 0.x =-)
dalek rrot: r46758 | whiteknight++ | trunk/NEWS:
Expand NEWS. more to do
moritz a few paragraphs of text, explaining some of the more fundamental changes in more than a single line
moritz and then bullet point list with some less important changes, for the technically more inclined reader 13:56
Coke moritz: yes. I have the next stable release; I'll try to set a trend there. =-)
s/stable/supported/
Whiteknight if developers want proper credit for their contributions, they should update NEWS
nobody does, so they get stuck with whatever disjointed shit I generate before the release
Coke I thin focusing on credit is the wrong approach. the release announcements are for the users, not for karma. 13:57
*think
Whiteknight Coke: it's not just credit, we do want the users to know what things have changed since last release
Coke I think that for the next supported release, I'll do a rollup of all 3 releases into a "since last supported". 13:58
Whiteknight that would be fine
so you can poorly aggregate the previous two poorly aggregated NEWS entries
:)
Coke for the dev releases, it doesn't matter as much, i guess. We should probably also have verbiage that says "developer release" etc.
13:59 skv joined
Whiteknight that would be fine too 13:59
Coke Whiteknight: no, I'd pretty much do it over from scratch. Hopefully 3.0 will have an easier time of it.
moritz Whiteknight: that's fine, as long as you have an amplifier loop that reduces noise :-)
Coke ... assuming that the intermediate releases look like the previous releases. 14:00
moritz: smolder.plusthree.com/app/projects/...reports/17 14:02
14:03 eternaleye joined, gbacon_ joined
Coke mikehh: if you're still testing partcl, it would be spiffy if you could throw a "make smolder" at the end. 14:04
can anyone suggest a good way to search for the offending parrot revision that broke partcl when I've already changed partcl to deal with other deprecations? 14:06
dalek rtcl: 0597fb5 | Coke++ | config/makefiles/root.in:
These slashes don't need to be escaped
 14:09
rrot: r46759 | fperrad++ | trunk (2 files):
[LWP] fix POST to smolder
 14:11
14:11 plobsing joined
atrodo Are there any good docs on how to embed parrot and run a HLL? 14:12
Coke atrodo: hurm. docs.parrot.org/parrot/latest/html/...g.pod.html is embarassingly drafty. 14:13
mod_parrot was, for some time, the leading example on how to do embedding. 14:14
I'm not sure if it's functional with latest parrot, though.
Coke hurls www.parrot.org/mod_parrot
(it probably requires apache to build) 14:15
atrodo Over a year since the last release. 14:16
Any idea how accurate pdd10 is currently? 14:18
Coke anything in [draft] is suspect, and the large number of TBD's there isn't reassuring. 14:20
Were I you, I'd ping the parrot-dev list and ask for the current best examples. (maybe there's something in examples/, btw.)
atrodo That's what I thought. Seems like a good place to start digging with 14:21
Coke (examples/embed, but it's generating warnings atm.)
dalek rrot: r46760 | fperrad++ | trunk/runtime/parrot/library/LWP.pir:
[LWP] clean up
 14:27
rrot: r46761 | fperrad++ | trunk/NEWS:
more NEWS
rrot: r46762 | NotFound++ | trunk/NEWS:
mention find_codepoint and unroll ops in NEWS
rrot: r46763 | NotFound++ | trunk/NEWS:
mention PackfileDebug PMC in NEWS
14:31 bubaflub joined
mikehh Coke: will give it a try later today 14:41
14:48 patspam joined
Coke mikehh: oh, it's completely broken, btw. =-) 15:01
(but it would be nice if I had a canary in the coal mine once it's working again)
15:08 snarkyboojum joined 15:12 TonyC joined 15:15 hercynium joined 15:18 nopaste joined
Whiteknight Okay, I'm ready to commit for the release. Anybody have any last sneaking updates to NEWS? 15:21
It's 11:20AM here. I'm going to lunch at 12:00, will commit back at 12:30, so about an hour from now 15:22
NotFound Go, go, go! 15:26
15:30 atrodo_ joined
Whiteknight is manually bisecting a bug in SVN-controlled software for work 15:31
bubaflub Whiteknight: probably not worth the time or effort, but i thought i saw a script for git-svn bisect somewhere 15:32
15:33 Hunger joined
Whiteknight bubaflub: yeah, I had seen something like it. But I'm on win32, using VisualStudio + AnkhSVN plugin 15:33
moritz if you have a git-svn clone, you can use normal git bisect 15:34
Whiteknight meh
Coke msg fperrad the problem with using LWP for smoking results is that you need a functional parrot to complain about your broken parrot. 15:35
purl Message for fperrad stored.
Coke use svn-bisect. 15:36
search.cpan.org/~infinoid/App-SVN-B...svn-bisect
Whiteknight no perl on that machine
15:37 preflex joined 15:38 Infinoid joined
dalek rrot: r46764 | fperrad++ | trunk/runtime/parrot/library/distutils.pir:
[distutils] now smoke works with LWP.pir (instead of curl)
 15:39
kudo: c515f5f | (Solomon Foster)++ | src/core/Cool-str.pm:
Implement :overlap for Any.match.
kudo: 6894296 | (Solomon Foster)++ | t/spectest.data:
Turn on overlapping.t.
15:41 bluescreen joined 15:49 TonyC joined
Coke reminder, parrotsketch today. 15:49
15:53 nopaste joined 15:55 bluescreen joined, Infinoid joined, hercynium joined, Whiteknight joined, clinton joined, baest_ joined, dmagnus_ joined, spinclad joined, mj41 joined, mikehh joined, AzureStone joined, pmichaud joined, Util joined, athomason joined, cognominal joined, Tene joined, krunen joined, Maddingue joined, nnunley joined, cotto joined, confound joined, wagle joined, treed joined, GeJ joined, silug joined
darbelo Net is splitty today. 16:09
16:10 bluescreen joined, Infinoid joined, hercynium joined, Whiteknight joined, clinton joined, baest_ joined, dmagnus_ joined, spinclad joined, mj41 joined, mikehh joined, AzureStone joined, pmichaud joined, Util joined, athomason joined, cognominal joined, Tene joined, krunen joined, Maddingue joined, nnunley joined, cotto joined, confound joined, wagle joined, silug joined, GeJ joined, treed joined
dalek rrot: r46765 | fperrad++ | trunk/t/harness.pir:
send smoke report to Smolder with LWP.pir (instead of curl)
 16:12
rrot: r46766 | fperrad++ | trunk (4 files):
[LWP] some server like Smolder use LF instead of CRLF
16:19 Andy joined
dalek rrot: r46767 | fperrad++ | trunk/runtime/parrot/library/LWP.pir:
[LWP] fix redirection when relative path
 16:29
cotto_work good morning 16:30
purl For you maybe.
Whiteknight ferrad: ping
hello cotto_work 16:31
fperrad: ping 16:32
(almost didn't notice that typo)
16:32 KatrinaTheLamia joined
dalek TT #1648 created by petdance++: Look at using restrict keyword on pointers 16:33
TT #1648: trac.parrot.org/parrot/ticket/1648
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#33918), fulltest) at r46767 - Ubuntu 10.04 amd64 (g++) 16:43
16:43 szabgab joined
Whiteknight Okay, I'm going to cut the release now. EVERYBODY STAND BACK!!! 16:44
:)
bubaflub boom.
Whiteknight r46768 16:49
as soon as the tarball passes all it's tests, I'll make the tag and put out the announcements
darbelo gets to work on breaking trunk.
tewk I have the starts of a Select PMC, anyone want to help me pretty it up? 16:59
Coke Whiteknight: ping.
Whiteknight: did you see that fperrad snuck some code in under the wire? Please consider rolling that back for the release.
tewk: why a PMC and not an opcode? 17:00
darbelo tewk: Yes please!
Whiteknight Coke: you think it's a big issue? I was going to ignore it
fulltest just passed in the tarball
Coke he just switched our smolder to use his brand new, experimental, untested LWP. 17:01
let's move it to post-release, then we can have a month of testing before the next release goes live with it.
17:01 Mokurai joined
dalek rrot: r46768 | whiteknight++ | trunk (8 files):
Update all metadata for the 2.4 release.
 17:02
Coke my 2�
cotto_work +1
purl 1
Whiteknight urg
cotto_work that's a bad idea
tewk Coke: its kinda modelled after IO::Select, its ment to give us simple working async io (on unix) now, until the concurrency scheduler version of async io is implemented. 17:03
Whiteknight okay, okay. now I have to try and remember how to roll back previous commits like that
Coke cotto_work: precedent failure.
tewk: ah, ok, so it's not just a bare select. Danke. 17:04
Whiteknight does any body here have the svn smarts to rollback fperrads last few commits?
cotto_work whiteknight, iirc it's svn merge -r 46765:46764 svn.parrot.org/parrot/trunk 17:07
Whiteknight I always end up being off-by-one in my revisions when I do reverse merges 17:08
darbelo Whiteknight: Also, remember to do that to the tag, not trunk.
Whiteknight I'm going to do it in trunk, but I'll un-redo it after the tag 17:09
sound good?
darbelo Ewwww. But I'0m not the one doing it, so...
Coke darbelo: are you suggesting applying changes to a /tag/ ? ewwww. 17:10
tags shouldn't be treated like branches, even if they really are. =-)
darbelo Hasn't he already tagged?
Coke so you rm the tag and recreate it.
easy peasy.
darbelo But that involves undo;undo-the-undo games with trunk. Which I dislike. 17:11
Coke or Whiteknight could just ignore me. 17:12
but I really don't like that people are ignoring the RM and slipping code in at the last minute. 17:13
darbelo Or ignore me. Or ignore us both.
Whiteknight I haven't made the tag yet
Coke I would probably actually drop the tag, branch from pre-fperrad's commits, apply the patch that came after them, tag the branch as the release, and go from there. but that will screw up rakudo (slightly).
Whiteknight final verdict: I'm undoing the changes in trunk, recreating my tarball, reapply the changes to trunk, and make my tag 17:14
darbelo Whiteknight: Then ignore us both and do what you said ;)
Coke er, tag before reapplying?
the tag should == the tarball.
darbelo Coke: Rakudo needs to start pulling from tags. Using a rev is wrong. 17:16
Coke darbelo: yes. we've already been down that path with them. =-)
darbelo If they are knowingly doing it wrong, it's not our problem. 17:18
dalek rrot: r46769 | whiteknight++ | trunk (4 files):
Undo recent changes by fperrad, so I have a clean revision from which to cut the release. Will re-apply these changes in a minute
Whiteknight if we used Git, I could have had my own release manger, been pulling in commits, and purposefully ignored those last few commits untill after the release
cotto_work sits in choir and listens 17:20
Whiteknight of course, as soon as I dictate my master plan I remember how much of a huge pain in the ass it is to do reverse merges in svn 17:23
okay, crap. I can't find a way to put fperrad's changes back into trunk right now 17:27
whatever, moving forward 17:28
Coke easiest way is to just reverse merge your reverse merge. 17:29
would anyone mind if the EVAL_X files generated with -D60 were actually EVAL_X.pir ? 17:30
darbelo I don't know what you are talking about but lend my full support to your plan. 17:31
17:35 davidfetter joined
tewk Is there an easy way to turn a local git branch into a svn branch? 17:37
Does anyone use git svn branch, to create branches in svn?
17:43 KatrinaTheLamia joined
Whiteknight fulltest runs much faster now with fewer runcores to test 17:44
Coke Whiteknight: let me know how the "update docs.parrot.org" task goes.
PacoLinux this seems interesting : news.squeak.org/2010/05/14/workshop...tems-2010/ 17:46
Whiteknight Coke: okay, will do 17:50
NotFound They don't mention UCSD Pascal? Shit.
dalek rrot: r46770 | whiteknight++ | trunk (4 files):
reapply changes by fperrad++
 17:51
Whiteknight Fulltest passes in the tarbal. we have ourselves a release 17:52
bubaflub Whiteknight++ 17:53
Whiteknight r46771
my login for the osuosl.org website doesn't work 17:54
darbelo tewk: bacek would know how.
Coke cotto_work: also, I have to rebuild docs.parrot.org once a month.
lemme know when the tarball is uploaded. 17:55
cotto_work Whiteknight: I can upload the tarball if needed 17:56
or add your public key to the ftp site 17:57
Whiteknight I had a public key added there, but I've since reformatted this computer
so, I suppose that's no long working 17:58
cotto_work easy fix
Whiteknight good, cause it was easy to break
cotto_work You can nopaste your rsa public key here. The worst that can happen is that people can allow you to log in to their systems. 18:00
Whiteknight THE HORROR! 18:01
purl the horror is a genre, or a collection of related genres.
nopaste "Whiteknight" at 192.168.1.3 pasted "public key for cotto++" (1 line) at nopaste.snit.ch/20596 18:02
cotto_work added 18:05
Whiteknight word up
18:05 cotto_work2 joined
cotto_work2 can you log in? 18:06
dalek rrot: r46771 | whiteknight++ | tags/RELEASE_2_4_0:
tagged release 2.4.0
 18:08
cotto_work2 also, why is it your username at gmail? It's usually the name of the host you're logging in from.
Whiteknight cotto_work2: yes
cotto_work2: I have no idea. it is what it is
cotto_work2 ok
Whiteknight I'm not very bright, so leave open the possibility that i messed something up at some point 18:09
cotto_work2 wmy, wfm
wfy, wfm
Whiteknight crow.pir is giving me problems 18:12
nevermind, had a typo in release.json 18:13
whatev
18:15 GodFather joined
dalek rrot: r46772 | whiteknight++ | trunk/tools/util/release.json:
fix two typos in release.json. Tried to do that step too early in the morning.
 18:24
Coke docs.parrot.org updated. 18:27
ugh. adding "supported" link to eventually replace /stable/ 18:28
so docs.parrot.org/parrot/supported/html/ now works.
stable--
ps in 2h 18:29
cotto_work Coke, do you have Khairul's cla?
Coke real name? 18:30
purl real name is Erich Muller.
Coke purl, forget real name.
purl Coke: I forgot real name
cotto_work Khairul is part of it.
Coke checking
cotto_work: rec'd on 4/30 18:31
cotto_work good news then
Does he need to go through the usual #ps nomination process or can he just get a bit now, since he's a gsoc student? 18:32
Coke I'd still go through the formality
cotto_work np
Coke as we can then be sure to tag his mentor as per usual.
don't want any slacker-mentors. 18:33
dalek website: Whiteknight++ | Parrot 2.4.0 "Sulfur Crest" Released!
website: www.parrot.org/news/2010/Parrot-2.4.0
cotto_work interesting quote choice 18:34
Coke ... I love ATHF but probably would not have selected it for parrot. =-)
Whiteknight Coke: need to live dangerously 18:36
Coke glados++
Whiteknight Coke: how does one ssh into parrotvm?
Coke 'ssh userid@parrotvm.osuosl.org' ?
ISTR we were vaguely limiting access to that box. 18:37
(it's the host of trac, svn, www, docs, etc.)
cotto_work where userid == parrot
nm. I'm thinking of the ftp site. 18:38
may be different there
Coke in my case it's not parrot, no. 18:40
which may be why access is more limited.
Coke checks who has access.
Coke wonders what has replaced /etc/password in this particular *nix variant. 18:41
ls ~/.. works. =-P 18:42
moritz ldap?
purl well, ldap is Lightweight Directory Access Protocol a networking protocol for querying and modifying directory services running over TCP/IP. or good at dealing with tree structures
moritz nis?
purl i heard nis was annoying. i mean, he can hack on module::build already instead of complaining about it all the time or New Israeli Shequel, but I prefer it when you say ILS
moritz yp?
purl hmmm... yp is yellow pages
Coke moritz: yah. pita. I just want "list of users on this box."
hey, I used to manage NIS+. you can't scare me.
so, users with access on that box include allison, me, cotto, particle, and jeff. (and some others who are not active parrot folks) 18:43
so, for the next release, bug someone who is not me for the doc release. =-) 18:44
moritz bugs Coke into automating the doc release
18:45 tcurtis joined
particle will be automating html doc generation from pod for openefs this week, but from git not svn 18:47
cotto_work I somehow didn't know that I have access to that box.
Coke cotto_work: you got it the last time you did a release, but probably not in time. 18:48
looks like you only ever logged in today. =-) 18:49
I am tempted to remove the "thank our sponsors" line. :P 18:50
someone go find us one! 18:52
18:56 bluescreen joined
Coke msg chromatic you mentioned some time ago that svn throws away commit information; I am wondering how the normal branch merging process for git avoids this, since it seems to do the same "lump all commits from this branch into master". (unless you mean to cherry pick them to master) 18:59
purl Message for chromatic stored.
tcurtis darbelo: I'm commenting on your first NFG blog post, but it's just a test to see if it works the same as when I did on my post. So, assuming you're actually able to see it, ignore it.
18:59 jhelwig joined
tcurtis sorear, do you remember if were you logged in on the parrot.org website when the comment on my post didn't show up for you? 19:01
19:03 chromatic joined
chromatic Only if you squash, Coke. 19:04
Coke ISTR if you don't squash, you get other issues.
chromatic Lots of commits, perhaps? 19:05
moritz well, 'git log' shows the commits of the merged branches intermixed
Coke perhaps it was just on git-svn.
moritz which is sometimes not awesome
chromatic git-svn may indeed not be awesome that way. 19:06
19:08 Myhrlin joined
tcurtis bacek++, thanks for "provides invokable". 19:10
darbelo Shouldn't that be 'invocable'? 19:11
tcurtis I'm not sure. wiktionary provides results for both, listing "invocable" as an alternate spelling of "invokable". Some derivatives of "invoke" use the k, but "invocation" and related use c. 19:15
moritz just spell it invo[kc]able :-)
moderator Parrot 2.4.0 "Sulfur Crest" Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | GSoC Students, please read trac.parrot.org/parrot/wiki/GSoCersStartHere 19:16
19:18 mmcleric joined
moderator Parrot 2.4.0 "Sulfur Crest" Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | GSOC Students: trac.parrot.org/parrot/wiki/GSoCersStartHere 19:19
Coke trims a few chars of that. 19:19
*off
dalek TT #1649 created by tcurtis++: Blog post comments not visible for non-logged-in users 19:20
TT #1649: trac.parrot.org/parrot/ticket/1649
Whiteknight I'll send out the rest of the release announcements tonight 19:23
that's my least favorite part of the whole event
19:23 Chandon joined
Coke tcurtis: I can twiddle the drupal settings if someone tells me which one to pus. 19:25
*h
Whiteknight seen Chandon? 19:26
purl Chandon was last seen on #parrot 45 days, 23 hours, 22 minutes and 48 seconds ago, saying: It really depends on the application and the algorithm. For some things, nothing, for others, everything. [Apr 2 20:03:14 2010]
Chandon Heya. Finally found my way out of the cave. 19:27
Whiteknight nice!
I was starting to worry
:) 19:28
Chandon Anything crazy happen in the mean time? Parrot get finished and no longer accepting new code perhaps? 19:29
Coke checks 'access comments' for anonymous users...
tcurtis: fixed. 19:30
Whiteknight Chandon: no worries, parrot has gotten better but is certainly not "finished" 19:31
in particular, our threading system is garbage :) 19:32
Chandon: send in your CLA yet? 19:33
Chandon Yes, I got that sent in.
Whiteknight Coke: have you received his CLA? 19:34
Coke real name? 19:35
Chandon Nat Tuck
dalek TT #1649 closed by coke++: Blog post comments not visible for non-logged-in users 19:37
TT #1649: trac.parrot.org/parrot/ticket/1649
Whiteknight Coke is the gate keeper. He holds all the keyes
keys*
19:47 joeri joined
Whiteknight Chandon: I saw your blog post by the way. Very nice 19:47
As a matter of persona inquiry, why did Ruby 1.9 drop green threads and go with the "wacky" CPython GIL design? 19:49
Coke I'm just /a/ gate keeper. we come in six packs.
yes, I have your CLA, Chandon
Andy Reading about the "restricted" keyword makes me drooly.
moritz Coke: you're the gate keeper with the highest IRC presence
Chandon I'm not actually sure why they did that, except that it's really simple to implement and reason about. 19:50
Whiteknight Andy: is "restricted" in C89?
chromatic No pointer aliasing, I believe.
Tene Whiteknight: I saw on github that you started following my old, untouched parrot GUI libs experiment.
Whiteknight Tene: I did a search for "parrot" and followed every result that turned up 19:51
Tene Ah. Nice.
purl Ah. Nice. is that considered 'dual sport' ?
Tene purl: forget Nice
purl Tene: I forgot nice
tcurtis chromatic: have you had a chance to look at my blog posts? 19:53
chromatic I have, but I haven't read them in sufficient detail to comment as they deserve.
I did have a thought the other day. 19:54
What if you tried to get only one optimization working as soon as possible?
It could be as simple as constant folding the addition of two integers.
Whiteknight one usually opens the door to others
chromatic That'd let us review the whole process and see what's feasible and what isn't, as well as what we like and what we don't. 19:55
I'd rather find out very early if your project blocks on anything we need to change in or add to Parrot.
Andy Whiteknight: I don't know. 19:56
Whiteknight Andy: looks like it's part of C99, or at least a proposal
Andy Whiteknight: OK, but what about it? 19:57
I would not implement anything that would make us any less cross-platform capable.
Chandon Isn't restrict just an optimization enabler, so you can macro it to "" on compilers that don't know about it? 19:58
NotFound restrict? The "son of noalias"?
tcurtis chromatic: That sounds like a good idea. And a good choice for the optimization. Simple but both somewhat useful and extensible into something more useful(more extensive constant-folding). 19:59
Andy Chandon: That could potentially be a solution.
NotFound Is a pesimization enabler, mostrly.
mostly.
chromatic Right, get one working then iterate.
Andy see the article linked to in the Trac ticket
Whiteknight has to pack up and head home now. Will try to catch the end of #ps. Later 20:01
20:06 iblechbot joined 20:09 mmcleric joined
dalek kudo: cf5afac | moritz++ | (2 files):
implement :x in Any.match; also test it
 20:20
20:22 allison joined
Util Does NQP allow heredocs? Like `my $block := <<END_OF_BLOCK;` ? 20:23
moritz nope
Util Are heredocs planned for NQP?
moritz no idea 20:24
Util OK, thanks
PerlJam Util: even if it did, they wouldn't have that syntax. 20:27
chromatic #ps in 1 20:30
20:30 smash joined
smash hello everyone 20:30
Util PerlJam: thanks; I would settle for any syntax; just used the Perl 5 syntax as an example.
moritz it would be Q:to<end_marker> 20:31
Util Hi, smash. #ps meeting just started, so attention here may be fragmented.
dalek kudo: 72ed16f | moritz++ | src/core/Cool-str.pm:
simplify .match(:x) implementation a wee bit
 20:42
20:46 mmcleric left 20:50 ash_ joined 20:53 iblechbot joined
Coke for the ops_pct merge, how does one regenerate the checked in generated files? 20:53
chromatic Speaking of Lorito, does anyone know anything about C-- as used in Haskell? 20:55
bacek Coke, make bootstrap-ops 21:03
Coke msg plobsing OOOH. here's an IMCC ticket for you: trac.parrot.org/parrot/ticket/906
purl Message for plobsing stored.
Coke bacek: ok. that's completely different from the "update the IMCC generated files" command. 21:04
let's make them both work the same way RSN.
GeJ Good morning everyone.
dalek kudo: 6aedf5e | moritz++ | src/core/Cool-str.pm:
[Any.match] remove some duplicated code
 21:05
bacek Coke, iwbn. Especially to avoid accidential regenerate of imcc files.
Coke I don't have a problem with "make bootstrap-imcc". lemme think for a bit about the best place for these sorts of things. 21:06
21:09 Whiteknight joined
dalek rrot: r46773 | tewk++ | branches/select_pmc:
[SelectPMC] Start of Branch
 21:11
Whiteknight starts backlogging 21:13
more information for the GC is a good thing, if we can get it inexpensively enough 21:22
chromatic With the split header/buffer or header/PMC_data scheme we have now, I'm not sure copying/compacting is super useful. 21:23
Whiteknight true 21:24
chromatic A generational system definitely is, but I want to see numbers on various approaches.
Chandon Does tools/dev/parrot_shell.pl work? 21:26
21:29 davidfetter joined
Whiteknight Chandon: Only one way to find out. 21:32
Chandon: dukeleto wrote that, I think
chromatic Where's he been, anyway?
Chandon Experimentally, the answer is no, assuming I'm not managing to use a repl wrong. 21:33
Coke Chandon: works here. 21:35
you have to actually print stuff.
tcurtis It seems to work. You have to use "." at the end of your code, too.
darbelo Chandon: perldoc it for usage info.
cotto_work Coke: to regenerate the C ops in ops_pct, you build with the checked-in code and run make bootstrap-ops 21:36
Coke or type "h" for help.
Chandon Nice. I *did* manage to use a repl wrong. Works great.
Coke cotto_work: yes, bacek covered that. =-)
Chandon: it's more of a REL.
dalek kudo: 980e03c | (Solomon Foster)++ | src/core/ (2 files):
Implement (generic) Numeric.succ and Numeric.pred. Remove the multi from Rat's
 21:37
purl .succ and .pred are already implemented.
kudo: 1bf6eab | (Solomon Foster)++ | src/core/ (3 files):
Create postfix:<i> for Numeric and Any, delete the old Complex and Any versions.
website: darbelo++ | UCS-4, NFG and how the grapheme tables makes it awesome 21:40
website: www.parrot.org/content/ucs-4-nfg-an...it-awesome
darbelo Hmm. Maybe I should de-promote that from the front page. 21:42
Chandon darbelo: How do you do that, anyway? 21:44
darbelo There's an option for that, somewhere in the edit tab. Can't recall exactly here now, but I've used it before. 21:45
Hah! found a better option. Made the release announcement stick at the top so we can blog with impunity. 21:46
Chandon That's a good plan. 21:47
darbelo I keep forgetting I have editor privileges on parrot.org 21:48
cotto_work note to self: backscroll all the way before responding 21:49
NotFound Somone is going to declare #ps officially finished? 21:50
chromatic yes 21:51
smash Whiteknight++ # release 21:52
GeJ darbelo: wouldn't it be worth pushing all the fp-promoted blog entries on www.parrot.org to planet.parrotcode.org?
NotFound Whiteknight++
Whiteknight smash, NotFound: thanks!!
GeJ Whiteknight++
Coke planet.parrotcode.org should already include everything on www.parrot.org; if not, bug.
GeJ Coke: then bug. 21:53
Is there someone I should contact?
darbelo GeJ: I think Coke was in charge of that. 21:54
;)
Coke bug. fixed.
GeJ Coke++
Coke need a new parrot_small logo. someone have a url on parrot.org with that images? 21:56
(otherwise planet.parrotcode.org is borked) 21:57
GeJ how small? www.parrot.org//files/parrotify_favicon.ico ? 21:58
Coke too small. =-) 21:59
something without the logo but same size as front page would be nice.
NotFound We need a vectorized version.
Coke if someone wants to GIMP that up for me, attach it somewhere for me.
danke. 22:01
(parrot.org now showing up on planet.parrotcode.org) 22:02
GeJ a1.twimg.com/profile_images/4829988...normal.png
(taken from the twitter box on the front page)
Coke ah, right. 22:03
danke. will fix this evening.
GeJ bitte.
22:08 pjcj joined
ash_ has 2.4 been released? or is that tomorrow? 22:23
ah, wait, silly question
darbelo Released today. 22:24
ash_ yeah, i just needed to read the irc channels message
darbelo It should show up on parrot.org too, unless the cache is acting up again. 22:25
tcurtis: ping. 22:28
tcurtis darbelo: pong.
darbelo tcurtis: Did you problem with comments get solved? I can delete the test comments for you when they are no longer needed.
22:28 bacek_at_work joined
tcurtis Yes. Coke++ fixed it. 22:30
darbelo Okay, two test comments axed. I think that's all of them. 22:31
22:32 davidfetter joined
tcurtis It is. 22:35
22:36 cognominal joined
darbelo tcurtis: I've just read your posts now, and I'm wondering about something. 22:39
tcurtis What's that, darbelo?
darbelo Where do you expect most uses of your library to happen? Inside the parrot toolchain (ie, as a standard HLLCompiler stage) or in HLLs as a customization step. 22:40
Rephrasing it: Will the optimizations be something HLLs get for free or something that the HLL dev will have to learn in order to gain something out of it? 22:43
22:46 joeri joined
chromatic Ideally they'll be free. 22:48
ash_ does parrot currently have an optimization setting for the command line? (akin to -O0 or -O3 in gcc) 22:49
darbelo ash_: Yes, but it might break your code.
bacek_at_work (optimization) It should be one of the phases in PCT. 22:50
darbelo bacek_at_work: Right now it's in IMCC and, surprising as it sounds, it doesn't really work. 22:51
What I really, really want is to get the PCT PIR compiler working again. 22:53
bacek_at_work darbelo, let's do it! 22:54
(apart from that it never "worked" before :) 22:55
darbelo bacek_at_work: And it's written in old NQP.
bacek_at_work darbelo, guess? I KNOW!
tcurtis darbelo: HLL devs won't have to know how to write optimizations, but they will(unless it is decided to make some optimizations default stages) have to add the HLLCompiler stages for any desire optimizations to their compilers. But that should be as simple as an "addstage(foo)" call. Something for turning PAST::Walkers and PAST::Patterns into HLLCompiler stages will be another thing I'll have to implement for that. 22:56
22:58 Psyche^ joined
darbelo bacek_at_work: Complete rewrite of code that's never worked before? Sounds like fun! 23:00
Once I get GSoC out of the way I'll be all over that ;) 23:01
bacek_at_work darbelo, start from scratch. Implement PIR grammar in NQP-RX. Converting to PAST will be easy.
cotto_work clock? 23:03
purl cotto_work: LAX: Tue 4:03pm PDT / CHI: Tue 6:03pm CDT / NYC: Tue 7:03pm EDT / LON: Wed 12:03am BST / BER: Wed 1:03am CEST / IND: Wed 4:33am IST / TOK: Wed 8:03am JST / SYD: Wed 9:03am EST /
tcurtis If you guys get that working and PCT adds support for anything in PIR that it can't currently output(is there anything left?), then everything targeting Parrot will be able to benefit from my GSoC project. We'd even be able to add optimizations at any given time to code after it's already been compiled(as long as we have the PIR output somewhere and not just the PBC) like LLVM can. 23:06
darbelo HLL >-(PCT)-> PIR >-(PCT)-> PIR >-(IMCC)-> PBC ? Too much overhead for my taste, but i can see it, yeah. 23:08
Still, I'll be investing all my tuits in NFG this summer.
chromatic PCT doesn't have to go to PIR.
darbelo True, but then our PIR -> PIR compiler gets left out of the loop :) 23:12
Coke updated planet's images. 23:14
tcurtis PCT could have separate stages for POST->PIR and POST->PBC/PASM/whatever it might use as a final product other than PIR. 23:15
bacek_at_work tcurtis, +1 23:18
darbelo +Inf
23:21 joeri joined
cotto_work That's been a request/plan for a while. 23:28
bacek_at_work cotto_work, trac.parrot.org/parrot/ticket/599 23:29
cotto_work allison, in your cft, I'd appreciate a review/correction of trac.parrot.org/parrot/wiki/LoritoRoadmap 23:30
ash_ anyone know why "Use of uninitialized value in pattern match (m//) at lib/Parrot/Configure/Step/Methods.pm line 124." would appear in a config file I wrote? 23:31
darbelo Maybe you are using an uninitialized value in pattern match (m//) at lib/Parrot/Configure/Step/Methods.pm line 124 23:32
;)
Or, your platform name is somehow blank. 23:33
Can you nopaste the file? 23:34
ash_ sure, one sec
gist.github.com/405722 i am mostly copying it off of how the glibc looks for glibc, since it was another c library that seemed relatively simple, this is for the config steps for libffi 23:35
23:36 joeri joined
darbelo Hm, and where on the list of steps did you place it? 23:37
ash_ right after glibc
darbelo Does moving it to the vincinity of auto::pcre help any? 23:38
ash_ no, still complained 23:39
oh, i see why 23:40
i didn't fill out the hash i am sending to _select_lib correctly
darbelo Another tip: You might want to do something like " $conf->options->get( without-libffi )" somewhere and bail out early if that option is passed. 23:42
23:43 eternaleye joined
darbelo Respecting the 'verbose' option is also a good practice. 23:43
And kid51 will like you very much if you add tests for your step. 23:44
ash_ kk, i'll keep those things in mind 23:45
is it bad practice to use pkg-config?
darbelo We don't rely on it being available. 23:47
Unless you want to prbe for it as well ;) 23:48
ash_ hmm, okay, its just i already see 2 different ways of libffi being installed (depending on linux vs os x)
i might
if i use pkg-config, its the same command to check for it on linux and os x, but i guess i could also check for OS specific locations of the file too
darbelo I'd take a page from other config steps and key off whatever is in "my $osname = $conf->data->get('osname');" 23:49
config/auto/pcre.pm is probably a better example of what you want to do than glibc
ash_ kk
i am going to go get dinner, i'll try to finish the config step tonight, and test it if i have time 23:51
although i might have to ask for someone with windows for windows tests, and maybe a non-ubuntu linux user...
darbelo auto::icu, auto::gmp, auto::readline, auto::pcre, auto::gettext and maybe auto::opengl will probably be of use to you. 23:52
ash_ thanks, food & 23:54
sorear tcurtis: I do not have an account on parrot.org and it wouldn't be visibile to wget anyway 23:56
Whiteknight I'm really enjoying reading all these GSoC blog posts
tcurtis sorear: turns out the problem was that anonymous users didn't have permission to see comments. Coke fixed it. 23:57
darbelo Whiteknight: Talking about code I haven't written yet is easy ;)
I can do that all summer.
Whiteknight darbelo: yeah, that's what my blog is all about!

Powered by Cro and the Raku® Programming Language.

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

Thank you!