Parrot 1.7.0 "African Grey" is out! | Fix issues caused by the pcc_reapply and context_auto_attrs merges | find out what's up with the slice opcode | Latest modified TT's: icanhaz.com/parrotbugs
Set by moderator on 1 November 2009.
kiwichris darbelo, No OpenBSD support. There is a NetBSD user so maybe that will help. 00:00
darbelo Toolchain should be close enough. I'll give it a go later in the week after I'm done with the bloody circuit analisys exam. 00:01
kiwichris darbelo, by support I mean existing packages or binaries 00:02
darbelo So, a build from source should work. Cool.
00:03 TiMBuS joined
kiwichris darbelo, yes it should 00:03
dalek TT #1187 created by mikehh++: t/src/warnings fails test 2 on optimized build with Ubuntu 9.10 amd64
darbelo kiwichris: Excelent! I'll still wait until after the exam though. 00:07
I have so much stuff in my brain right now I'm afraid I'll start forgetting it if I learn anything new.
Oh, look at this cool shiny RTOS! and *poof* I don't know anything about tchevychev filters anymore. 00:10
Whiteknight: ping 00:11
Whiteknight darbelo: pong
darbelo I'm working on TT#1072, and I'm one line away from a correct solution. 00:13
Coke chevy chase filters?
huh.
chromatic They forget everything after Caddyshack, Ghostbusters, and the first Fletch movie. 00:14
Coke that's probably for the best. (Chevy was in ghostbusters?)
darbelo do you know what the VTABLE_instantiate_str() call in compilers/imcc/pbc.c should be repolace with?
chromatic No, but it's a general principle.
darbelo s/repolace/replaced/
Coke darbelo: hah. good luck with that. =-) 00:15
chromatic Ouch, that's a tough one.
darbelo It's the one remaing use I have of thet vtable. Removing the rest was easy enough. 00:16
Coke I suspect this one will be much more difficult. =-)
darbelo But I (fortunately) don0t know enough about symregs to know what's TRT here. 00:17
chromatic Only FIA, Float, Integer, and String implement that vtable anyway.
FIA will be the tough one.
darbelo Yeah, The others just get and ATTR set. 00:18
00:18 zak_ joined
darbelo But I wonder, what make this types special? Why can the other go by without this vtable. 00:19
chromatic I think it's to handle the get_params and similar opcodes. 00:20
darbelo add_1_const() isn't making that distinction. Icall there fora any pmc it recieves. 00:21
00:21 Andy joined
Coke darbelo: wasn't this for constant generation? 00:23
00:23 Zak joined
darbelo Creates a constant PMC, given a SymReg. Allegedly. 00:23
chromatic Any of the other PMCs gets the default behavior, which throws an exception. 00:24
dalek rrot-linear-algebra: 20c6e46 | Whiteknight++ | src/pmc/nummatrix2d.pmc:
fix whitespace, and actually get the PMC to build.
 00:25
darbelo chromatic: Exactly. How come we get away with that?
darbelo We never create constants of any other type?
Coke pretty sure FIA was for PCC 00:27
chromatic As far as I know, darbelo.
darbelo Hmm. I wonder how pirc does this... 00:30
Whiteknight urg, how do I compile something with debug symbols?
is it -DDEBUG?
darbelo put -g in CFLAGS ?
Whiteknight -g is debug? 00:31
darbelo debug symbols, last time I looked.
I think Configure already puts it there if you don't --optimize 00:33
Coke you can do both with gcc.
perl Configure.pl --optimize --ccflags=-g --prefix=/Users/coke/bird 00:34
00:34 zak_ joined
darbelo Coke: Exactly, but you don't need to add -g if you don't --optimize. 00:34
Coke yes, but you should always optimize. =-) 00:35
allison FIA is for IMCC, PCC doesn't use instantiate_str to create the signature arrays 00:36
kiwichris What is src/parrot_config.c ?
Coke at a guess, the source of parrot_config (an executable) ? 00:37
allison really, that could move to a subroutine in IMCC instead of a vtable function on FIA
darbelo kiwichris: That is a comand-line interface you can use to query the parrot config settings.
kiwichris kiwichris, config, as in how it was configured and built ? 00:38
darbelo, oops for you :)
00:38 allison joined
darbelo Yes, it reports all of the data generated by the configure probes, and hints files. 00:39
darbelo forgets what a transfer function is.
kiwichris What are src/dynoplibs ?
cotto_work That's where dynops live, which are ops that can be loaded at runtime. 00:40
kiwichris cotto_work, thanks 00:41
I am looking for a away to diff to built trees so I am get an RTEMS specific patch. Once Configure.pl can support RTEMS is not less important. 00:43
msg Should concentrate on one thing at a time.
purl Sorry, I've never seen should before.
darbelo allison: Okay, I was looking for a cleaner way out, but I can do that too.
dukeleto back again 00:44
allison darbelo: on what?
darbelo <@allison> really, that could move to a subroutine in IMCC instead of a vtable function on FIA 00:45
allison ah, right
darbelo I'm working on killing the VTABLE. TT#1072
allison if it helps any, that whole generation of a FIA is likely to go away with the next refactor of PCC
and just create a CallSignature directly instead 00:46
darbelo: (after looking at ticket)yes, makes sense 00:47
dukeleto kiwichris: dynops might prove to be hard on RTEMS
kiwichris: they are dynamically loadable opcodes. rtems only supports static linking 00:48
00:48 Zak joined
kiwichris dukeleto, Agreed. A gsoc project this year added dlopen and friends so in time it could be added. What would we miss out on ? 00:49
darbelo kiwichris: HLL-provided ops.
dukeleto kiwichris: not very much, right now. just a few special purpose math ops. but you would miss out on random number generation, which is implemented as a dynop 00:50
darbelo Also, you won't have dynpmcs.
dukeleto kiwichris: and more things are planned to be implemented as dynops in the future
darbelo: what dynpmcs exist now? 00:51
kiwichris dukeleto, interesting, so this just stops bloating the base VM ?
darbelo dukeleto: Languages that need dynops won't work either.
kiwichris: In theory :) 00:52
dukeleto: In parrot core? Not many important onw I can think of righ now.
00:52 tetragon joined
dukeleto kiwichris: it allows swapping in/out opcodes at runtime. mostly for ease-of-customizability, but it also keeps the core small 00:52
dalek TT #1188 created by allison++: [RFC] Modify build directories to more closely match the install ... 00:53
dukeleto kiwichris: parrot does not need to be recompiled to change out dynops
kiwichris dukeleto, I like it. Some RTEMS users insist on static everything (easier to certify) so could that they built in ?
dukeleto kiwichris: it should be possible to just choose which dynops people wanted in a static build and just include those. like 'all' or 'rng' 00:54
kiwichris dukeleto, great.
cotto_work Couldn't the dynopslibs be statically linked? I don't think parrot will care as long as Parrot_load_foo_lib gets called to initialize the ops/pmcs. 00:56
dukeleto cotto_work: yep
darbelo goes on the hunt for foods.
See t'aal later!
japhb Has there been a final decision on the NQP-rx .pir file snapshot-into-parrot-core request? 00:59
dukeleto where are the docs for fiddling with src/parser/grammar-oper.pg in an HLL? 01:01
msg Infinoid can we get the kea HLL working on dalek? dalek hates me. 01:04
purl Message for infinoid stored.
dalek rrot: r42230 | allison++ | trunk/docs/book/draft/ch11_directive_reference.pod:
[cage] Remove some references to "compilation unit" in draft book chapters.
 01:08
Whiteknight incoming 01:10
purl i heard incoming was pause.perl.org/incoming/
Whiteknight purl forget incoming
purl Whiteknight: I forgot incoming
Whiteknight purl incoming is <reply>oh shit! hit the decks!
purl OK, Whiteknight.
Whiteknight incoming
purl oh shit! hit the decks!
Whiteknight purl++ 01:11
dalek rrot-linear-algebra: 32890b0 | Whiteknight++ | src/pmc/nummatrix2d.pmc:
get initialize_from_array method working.
rrot-linear-algebra: bf202a1 | Whiteknight++ | t/10-nummatrix.t:
add tests for initialize_from_array method
rrot-linear-algebra: b7010f2 | Whiteknight++ | (2 files):
redo initialize_from_array to be cleaner and more sane
rrot-linear-algebra: 1244fc5 | Whiteknight++ | src/pmc/nummatrix2d.pmc:
get_block now appears to work correctly
rrot-linear-algebra: d0b6840 | Whiteknight++ | t/10-nummatrix.t:
add tests for get_block
rrot-linear-algebra: 09114a5 | Whiteknight++ | src/pmc/nummatrix2d.pmc:
set_block now appears to work
rrot-linear-algebra: 2e4f6f1 | Whiteknight++ | t/10-nummatrix.t:
add a test for set_block
rrot-linear-algebra: 650ccbb | Whiteknight++ | t/10-nummatrix.t:
fix the test plan
Whiteknight ...and with that, I think parrot-linear-algebra contains all the functionality that Matrixy needs
at least, for basic operation
dukeleto Whiteknight++ 01:13
dalek rrot: r42231 | allison++ | trunk/docs (3 files):
[cage] More documentation updates to remove references to "compilation unit".
 01:15
rrot: r42232 | allison++ | trunk/t/pmc/sub.t:
[cage] Changing a comment to remove "compilation unit".
 01:22
dukeleto msg mikehh i am seeing that same failure in t/src/warnings.t on darwin/x86 01:24
purl Message for mikehh stored.
plobsing hi #parrot 01:34
dukeleto plobsing: hello! are you working on jit-stuff?
plobsing: if so, you need a commit bit. have you thought about sending in a parrot CLA so that you can drink the koolaid straight from the jar? 01:35
plobsing yes. I didn't have much time last week, but I'm going to try to get the auto_libjit working with pcc_reapply.
dalek rrot: r42233 | dukeleto++ | trunk (3 files):
[t][TT #1183] dies_ok() for Test::More, bubaflub++
plobsing dukeleto: I don't have nearly enough good patches to be a parrot commiter
all the pathes I've sent in have been horrible and needed massive fixes 01:36
dukeleto plobsing: perhaps you should focus more on tests
plobsing I'll experiment in github and if it looks good, then we'll try to commit it
I'm working on drinking the testing coolaid, but it's a pretty bitter brew 01:37
dukeleto plobsing: perhaps you can write down your ideas on the wiki? what are the first steps? frame builder or L1 ops for llvm?
plobsing here's what I think (may be a little off): 01:38
dukeleto plobsing: we kind of need a "spec" for our jit. what are the most basic things that we want it to do?
plobsing get a framebuilder, rewrite pmcs in nqp/whatever, bootstrap with framebuilder
dukeleto ok, so a framebuilder is first 01:39
what is the first step in making a framebuilder?
01:39 Andy joined
plobsing it's the easiest piece of the puzzle 01:39
dukeleto plobsing: do you know about the parrot git mirror on github?
plobsing just found out about it yesterday
dukeleto plobsing: how do we properly test a frame builder?
chromatic Run some NCI tests. 01:40
plobsing see TT#1147 tests
dukeleto plobsing: github.com/leto/parrot
plobsing thanks
dukeleto chromatic: which NCI tests?
01:40 nbrown joined
dukeleto chromatic: and is that testing-at-distance ? 01:40
plobsing t/pmc/nci.t is pretty comprensive
dukeleto plobsing: it should be updated every 2 hours. bug me if it ain't :) 01:41
plobsing dukeleto: one patch in #1147 tests the dynamic frame builder against the static ones generated in nativecall.pl
dukeleto plobsing: that is cool. is the test in perl or PIR?
plobsing currently perl
I dislike the idea of moving to all PIR based tests
dukeleto plobsing: do tell 01:42
plobsing if the harness doesn't work (because I messed something up) how do I know whether it is the test or the harness that fails?
there are a couple of areas that are sensitive to this: JIT, NCI for examples
dukeleto plobsing: i don't see how that is a Perl vs PIR issue 01:43
kiwichris is away: I'm away but logging
plobsing it isn't really, its an issue of using the unit under test to test itself
dukeleto plobsing: parrot's test harness is in perl. we are talking about the tests being written in PIR 01:44
allison plobsing: yes, I ran into that while working on pcc_reapply, where a large number of the failures were simply because the PIR version of Test::More couldn't make one subroutine call
plobsing: but, it is just a general problem of bootstrapping tests
plobsing: and the answer is to make the test suite smarter, testing core behaviors before more complex behaviors 01:45
chromatic Test::More relies on some complex behaviors.
dukeleto plobsing: i looked at the tests in that patch. they could be rewritten in PIR if need be.
chromatic: Test::More relies on PGE 01:46
plobsing but nativecall.pl is in Perl
allison chromatic: aye, but printing "ok 1" doesn't require particularly complex behaviors
chromatic I meant besides PGE, but yes.
plobsing the point of the tests is to make the NCI => PCC mappings synchronize
if we have 2 sources 1 for t/pmc/nci.t and 1 for nativecall.pl its just as bad as having 1 for src/pmc/nci.pmc and 1 for nativecall.pl 01:47
dukeleto plobsing: perhaps that stuff would be better implemented in NQP ?
plobsing the test or nativecall.pl?
dukeleto needs to wander home 01:48
plobsing: nativecall.pl
purl nativecall.pl is in Perl
dukeleto purl, go play in traffic and die in a fire
purl go play in traffic and die in a fire is probably on the double!
plobsing thats a good idea. nativecall.pl is a little complicated for my skills though
allison dukeleto: NQP is non-core, can't depend on it 01:49
dukeleto: also, small bootstrapping problem, implementing core build tools in a language that can't run until parrot is built
plobsing allison: does this mean that to get rid of the Perl testing dependancy, all tests should be written in PIR? 01:51
allison plobsing: eventually, yes
plobsing: though, to be more accurate that's "tests of PIR features should be written in PIR" 01:52
plobsing and tests of PMC features?
allison plobsing: we also have some C tests
plobsing: PMCs are accessible from PIR, so count as PIR features 01:53
japhb allison, wait, can't depend on NQP? Do you mean in general, or just before Parrot has been installed?
allison japhb: they were talking about using NQP for the build process
japhb: that is, before Parrot exists
japhb OK, *phew*
It caught my eye from across the room, and I had a panic moment there. 01:54
plobsing what if I added a flag to nativecall.pl that just caused it to dump its info to stdout and then read that from PIR?
japhb Talking of which ... NQP-rx .pir file snapshot: decided?
allison japhb: you can always install NQP as a dependency before something else, but Parrot is an absolute dependency before NQP
japhb nod 01:55
allison japhb: decided how?
japhb allison, meaning, can we go and do it, so that Plumage can be unblocked?
plobsing why does parrot have to ship nqp-rx for plumage? why can't plumage ship it? 01:56
allison sounds like you're referring to a ticket or email I haven't read yet... pointer?
kid51 grits his teeth at the time he just lost doing a SeaMonkey upgrade
allison is it "Proposal for using NQP-rx"?
japhb allison, give me a sec to find the archives, I've just been watching the discussion in email 01:57
allison, yeah, there's been several emails in that thread (and perhaps another of similar name, I forget at the moment)
plobsing, because a snapshot of plumage is going to be shipped with Parrot, so shipping it as part of plumage is effectively shipping it as part of parrot. 01:58
... except with more schedule-coordination pain
allison japhb: I have to say, I'm inclined toward leaving NQP-rx out of Parrot core and making it a module install
japhb: but for that we'd really be heavily dependent on Plumage being usable 01:59
japhb allison, we're just talking about the snapshot of the three or four core .PIR files, not the whole source and all.
And basically Plumage is blocked on being able to move to NQP-rx. 02:00
allison aye, but they're really separate modules
japhb I could just do as plobsing suggests, but as I said, it would have no effective difference except to make life worse.
allison -- sure, but we've got a bootstrap problem.
allison and the source isn't in Parrot, so we don't have an effective way to maintain them, make bug fixes etc
japhb Plumage won't be able to run without NQP-rx. 02:01
allison well, not until it's installed
japhb allison, how would you install NQP-rx without Plumage?
allison but, you'll be installing your own modules as well
(at least I'm assuming Plumage will have some installation process to get started) 02:02
japhb allison, the snapshot of plumage (plumage-mini, or whatever) that will ship with Parrot will be installed with Parrot.
Otherwise again we have a chicken and egg problem.
... Or we force our users to manually install three different things, always. Which seems kinda awkward. 02:03
allison japhb:... wget ...nqp-rx.tgz; tar -gxvf; perl Configure.pl; make; make test; make install?
japhb allison ... what language will that script be written in? 02:04
allison japhb: I'm assuming there will be a something that installs Parrot, NQP-rx, and Plumage
japhb Right, so *what is that written in*
allison japhb: but I'm also assuming that something isn't the Parrot tarball that ships monthly
japhb that "something"
allison japhb: Plumage is written all in NQP? 02:05
japhb allison, NQP with a bit of PIR
Even the configure is written in NQP 02:06
allison japhb: which was fine as long as NQP was a core language 02:07
japhb: not sure what the answer is here 02:08
plobsing maybe I've missed something, but are we always going to depend on perl being present for configure-time?
allison plobsing: no, we explicitly plan to eliminate the perl dependency by 3.0
plobsing so how will we do nativecall.pl then? 02:09
japhb I'm a little frustrated here, so please bear with me for a moment ... but I just put in a LOT of effort based on the assumption that I could depend on PIR and the runtime part of NQP (the regex libraries) as an absolute minimum available set, always. Was I completely mistaken?
allison japhb: it wasn't my choice to move NQP out of the Parrot repo, but moving it out does have consequences. Will have to see how it shakes out. 02:10
japhb: We'll do the best we can. 02:11
japhb: but, we can certainly find a way to supply the necessary dependencies for Plumage 02:12
chromatic I'm not sure that there's anything more simple than including three PIR files.
japhb allison, OK, fair enough. It seemed to me that the right thing was to include those three PIR files. I am open to suggestions.
allison japhb: it may be a simple matter of doing an initial configuration step in simple PIR, before you depend on NQP
japhb I need those regex libraries. 02:13
Or hand code the regex parts of Configure in raw PIR.
allison chromatic: I'm concerned about maintaining generated PIR code in the repository when we don't have the source code to regenerate it from
japhb Which means rewriting the regex libraries. I'm not qualified, sad to say.
allison chromatic: what do we do if we find a simple bug, but aren't ready to pull in a full update because NQP has made changes that don't fit the Parrot deprecation policy? 02:14
02:14 Zak joined
japhb (fork if necessary), branch, fix, reimport 02:15
chromatic Does our deprecation policy cover PIr libraries? I don't recall that it does.
02:15 JimmyZ joined
allison chromatic: core PIR libraries, yes 02:15
chromatic: if they're shipped in the tarball, they're covered
japhb It's all public-hosted code, with compatible license. Fork your friend. No worries. 02:16
allison chromatic: (several, completely unrelated to this bit, need to be removed as soon as they can be deprecated)
chromatic The only thing I see documented is:
=item * API changes in the compiler tools
japhb Being pulled away by $family 02:18
bak later
allison chromatic: the support policy says "feature" it doesn't specify where it's implemented
chromatic It does, in =head2 Deprecation Candidates 02:21
The question is whether that is an exclusive list or an example list.
dukeleto chromatic: indeed.
chromatic I can't imagine how that could be anything other than exclusive.
dukeleto can someone help me update the "latest release" link to point to 1.7.0 on parrot.org ? 02:22
it is pointed at 1.6.0 right now, which sucks
allison chromatic: I could invoke the Wikipedia rule, but from a more practical perspective, ripping out a PIR library from under a core user is every bit as bad as changing a core API 02:23
dukeleto allison: what is the wikipedia rule?
allison chromatic: that is, if it's not in that list, it should be 02:24
dukeleto: the "I'll prove it! Just give me 5 minutes to modify Wikipedia..."
(or in this case, the support policy document) 02:25
dukeleto allison: aha. gotcha
allison: · API changes in the compiler tools <-- is on the list of Deprecation Candidates in support_policy.pod 02:26
allison: so I very much agree with you.
i think we are all sitting in the dark a bit, because we don't know how much pmichaud plans on changing/modifying nqp-rx, going forward 02:27
is it 80% done? 99% done? will it have backward-incompatible changes? these are big questions in my mind
cotto dukeleto, iirc rakudo needed only one minor change to switch to it, though that's without using its new features. 02:33
plobsing link to the diff?
cotto I'm curious about that myself. 02:34
dalek TT #1189 created by jkeenan++: t/steps/*/*.t: Replace reliance on init::defaults 02:35
dukeleto plobsing: nqp-rx lives here: github.com/perl6/nqp-rx
02:35 patspam joined
dukeleto plobsing: nqp-rx is being written by pmichaud and installs a 'nqp' binary instead of 'parrot_nqp', currently 02:35
plobsing dukeleto: I am familiar with nqp-rx (I tend to read the irclogs a lot) 02:36
dukeleto plobsing: there is much discussion of how and when to sync nqp into parrot core, since it has always been in core previously
plobsing: ok, just making sure
plobsing dukeleto: what I am asking for is the commit number on rakudo that made the fix
what is this "minor change" that I keep hearing abou 02:37
t
but I have more important questions: 02:38
dukeleto plobsing: nqp-rx has proper perl 6 regexen
plobsing 1) where should I put my JIT/NCI ideas on the wiki? JITrewrite? L1Recap?
dukeleto plobsing: and string interpolation
purl string interpolation is there more than one kind of interpolation?
dukeleto plobsing: start a new page
plobsing 2) t/pmc/nci.t is currently in Perl. I understand that this should be moved to PIR "soon". Does this block making any changes to the perl version? 02:39
dukeleto plobsing: or if you want, if they fit, on one of the other JIT wiki pages
plobsing: no, you can add perl tests if you need. but I may split that file and start to write some of the tests in PIR :) 02:40
02:40 Zak joined
dukeleto plobsing: we use foo-old.t for perl tests and foo.t for pir tests 02:40
plobsing 3) writing tests in PIR is going to require re-ordering tests to satisfy subsystem dependancies. this seems like a major undertaking. is this reordering imminent?
dukeleto plobsing: not sure about that. I would need more details. 02:41
plobsing: write tests in any way that you need to. perl tests are better than no tests 02:42
02:42 allison joined
plobsing for example: right now, nci functionality gets tested halfway through the test suite. When I mess something up, I don't even get to start tests. 02:42
because it tries to compile PGE first
dukeleto plobsing: do you know about coretest?
plobsing: 'make coretest' runs a minimal test suite 02:43
plobsing no. if it does what I think it does, that would be really handy
dukeleto plobsing: i think it does what you want it to
dalek rrot: r42234 | allison++ | trunk/docs/project/support_policy.pod:
[cage] Make the API listing explicit in the support policy.
dukeleto plobsing: no pge in coretest
plobsing sweet
dukeleto plobsing: allison made coretest when she was working on pcc_reapply, iirc 02:44
plobsing: there is also 'make corevm' for just compiling the basic core vm
plobsing should have looked at the makefile more carefully 02:45
dukeleto plobsing: github.com/rakudo/rakudo/commits/usenqprx 02:48
plobsing: they have a branch going for using nqp-rx 02:49
parrot is down to 108 RT's! 02:50
i am on a mission
plobsing this would be good to show ppl to justify including nqp-rx in core
dukeleto plobsing: the branch? 02:51
purl the branch is ready to go, I want to update it to trunk now and maybe run rakudo spectest run with it first to double-double check
dukeleto kicks purl in the face 02:52
plobsing dukeleto: yes, or a selection of commits from the branch
kid51 purl: touch�
purl kid51: i'm not following you...
pmichaud (rakudo and nqp-rx) -- the commits that were needed to get rakudo running with nqp-rx are the ones on 2009-10-30 02:53
that's it, nothing more
first commit was "Add an s1_setting.pir file", last one was the one that is at the top 02:54
dukeleto pmichaud: ok
pmichaud at that point I believe that rakudo was passing 100% of its spectests using nqp-rx instead of nqp
chromatic If plumage is capable of installing a newer NQP than the one bundled with Parrot, there's little problem depending on an older, bundled version of NQP for backwards compatibility concerns.
pmichaud and the last commit was just optimization, not really needed to get things working 02:56
i.e., it was enabling rakudo to take advantage of a nqp-rx feature that nqp didn't provide
dukeleto pmichaud: seems like minimal changes 02:57
pmichaud they were very minimal
chromatic We're really talking about the risk of NQP making backwards-incompatible API changes versus the pain of plumage having to design, reimplement, debug, and support part or all of NQP. 02:58
dukeleto chromatic: yes.
pmichaud more to the point, NQP's intent has always been to make it possible to create libraries that are written in PIR that could be bundled with Parrot and used even if NQP wasn't present.
if parrot refuses to include libraries where the original source isn't in the parrot repo... well, I'm not sure what to say to that.
chromatic If you want regexes, you have to have *something* present.
dukeleto pmichaud: ok, so should I just add a check in plumage to use 'nqp' before 'parrot_nqp', if it is available? 02:59
pmichaud dukeleto: I don't have any answers at the moment.
dukeleto: I don't know what parrot's answer is going to be.
dukeleto pmichaud: what would you prefer?
pmichaud: i am asking about plumage's answer
pmichaud I would think that plumage's answer will depend on what nqp does, which will partially depend on what parrot decides
if a decision has been made, I just haven't read it yet (I've been pretty distracted today by other concerns) 03:00
dukeleto pmichaud: wheels within wheels
purl Wheels within wheels in a spiral array. A pattern so grand and complex. Time after time we lose sight of the way. Our causes can't see their effects.
dukeleto pmichaud: all comments on parrot-dev have been positive, that i have seen
pmichaud: but more people need to weigh in 03:01
pmichaud: i want to test plumage with nqp-rx. now. what is the easiest way to do that? just hack plumage to use an installed 'nqp' instead of 'parrot_nqp' ? 03:02
pmichaud dukeleto: yeah, that'd be easiest 03:03
just install parrot, install nqp, have plumage use 'nqp' instead of 'parrot_nqp' 03:04
(er, install nqp-rx)
japhb bak for a bit 03:05
dukeleto pmichaud: does 'nqp' get installed into PARROT_BIN or the users path ? 03:06
japhb I can't sanely do some of the stuff I want to do in Plumage without nqp-rx. If Parrot refuses to ship nqp-rx, I will probably just put it into plumage, and it will get into parrot anyway -- except in the wrong way. I'm going for right action here, and end-running is not that. 03:08
pmichaud dukeleto: parrot_bin, of course :)
allison japhb: you could include the .pir files in Plumage, that would work
pmichaud japhb: I've made my proposal for right action. If parrot doesn't want to follow that, nqp will create its own.
allison japhb: that would actually solve the problem all around 03:09
japhb: as in "the Parrot tarball doesn't include any modules at all, just install Plumage, then pick what you want"
pmichaud japhb: I wouldn't be opposed to having plumage as the official mechanism for people to obtain nqp 03:10
(depending on how its policies end up working out)
japhb pmichaud, OK.
pmichaud I mean, ultimately whatever happens has to be convenient for library and hll developers
dukeleto pmichaud: really, in PARROT_BIN ? 03:11
pmichaud: is that kosher ?
japhb I still think this is not ideal, but I am nothing if not willing to compromise.
pmichaud dukeleto: aren't languages supposed to install themselves in parrot_bin ?
japhb dukeleto, that's where all the languages go.
allison yes, that's where languages go
pmichaud dukeleto: where else would they go?
allison (with any supporting libraries in lib/parrot/...) 03:12
dukeleto pmichaud: don't know. just askin'. if that is where languages go, sounds good to me.
pmichaud: means I need to tweak my patch tho...
dukeleto is compiling nqp-rx 03:13
allison japhb: since Plumage is a module installer, you would have the option of either including the NQP .pir files, or pulling them in like you do other module dependencies, whatever works out to be easiest 03:15
03:16 zak_ joined
allison japhb: I don't know what to do in the case where two different modules or languages depend on two different versions of NQP 03:16
japhb: (that's a problem that core Parrot really wouldn't be equipped to handle, but Plumage probably could)
dalek tracwiki: v16 | plobsing++ | JITRewrite
tracwiki: put tasks at the top
tracwiki: trac.parrot.org/parrot/wiki/JITRew...ction=diff
dukeleto japhb: i think plumage is the best solution to the nqp-rx issue 03:17
pmichaud: (kadath)(~/git/nqp-rx master )$ gmake test
prove -r --exec ./nqp t/nqp
Unknown option: exec
gmake: *** [nqp-test] Error 1
$ prove --version 03:18
prove v2.64, using Test::Harness v3.16 and Perl v5.10.0
japhb OK, I will discuss in more detail with pmichaud about how the details, and we'll try to get plumage back on the fast track again.
dukeleto japhb: i think plumage should use nqp-rx as a git submodule
japhb dukeleto, That will definitely be on the list as a possibility. 03:19
dukeleto, right now I just want to be in a least-friction basis between the two projects.
pmichaud is AFK at the moment, but we'll be talking later. 03:20
dukeleto msg pmichaud nqp --version does not give a version number in nqp-rx 03:21
purl Message for pmichaud stored.
plobsing question: how does Lorito relate to PMCs? 03:24
nopaste "dukeleto" at 69.64.235.54 pasted "Plumage's Configure.nqp fails when using nqp-rx" (23 lines) at nopaste.snit.ch/18538
plobsing I would guess we'd want them accessible to jit for inlining etc...
dukeleto plobsing: very carefully
allison plobsing: Lorito doesn't exist yet, but hopefully will be a language used for defining PMCs
(instead of defining them in C)
plobsing i thought lorito was for ops
allison that too 03:25
dukeleto plobsing: pmcs would have to be defined in lorito ops
allison plobsing: basically, it's a small core op set
dukeleto msg pmichaud Plumage's Configure.nqp fails when using nqp-rx" (23 lines) at nopaste.snit.ch/18538
purl Message for pmichaud stored.
allison plobsing: we jit the small set and jit larger chunks composed of the small set 03:26
plobsing couldn't we implement most of our PMC functionality now with NQP
and have NCI pick up the remainder?
then when lorito rolls around its already there 03:27
allison plobsing: we need the PMCs to implement NQP
plobsing: NQP is actually a high-level language, on the same tier as Perl
(it is Perl, basically, just smaller syntax set) 03:28
plobsing ok what about PIR then? or something pretty close to PIR?
allison yes, Lorito is basically a next-generation PIR
japhb plobsing, Note that NQP (and especially NQP-rx) pick the Perl subset that it easiest to compile down to efficient PIR.
plobsing ok. i must have missunderstood nqp 03:29
japhb NQP is efficient and high level at the same time, which makes it nice for writing other HLLs in. :-)
dalek tracwiki: v17 | plobsing++ | JITRewrite 03:30
tracwiki: add note about frame builder
tracwiki: trac.parrot.org/parrot/wiki/JITRew...ction=diff
dukeleto what relation does Lorito have to PIRC ? 03:31
allison dukeleto: no relation 03:32
dukeleto: PIRC is a cleaner implementation of PIR, but just straight PIR
plobsing could it be forked and augmented to support lorito? 03:33
dalek tracwiki: v18 | dukeleto++ | JITRewrite
tracwiki: trac.parrot.org/parrot/wiki/JITRew...ction=diff
dukeleto allison: thanks for the clarification
plobsing: i have serious reservations about this: "Here are some options for backends we can support. We should be able to target all of these, depending on which the user has installed on their system. "
allison plobsing: maybe, the spec for Lorito is not at all defined yet, just an idea 03:34
dukeleto plobsing: i talked to many open source projects and all indicated that supported too many jit backends too early is an induced drag on the project that is not worth it, especially since everyone throws the first *few* jit implementations away 03:35
s/supported/supporting/
03:35 janus joined
allison dukeleto: 8 years in is probably past "too early" 03:35
plobsing I would tend to agree. We should get one working first.
LLVM seems to have a lot of hype. not sure if thats good or bad. 03:36
japhb dukeleto, I just realized: you may be having problems with running Plumage unchanged on nqp-rx because nqp-rx does not ship a small library of helper functions that the original NQP does include. Mind you, it turns out users weren't supposed to directly use that NQP library (it was for internal use), but I had not gotten around to doing as pmichaud asked and copying that library code into Glue.pir. (Mostly because I was waiting for detai
ls on the resolution of the nqp-rx-in-core question.)
plobsing note the backends wording is *not mine*
allison ignore the hype and decide based on features
plobsing: note that LLVM has a *lot* of limitations as a JIT 03:37
plobsing yes. I consider it to be a compiler with some JIT-ish capabilities
dukeleto allison: what are the most limiting factors of jit on llvm?
allison plobsing: and that the LLVM developers freely admit that they haven't put much work into the JIT code
dukeleto: platform support, features, especially features for dynamic languages 03:38
japhb dukeleto, so, as soon as pmichaud and I are online at the same time again, we'll probably get that resolved pretty quickly.
dukeleto japhb: ok, sounds good.
purl sounds good. is there a good way for me to find out when branches are merged, other than read every svn commit?
allison dukeleto: but even in general, their JIT base supports a fraction of the features that their static compilation supports
plobsing I have heard about exponential behaviour in llvm. is there a guaranteed way to prevent this?
dukeleto allison: so why did we decide on llvm for jit? or did we? 03:39
allison dukeleto: but, LLVM still seems to be the farthest ahead of the available options
dukeleto allison: i see.
allison dukeleto: it's a "try it and see" kind of decision, rather than any firm committment
dukeleto allison: the llvm developer that I met was very helpful and gave me lots of good ideas. I still need to get my notes from the gsoc conf organized 03:40
allison: he said that llvm has a way of attaching arbitrary metadata to ops and then allowing a custom stage to optimize based on that metadata
plobsing dukeleto: once you have those notes organized (or even before that), can you throw them up somewhere?
dukeleto allison: so if llvm does not have good jit support, it seems like it should have the tools to make good jit support 03:41
plobsing: yes
allison dukeleto: it has the beginnings of good jit support, plenty to build on
03:42 patspam joined
nopaste "dukeleto" at 69.64.235.54 pasted "very rough notes from the GSoC Mentor Summit on Dynamic Language Interoperability" (88 lines) at nopaste.snit.ch/18539 03:42
dukeleto that should be on the gsoc wiki in session notes, but at least y'all can see it until I get it posted there 03:43
plobsing dukeleto++
I see FFI in there. I was thinking of making a framebuilder with either libffi or ffcall after libjit. 03:44
would this be of interest to parrot?
allison I swear, if this keeps up I'm going to do my PhD on dynamic JIT's instead of dynamic automatic parallelism
dukeleto plobsing: how does Factor do ffi? we should emulate them. I think they use ffcall ? 03:45
allison dukeleto: good notes, thanks
dukeleto plobsing: factor has the easiest-to-use ffi i have ever used
allison: no problem
plobsing have not looked at factor ffi. will do. 03:46
nopaste "dukeleto" at 69.64.235.54 pasted "rough notes from GPU session" (82 lines) at nopaste.snit.ch/18540
dukeleto the GPU session was quite enlightening 03:47
plobsing: factor is a really cool language. I am writing Kea: Factor on Parrot: github.com/leto/kea
plobsing I've been meaning to take a look at it. basically FORTH w/ closures and gc right? 03:48
wrt factor ffi: while i'm sure a more usable FFI system would be cool, I'm looking to implement more backend to our current FFI. 03:51
s/backend/backends/ 03:52
in fact, I'm pretty sure alternate FFI systems would be pretty easy to do as PMCs 03:54
dukeleto plobsing: that is a cool idea 03:55
plobsing: factor is like forth with an interactive IDE, GC, a fast VM and lots of libraries, batteries included
plobsing: currently factor has a VM in C and everything else is written in factor. mostly self-hosting. 03:57
plobsing: the creator's blog is full of good info: factor-language.blogspot.com/ 03:58
plobsing does factor do JIT? 03:59
04:00 JimmyZ_ joined
plobsing recoils in terror uppon seeing C++ 04:00
dukeleto plobsing: oooh. is the VM in C++ ? 04:01
plobsing: i am not sure about their JIT. they are serious about performance, tho. 04:02
plobsing it would appear so. lots of links to .hpp and .cpp files
I pretty much flat-out refuse to do c++
karma c 04:03
purl c has karma of 8170
plobsing it comes up a lot apparently
dukeleto well someone should write c-- 04:04
plobsing: i loath c++ as well
plobsing there is c--. it was being worked on by haskell people as an alternate to LLVM i think. 04:05
www.cminusminus.org/
dukeleto plobsing: that makes me chuckle.
pmichaud back again 04:06
dukeleto plobsing: funded by Microsoft Research and by the National Science Foundation. interesting. 04:07
pmichaud: see your msg's
pmichaud purl messages
chromatic Factor's well worth considering. I've looked at their FFI; we should steal as many ideas as we can.
dukeleto chromatic++
pmichaud I haven't had a chance to add a revision number to nqp yet.
plobsing are we looking to revamp the NCI API?
chromatic I think we should consider it. 04:08
dukeleto plobsing: if it sucks, yes.
pmichaud: most importantly: nqp used to come with some convenience functions, now it doesn't, so plumage's Configure.nqp fails 04:09
pmichaud: japhb said he knew what was up and was going to contact you
plobsing what if it only sucks a little bit?
pmichaud plumage's Configure.nqp should use pir::load_bytecode(...) instead of load_bytecode(...)
then no external library is needed :)
where is plumage hosted?
plumage? 04:10
purl plumage is, like, 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
pmichaud okay, got it.
dukeleto pmichaud: i am working on it now 04:11
pmichaud: we have a lot of load_bytecode's :)
pmichaud that's at least one of the advantages of the new NQP, one can get to pir opcodes without having to have any extra libraries :) 04:12
dukeleto pmichaud: Null PMC access in invoke()
current instr.: 'rx' pc 264 (./src/lib/Glue.pir:326)
pmichaud weird 04:13
that shouldn't be a problem -- PGE is still present
I'll try building it on my system locally 04:14
dukeleto pmichaud: that is happening with my normal parrot_nqp. something is awry.
pmichaud then I have no idea.
dukeleto pmichaud: no wait. forget what I just said. 04:15
nopaste "dukeleto" at 69.64.235.54 pasted "plumage attempting to use nqp-rx" (13 lines) at nopaste.snit.ch/18541 04:16
dukeleto pmichaud: is rx a nqp-rx keyword?
pmichaud not yet
it will be, yes. but shouldn't be the problem.
svn checkouts of parrot are sooooo slooooooow!!! :-( 04:17
dukeleto pmichaud: use this branch: gitorious.org/parrot-plumage/parrot...mits/nqprx 04:18
pmichaud: yes. like running in molasses
we are fast approaching the need to write nqp_debugger 04:23
pmichaud I haven't had time to add good error messages or debugging into nqp 04:24
but yes, this is a good opportunity to start doing that 04:25
dukeleto pmichaud: i don't know how to debug this "Null PMC access in invoke()"
pmichaud try it with --trace=1
or --trace=4
04:26 mikehh_ joined
dukeleto holy secret back door, batman! 04:26
nopaste "dukeleto" at 69.64.235.54 pasted "/Users/leto/git/nqp-rx/parrot_install/bin/nqp --trace=4 Configure.nqp" (37 lines) at nopaste.snit.ch/18542 04:27
dalek rrot-plumage: 00fde77 | leto++ | :
Change all remaining references to parrot_nqp to nqp
purl dalek: that doesn't look right
pmichaud oh
it needs to load PGE
dukeleto pmichaud: how? 04:28
pmichaud patch coming
dukeleto looks like dalek is picking up random gitorious commits. odd.
nopaste "pmichaud" at 72.181.176.220 pasted "patch to get plumage's Configure.pl to run" (16 lines) at nopaste.snit.ch/18543 04:29
dukeleto pmichaud: that worked! 04:30
TimToady pmichaud: Wolfman2000 has 256m on #perl6; is he thrashing?
dukeleto pmichaud: 'it' being Configure.nqp. the build blew up! 04:31
pmichaud TimToady: switching to #perl6
nopaste "dukeleto" at 69.64.235.54 pasted "building plumage goes 'boom' with nqp-rx" (28 lines) at nopaste.snit.ch/18544
pmichaud nqp-rx doesn't recognize pod comments yet 04:32
I can probably add that quickly.
dukeleto pmichaud: THE HORROR ;) 04:33
pmichaud: can I help in any way?
dalek rrot-plumage: b670d08 | leto++ | :
This makes Configure.nqp compatible with nqp-rx, thanks pmichaud++ !
04:37 petdance joined
dukeleto most of llvm's docs give examples in c++ 04:40
this concerns me
plobsing apparently there are C bindings 04:41
dukeleto plobsing: yes, but most of the docs give C++ examples. that means using the C bindings will be harder to learn
plobsing still better than libjit: no stack allocation except on x86 04:42
dukeleto perhaps we should just focus on being a well-tested, functional frame-builder, and see if that governs our decision for how to do jit 04:43
s/being/writing/
plobsing its a more immediate priority
for sure
purl like totally!
plobsing die purl die
purl NYOOMMMMM SPLAT
plobsing I should have a functional libjit+pcc_reapply framebuilder by the end of the night. Feel free to steal any good ideas and re-implement with llvm/whatever 04:45
dukeleto plobsing: how can I help you with the framebuilder? do you want to create a frame-builder branch on github?
plobsing: wow. cool stuff.
04:45 patspam joined
plobsing dukeleto: I already have a github repo with the initial stuff. I am working on merging auto_libjit (kid51++ helped me with that) to work with trunk 04:46
github.com/plobsing/parrot-libjit 04:47
dukeleto llvm.org/docs/ReleaseNotes.html#whatsnew has lots of interesting jit-related stuff
plobsing currently wrapping my head around how pcc_reapply does things
dukeleto plobsing: there are a few wiki pages about it 04:48
plobsing: trac.parrot.org/parrot/wiki/Callin...nsOverview
plobsing docs lie. reading the code. 04:49
dukeleto plobsing: it has a pseudocode description of what is going on at the bottom.
plobsing: but yes, you are correct.
"Use The Source, Duke" is my motto
plobsing good grief! more NCI=>PCC mapping fail! 04:50
what happens when my generated frame gets called with the wrong PCC signauture? probably segfault. 04:51
and it will look like its my fault
dukeleto plobsing: no good deed goes unpunished 04:56
kiwichris is back (gone 03:26:44) 05:10
In 1.7.0 in file compilers/imcc/pbc.c:427 the test fails so I exit. Why might this be happening ?
plobsing error msg? 05:11
purl i heard error msg was englisch anyways
kiwichris plobsing, none that I can see. Just a call to Parrot_on_exit. It is the second check after the allocation in e_pbc_open. 05:13
05:13 cotto joined
plobsing I'm at a loss. IMCC isn't really my thing, sorry. 05:15
kiwichris plobsing, no problem and thanks anyway 05:16
chromatic I don't understand the diagnosis. 05:17
dukeleto kiwichris: hello! 05:23
kiwichris: Parrot_on_exit is giving you issues?
kiwichris dukeleto, no it is deeper. stepping in to find the problem. 05:33
dukeleto, went into PF_create_default_segs and never came back 05:34
05:34 mokurai joined
kiwichris dukeleto, sorry it segfaulted 05:34
dukeleto kiwichris: interesting
pmichaud I have a patch that allows plumage to compile with nqp-rx 05:41
(I don't know if it runs, but it compiles)
where should I send the patch? 05:42
dalek p-rx: 4bb9313 | pmichaud++ | src/NQP/Grammar.pm:
[nqp]: Enable Perl 6 pod comments.
nopaste "pmichaud" at 72.181.176.220 pasted "patch to allow plumage to compile (with latest nqp). for dukeleto++ and japhb++" (281 lines) at nopaste.snit.ch/18545 05:43
"pmichaud" at 72.181.176.220 pasted "complete plumage build with nqp-rx and #18545 patch (for dukeleto++ and japhb++)" (58 lines) at nopaste.snit.ch/18546 05:46
pmichaud looks like the only changes needed to plumage were to add the pir::load_bytecode() lines and to update the pod comments.
(the pod comments were using an obsolete pod syntax) 05:47
afk for a bit
japhb Holy cow, I disappear for two hours, and you guys go nuts. ;-) 05:53
dukeleto, If I'm reading all the backlog right, it looks like the only thing left to commit to the plumage repo is pmichaud++'s changes to the pod formatting, yes? 05:54
pmichaud afaict, plumage runs just fine locally
japhb pmichaud, did you submit a PaFo CLA? 05:55
If so, I can just give you a commitbit. :-)
pmichaud of course I have a PaFo CLA :-) 05:56
I've been a parrot committer since 2004
kiwichris What is a "cross-interpreter" ? 05:59
japhb pmichaud, A lot of people only did the Perl CLA, and the grandfathering has been brought up as an issue. I had to check that you weren't in that case. :-) 06:01
pmichaud japhb: I'm on the Parrot Board of Directors... I'm pretty sure I'm okay :) 06:02
japhb pmichaud, never hurts to ask. ;-) Anyway, what's your gitorious id? pmichaud there as well? 06:03
pmichaud yes
06:04 fperrad joined
pmichaud afk for a while 06:17
dukeleto japhb: yes, looks that way 06:52
i just saw the backlog 06:53
06:59 uniejo joined
dalek rrot-plumage: f9f47a8 | leto++ | :
Attempt to convert over to nqp-rx-friendly POD. Something is still wrong...
 07:14
07:18 bacek joined
dukeleto msg pmichaud how do i use nqp-rx to read/check pod in an nqp file ? 07:20
purl Message for pmichaud stored.
bacek o hai
pmichaud dukeleto: we don't have that tool presently 07:22
just a sec, I'll update and check
dukeleto pmichaud: i am getting odd errors because my pod is messed up somewhere
japhb dukeleto, it looks like you've got some =begin after the actual beginning. 07:23
Like at top of src/lib/Glue.pir
pmichaud ???
nopaste "dukeleto" at 69.64.235.54 pasted "current error that plumage gets with nqp-rx" (13 lines) at nopaste.snit.ch/18547
dukeleto japhb: what do you mean by that? 07:24
pmichaud: =begin is needed after each =end block, correct ?
pmichaud other way around
purl other way around is not better
dukeleto pmichaud: or only at the beginning of the file?
pmichaud =begin and =end are pairs
dukeleto pmichaud: yes, that is what I meant 07:25
pmichaud =begin starts a pod block, that is terminated by the following =end
you can also tag them, thus
=begin XYZ
...
dukeleto pmichaud: so if you look at gitorious.org/parrot-plumage/parrot...228d727b9, that is what I tried to do
japhb You can't have a =head before the =begin
pmichaud =end XYZ
dukeleto pmichaud: gotcha. I just pushed the fix for the top of src/lib/Glue.pir 07:26
but stuff is still broke
pmichaud note that the pod that is in *.pir files isn't the same as the pod that is valid in Perl 6
PIR doesn't understand perl 6 pod
dukeleto pmichaud: oy vey 07:27
pmichaud so yes, if you put perl 6 pod into a .pir file, it will not compile. Or it will think the entire file is a comment.
japhb OK, so perl 5 pod in the .pir files, and perl 6 pod in the .nqp files, yes?
pmichaud yes
and hopefully we can get rid of the .pir files altogether 07:28
should just make them nqp :)
(I'll help with that conversion, too )
do I need to push my patch, or has that been done already? 07:29
japhb I think duke is looking for help with his latest commit, but he may be fixing it himself right now. 07:30
dalek rrot-plumage: 8ce7dad | leto++ | :
Fix a POD issue
rrot-plumage: ac091d4 | leto++ | :
Make src/lib/Glue.pir Perl 5 POD, not Perl 6 POD
japhb Well, that may be our answer. 07:31
07:31 theory joined
pmichaud why are things being pushed without testing first anyway? ;-) 07:31
dukeleto pmichaud: that is on a branch 07:32
pmichaud ahhh
japhb dukeleto, Is everything working on the branch now?
pmichaud even on a branch I'd expect things to get pushed only when working, though :) 07:33
japhb pmichaud, that prohibits getting help branch interaction ...
er, help *via*
as opposed to via pastebins and emails. :-) 07:34
dukeleto japhb: no
japhb dukeleto, care to refine that statement? 07:35
pmichaud I think "no, everything is not working on the branch now"
japhb ah
dukeleto still the same error about not finding slurp 07:36
japhb paste?
purl paste is (see: nopaste) or like glue but a little safer to sniff. or nopaste.snit.ch:8001/ or scsys.co.uk:8001/ anywhere shadowpaste is or mmm, sticky paste or You there! Eating the paste. or <see> 2 girls, 1 paste or App::Nopaste
pmichaud just a sec, trying locally (in the nqprx branch)
nopaste "dukeleto" at 69.64.235.54 pasted "plumage+nqp-rx error" (12 lines) at nopaste.snit.ch/18548
pmichaud so, are we working in the nqprx branch or in master or ... ? 07:37
dukeleto pmichaud: i installed nqp-rx 1ee030e7d0a87
pmichaud okay, that one has the load_bytecode error in it for me
dukeleto pmichaud: and I am on the nqprx branch of plumage
pmichaud: and I did a --gen-parrot on the nqp-rx install 07:38
pmichaud just a sec, I'll start over
japhb rebuilding parrot ...
pmichaud same here
oh, I need to set up authentication on gitorious 07:39
dukeleto pmichaud: there should be a read-only clone url, no? 07:40
pmichaud: or did you want to commit?
pmichaud yes, but I want to be able to push fixes andupdate also
dukeleto pmichaud: japhb can give you a commit bit, i think
japhb dukeleto, already done. He needs to set up his keys
dalek rrot-plumage: 3853bd8 | leto++ | :
Make Plumage::Downloader Perl 5 POD
 07:41
pmichaud okay, got my keys set up, now building stuff 07:43
japhb uses plumage master to install nqp-rx, before switching to branch ... :-) 07:45
pmichaud okay, I'm confused :) 07:46
which branch has the current changes in it?
japhb nqprx
pmichaud okay
It'll take me a bit to catch up, I want to improve the --gen-parrot options in nqprx and rakudo 07:47
07:50 bacek joined 07:51 mikehh joined
japhb Well, I can certainly confirm dukeleto's problem with the 'Could not find non-existent sub slurp' error ... it's as if it's not successful in loading the Glue.pir library. 07:58
OK, I need to sleep, I'll try to check back in the morning. 07:59
dukeleto++ and pmichaud++
g'night all.
pmichaud aha! "make test" 08:01
Glue.pir is missing the =cut lines for the pod
I have it fixed locally, compiling now
pushed. 08:02
dalek rrot-plumage: 174ea49 | pmichaud++ | :
Put back the =cut lines in Glue.pir. Otherwise it thinks the entire file
 08:03
japhb (not quite gone yet) 08:04
Fixing Metadata.nqp
pmichaud I already fixed that. 08:05
and pushed it.
I'm looking at the tests now.
fixed the test harness 08:06
japhb OK, really really going to bed then. :-) 08:07
pmichaud I'm getting...
pmichaud@plum:~/plumage$ make test
/home/pmichaud/parrot/install/bin/nqp t/harness t/*.t
t/glue.t...INVALID PLAN: /bin/sh: nqp: not found
t/sanity.t...INVALID PLAN: /bin/sh: nqp: not found
FAILED 2 files, PASSED 0 tests
08:07 iblechbot joined
pmichaud does nqp have to be in my PATH ? 08:07
dalek rrot-plumage: 3ef7389 | pmichaud++ | :
Put =begin markers back at top of *.nqp files.
 08:09
pmichaud message japhb okay, I have plumage such that it passes all of the tests in t/glue.t and t/sanity.t when run directly with nqp. It doesn't run t/harness properly for some reason though -- I'll let you track that done. 08:17
purl Message for japhb stored.
08:17 TiMBuS joined
pmichaud message japhb s/done/down/ 08:17
purl Message for japhb stored.
dalek rrot-plumage: cab5aca | pmichaud++ | :
Update t/glue.t, t/sanity.t, t/harness for nqp-rx.
 08:20
08:36 zak_ joined 08:49 xenoterracide joined 08:52 xenoterracide joined 08:58 xenoterracide_ joined 09:17 viklund joined 09:24 plobsing joined
dalek rrot: r42235 | mikehh++ | trunk/docs/pdds/pdd19_pir.pod:
fix codetest failure - pdd_format - lines > 78 chars
 09:27
09:38 gaz joined 09:40 fperrad_ joined
dalek p-rx: f1cda32 | pmichaud++ | (7 files):
Rework the build process a bit; remove some outdated files,
 09:43
p-rx: 23cd363 | pmichaud++ | build/Makefile.in:
Makefile fix.
 09:44
p-rx: 24654af | pmichaud++ | src/ (3 files):
Convert 'action' to 'actions', per recent S05 change.
09:47 frodwith joined
mikehh All tests PASS (pre/post-config, smoke (#29616), fulltest) at r42235 - Ubuntu 9.10 amd64 (no --optimize) 09:47
with --optimize t/src/warnings.t fails test: 2 - smoke (#29617), fulltest at r42235 - Ubuntu 9.10 amd64 09:53
all other tests PASS - TT #1187
dalek p-rx: 5910ede | pmichaud++ | src/Regex/P6Regex/Grammar.pm:
Add some appropriate nyi warnings for :: and :::, rather than
 10:07
purl it has been said that silently failing is the worst possible option
10:10 frodwith joined 10:28 khisanth_ joined 11:41 kiwichris joined 11:56 kid51 joined 12:32 Coke joined 12:39 masak joined 12:49 petdance joined
dalek rrot: r42236 | jkeenan++ | branches/configtests:
Create branch to explore whether trac.parrot.org/parrot/ticket/1189 can be implemented.
 13:04
13:10 Psyche^ joined 13:14 whiteknight joined 13:15 Psyche^ joined 13:21 Psyche^ joined
whiteknight good morning #parrot 13:27
13:34 tetragon joined 13:38 fperrad_ joined
dalek rkdown: f5e33c6 | unknown++ | (22 files):
convert all tests to PIR (from Perl)
 13:49
purl I don't know how to convert all tests to PIR (from Perl).
moritz somebody should git-config his/her user name :-) 13:50
(re: last markdown commit by unknown)
whiteknight somebody should teach purl to convert tests to PIR 13:58
:)
13:59 patspam joined 14:05 Psyche^ joined 14:14 iblechbot joined
dalek rkdown: cc9d938 | fperrad++ | t/ (19 files):
chmod +x *.t
 14:18
rkdown: f87b15e | fperrad++ | config/makefiles/root.in:
fix on linux
14:55 ilia joined 15:00 fperrad joined 15:03 bubaflub joined 15:12 fperrad joined 15:19 fperrad_ joined 15:25 fperrad joined 15:29 Psyche^ joined 15:31 kj joined
kj good afternoon #parrot 15:34
15:35 theory joined
Coke RT: 107 tickets. 15:37
15:37 fperrad joined
Coke whiteknight, kid51: I'm taking 37934, since i think it's worth keeping as a todo, i'll copy it over and change it into something a little more actionable. 15:38
whiteknight Coke: okay. That's fine by me. As is, it's really too vague 15:39
but if you think you can make it actionabe, I have no objections
Coke moritz: can you look into closing/migrating/resolving RT #57190? 15:42
15:45 fperrad joined
moritz takes a look 15:47
I've seen HLL interop between rakudo, cardinal and blizkost/perl5 15:48
that's enough for me to close the ticket, if nobody objects
dalek rrot: r42237 | coke++ | trunk/docs/pdds/pdd07_codingstd.pod:
Rejected ticket, removing comment.
 15:49
Coke parrotcode@gmail.com?
kjs?
purl kjs is Klaas-Jan Stol <mailto:parrotcode@gmail.com> from The Netherlands or KHTML (read Safari/Koqueror)'s JavaScript engine... or called kj these days.
kj hi 15:51
Coke: lookin for me?
Coke kj: yes. you now have a ticket assigned to you in RT. 15:53
(one you had opened)
can you review it, make sure you still want it or see if it's resolved or move it to trac?
kj i'll have a look
yes
Coke danke.
moritz RT #57190 resolved, Coke++ for bothering me 15:54
16:01 fperrad joined
dalek TT #1190 created by kjs++: Replace .globalconst/.const directives by .const/.localconst ... 16:01
kj Coke: should i resolve RT ticket? 16:03
how to close it?
or doesn't it matter?
kj is afk # compulsory coffee break 16:04
dalek rkdown: da63cbc | unknown++ | (2 files):
clean up
 16:06
purl clean up is a breeze
16:09 fperrad joined
dukeleto zomg good morning 16:10
16:13 jhelwig joined 16:14 davidfetter joined
Coke kj: if you moved it, reject it on the old system, it's the best choice of a bad lot. 16:17
(since the only 2 choices are reject or resolve, and we didn't resolve it.)
kj++
16:24 fperrad_ joined
Coke kj: you should also close out the original ticket with a comment pointing to the trac ticket, cc'd to the list. 16:25
danke.
16:32 buildbot joined
Coke kj: whoa, you actually converted the individual comments. awesome. 16:34
16:34 fperrad joined
Coke japhb? 16:38
purl rumour has it japhb is Geoffrey Broadwell, mailto:geoff@broadwell.org
Coke msg japhb: I assigned two RTs to you as the original requestor; please review them and see if we still need them, and if so, move them over to trac, ala my posting to parrot-dev; danke. 16:39
purl Message for japhb stored.
Coke notfound? 16:41
purl rumour has it notfound is Julián Albo <mailto:julian.notfound@gmail.com>
NotFound Yes?
Coke was double checking your email. can you examine/migrate/resolve/reject RT # 58988 ? 16:42
NotFound Looking...
Coke seen infinoid 16:43
purl infinoid was last seen on purl 5 hours, 51 minutes and 42 seconds ago, saying: <private message>
Coke infinoid?
purl hmmm... infinoid is Mark Glines <mailto:mark@glines.org> or likes shiny things or the master of the universe
Coke msg Infinoid I assigned 3 RTs to you that you originally opened. if you can close/transfer to trac/resolve them, per my msg to parrot-dev, it woudl be a great help. Thanks. 16:46
purl Message for infinoid stored.
Coke barney?
purl barney is probably a big, purple piece of shit, or gettin' jiggy wid it or see 'grimace'. or purple dupa or o/` I love you / You love me / We're a happy family / with a great big hug /and a kiss from me to you / Won't you say you love me TOO! /`o
Coke bernhard?
purl bernhard is probably mailto:Bernhard.Schmalhofer@gmx.de
Coke msg bernhard I assigned 3 RTs to you that you originally opened. if you can close/transfer to trac/resolve them, per my msg to parrot-dev, it woudl be a great help. Thanks. 16:47
purl Message for bernhard stored.
Coke wonders if jonathan@jwcs.net is also jonathan. 16:57
NotFound Coke: done 16:58
Coke NotFound++ 16:59
NotFound Coke++
dalek TT #1191 created by NotFound++: Deprecate Parrot_get_runtime_prefix 17:00
dukeleto good moroning 17:22
is bernhard still our current release manager? I haven't seen him in a while. 17:23
dalek rrot: r42238 | NotFound++ | trunk (2 files):
[nci] add a few more signatures, add a function to Xlib example module
 17:24
rkdown: 889fa2a | unknown++ | (3 files):
convert Configure in PIR (from Perl)
 17:42
rkdown: 3306295 | unknown++ | config/makefiles/root.in:
remove $(PROVE)
fperrad ping japhb 17:47
purl I can't find japhb in the DNS.
whiteknight fperrad: is that you committing to markdown? 17:51
fperrad msg japhb could you update markdown.json, now Configure & tests are in PIR (remove Perl module dependencies) 17:52
purl Message for japhb stored.
fperrad Whiteknight, yes, after an update of git, I lost my configuration, so I am unknown 17:53
whiteknight fperrad: okay, just making sure you are aware :)
17:54 barney joined
dukeleto barney: hello! 17:54
barney hi 17:55
Any secret hints for doing a Parrot release? 17:56
moritz 1) read the release manager guide 17:57
pmichaud #ps in 33.
moritz 2) ...
3) profit
allison barney: start early
purl start early is their motto.
allison barney: as in, there's stuff that's really best done a couple weeks in advance, and it's a lot easier to do it in little steps 17:58
barney: (though, it's also perfectly possible to do it all in one day, just harder)
cotto_work barney, especially NEWS updates 17:59
barney Yes, I'll start on NEWS this weeking, being a bit out of touch in the last months 18:00
cotto_work mining #ps for NEWS wouldn't be a bad idea 18:01
actually, I wish I'd thought of that while I was doing the release
Coke barney: you have old rts assigned to you. (i messaged bernhard, not barney) 18:03
barney I'll take a look at the old RT 18:04
Coke danke. 18:05
dukeleto barney: i was the most recent release manager. i can give you some tips, but I need to go right now. I will be back in ~1 hr
barney: perldoc docs/project/release_manager_guide.pod 18:06
barney: there are some places where it is wrong or incomplete, though.
barney dukeleto: k
Coke RT: 102 18:12
18:13 joeri joined 18:20 kj joined
Coke anyone using perl6 on windows? 18:22
I think RT #60098 is closable.
pmichaud Coke++ # closing RT tickets 18:26
#ps in 3 18:27
Coke pmichaud: you have several assigned to you. 18:30
;=)
RT: 100 18:31
pmichaud #ps in -1
Coke seen chromatic? 18:32
purl chromatic was last seen on #parrot 13 hours, 15 minutes and 19 seconds ago, saying: I don't understand the diagnosis.
Coke seen allison? 18:34
purl allison was last seen on #parrot 35 minutes and 15 seconds ago, saying: barney: (though, it's also perfectly possible to do it all in one day, just harder)
18:42 theory joined
Coke msg chromatic - you have a roadmap ticket open for 1.8 18:46
purl Message for chromatic stored.
18:47 darbelo joined
pmichaud I'm afk for a while -- gotta fetch food for the family. 18:51
whiteknight family++
jonathan food++ 18:52
whiteknight I suspect a lot of other people will show up to the meeting in 1 hour 18:55
Coke too late for them!
darbelo DST?
purl DST is political, hence is localized all over the world. or daylight savings time or Devel::SimpleTrace or Daylight Saving Time or Direction de la Securite du Territoire ...
whiteknight (daylight savings)--
darbelo just plain ol' forgot. 18:56
Coke ah, but there IS NO DST NOW! 18:57
MUAHAHAHA
whiteknight now the clock in my car is going to be wrong for the next 6 months 19:00
...and the clock on my microwave is still going to be wrong like it always is 19:01
Coke whiteknight: I'm sure google can help you. =-)
Util whiteknight: had the same problem. The Interwebs might know your car manual.
Coke thankfully has memorized how to update his saturn.
NotFound Coke: With the new ring?
cotto_work just push buttons randomly until something happens 19:02
whiteknight I think I know how to update the clock in the car, I just can't be bothered
dalek rrot: r42239 | allison++ | trunk/src/extend.c:
[cage] Refer to actual ticket for deprecated calling API functions.
whiteknight plus, it's hard to do it when I'm screaming down the highway at 80mph
Util Saturn++. GM sale FAIL. :(
19:02 bacek joined
cotto_work I'll take a post-2.0 release if someone wants to pencil me in 19:04
whiteknight cotto_work: sure, which one?
darbelo: and when would you like to try it?
cotto_work 2.1 or 2.2, either is fine 19:05
whiteknight 'salright
darbelo which month is which? 19:06
whiteknight 2.1 = feb, 2.2 = march
I think
cotto_work yes
darbelo Is feb available?
whiteknight yes 19:07
darbelo Sign me up, then.
whiteknight done
cotto_work: you're march
darbelo goes looking for questionable names
Coke whiteknight: please also update the calendar. =-) 19:08
mikehh I.m available to help - I don't think I want to take on the role though
Coke mikehh: running fulltest frequently is a big help. =-)
cotto_work whiteknight, awesome 19:09
dalek rrot: r42240 | whiteknight++ | trunk/docs/project/release_manager_guide.pod:
[release] darbelo++ has 2.1, cotto++ has 2.2
 19:12
tracwiki: v1 | allison++ | PCCMigrationNotes
tracwiki: trac.parrot.org/parrot/wiki/PCCMig...ction=diff
rkdown: 10ebc01 | fperrad++ | Configure.pir:
better handling error with exception
 19:13
allison oh, rats, did I get the timezone wrong? 19:17
whiteknight an architect is never late
she always shows up precisely when she means to
allison ah well, I'll paste my report anyway
whiteknight allison++
(/me has fallen for this trick before, I think we should post an announcement when daylight savings happens) 19:18
cotto_work is glad his phone can schedule events according to UTC 19:19
allison so, to double-check, was it 1 hour ago? 19:20
pmichaud yes
allison (well, 40-something minutes)
whiteknight yes
pmichaud 18:30 utc
purl 18:30 utc is not always 12:30 CST
pmichaud right now it's 19:20 utc
whiteknight truer words were never spoken, purl
purl whiteknight: i'm not following you...
NotFound even purl knows ;)
allison okay, so my google calendar is right, it just didn't alert me as usual for some strange reason
cotto_work clock?
purl cotto_work: LAX: Tue 11:20am PST / CHI: Tue 1:20pm CST / NYC: Tue 2:20pm EST / LON: Tue 7:20pm GMT / BER: Tue 8:20pm CET / IND: Wed 12:50am IST / TOK: Wed 4:20am JST / SYD: Wed 6:20am EST /
allison (and my brain is wrong)
particle uses foxclocks 19:21
allison particle: looks interesting
darbelo drives a stake into the ground and compares the angle fo the shadow to his compass. 19:22
Coke particle: you have several tickets assigned to you in RT; if you can fold/spindle/mutilate them, it woudl be very helpful. Thanks.
NotFound Has the type of n_regs_used changed recently?
pmichaud if you don't see a shadow, does spring come early? ;-)
allison what's our priority this week? the one listed in #parrotsketch channel is old
cotto_work darbelo, speaking of which I saw an interesting non-computer implementation of a tic-tac-toe AI: shorttermmemoryloss.com/menace/ (scroll down to the tic-tac-toe picture) 19:23
allison NotFound: shouldn't have
Coke allison: by default, closing out RT.
NotFound There is a problem with mixing INTVAL and UINTVAL. c++ is intolerante to mixed pointer to those. 19:24
The problem is in Parrot_pcc_calculate_registers_size 19:25
allison Coke: okay, set
Coke: (and I closed several off my list this week)
darbelo pmichaud: spring every night! 19:26
cotto_work: Ohh. Clever.
19:27 chromatic joined
NotFound allison: Some objection to change the signature of that function for INTVAL* to UINTVAL* ? 19:28
Coke NotFound: is it in the public API? =-) 19:29
allison NotFound: we tend to avoid UINTVAL whenever possible, the motivation for the change?
19:29 athomason joined
NotFound allison: just that the values are UINTVAL 19:29
And I tend to avoid add castings to fix mistakes. 19:30
allison NotFound: they're going to be cast at somepoint, since INTVAL is the standard type, might as well be cast right there at the source
NotFound: or, are they immediately cast back to UINTVAL? 19:31
NotFound allison: They are UINTVAL always.
dukeleto guess I missed #ps
Coke dukeleto: yes. post your report anyway. =-) 19:32
dukeleto Coke: i posted it before :)
NotFound This is not a user friendly function, is an internal affair.
allison NotFound: I'm asking what context the function is used in, if it returns an INTVAL that is always immediately cast as a UINTVAL, then it might be worth changing
NotFound: if it's returning an INTVAL that's always used as an INTVAL, then it makes sense as is 19:33
NotFound allison: is a function that takes a pointer to an array of UINTVAL, usesd as UINTVAL
bacek Good morning
Coke cotto_work: ww?
purl ww is probably wrong window or 2/3 of www
NotFound And mistaken types just forces to uses cast for no good reason. 19:34
cotto_work Coke, ww indeed
barney Anybody cares for runtime/parrot/library/YAML/Parser/Syck.pir ?
dukeleto barney: what do you mean "care" ? 19:35
Coke no. let's rip YAML out of the repo.
allison NotFound: Parrot_pcc_calculate_registers_size?
NotFound allison: yes
barney care to keep it
allison NotFound: it's a static function called once in src/call/context?
dukeleto barney: also, I am back, if you have questions about releasing
NotFound I don't think someone wants to use a negative register size.
allison NotFound: it returns a size_t
NotFound allison: the return value is not a problem, the registers used are. 19:36
allison doing some code spelunking 19:37
Coke allison: you own more than 6% of /all/ open RTs. =-)
allison NotFound: what is ctx->n_regs_used typed as?
dalek rrot: r42241 | barney++ | trunk (5 files):
Removed runtime/parrot/library/YAML/Parser/Syck.pir, nobody needs it
 19:38
allison Coke: hey, it's about 1/4 fewer than last week :)
NotFound allison: the number of registers used in a context, wich are UNITVAL
Coke allison: but your percentage is going up!
allison Coke: then I'll have to get cracking
NotFound Better I'll svn log that file to see why changed. 19:39
Coke allison: you also have 53452 and 57028 that you opened that are unowned. if you could handle those too, that would be helpful. 19:40
allison NotFound: okay, it's a UINTVAL array
Coke: sure, I'll add them to my list
barney dukeleto: I'm just reading release_managers_guide.pod. Can you give me, Barney, admin privs on www.parrot.org/ ? 19:41
dukeleto barney: i don't know if I have that power 19:42
allison barney: it's editor privs you need
barney: I'll wave the magic wand
barney: are you 'barney' on www.parrot.org?
dukeleto allison: can you help us update the latest parrot link? it is currently pointed at 1.6.0
NotFound bacek: ping 19:43
allison dukeleto: yes, will do
bacek NotFound, pong
dukeleto allison: thanks.
allison: i would like to add "live release" tests that we can run after a release, that verify that certain actions have happened. such as the latest link pointing to the correct place.
allison dukeleto: did we not get you set up with editor privs for the release? that should be all you need to update that link.
NotFound bacek: some reason to change UINTVAL* to INTVAL* in Parrot_pcc_calculate_registers_size at r42197 ? 19:44
dukeleto allison: i don't think i had them for the release, or I couldn't figure out how to follow the instructions in the release manager guide
barney allison: It's Barney, with capital B
allison dukeleto: updated release/current and release/developer
dalek rrot: r42242 | barney++ | trunk/docs/project/release_manager_guide.pod:
Editor privs are needed on �www.parrot.org/.
allison barney: okay
dukeleto allison: thank you very much. 19:45
allison barney: I don't see a Barney, but I do see a barney (with member privs)
barney: so I'll grant that user editor privs 19:46
bacek NotFound, erm... Looks like my mistake. n_reg_used should be UINTVAL. OTOH in many places in core we inconsistently use INTVAL in call.
NotFound bacek: that breaks c++ build.
dukeleto barney, allison: we need to add sending out messages to the @parrotvm account on twitter and @parrot on identi.ca, via ping.fm . I will write a script to mostly-automate this
NotFound bacek: I'll fix it, then.
dukeleto barney, allison: that needs to be added to the release manager guide, i mean 19:47
bacek NotFound, you'll have to fix few other places as well
allison dukeleto: good idea
purl allison: Good Idea: Playing cops'n'robbers in the park. Bad Idea: Playing cops'n'robbers in the bank.
barney allison: yes 'barney' is it
NotFound bacek: yeah, that's why I was asking.
allison barney: okay, done, try it out
dukeleto barney: i think i fixed most errors in the release manager guide, but there are probably still some omissions
allison barney, dukeleto: I also checked, and I hadn't enabled permission for editors to change redirects (those custom links). I added that permission now. 19:48
dukeleto barney: also, I wrote some scripts to automate some of the steps. I will clean these up and add them to the repo
barney allison: looks good, tnx
dukeleto allison: yeah, that is why the directions didn't make any sense. I didn't have any of those links available to me :)
allison: thanks for fixing that.
Coke Anyone want a google voice invite?
dukeleto Coke: sure
allison dukeleto: you have editor, member, and poster privs
bacek allison, I've got question about Context and CallSignature merging. Who have to create and push new "CallContext"? build_sig_object, invoke_from_sig or Sub.invoke?
dukeleto allison: awesome!
purl i think awesome is a window manager or at awesome.naquadah.org or awesome!
Coke duke, preferred email? 19:49
allison dukeleto: (just checking to make sure while I was there)
dukeleto Coke: jaleto@gmail.com
Coke naquadah? like from stargate?
dukeleto: done.
dukeleto Coke: many thanks.
allison bacek: at that point, it would be build_sig_object that pushes the context
bacek: since it would have to exist much earlier 19:50
bacek: and also set_args and set_returns (instead of creating a FIA signature)
bacek allison, ok, I'll try to implement it.
allison, set_args invoke build_sig_object, so it will push new Context
allison bacek: that likely can't go in until after 2.0 (depending on how conservative you make it) 19:51
bacek: that'll work for now, and conservative enough to be safe
bacek allison, I'll preserve external APIs anyway.
allison, so it should be pretty "conservative" :)
japhb fperrad, does markdown need perl5 at all? Are you using it during build or install?
allison bacek: sounds good 19:53
bacek allison, I just need little bit more free time...
allison NotFound: since it's a static function, only called once, and always called passing in a struct member that's a UINTVAL *, I can't see any reason not to change the function signature to UINTVAL * 19:54
NotFound allison: already done
dalek rrot: r42243 | NotFound++ | trunk/src/call/context.c:
[cage] fix a INTVAL* / UINTVAL* mistake in registers size calculations
rrot: r42244 | barney++ | trunk/ChangeLog:
Consistently use tabs in ChangeLog.
rrot-plumage: 36f0730 | japhb++ | :
[METADATA] Update markdown.json to reflect Perl 5 -> PIR conversions,...
allison NotFound: great minds think alike :) 19:55
NotFound allison: I was asking just because there was other function implied, and because maybe there was a reason for that change.
I must have done the svn log before. 19:56
bacek NotFound, check Parrot_push_context. 19:58
allison NotFound: makes sense
bacek NotFound, we have to check few other places to use UINTVAL for regs consistently. 19:59
allison Coke: any thoughts on where to put a link to the PCC Migration Guide wiki page?
NotFound bacek: I'm checking only c++ build ATM, to be able to deal with TT #1110
Coke allison: from Deprecation ? 20:00
dalek rrot-plumage: 8d8c2f1 | japhb++ | :
[META] TODO: Convert Configure to PIR
Coke which is a generic page for that sort of thing? (switching from old usage.)
allison Coke: not sure we have one 20:01
dalek rrot: r42245 | barney++ | trunk/docs/project/release_manager_guide.pod:
Editor rights on www.parrot.org.
allison Coke: but Deprecation seems closest, will put it there 20:03
chromatic Did anyone volunteer for 2.0? 20:04
NotFound Problem in TT #1110 is: Parrot_str_unescape takes a const char * pointer to enc_char, and *writes* to it. The strchr issue is a symptom, not the real problem.
darbelo chromatic: Nopes. Coke indicated that maybe somebody on the board would handle that one. 20:05
chromatic I'm available.
dalek tracwiki: v10 | allison++ | Deprecation
tracwiki: trac.parrot.org/parrot/wiki/Deprec...ction=diff
whiteknight chromatic want it?
Coke allison: no, Deprecation, which is... 20:06
(i was answering, not asking. =-)
chromatic I'll do it.
whiteknight done
Coke chromatic++ #taking one for the team.
allison Coke: got it
chromatic: as release manager? I thought I did, but not tied to it 20:07
chromatic: in fact, I think I have exams that week
NotFound Fixing Parrot_str_unescape for TT #1110 requires changing his signature or allocate a temporary C string. What's the prefered way?
allison chromatic++ #taking one for the team. 20:08
dalek tracwiki: v11 | allison++ | Deprecation 20:09
tracwiki: trac.parrot.org/parrot/wiki/Deprec...ction=diff
chromatic What writes to enc_char there, NotFound?
Line 2803?
NotFound *p = strchr (enc_char ...... *p= '\\0'
chromatic: yes
chromatic I realized that as soon as I typed the question. 20:10
NotFound p can't be const char * because of that, and thus the strchr problem
chromatic Pointer aliasing bites again! 20:11
darbelo Pointer aliasing bites. 20:12
NotFound The easy solution is to change the function signature. In fact is being used as non const right now.
chromatic I don't think that's the right solution.
I don't want to change someone's C string out from under them.
NotFound The right solution is to kill that nasty function and write a sane way to unescape 20:13
darbelo NotFound: Ticket it!
Coke RT?
purl RT is just RT (bestpractical.com/rt) or (:rt3) or (: rt bugs) or Obra's trouble ticketing system or the first IBM RISC workstation (www.contrib.andrew.cmu.edu/~shadow/ibmrt.html) or the bombsquad or the Right Thing or very very capable and open-source or an application framework that bundles a ticketing system or obra's baby or SOOOO slow :-S or email mailto:perlbug-owner@perl.org for access
NotFound chromatic: I don't want to, is the function whose doing it.
chromatic I don't see a way to fix this without splitting the string into two parts. 20:14
Have I mentioned that C's string handling is the source of much pain and many bugs?
dalek rrot: r42246 | whiteknight++ | trunk/docs/project/release_manager_guide.pod:
[release] chromatic++ has 2.0
NotFound chromatic: I was thinking the same.
whiteknight chromatic: I'm sure you're not the first person to mention that 20:15
chromatic Sometimes these things can go hidden for 40 years.
whiteknight shit, there are entire classes of exploits named after C's shortcomings
dalek TT #1192 created by japhb++: [RFE] Redesign Parrot NCI callback functionality
NotFound Well, I'm going to the way of allocating a temporary C string, then. If that slow down pir parsing, maybe someone gets motivated to improve things }:)
chromatic Now I have an old gospel song stuck in my head about Moses and the desert.
A better approach is for the PIR compiler to split the names of encodings and character sets into separate parts of string metadata. 20:16
whiteknight string.h should have been named <bufferoverflow.h>
chromatic ... but this is not me volunteering to fix that in IMCC.
NotFound chromatic: anyway, this function needs to be fixed while it exists,
chromatic I agree. Are you going to do it or am I? 20:17
NotFound I'm on it right now.
chromatic Thanks!
20:19 mokurai joined
NotFound Provided that encoding names aren't supposed to be infinite long, we can avoid allocating a temporary by using a fixed size buffer. 20:19
chromatic As long as there's a "Please don't overflow!" check, that's fine with me. 20:22
Coke didn't we have a ticket for adding --pirc==<imcc|pirc> to Configure.pl ?
chromatic I thought we did, Coke. 20:23
Coke cannot find it.
Coke wasn't looking in new tickets, for some reason.
japhb Coke: I think I've dealt with the RTs you assigned to me. 20:24
dalek rrot: r42247 | barney++ | trunk/PLATFORMS:
Report success on linux-x86-gcc4.4.1.
Coke japhb: excellent.
now pick 2 more and do those. =-)
japhb++
japhb Coke, nice try. :-) 20:26
Coke japhb: excellent.
now pick 1 more and do that one!
japhb++
japhb LOL 20:27
dukeleto, OK, I'll be a little distracted by $work (shocking, I know), but I've got a few spare cycles now to discuss the Plumage nqprx branch. 20:28
NotFound Same strchr problem in IMCC_string_from__STRINGC 20:30
20:30 desertm4x joined
chromatic Could that function use Parrot_str_unescape() directly, NotFound? 20:32
I skimmed it and it looks possible.
NotFound chromatic: it does
That functions picks the charset and unescpae picks the encoding, both by writing to memory supposedly const 20:33
chromatic Oh, I see where I went wrong. 20:34
dukeleto japhb: ok 20:37
whiteknight writing to const memory could wreak havoc with some optimizations 20:40
NotFound whiteknight: or just kill the process if someone pass memory really const to that function. 20:41
japhb dukeleto, well, OK then. Um ... the first item is that pmichaud pointed out the test harness won't work unless nqp is in the path. Should probably fix that to use the parrot bin dir.
whiteknight right
20:41 AndyA joined
japhb dukeleto, can you do that one? 20:42
20:43 AndyA joined
Coke AndyA: cheers. 20:44
kiwichris dukeleto, still look for the segfault with RTEMS 20:45
dukeleto kiwichris: what is up?
purl A direction away from the center of gravity of a celestial object. or the y-axis, unless you're using a strange coordinate system.
dukeleto swings an axe in the direction of purl's face 20:46
kiwichris dukeleto, running fir.pir there is a segfault using 1.7.0
dukeleto kiwichris: but not with 1.6.0 ?
kiwichris dukeleto, correct
dukeleto kiwichris: or segfaults all around?
kiwichris: INTERESTING
kiwichris dukeleto, no idea just one at the moment
dukeleto kiwichris: that is a bit unexpected. what about any of the other benchmark scripts? 20:47
kiwichris dukeleto, the trail is break on first call to create_seg
dukeleto kiwichris: can you nopaste some of the output?
kiwichris dukeleto, no idea but will try
dukeleto, no real output at the moment.
dukeleto, it is in the call string/api.c:572 20:48
dukeleto kiwichris: does ./parrot --version work on 1.7.0 ? 20:49
kiwichris dukeleto, this leads down to to_encoding and the iterator init
dukeleto, will try and report back
20:49 mikehh joined, masak joined
dukeleto kiwichris: hmmm 20:49
20:51 bacek joined
NotFound IMCC_string_from__STRINGC can't be easily fixed. Given that is internal to IMCC and used only in the same file that defnes it, the best way is to change his signature. 20:53
chromatic It's probably copied the strings anyway. 20:54
kiwichris dukeleto, the --version option works
NotFound Don't worry, g++ will prevent mistakes
dukeleto kiwichris: that is good 20:55
kiwichris dukeleto, tried parrot examples/pir/life.pir and the same fault
dukeleto kiwichris: what about ./parrot examples/tutorial/01_temp_var.pir ? 20:56
kiwichris: that is a simpler script
20:56 mikehh joined
kiwichris dukeleto, the call string/api.c:572 dies down in string/encoding/utf8.c:601 20:56
dukeleto, same 20:57
dukeleto kiwichris: ENCODING_ITER_INIT is blowing up? 20:58
kiwichris: can you provide a backtrace?
kiwichris: also, are you using libicu ?
kiwichris: parrot can be configured with/without libicu
kiwichris dukeleto, yeap. No the thread faulting thread is deleted before I can get to it. Been getting closer to itt
dukeleto, what is libicu ? 20:59
darbelo kiwichris: unicode lib.
dukeleto kiwichris: library for unicode+friends
darbelo You can live without it.
On an embedded OS, you're better off without it.
kiwichris dukeleto, I doubt we have this. Maybe I have hacked out some config and broken it.
darbelo, agreed. 21:00
dukeleto kiwichris: perhaps parrot thinks it has libicu, but it doesn't
darbelo dukeleto: could happen, but I'm not sure how to determine that without parrot_config 21:01
kiwichris dukeleto, maybe, just checking the config files
darbelo kiwichris: Ah, I remember now, is PARROT_HAS_ICU #defined to anything?
kiwichris darbelo, parrot_config is the original Linux 64bit one so is not current
dukeleto kiwichris: grep has_icu config_lib.pasm
darbelo dukeleto: I'm no entirey sure that's reliable info when cross compiling 21:02
kiwichris dukeleto, set P0["has_icu"], "0"
dukeleto darbelo: i am not sure
kiwichris: so it seems that parrot thinks it does not have libicu. i am not sure if the correct codepaths are still being taken, tho 21:03
kiwichris: what about ./parrot editor/skeleton.pir ? 21:04
kiwichris dukeleto, should it end up in string/encoding/utf8.c
dukeleto kiwichris: that is a file with only a main function declared, nothing else
kiwichris: I really don't know. that is deep in the guts, which I haven't explored before
Coke allison: you're up to 6.25%!
kiwichris dukeleto, hee I am :)
dukeleto, same fault with the skeleton.pir 21:07
Coke who was hacking on the debugger in the last year. dukeleto?
dukeleto Coke: yes.
Coke dukeleto: you get RT# 46123!
21:07 chromatic joined
dukeleto kiwichris: so it seems that we have a low level issue with opening a pir file, on RTEMS 21:08
kiwichris: the commandline options work, but running an essentially empty file blows up
allison Coke: damn, you're good!
Coke allison: ?
kiwichris dukeleto, it gets all the way to create_seg then dies in that. It has created the interp
Coke if (RT) danke. 21:09
allison Coke: 6.25%, I have to catch up
Coke: aye
dukeleto who knows what create_seg does?
Coke OOOOH does it segfault?
darbelo dukeleto: Ugly shit, if it's the one in packfile land. 21:11
kiwichris darbelo, I think it is. 21:12
21:13 mikehh joined
darbelo shakes his magic eitght ball. 21:13
dalek rrot: r42248 | NotFound++ | trunk (4 files):
[cage] fix mistakes in char * constness manifested by g++ 4.4.1, TT #1110
 21:14
darbelo "Check for sizeof(INTVAL) and sizeof(opcode_t) mismatches."
kiwichris dukeleto, what was the nopaste link ? 21:15
21:17 mikehh joined
darbelo kiwichris: Does it segfault in that function or in one called from there? 21:18
dukeleto nopaste?
purl somebody said nopaste was at nopaste.snit.ch/ (ask TonyC for new channels) or poundperl.pastebin.com/ or paste.scsys.co.uk/ or App::Nopaste or tools/dev/nopaste.pl or at www.extpaste.com/ or paste.scsys.co.uk (for #catalyst, #dbix-class, #moose and others) or gist.github.com/ or paste or gtfo or tools/dev/nopaste.pl or trac.parrot.org/parrot/browser/tru...nopaste.pl
kiwichris darbelo, just pasting a bt from gdb
darbelo kiwichris: Okay, I'll wait.
fperrad japhb, markdown uses perl5 during build, install & test (via prove) 21:19
darbelo Also, I remember uglier shit being in this file. (Whoever cleaned it)++
dukeleto fperrad: for any good reason? it could probably easily be switched over to NQP for build/install/test
japhb dukeleto, probably uses EU::MM for stuff like cp, ln, and rm 21:20
darbelo japhb: We can fix that in no time, for all of parrot. Except windows. 21:21
chromatic LuaJIT 2.0 features and research: thread.gmane.org/gmane.comp.lang.lu...eral/58908
japhb darbelo, That's a big except. :-)
Coke I'd be happy if someone removed the explicit perl dep from partcl. 21:22
(but it doesn't really matter since you need perl to build parrot.)
dukeleto chromatic: that is really cool stuff.
Coke: does partcl have a perl-based test suite? 21:23
nopaste "kiwichris" at 58.172.128.7 pasted "RTEMS segfault on 1.7.0" (50 lines) at nopaste.snit.ch/18550
dalek rrot-plumage: d6e58d0 | japhb++ | :
[METADATA] markdown still uses prove for testing, and thus perl5
Coke dukeleto: yes, though it's mostly migrated to pure tcl.
darbelo Coke: you don't need perl to run parrot. And if we get cross-compilation going, that means very little at run time.
fperrad dukeleto, if we want remove Perl5 dependencies, we need remove Makefile usage. 21:24
dukeleto kiwichris: wow, that looks fun.
Coke fperrad: ?
fperrad And write a Module::Build equivalent with NQP (Python DistUtil equivalent)
darbelo fperrad: why?
kiwichris dukeleto, handle_flags is looking guilty at the moment
darbelo kiwichris: Hmm, you got sucked into src/string/charset/unicode.c somehow. That feels wrong. 21:26
21:27 mikehh joined
dukeleto kiwichris: you shouldn't be going into unicode if you don't have libicu 21:27
fperrad darbelo, with make, all features which depends on shell come
Coke does git support something like svn:keywords? 21:28
21:28 theory joined
dukeleto Coke: what do you want it to do? 21:28
kiwichris: looks like the call at src/spf_render.c:853 which calls Parrot_str_append is where the badness begins 21:29
Coke ... the same thing that svn:keywords does?
e.g. expand $Id $ or something like it.
kiwichris dukeleto, agreed
dukeleto, what if I remove unicode.c from the build and see what breaks ?
fperrad darbelo, with Perl5 module, Makefile.PL generates a Makefile with many EU::MM call 21:30
darbelo kiwichris: everything ;)
fperrad but Build.PL is more clean
dukeleto Coke: those things have cooties. and the answer is 'no'
kiwichris darbelo, so I need it but cannot use it ? ;)
dukeleto Coke: stackoverflow.com/questions/39742/d...mmit-hooks
kiwichris: go for it
kiwichris: but darbelo may be right :) 21:31
darbelo kiwichris: not really, you can use it, it's just that big chunks get #ifdefed out without icu
kiwichris All that failed was the call Parrot_charset_unicode_init. 21:32
darbelo But it still works, in a way.
dukeleto kiwichris: you might be able to get away with commenting out the check at src/spf_render.c:849
kiwichris: it does have a /* XXX Silently ignore? */ comment
kiwichris dukeleto, yes I did wonder about that.
dukeleto, just looking src/string/charset.c at the unicode.c init call 21:33
dukeleto kiwichris: you are in the bowels now. here be parrots.
kiwichris dukeleto, droppings ?
dukeleto kiwichris: dragon-eating parrot droppings. Watch your step ;) 21:34
Coke dukeleto: there are about 10 test files left in partcl that use perl not tcl. I could use help converting them.
(because they do things like test that 'puts' (tcl's say) works.)
dukeleto Coke: do you have a test library written in partcl? or you use Parrot's Test::More ?
Coke dukeleto: both.
kiwichris dukeleto, big boots on and socks tucked into pants !!
dukeleto Coke: yes, testing i/o is not simple in Parrot's Test::More, *yet*
Coke plus tcltest.tcl, which is tcl's non-tap testing harness.
dukeleto non-TAP test harness? THE HORROR. 21:35
Coke (I have 2 files written in PIR)
then most written in tcl using a test-more-like handrolled tcl module, then some written using perl's test::more.
kiwichris dukeleto, remove call to Parrot_charset_unicode_init in Parrot_charsets_encodings_init. Now have the same failure as 1.6.0. Different problem. 21:36
dukeleto, if unicode is not configured should that call be removed ?
darbelo I don't think so. 21:37
dukeleto kiwichris: what does the current backtrace/error look like? 21:38
Coke dukeleto: even if i rewrite all these tests in pure tcl, I'm still going to need prove to get the test harness, yes? (which means I still need perl for make test.) 21:40
(tempting to switch to tcltest.tcl entirely, but that is big and slow.) 21:41
japhb Coke: dukeleto wrote a pure-NQP test harness for Plumage. In fact, if I could have your attention for a sec dukeleto ... ;-)
Coke japhb: does it support smolder? 21:42
japhb Coke, I have no idea. I pretty much let dukeleto run with the testing side of things.
kiwichris dukeleto, bit harder this one as qemu is stopping with an invalid memory area call. Need to pop out then back looking. 21:43
japhb Coke, but given that we've floated the idea of letting projects directly use the Plumage configure, test harness, etc. so as to reduce inter-project copy pasta, we'd probably want to make it smolder-compatible even if it wasn't already. 21:44
21:44 hudnix joined
dukeleto Coke: i have a test harness written in pure NQP 21:44
japhb: what now? 21:45
purl now is the time for the Kwisatz Haderach to arise and conquer America on a sandworm army
dukeleto botsnack
Coke now++
purl thanks dukeleto :)
dukeleto karma now
purl now has karma of -33
dukeleto lulz
japhb dukeleto, I was trying to find out if you were planning to fix the harness requirement that nqp is in your PATH, or if I should do so.
(in the nqprx branch) 21:46
dukeleto Coke: the test 'harness' has nothing to do with smolder
japhb: i haven't touched the branch since I passed out last night
japhb dukeleto, OK, I'll throw some cycles at it.
dukeleto japhb: feel free to fiddle with it, I won't get a chance to work on it for a few more hours
Coke dukeleto: currently or ever? in parrot, the harness is responsible for generating the output that goes to smolder.
japhb dukeleto, roger that 21:47
dukeleto Coke: actually, the parrot Makefile target "make smolder_test" does that, which runs the test harness, then captures the TAP output
japhb Coke, if it doesn't now, it probably will, as I said. If dukeleto is uninterested in that feature, we welcome your contribution. ;-)
dukeleto Coke: so it is not technically part of the test harness
Coke, japhb: I will look into how hard it would be to submit to smolder from pure NQP 21:48
mikehh NotFound - ok that works for me - g++ 4.4.1, will it still work with g++ 4.3.3?
japhb dukeleto, EXCELLENT
NotFound mikehh: yeah 21:49
japhb dukeleto, remember that we have the beginnings of an HTTP client in pure PIR already ...
Coke dukeleto: no, it doesn't run the test harness and capture the output. it runs the test harness with a special --archive option.
dukeleto Coke, japhb: we would have to implement TAP::Harness::Archive in NQP
Coke dukeleto: right.
dukeleto Coke: like I said, it catches the output with TAP::Harness::Archive :)
mikehh I'll have to reboot to test it - maybe later
Coke but it is /part/ of the harness, currently.
21:49 xenoterracide joined
Coke I think we're both on the page. 21:50
dukeleto Coke: the harness calls out to other code, so I guess it is a matter of opinion whether it is part of the harness or what the harness uses
Coke: yep.
Coke: i consider that a "plugin" for the harness to use
Coke: but not quite part of the "core harness" 21:51
Coke: if such a thing exists
Coke dukeleto: does your nqp harness let you run code that isn't pir but requires a HLL runtime? 21:53
21:53 hudnix joined
dukeleto Coke: the harness just read TAP output. that is all 21:53
s/read/reads/
Coke dukeleto: er... how do you run the tests?
dukeleto Coke: in which repo? 21:54
Coke say, mine.
if I have a test written in tcl, how do I get the nqp prove to run it with "parrot tcl.pbc t/foo.t" ? 21:55
NotFound Coke: maybe you can adapt Winxed harness 21:58
Coke winxed?
purl hmmm... winxed is not javascript, just uses a familiar syntax.
Coke url? 21:59
NotFound Coke: my new toy
code.google.com/p/winxed/
Coke winxed is also code.google.com/p/winxed/
purl okay, Coke.
Coke NotFound: did you write your own harness in <HLL> ? 22:00
dukeleto Coke: that is given in the shebang line of the test
Coke: prove runs any test under the binary given in the shebang line
NotFound Coke: yeah
Coke dukeleto: is your tool replacing prove? 22:01
NotFound Coke: not so bad for a language less than a month old :)
dukeleto Coke: not in any way.
purl not in any way is it my right.
Coke so even if I switch to your stuff, I still need perl.
?
dukeleto NotFound: how do you pronounce 'winxed' ?
NotFound Coke: no, you can compile it to pir
dukeleto Coke: i am not sure. you have your tests written in a mixture of stuff. 22:02
NotFound dukeleto: I just write it ;)
dukeleto Coke: for instance, plumage does not require prove, but you can use it if you want
Coke dukeleto: if you are not replacing prove, how are you running the tests?
dukeleto Coke: but plumage's entire test suite is in PIR/NQP 22:03
Coke: with nqp
Coke yes, but how?
purl hmmm... but how is it you know.. red?
Coke do you have a prove like utility? how do you decide which tests to run?
dukeleto NotFound: did you start winxed with the mk_language_skeleton script ?
NotFound dukeleto: no, is entirely written in C++ 22:04
dukeleto NotFound: holy $expletive.
NotFound: are you trolling me?
NotFound It uses parrot only at runtime
dukeleto: see the source if you don't believe me
dukeleto Coke: plumage has only 1 test suite and you currently can either run the entire test suite or individual tests, manually 22:05
NotFound: i believe
Coke: we don't have all the complications of partcl's test suite, yet.
Coke gives up and considers checking the source later.
dukeleto Coke: but you can run a set of tests with prove from the command line: prove -v t/foo*.t 22:06
Coke: this does not mean that our test suite *relies* on prove. prove just makes it slightly easier to run arbitraries subsets of the test suite
s/arbitraries/arbitrary/ 22:07
NotFound winxed harness can also do that
dukeleto NotFound: the name winxed makes me think it is windows-related. just sayin' 22:08
NotFound dukeleto: If I say you the origin of the name, you won't believe me X-)
Coke NotFound: I can build it on osx/x86 10.4
but running 'make test' results in an error. 22:09
./winxed t/harness -r t
ERROR: exec failed: No such file or directory
:Run failed
make: *** [test] Error 1
NotFound Coke: you must put parrot bin in the path
Coke yah, just found that. 22:10
a slightly better error message would help. =-)
NotFound takes a note
Coke needs to rebuild parrot and will then give you a platform report. but later. 22:11
NotFound It doesn't need parrot to compile, but hardly can run the generated pir whithout it ;) 22:12
chromatic Failed to build 't/src/warnings_2.o': t/src/warnings_2.c: In function ‘main’:
# t/src/warnings_2.c:15: error: dereferencing pointer to incomplete type
Was RTEMS seeing that too? 22:13
dukeleto chromatic: that is a known failure on trunk
chromatic: i am not sure if it is effecting the rtems port right now 22:14
chromatic: mikehh created a ticket for it
dalek rrot-plumage: 3b0eb83 | japhb++ | :
[CORE] Glue.pir: Add support for global
 22:18
rrot-plumage: 24a4fa9 | japhb++ | :
[TEST] t/harness: use rather than hardcoding 'parrot_nqp'
 22:19
22:19 Whiteknight joined
dukeleto japhb: you have been bitten by shell-expansion in your commit message. 22:20
japhb dukeleto, yeah, I noticed that after the push. :/
dukeleto, OK, as of the push I did just now, I believe all is good in the nqprx branch. 22:21
dukeleto japhb: it happens to the best of us
japhb: nice!
japhb: i have started using single quotes unconditionally in my commit messages, so that I don't have to think about the shell doing the wrong thing 22:22
japhb I stopped doing that because I had a habit of using contractions in my commit messages ...
dukeleto japhb: yeah, that is the kicker.
chromatic Ah, there's the problem.
dukeleto japhb: i just use bad grammar and speling instead :) 22:23
chromatic: which problem?
chromatic A libparrot built with -DNDEBUG won't export Parrot_pcc_warnings_on().
See include/parrot/context.h.
dalek rrot-plumage: 6400e90 | japhb++ | :
[TEST] t/harness: Resolve conflict between branches
 22:24
dukeleto chromatic: do you know how to fix t/src/warnings.t ? 22:25
chromatic Sort of. It's a little trickier than you might think.
dukeleto japhb: you need a spiffy avatar on gitorious :) 22:26
japhb dukeleto, know any good avatar creator sites? I don't feel like using a real pic ...
dukeleto japhb: avatar creator, or image cropper ? 22:27
japhb: mypictr.com/ is decent for online cropping
japhb Oh, I can do image editing locally, I've got freshman skills at The Gimp. 22:28
I just don't feel like using a real pic, because I've got one of those faces that looks REALLY generic at small pixel sizes. ;-)
"Oh, it's Generic White Boy. I think I know him from ... uh ... somewhere."
dukeleto japhb: www.doppelme.com/create/ 22:29
japhb Oh yeah, I managed to blow that up in no time flat. 22:30
Seriously? Some hair styles not available without login?!? 22:33
doppelme, you lose.
dukeleto japhb: yeah, I just found it from googling "create avatar"
japhb np
dukeleto japhb: it looks kind of lame
japhb If I could think of a really good 3D image that would survive shrinking down to 64x64, I'd probably use that. But I haven't had any brilliant ideas yet. 22:35
dalek rrot: r42249 | chromatic++ | trunk/src/string/api.c:
[string] Avoided dereferencing an uninitialized pointer in
 22:36
rrot: r42250 | chromatic++ | trunk/t/src/warnings.t:
[t] Cleaned up some code in t/src/warnings.t. The failing tests (TT #1187)
22:45 kiwichris joined 23:18 plobsing joined
plobsing hi #parrot 23:18
dukeleto plobsing: hola 23:19
japhb: someone mentioned aviary.com/ when I asked about avatar creation. looks like a web-based gimp or something 23:21
darbelo Name stealers! 23:23
23:23 zak_ joined 23:26 Zak joined
darbelo jumps on the user pic bandwagon. 23:26
dukeleto darbelo: yeah, you need one too 23:27
darbelo: congrats on signing up for release manager on 2.1 23:28
darbelo: I dare you to code name it "Enterprisey"
japhb dukeleto, hmmm, aviary looks interesting. Doing $day_job right now, but will look more later. 23:29
darbelo dukeleto: Only if someone steals NCC-1701 for 2.0 23:30
japhb darbelo, which one?
darbelo japhb: What do mean which one? 23:31
japhb darbelo, there are at least a half dozen NCC-1701 models. :-) 23:32
en.wikipedia.org/wiki/Starship_Enterprise 23:34
darbelo japhb: the one with no suffix, clearly.
Unless you are asking me about a particular visual? (TOS, Movies, new movies?) 23:37
japhb Basically, yeah, but I was also just joking too. 23:38
darbelo I'd probably go with shatner-era movie look, then. The nostalgia factor and all. 23:40
It's also the picture get in my head when people say 'Enterprise' 23:41
23:41 tetragon joined
dukeleto darbelo: that gave me the lulz 23:41
23:45 theory joined
darbelo dukeleto: My backup name is "KHAAAN!" 23:51
dukeleto darbelo: I like it.

Powered by Cro and the Raku® Programming Language.

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

Thank you!