Parrot 3.1.0 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Goals: Get more GSoC ideas on wiki; close tickets; stable 3.2 release; assess status of roadmap goals for 3/15 meeting.
Set by moderator on 9 March 2011.
kid51 ~~ 00:06
00:08 Andy joined 00:18 jsut joined 00:23 jsut_ left 00:33 whiteknight joined 00:52 kid51 left 01:00 mtk left 01:04 woosley joined 01:07 mtk joined
cotto ~~~ 01:55
01:56 davidfetter left 01:57 particle1 joined 02:01 particle left 02:05 Drossel joined 02:11 Kulag left, kid51 joined
dalek sella: b2c42a7 | Whiteknight++ | / (4 files):
remove call_parrot_method. it was from an earlier time
 02:19
sella: 4bd87d0 | Whiteknight++ | src/core/Rosella.winxed:
use 'can' in find_named_method instead of find_method inside a try/catch. At the very least this is cleaner, if not faster
sella: a7c3645 | Whiteknight++ | src/ (8 files):
some cleanups and docs for the action library
sella: 26b579c | Whiteknight++ | src/container/ (6 files):
small fixes/cleanups to the container files
sella: 3cb1ae5 | Whiteknight++ | / (2 files):
properly namespace EventManager-> Event.Manager
atrodo NotFound: ping
02:20 whiteknight left
sorear kid51: Have you seen PIRATE? It can emit .pbc files without going though Parrot's command line. Is it unsufficiently reliable? 02:29
cotto sorear, it's incomplete atm and slow.
Andy OK, now ready for another andy++ 02:39
See line 3519 of src/debug.c
kid51 Ah, yes, I remember it well.
Andy No, seriously, go look at it. 02:40
kid51 Yes ... 02:41
Andy That should be *string
PerlJam you added a * ?
Andy Hooray for splint!
I haven't yet.
PerlJam andy++ there you go :)
Andy But splint catches stuff GCC didn't.
A teeny bug, but it gives me a little validation for the time I put into the splint stuff. 02:43
dalek rrot: 45c6657 | mikehh++ | src/gc/alloc_memory.c:
fix codetest failure - add ASSERT_ARGS
 02:49
rrot: 0afcb30 | mikehh++ | src/hash.c:
fix codetest failure - line length
rrot: 089e7cb | petdance++ | config/gen/makefiles/root.in:
more splint updates
 02:59
rrot: 514a578 | petdance++ | src/gc/gc_gms.c:
fix up size_t argument in failed_allocation
atrodo NotFound: unping 03:04
03:21 kid51 left
dalek rrot: f8e6788 | petdance++ | / (2 files):
two functions actually can return NULL
 03:49
04:06 bacek left
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#11988) fulltest) at 3_1_0-780-g514a578 - Ubuntu 10.10 i386 (g++-4.5) 04:08
dalek rrot: a8d50a5 | petdance++ | / (2 files):
annotated Parrot_compile_string
 04:28
rrot: f31d6e9 | petdance++ | / (3 files):
fix a decorator. Assign NULLs to pointers, not zeroes.
04:46 ShaneC left 04:50 benabik left, benabik joined 04:54 woosley1 joined 04:57 woosley left 04:59 woosley1 left 05:09 ShaneC joined
dalek rrot: 34dda14 | petdance++ | / (9 files):
Annotating all the GC function pointers. Now to make the functions match the typedefs.
 05:34
05:48 cosimo joined 06:04 Andy left 06:21 bacek joined, fperrad joined 06:29 bacek left 06:35 cosimo left 06:40 bacek joined 06:42 jrt4 joined
bacek ~~ 06:52
06:59 bacek left 07:00 rurban_ joined 07:02 rurban left, rurban_ is now known as rurban 07:11 bacek joined 07:49 jrt4 left
dalek rrot/opsc_llvm: 21594e1 | petdance++ | / (2 files):
Parrot_hash_put can return NULL
 07:52
rrot/opsc_llvm: ddf8ae8 | bacek++ | NEWS:
Update NEWS for opsc_full_parse branch merge.
rrot/opsc_llvm: 7d79837 | petdance++ | config/gen/makefiles/root.in:
ratcheting down some splint warnings
rrot/opsc_llvm: 492de04 | bacek++ | t/pmc/capture.t:
Add todoed tests for Capture.freeze/thaw.
rrot/opsc_llvm: cdecdf8 | bacek++ | / (2 files):
Implement Capture.freeze/thaw. Closes #2047
rrot/opsc_llvm: afa7674 | bacek++ | src/pmc/capture.pmc:
Set custom_mark flag in Capture.thaw when needed.
rrot/opsc_llvm: e24da1f | bacek++ | compilers/opsc/ (4 files):
Add few utility methods for workaround of P6object freeze/thaw bug
rrot/opsc_llvm: 763edbf | bacek++ | compilers/opsc/src/Ops/Compiler/Actions.pm:
Deep clone of Ops variant.

It will be used for generating of dump files with parsed Ops.
rrot/opsc_llvm: d871b26 | bacek++ | compilers/opsc/src/Ops/File.pm:
Force boolify "experimental" flag.
 07:53
rrot/opsc_llvm: fa88183 | bacek++ | compilers/opsc/src/Ops/Util.pm:
Clone scalars in process_recursive.
rrot/opsc_llvm: 1315017 | bacek++ | compilers/opsc/src/Ops/Op.pm:
Simplify Op.arg_type
rrot/opsc_llvm: 82978da | bacek++ | t/compilers/opsc/03-past.t:
Rewrite test using "is" instead of "ok".
rrot/opsc_llvm: 01a98eb | bacek++ | compilers/opsc/src/Ops/Util.pm:
Skip undefined elements in process_recursive. Fix last broken test in t/compiler/opsc
07:58 woosley joined 08:03 theory left 08:22 M_o_C joined 08:35 sri left, sri joined 08:47 mtk left 08:53 mtk joined 08:55 dawg joined 08:57 woosley left 08:59 dawg is now known as lucian 09:26 clunker9 joined 09:30 lucian left 09:56 cosimo joined
dalek rrot/opsc_llvm: 36e4a6f | bacek++ | compilers/opsc/src/Ops/Util.pm:
Fix stupid bug with process_recursive.
 10:14
rrot/opsc_llvm: e020e81 | bacek++ | src/pmc/class.pmc:
Add another workaround for thawed Class/Namespaces
rrot/opsc_llvm: 9810967 | bacek++ | compilers/opsc/src/Ops/File.pm:
Made :core optional. Otherwise we can thaw Ops::File.
rrot/opsc_llvm: cb5a75d | bacek++ | compilers/opsc/src/Ops/Util.pm:
Few minor bugfixes.
rrot/opsc_llvm: ca64257 | bacek++ | compilers/opsc/ops2c.nqp:
Implement dumping of parsed ops
10:18 dafrito joined 10:27 M_o_C left
dalek rrot/opsc_llvm: e51f9d6 | bacek++ | compilers/opsc/src/Ops/Op.pm:
Revert "Simplify Op.arg_type". It was really bad idea.

This reverts commit 1315017b4be28f7425941ce3858710072a5b43eb.		 10:33
10:51 cosimo left 11:44 woosley joined 13:03 clunker9 is now known as clunker3, jsut left 13:04 jsut joined 13:43 whiteknight joined 13:47 woosley left
whiteknight good morning, #parrot 13:49
tadzik good morning zebras 13:50
rurban Has duke submitted tpf to gsoc 2011? today is deadline? 13:51
whiteknight I think they said he had on the mailing list 13:54
14:01 rurban left 14:06 ambs joined 14:09 mtk left 14:11 mtk joined
whiteknight blah. In updating my branch from master I get a conflict in core_thunks.c but not in core_thunks.nci 14:12
dalek Heuristic branch merge: pushed 266 commits to parrot/whiteknight/imcc_compreg_pmc by Whiteknight 14:32
whiteknight msg plobsing I updated whiteknight/imcc_compreg_pmc from master. There were some nci-related conflicts as expected. The build fails running parrot_nci_thunk_gen. Can you help take a look at it? Thanks 14:33
aloha OK. I'll deliver the message.
ttbot Parrot b12bcf3e i386-linux-thread-multi make error tt.taptinder.org/cmdinfo/54478 14:34
14:37 rurban joined
ttbot Parrot 91c2194a MSWin32-x86-multi-thread make error tt.taptinder.org/cmdinfo/54498 14:39
Parrot b12bcf3e MSWin32-x86-multi-thread make error tt.taptinder.org/cmdinfo/54499
14:41 woosley joined
ttbot Parrot 91c2194a darwin-thread-multi-2level make error tt.taptinder.org/cmdinfo/54533 14:51
Parrot b12bcf3e MSWin32-x86-multi-thread make error tt.taptinder.org/cmdinfo/54536 14:53
14:59 rurban_ joined 15:02 rurban left, rurban_ is now known as rurban
Tene fwiw, getting HLL interop into a good state again would be a great GSOC task. 15:09
whiteknight Tene: that might make a good project. Do you have anything like a tasklist or checklist that would need to be completed? 15:10
or, at the very least, do you have a list of blockers?
Tene whiteknight: not currently, but I could put one together sometime after I sleep. 15:11
15:11 lucian joined
Tene There aren't any real blockers. 15:11
whiteknight Tene: If you could put together some kind of tasklist for that, it would be a huge help
Tene HLL smoke testing has been kind of a blocker for me personally, but I don't expect that to generalize.
whiteknight yeah, our smolder server has been less than optimal lately 15:12
Tene new.leto.net:3000/project/parrot is supposed to handle HLL testing eventually, but doeosn't currently afaict. 15:13
whiteknight we can add other projects to smolder too. We've added things like Rakudo, PL/Parrot, and PLA to it already 15:14
jnthn_ Oh, Rakudo is already added to smolder? 15:18
jnthn_ didn't know that :)
oh... new.leto.net:3000/project/rakudo 15:19
15:19 woosley left
jnthn_ Clearly somebody needs to start smoldering. :) 15:19
Tene The high-level tasklist is approximately: look at the previous HLL interop API (should be in a PDD, and should be able to find previous impls in a few HLLs), evaluate current level of implementation, make sure the 'parrot' language (or pir or whatever) is set up correctly and load_language-able (there's at least one current issue with some part of PCT registering something else in its place), get reasonable PCT support for the HLL interop ... 15:21
... API, and write patches for N HLLs.
So, a bit less work than the previous N times I've done it.
Also, add tests to parrot for it, and get something in leto's smoker that tests interop between multiple HLLs. 15:22
I've been reassurred by a few people that if an HLL interop API gets established again, it's more-likely to stick this time. 15:23
The last time I did this, it was something like a couple of weeks of work, maybe? 15:24
I don't remember too clearly.
15:24 hercynium joined
Coke lack of partcl testing by not-me has been a blocker for me, ;) 15:25
Tene I haven't paid much attention to GSOC; does that actually look like an appropriate task? 15:26
15:27 estrabd_ left, estrabd joined
whiteknight Tene: maybe, probably not. GSoC needs to be a task a student can devote to for 2-3 months 15:27
HLL interop could probably be a portion of a larger task
Tene Ah. 15:30
Maybe someone else can come up with work that can build on this.
I need to get around to documenting it anyway.
whiteknight yes, I will definitely keep it in mind
15:39 Andy joined 15:45 whiteknight left, whiteknight joined
whiteknight I'm overhauling the GSOC2011 page right now, so don't edit it 16:01
atrodo Did the marketing team ever take off like the other temas have? 16:02
whiteknight no 16:06
dalek tracwiki: v9 | whiteknight++ | GSoc2011 16:09
tracwiki: Remove ideas which are no-go (documentation-only, etc) Reformat page to use a new template for project ideas
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
tracwiki: v10 | whiteknight++ | GSoc2011
tracwiki: cleanup and +a few new ideas
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
atrodo I wonder if anything marketing-like (like a fun website or the like) would be a usable GSoC idea 16:10
16:19 Patterner left, Psyche^ joined, Psyche^ is now known as Patterner
dalek tracwiki: v11 | whiteknight++ | GSoc2011 16:25
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
tracwiki: v12 | whiteknight++ | GSoc2011
tracwiki: new introduction
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
tracwiki: v13 | whiteknight++ | GSoc2011
tracwiki: +some emphasis
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
tracwiki: v14 | whiteknight++ | GSoc2011
tracwiki: emphasize all fields in all templates. Why doesn't trac wiki allow template transclusion? That would be quite the handy feature right about now.
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
16:36 theory joined 16:41 mj41 left 16:44 mtk left, mtk joined
Tene whiteknight: Make sure to remember to mention in the channel when it's safe to edit again. 16:45
whiteknight it's more or less safe to edit again 16:46
although I'm still making frequent edits
cotto_work ā˜ƒā˜ƒ 16:47
whiteknight lolwat? 16:48
Tene whiteknight: I don't know how to calibrate my difficulty scale for x/5 for project difficulty. Please advise. 16:49
whiteknight Tene: just best guess. The difficulty field is just a rough estimate 16:50
16:52 mj41 joined
dalek tracwiki: v15 | whiteknight++ | GSoc2011 16:56
tracwiki: +details about a few projects
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
tracwiki: v16 | whiteknight++ | GSoc2011
tracwiki: +more info about projects
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
tracwiki: v17 | whiteknight++ | GSoc2011
tracwiki: +ANTLR backend
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
whiteknight needs food, then will dump more ideas 16:59
17:02 sigue left
Tene I expect I could dump more ideas if I remember to think about it after I wake up. 17:03
Going to bed now. Open invitation for anyone to harass me about gsoc proposals later today.
aloha: tell Tene to post gsoc proposals. 17:04
whiteknight Tene++ 17:09
don't worry, we'll remind you
dalek tracwiki: v18 | whiteknight++ | GSoc2011 17:12
tracwiki: +LAPACK bindings for PLA
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
tracwiki: v19 | tene++ | GSoc2011
tracwiki: Add minor notes about hll interop task.
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
Coke possible ideas: "resurrect mod_parrot" 17:14
(for GSOC)
seen jhorwitz?\\
clunker3 Sorry, Coke, I haven't seen jhorwitz?\\.
aloha Sorry, I haven't seen jhorwitz?\\.
Coke seen jhorwitz
clunker3 jhorwitz was last seen on #parrot 2 years, 6 months, 14 days, 51 minutes and 1 second ago, saying: and probably not the last...
aloha Sorry, I haven't seen jhorwitz.
Coke clunker3, owner?
GSOC idea: get partcl-nqp to the point where it passes spec tests. 17:15
GSOC idea: improve profiling tools. 17:16
GSOC idea: speed up rakudo benchmark by x%. (or any HLL, really. "get within only 100X slower than native version) 17:17
whiteknight Coke: did you see my most recent email to parrot-dev? Google far prefers projects which produce something new. Not projects which simply fix or vaguely improve something which already exists
so "fix partcl" is not an acceptable GSoC task
but "implement Tk bindings for partcl" would be 17:18
actually, that's a pretty damn good idea
I have one task in there to add new sub-level profiling tools 17:19
if we can find a specific enabling technology that the student can implement to bring performance gains, that would be a good project 17:20
Coke whiteknight: I didn't say "fix partcl". sheesh. 17:27
whiteknight "resurrect mod_parrot" and "get partcl-nqp to the point where..." are both along those lines 17:28
Coke They are both easily updatable to have achievable, delineated goals. Which is why I dumped them here in as an idea. If I had fleshed them out, I'd have bothere dto put them on the web site. 17:31
I dispute that "mod_parrot" exists, btw.
So, anyone have any bright ideas on TT #1886? 17:35
I have zero interest in Tk bindings for partcl, btw. Happy to mentor someone if they want to do that, but at the moment, the state of the parrot it's sitting on needs a lot more attention than an external library. 17:37
plobsing attention as in "Tk bindings for Parrot" or more along the lines of "fix the damned thing"? 17:38
Coke speed, memory usage of core parrot is too slow, too bloated. can't even run /tcl/, let's get that working before we worry about Tk. 17:41
(just running tcl's "tcltest.tcl" (and not running any tests) is currently about 300x slower than tclsh on my OS X box.) 17:42
whiteknight creating a new apache plugin for Parrot to replace the old, dead mod_parrot would be a good idea
plobsing how about a FastCGI interface?
Coke (with partcl old. partcl-nqp chokes on that file still.) 17:43
whiteknight I'll write up an idea for an "Apache plugin". However the student wants to implement that is fine
Coke whiteknight: funny, I thought you just shot that down.
whiteknight mod_parrot, or FastCGI, or whatever is all fun
what did I just shoot down?
dalek tracwiki: v20 | whiteknight++ | GSoc2011
tracwiki: +more details from a few projects. +idea for a code metrics toolset
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
tracwiki: v21 | whiteknight++ | GSoc2011
tracwiki: volunteer to mentor several ideas. +details about profiler
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
whiteknight fixing mod_parrot is one thing. Writing a new apache plugin is different
plobsing Coke: there's a difference "fix up old and busted" vs "shoot old and busted behind the shed and reimplement"
Coke plobsing: my point was that whiteknight shot down my specific one liner description of that very project. I know there's a difference. 17:45
whiteknight Coke: I apologize for shooting it down. What I understood from your one-liner above was inappropriate for a project
I have more information about it now, and can see that it has merit
Coke <shrug> You and I clearly have some kind of communication gap. This isn't the first time. ;) 17:46
plobsing arg. who knew MULTIs could be used to sorta kinda do multi-dispatch for VTABLEs? 17:55
whiteknight can they?
plobsing not well
whiteknight I'm not sure if that's bad or good
oh, you mean a MULTI declaration in a .pmc file? 17:56
plobsing yep. For example, complex has a bunch of MULTIs with the same name as vtables (add, i_add, etc)...
whiteknight right, okay 17:57
yes, those things are stupid and hideously inefficient
at one point I think they were either deprecated or strongly suggested to be removed
PLA still uses a few of them, but I've been meaning to have them removed
plobsing what 'MULTI' in .pmc s? or multis standing in for vtables?
whiteknight ...does not parse 17:58
plobsing what exactly are you saying is 'stupid and hideously inefficient'?
whiteknight MULTIs in .pmc files
regular MultiSubs are not so bad
dalek tracwiki: v22 | whiteknight++ | GSoc2011 17:59
tracwiki: two new projects. An apache plugin (replace mod_parrot) and a text templating engine
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
whiteknight Parrot used to have a lot more MULTI declarations for its core PMCs, but many of them were removed and replaced with switch/case statements
plobsing if you mean "dispatch based on type of one argument", don't use multi. 18:00
if you mean "find the most constrained method for all the types at once", then use multi. 18:01
I'm not sure we need to be providing sugar for that in pmc-lang.
whiteknight I'm certain we do not
I'm also certain that we could remove 50% of our arithmetic vtables, the worst offenders with MULTI, and nobody would ever notice the difference 18:02
but that's neither here nor there
plobsing whiteknight: I can't seem to find a deprecation for MULTI. 18:04
whiteknight I doubt it's a deprecation
Coke whiteknight: oh, I'd notice the difference. please don't change that again without a very good plan. 18:05
plobsing should we make it one?
whiteknight I'm not changing anything. Only musing
Coke I cannot tell you how many times partcl has broken because someone changed something related to multi vtables.
plobsing Coke: is it feasible to avoid using multi vtables? 18:06
I understand the sugar is sweet, but the mechanism is rotten.
18:08 lucian left
whiteknight plobsing: trac.parrot.org/parrot/ticket/452 18:08
plobsing tl;dr, what is the thing that is being deprecated? 18:10
whiteknight MULTI
plobsing completely?
or "only when not needed"?
whiteknight the ticket is a little wishy-washy
plobsing which is a really loose and vague statement
whiteknight it is marked as a deprecation 18:11
18:11 sigue joined
Coke whiteknight: if it's vague, it needs to be rejected. Just like with any other ticket that has no details. 18:12
whiteknight there are plenty of details. It just doesn't specifically say "remove all these things"
plobsing I'm with Coke on this. It is not a clear mandate to perform any particular action.
whiteknight Whatever, i'm not going to stop anybody from closing it 18:13
I'm only pointing out that ticket as prior discussion of the issue of MULTIs, which seems to suggest we deprecate and remove them 18:14
Coke in any case, partcl only has one place where we're using multi on a vtable. as long as it doesn't break in whatever future cleanup happens, I'm fine with that. 18:15
whiteknight if it doesn't break by being reimplemented to something nicer, would that count?
18:18 ShaneC left
Coke I have 2 points of view here: don't screw up HLL developers by ripping out things just because. The other is: as long as partcl works, I don't care. 18:21
so from the latter standpoint, sure, that's fine. 18:22
whiteknight we wouldn't be ripping it out just because. We're not nearly so frivolous 18:23
plobsing no honour among thieves or HLL developpers eh?
Coke whiteknight: it's hard to get that impression from reading your blog and listening to you on IRC. 18:24
whiteknight it's hard to convincingly say that we both want parrot to be faster, and we think it's perfectly fine for e.g. the Integer PMC to use full multidispatch semantics on every arithmetic operation
Coke I think it's hard to say that that's a bottleneck. 18:25
but if it /is/ a bottleneck, by all means, let's consider it.
whiteknight there's a difference between a bottleneck and a component of a larger performance problem
Parrot's performance issues are not just related to a small handful of distinct bottlenecks 18:26
there are significant architectural issues at play, complexity problems throughout, etc
And if fixing every little problem is going to be treated as a big imposition by our HLL developers, we're not going to get too far 18:27
18:27 preflex left
plobsing there's too much confounding code. code that prevents other code from being performant, or from being rewritten, etc... 18:27
whiteknight I don't really care about MULTI, it's not anywhere on my list of priorities. But that doesn't mean it can't be recognized as the bad code that it is 18:28
plobsing pmc2c is particularly bad when it comes to NCI. they shouldn't really be related at all. but whenever I go to change NCI, it's always some dumb thing pmc2c does that gets broken.
which is why I care about MULTI. it gets in the way of stuff it shouldn't even be touching. 18:29
dalek TT #452 closed by plobsing++: Don't use MULTI when not needed 18:30
TT #452: trac.parrot.org/parrot/ticket/452
jnthn_ added a GSoC suggestion for Rakudo to epo.means.no/gsoc2011/ideas (mentioning it in case folks show up here looking for a Rakudo GSoC task) 18:31
Coke whiteknight: I'm not saying "don't make parrot better". I obviously want that. I just have to deal with a lot of stuff getting change, modified, ripping out, improved, broken, whatever. So when I see someone saying something is "bad code" without qualification, my hackles go up. 18:32
18:32 preflex joined
whiteknight jnthn: thanks! We'll be sure to pass that note along 18:33
jnthn: If you have any project ideas which would be mutually beneficial to Parrot and Rakudo, we'll add them to our list too
plobsing and then, once I look under the covers at MULTI, I see that it generates incorrect code, but on uncommon code paths. I could file a ticket, but I doubt that would get fixed any time soon. 18:34
jnthn_ whiteknight: I think I saw somebody mention sub level profiling; if that could make use of HLL annotations too and then generate useful output when run with Rakudo programs, it could be a big help. 18:35
dukeleto ~~
jnthn_ I didn't try the profiler recently, maybe I should at some point soon to see what the latest state of it is. :)
whiteknight I'm a big proponent of being honest about the quality of our code. When I see bad code, I call it bad. That doesn't mean I have the time or the inclination to personally change or fix everything that ends up on the shit list
just because it's bad doesn't mean it's going to change
that's a little bit of honesty which is more than a little bit depressing to have to type 18:36
Coke yes, but "bad" tells me nothing. it doesn't help anyone fix whatever you might think is wrong with it.
whiteknight jnthn: yeah, I added the idea about the sub-level profiler. It's on my list of things pmichaud told me you guys wanted
jnthn_ whiteknight: Excellent.
Coke (don't we already have .sub level profiler??)
plobsing we have a line-level profiling runcore 18:37
whiteknight Coke: it does. It identifies it as something which is worthy of attention. Other developers may have radically different priorities than I have, and might want to look at bad code
Coke I thought cotto's stuff was sub level.
plobsing which can now make use of HLL annotations without any noticeable cost
Coke whiteknight: "Multi vtables are bad"... how is that helpful?
plobsing Coke: I provided at least 2 good backing reasons for that statement. 18:38
whiteknight Coke: because it tells a person looking for a task that it's a particular area worth looking at
Coke plobsing: yes, but I'm not arguing with /you/ ;)
whiteknight plus, if we have to fix a bug that touches code we know to be bad, maybe we make different decisions than if the code were sacrosanct
Coke Yes, but I don't know that it's bad. I know it smells funny to you. that's it. 18:39
whiteknight Maybe we put it on the deprecation chopping block if it's recognized as a big enough pain point. Maybe we refactor. Maybe we can do a lot of things. The first stage in all of that is recognizing there to be a problem
Coke yes, but you haven't said what the problem is.
18:39 ShaneC joined
whiteknight Coke: so what are you saying? That I haven't made my case compelling enough? or that I haven't compelled you personally? 18:39
Coke I can only speak to me personally, of course, but yah. "bad" is not compelling nor informative. 18:40
whiteknight plobsing took one look at it today and recognized it as being a bad mechanism. Hell, he's already found bugs in it. At the very least that's some kind of vindication 18:41
but like I said, it's nowhere on my list of priorities. I'm not going to analyze it or write huge blog posts about it, because it's not worth my time
cotto_work really wishes he didn't have a fire to put out 18:45
18:54 dodathome joined
dalek sella: 7d1cb71 | Whiteknight++ | README.md:
Some small fixes. Separate libraries which I consider "stable" from those which are still more experimental.
 18:55
19:11 davidfetter joined 19:53 plobsing left 20:04 plobsing joined 20:28 jrt4 joined
dukeleto davidfetter: severely delayed pong 20:37
20:38 ambs left
dalek sella: 139f2e3 | Whiteknight++ | s (3 files):
add in a new decorator library. A decorator is a transparent pass-through proxy which holds extra hashes of extension methods and attributes.
 20:39
sella: 4d01976 | Whiteknight++ | src/decorate/ (2 files):
add in some code, more of a novelty at this point, to implement an immutability proxy. it only works for direct attribute accesses on the proxy, not attribute accesses inside methods on the target
sella: bdbef2b | Whiteknight++ | src/decorate/Builder.winxed:
fix find_method to invoke the method on the proxy, not the target. This extends immutability all the way through to methods on the underlying class
sella: 9c5b872 | Whiteknight++ | README.md:
Merge branch 'master' of github.com:Whiteknight/Rosella
sella: a805977 | Whiteknight++ | / (2 files):
add an entry in README. Add a TODO note in the immutability section
whiteknight I wasn't sure if anything like cheap decorators or immutability wrappers was even going to be possible 20:41
but, it's looking like they both are, if we work around a few MOP warts 20:42
I like being pleasantly surprised by these kinds of things 20:44
I am reminded about how lousy it is to inherit a Class from a PMCProxy, however 20:50
21:03 fperrad left
dukeleto whiteknight: thanks for forging ahead with GSoC stuff. The last week has been a bit insane for me. Still getting back into my groove. 21:07
...o( How Duke Leto Got His Groove Back )o... 21:08
whiteknight dukeleto: No problem. I'm sorry it's taken me so long to get into gear myself
dukeleto whiteknight: perhaps we should have a "gsoc ideas page" virtual hackathon this weekend 21:11
whiteknight: even if we don't get into gsoc, an ideas page is a huge resource
whiteknight let's think positive 21:12
we really really want to get in
davidfetter what's shakin', dukeleto ?
whiteknight because it would be a kick to the jimmies if we don't get accepted but TPF does
dukeleto davidfetter: not much. working in Ithaca, NY until next Friday 21:15
whiteknight: yes, it would. Which is why we should really improve our Ideas page :)
davidfetter dukeleto, getting snowed on ?
dukeleto whiteknight: i will add a few ideas in a few minutes
davidfetter: there was 16inches on Monday, but it has mostly melted by now
davidfetter whee! 21:16
dukeleto whiteknight: i vouched for Parrot Foundation from the TPF org app, so there is that
davidfetter vaguely cognizant of the fact that there's been enormous changes to parrot lately
is there still a PIR?
whiteknight davidfetter: yes, unfortunately
dukeleto whiteknight: also, we were chosen as one of the 20 GCI orgs and have a good track record, so that is good.
whiteknight we move quickly, but not that quickly
davidfetter "unfortunately?" 21:17
dukeleto davidfetter: whiteknight is not a fan of PIR
whiteknight dukeleto: I know. I'm optimistic. It would just be very bad if we dropped the ball at this point
davidfetter:
dukeleto whiteknight: yes, it would.
whiteknight We can do better than PIR
dukeleto whiteknight: which is why I wanted to split PaFo/TPF because it was really hard to manage 2 umbrella orgs
davidfetter doesn't have a dog in this fight. there's "legacy" PL/PIR, but since there's no prod deployments...
dukeleto whiteknight: i can now concentrate on parrot stuff and I think I will be more effective
whiteknight dukeleto: Yes, I know. You were a saint for ever attempting it
plobsing I always enjoy how PIR is pronounced as "worse" in French 21:18
dukeleto whiteknight: or a fool, but thanks :)
whiteknight dukeleto: a saint for doing it. A fool for starting the summer thinking it was a good idea
davidfetter heard from a french speaker that ENOCLUE is pretty close to a rude verb
dukeleto I have also seen stickers on cars that say "PIR" which stands for "Portland International Raceway"
davidfetter (or at least one best mentioned only in private)
dukeleto whiteknight: we should make sure we have a spectrum of easy, medium and hard project ideas 21:21
whiteknight dukeleto: I've been trying to have several different levels. I gave each idea a ranking out of 5 21:22
there are a few 2s, 3s, 4s, 5s
and those are baseline difficulties. The student may propose something with a smaller scope 21:23
dukeleto whiteknight: cool 21:25
whiteknight: also, we can troll the "gci" trac tickets for ideas
whiteknight yes, although many of those may be too small
dukeleto whiteknight: we can add some stuff to gci tasks to make them "gsoc size"
whiteknight: or package a few together
whiteknight super size it
dukeleto whiteknight: scope creep is always free :)
whiteknight dukeleto, davidfetter: Are there any PL/Parrot-related tasks which could go in the list? 21:26
davidfetter whiteknight, there's always something ;)
whiteknight well add it to the list! We don't have a lot of time!
davidfetter linky? 21:27
21:29 perlite left
whiteknight trac.parrot.org/parrot/wiki/GSoc2011 21:29
whiteknight has to pack up and go home now. Later
21:29 whiteknight left 21:31 perlite joined
davidfetter what languages are implemented on top of the current parrot? 21:33
plobsing davidfetter: it depends on your criterion for "implemented" 21:34
davidfetter let's cast a narrow net to start
"passes the original language's own test suite," let's say 21:35
plobsing "works as planned", "works with minor documented bugs", "works with major problems", "skeleton exists and is useable", "skeleton exists and is detrimental to all who look upon it"
davidfetter how many in the 1st category?
plobsing most of the older languages fall in the last category and deserve complete rewrite
davidfetter: 3 - lua, rakudo, winxed 21:36
davidfetter that's actually pretty awesome. where's LOLCODE?
plobsing unmaintained so probably in the last or second last category
davidfetter that's a shame. i guess winxed is small 21:37
tadzik LOLCODE almost passes its tests
plobsing if you're looking for a particular list of languages, why not use the recent postings to parrot-dev? 21:38
s/parrot-dev/parrot-users/
C# and/or GCC backend
both do not yet exist
davidfetter
.oO(PL/winxed)
plobsing alternatively, scrap Perk and start a new Java on Parrot implementation
dalek p/ctmo: 32f2a59 | jonathan++ | src/ops/nqp.ops:
Fix error message typo.
 21:47
p/ctmo: 13b2478 | jonathan++ | src/pmc/serializationcontext.pmc:
Add quick and dirty method for getting the index of an object in an SC.
p/ctmo: 929093e | jonathan++ | src/ops/nqp.ops:
D'oh, another typo.
p/ctmo: d17eb00 | jonathan++ | src/metamodel/ (3 files):
KnowHOW and it's HOW need to live in a serialization context, so we make a core one.
dukeleto davidfetter: yes, PL/winxed could be cool 21:48
21:51 hercynium left
davidfetter GSoC project? 21:52
the thing is, for a real PL, you want some language that could already call other languages
plobsing winxed can call NQP. Rosella has some examples of exactly this kind of back and forth 21:53
davidfetter of course the language most in play is SQL 21:55
are those in the parrot distribution along with winxed?
plobsing davidfetter: winxed is a separate project developed by NotFound++. rosella is a separate project developed by whiteknight++. 21:56
davidfetter k 21:57
22:16 dodathome left 22:24 cosimo joined
dalek tracwiki: v23 | plobsing++ | GSoc2011 22:26
tracwiki: Add some interesting HLL implementation tasks.
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
Heuristic branch merge: pushed 16 commits to parrot/pmc2c-multi-nativepccmethod by plobsing 22:30
22:49 Andy left 23:00 rurban_ joined 23:02 rurban left, rurban_ is now known as rurban
dalek p/ctmo: c8c7a6f | jonathan++ | build/Makefile.in:
Seems that we have got the meta-objects and setting reified into the wrong levels of the bootstrap. :/ As a first step, this fixes up the Makefile dependencies, but it's probably not enough.
 23:10
p/ctmo: 80b8275 | jonathan++ | src/HLL/SerializationContextBuilder.pm:
First bits towards SC cross-referencing. Not able to use it just yet, due to issues with bootstrap layer reificationa and separate compilation. :/
p/ctmo: 765695a | jonathan++ | src/HLL/SerializationContextBuilder.pm:
Little more sketching out in the SC builder.
p/ctmo: 3100f3d | jonathan++ | src/NQP/Actions.pm:
Resolve lexically installed items (e.g. don't assume they're all package scoped).
23:16 cosimo left 23:20 whiteknight joined 23:27 bubaflub joined
plobsing whiteknight: I'm trying out your imcc_compreg_pmc branch, but I am sofar unable to reproduce the issues you reported seeing. 23:30
whiteknight plobsing: damnit. I need to get on whatever computer you are using 23:31
davidfetter it's a zuse z4 23:33
bubaflub hey whiteknight, i saw that you'd mentor someone to get LAPACK bindings for PLA for GSoC 23:35
i'm interested, but would like to do some investigation before i submit a proposal
whiteknight bubaflub: yeah, PLA has been mostly a solo project by me so ar
far
bubaflub whiteknight: i remember looking at it awhile back. my only hesitancy is that i haven't really done anything related to generating bindings to foreign or C libraries. 23:36
whiteknight: anywhere you'd recommend me to start looking / research to see if this is feasible for me to do? 23:42
whiteknight bubaflub: good question. checkout and build PLA if you haven't yet. I've been working on the test suite, so that might not be working right now 23:43
also check out the lapack website
www.netlib.org/lapack/ 23:44
brb
bubaflub whiteknight: ok. i'll do that. i've gotta head ot class but will be back on in 20 or so
23:46 bubaflub left

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!