Vision for 2.0: Production Users
Set by moderator on 21 July 2009.
04:45 japhb joined 09:22 einstein joined 13:02 Whiteknight joined 13:22 masak joined 15:02 quek joined 16:19 NotFound joined 16:36 Whiteknight joined 17:40 darbelo joined 17:59 einstein joined
darbelo put his report at www.parrot.org/content/week-decnum-dynpmcs-0 18:01
18:03 moritz joined 18:05 Util joined
Util # Done 18:06
* Tried compiling Parrot with -ObjC on Darwin, for possible iPhone embedding. Compiles and tests OK.
* Presented YAPC Recap at Atlanta PerlMongers. Particular interest was expressed in the PVMW, which I reported as a big success. 18:07
- At the after-meeting dinner, I tried to explain L1 to several interested Parrot observers. FAIL.
# Plan for next week:
* Change my testing scripts and parsing from `make fulltest_all` to `make -k fulltest`.
* Resume stalled work
# Blockers:
* Minimal tuits this week.
.end
cotto # What I did:
* started documenting the subset of callgrind's output format that we'll need to profile PIR
- trac.parrot.org/parrot/wiki/CallgrindFormat
# What I hope to do and how many tuits I expect to have:
* To avoid the runtime overhead of maintaining data structures representing statistics, the runcore will use output a format that can be written incrementally.
* I'll spend time specifying that format.
* also, I hope to get some opsc work done.
* tuit outlook is good
# What could block my progress:
* no external blockers
.eor
18:10 jonathan joined
jonathan Rakudo 18:10
* Lots of metamodel refactoring
* Rakudo now has a metaclass distinct from the default p6object one, which means we now poke less stuff into Parrot namespaces
* Hunted down a couple of nasty issues; one in the narrowness analysis, another where Any ~~ Object and Object ~~ Any were both true, and another in which we delegated to .^isa in Object.does - not sure how we got away with that for so long
* Can introspect what roles a class does and what attributes a class has, which was two of the main remaining missing bits of the introspection interface
* Improved .^methods in a few small ways
* You can now also introspect roles as well as classes
* All in all, I'm close to being able to wrap up the introspection stuff
Next week
* Need to write my YAPC::EU talk. Started it at least...
* Want to focus on trait application and improve our testing situation there
* Will try and figure out methods declared on the metaclass, and implement support for that; it's the last deliverable on my Hague Grant that doesn't have significant progress
Blockers
* Brain cycles
.end
18:13 allison joined
NotFound Last week lacked time to write a report, this one covers both weeks. 18:15
* Show size in Out of mem panics
* Error on duplicate locals with different type, closing TT #767
* Added include file libpaths.pasm
* Added example of opengl lib usage from HLLs
* Miscellaneous fixes and cage cleaning
* Some old patches applied
EOR
allison - Prepared Debian and Ubuntu packages for 1.4, and worked them through the process. We're updated in Debian, and 1.4 is included in Ubuntu Karmic (the 9.10 release, coming up in October). 18:17
- Unfortunately, I won't be able to attend YAPC::EU, as I'm still waiting on my UK visa. I'm hoping to make it to Portugal in time for the Thursday hackathon.
- My plan for this week is to revive the pcc_rewiring branch (merging the changes with a fresh branch from trunk).
EOR
18:19 pmichaud_ joined
pmichaud_ here's my report for today: 18:27
What I did (since 2009-07-14):
* Rakudo now passing 12,043 spectests (+259 since 07/14)
* Delegated the July Rakudo release to moritz++
* Presentations on Rakudo and Parrot Compiler Toolkit at OSCON
* Fixed PGE so that calls to undefined regexes throw a better exception than "Null PMC Error"
* Create a "ins2" branch in Rakudo that can build Rakudo and do "make install" using install-dev Parrot.
* Fixed a problem where Rakudo was relying on ordering of attributes.
* Fixed a variety of Rakudo bugs in quoting, operator precedence, stringification, etc.
* Fixed Code.ACCEPTS.
What I'm doing this week:
* Working on REPL loop for Rakudo PCT::HLLCompiler
* PGE refactors
* Adding grammars and regexes to NQP
* Continuing work on "make install" in Rakudo's ins2 branch
What I'm blocking on:
* Insufficient waking hours
pmichaud EOR
Also for "What I'm doing this week": Making preparations for the Rakudo BOF at YAPC::EU 18:29
(real EOR)
18:34 eternaleye joined
Whiteknight What I did: 18:34
pmichaud #ps in -4?
Whiteknight Ripped out stacks and stack related ops
EOR
cotto HELLO
NotFound Hola
Whiteknight is there #ps today?
allison hi
Util Hi
darbelo Hi
Whiteknight hi
allison yes, I was just waiting for chromatic, but we can get started
pmichaud Hi 18:35
allison has everyone reported?
cotto q1q
Whiteknight reported too little, too late
allison okay, questions. cotto?
cotto I don't think we ever went through the roadmap items for 1.4.0.
japhb q1q
allison cotto: we reviewed them every week in the month before the release. Or do you mean something else? 18:36
japhb actually, nevermind, I'll just discuss it in #parrot instead.
cotto the milestone is still active. istr that I didn't have permission to close it.
(or couldn't figure out how to close it) 18:37
18:37 chromatic joined
allison cotto: ah, I can close it, but you should have permission too 18:37
I'll msg you now
cotto k
eoq then
allison other q's? 18:38
pmichaud cotto++ # excellent release
Whiteknight cotto++
NotFound I don't remember if I already asked this: can str_dup be deleted?
chromatic Did everyone see that we have Coverity reports again? 18:39
pmichaud chromatic: I saw it, yes.
Whiteknight q1q
Util q1q 18:40
chromatic If anyone wants access, let me know and I'll forward that to Coverity. EOQ
Util cancel query 18:41
allison ok, Whiteknight, question?
allison forgot to mention that the PIR book is up on Amazon now 18:42
pmichaud url?
Util NeonOnyx++
allison www.amazon.com/dp/0977920127 18:43
pmichaud orders a copy. 18:44
I've _always_ wanted to learn how to program in PIR!! :-)
Whiteknight I wanted to ask about pushaction, pushmark, popmark
I don't think anybody is using them, and I would like to deprecate them
pmichaud Whiteknight: I plan to use them, if we don't end up with a better rollback facility soon.
Whiteknight okay, that's all I needed to hear. They still exist but were reimplemented to not use the stacks system (which is now gone) 18:45
allison Whiteknight: they're changing to a context local stack rather than a global stack, but they stay
Whiteknight okay. Awesome
allison Whiteknight: we could likely come up with clearer names, though
pmichaud (clearer names)++ 18:46
Util q1q
Whiteknight We should add these kinds of tasks to TT #876
EOQ
pmichaud (Allison: If you're not going to be at YAPC::EU, how am I going to get you to autography my copy of the book? ;-)
*autograph
allison pmichaud: :)
Util, go ahead
Util Is there a good explanation of the L1 initiative anywhere? I am getting questions from Atlanta.pm, and am not up to the task. 18:47
Whiteknight There was something up on the wiki, but probably needs more
moritz Util: purl's reply in #parrot contains some useful links
Whiteknight it's not an "initiative" so much as a "nice idea" right now 18:48
Util thanks; eoq
allison Util: aye, I saw your report. It's tough for folks outside the project to get excited about low-level internals like Lorito (L1) 18:49
Util: they're mostly still at the level of getting excited about the general idea of Parrot
Util: fortunately, there's lots to talk about there 18:50
Whiteknight right, it's definitely not a topic that will interest beginners
Util No, at least one of them follows all the #parrot daily
NotFound Lorito?
pmichaud "Lorito" == tentative name for L1
chromatic Allison doesn't like the name L1.
allison chromatic picked the name Lorito
NotFound Then I must rename examples/embed/lorito.c to avoid confusions.
Util allison: but yes, for the most part. 18:51
chromatic NotFound's Spanish is too close to my Spanish!
allison L1 is unmemorable, and too confusing with general parser terminology, LL(1), etc 18:52
other questions?
NotFound I'll be happy to give up the name for that better usage :)
allison roadmap/milestone review 18:53
trac.parrot.org/parrot/report/14
export conventions (cross hll) 18:54
is that done? or, still work ahead?
pmichaud still work ahead, I think. 18:55
I didn't finish documenting in time for the release.
allison combine that with pdd31-hll interop, and hl interop
should we move it to 1.5 or 1.6?
pmichaud (instead, I fixed Rakudo so that it would continue to work with the release)
Let's say 1.6
if only because I already know how busy August is.
allison yes, that was a more important priority
okay, moving all three to 1.6 18:56
how about "implement pdd14-numbers"?
Whiteknight who is even working on that? 18:57
allison it says particle, but I don't remember it being on the 1.4 list (it's not a high priority) 18:58
quick show of hands, do we remove it from the roadmap since it has no people? or, reassign it later?
cotto It sounds like a reasonable task for a dedicated newcomer. 18:59
particle actually, i think that task is mostly modifying pdd14 to reflect the implementation 19:00
cotto up until the "implement it" part
chromatic We need someone to write up how to implement it in that case.
allison cotto: yes, the ticket stays, but "a wishlist task a newcomer might take on" isn't a roadmap task
jonathan How does this relate (if at all) to the work on the decnum things?
pmichaud is it important that it be on the roadmap? 19:01
I don't know that it affects me directly (from a Rakudo or PCT perspective)
jonathan We need to be able to support bignums and rationals at some point.
But it's not exactly Rakudo's most pressing issue right now.
cotto jonathan, The decnum pms are dynpmcs because we can't redistribute the decNumber code (and it's not designed to be packaged as a dynamic librarty).
s/rart/rar/
19:02 quek left
jonathan Does that mean that Parrot plans to implement some fallback to them? 19:02
einstein may i put some attention on the ticket #880 and the comment of Whitenight
darbelo jonathan: I based a big chink of the original proposal on it, but that was before deciding to make decnum aseparate package. They have diverged a bit now.
allison pmichaud: I don't see it as that important (though nice to have someday), so happy to remove it from the roadmap
darbelo s/chink/chunk/
jonathan Would there be any licensing issues arising if, say, Rakudo was to use the decnum dynpms and have them as a dependency? 19:03
(Or <insert HLL here>?)
Whiteknight jonathan: no, unless you're including it in your repo 19:04
jonathan OK, thanks.
cotto relatedly, where's a good place to read up on such issues? I don't understand the reason Parrot can't distribute decNumber. 19:05
allison okay, I'm unflagging pdd14 on the roadmap
(can be reinstated later if needed)
pir profiling tools? 19:06
not done, schedule to 1.7?
chromatic cotto and I are working on those. We hope to have them by 1.5, plan to have them by 1.6.
The sooner the better for Rakudo, Lua, and Partcl.
allison okay, will schedule at 1.5, and can move them later as needed
pir debugger, docs? 19:07
not done, reschedule to when? 19:08
NotFound I plan to so some work with the debugger when the pcc rewiring is done.
allison okay, we'll say 1.7
standard seed libraries (postgres, opengl, etc)? 19:09
japhb allison: what about them?
allison I saw good work going in on opengl
japhb: these are roadmap items
this is a task that really needs to be split into monthlies
japhb chuckles ... I had no idea my code had become a roadmap item .... 19:10
allison like, pick 6 seed libraries and do one a month between now and 2.0
for now I'll reschedule to 1.5 so we keep thinking about it
Whiteknight speaking of 1.5, I'll play release manager unless somebody else wants it 19:11
japhb allison: If the idea is that they be *complete* at some point ... OpenGL completion is blocked on stuff Parrot NCI can't yet do.
(And has been for ages)
allison japhb: (it was on the roadmap since last november. It's a good sign of a good roadmap if a task seems like the right thing to do at the time even if someone isn't looking at the roadmap)
japhb :-) 19:12
NotFound According some quick discussion and fixing before the release, the only HLL that does something with it is lua.
it = opengl
japhb When is Parrot NCI rework scheduled? After PCC rewire is complete?
Util japhb: is there a ticket for that missing NCI stuff?
allison japhb: okay, well the fact that it's on the roadmap gives higher priority to fixing those sorts of roadblocks
japhb NotFound: ... and Rakudo, where it Just Works.
NotFound japhb: I think someone said it uses by mean of embedding pir, or something like that. 19:13
japhb Util: ISTR there was something in RT for at least one of the problems, .... but that's clearly != all of them.
NotFound: Lua re-wraps the OpenGL binding to fit Lua's OpenGL binding, which is very DWIMy, I believe. 19:14
allison On to 1.5 roadmap tasks (a quick triage)
packfile pmcs?
do we have anyone working on it? reschedule or remove from roadmap?
japhb NotFound: If you're referring to Rakudo, Rakudo has not needed Q:PIR to access OpenGL for a bit now, thanks to the nascent hll-interop. 19:15
allison any objections to moving packfile pmcs to 2.6? 19:16
NotFound japhb: fine, I must have misunderstood something.
allison (that is, it needs to be done, but not urgently, and not a priority for 2.0)
chromatic I thought we had them now. 19:17
allison mmmm... looking the ticket it might be closable
chromatic I think they're also a prerequisite for PIRC and PCT -> PBC.
pmichaud my understanding is they were needed for pirc, yets. 19:18
*yes
allison and are they done?
pmichaud not as far as I know... the pirc folks can answer that. 19:19
chromatic In kj's absence, let's find the author and ask him.
jonathan I think that a lot of work has been done by bacek++ and Infinoid++ and maybe others on packfile PMCs.
allison okay, will hold the ticket in 1.5 for now, this week we can get more details from people who aren't here at the moment
vtable swap (e.g. bignum)? 19:20
Do we plan to complete this for 2.0?
or, after 2.0?
chromatic I think it's doable. I'll talk to bacek about his Key changes; it might be less useful now.
jonathan Is there a write-up of what this really means?
I'm guessing there's more to it than just "change a PMCs vtable point" :-) 19:21
erm, pointer
allison chromatic: doable, but is it a priority?
chromatic It could simplify vtable code substantially.
I *know* there are bugs and malfeasances in the Key code due to their complexity.
allison does that work toward our "production users" vision? 19:22
chromatic Only in the quality of implementation and ease of writing new code sense.
allison (I'm happy either way, but I don't want to keep roadmap tasks that will distract from the goal rather than aiding it) 19:23
chromatic Let's leave it untouched until I talk to bacek.
allison sounds good
chromatic We can reschedule or remove it if it's less necessary now.
allison pirc?
do we expect that for 2.0? 19:24
it's an ongoing task, but doesn't really need to be tied to a particular deadline 19:25
chromatic The bus number is too low for my comfort there.
allison and it's more of an internals cleanup than user facing
reschedule to 2.6?
chromatic We need to talk to kj first. 19:26
allison okay, can you talk to kj and bacek about those two items this week? 19:27
(want to make sure someone has it on their todo list)
chromatic Will do.
allison the last item "pluggable runloops"
is that in progress now? 19:28
chromatic Yes.
cotto That'll happen as part of profiling.
particle JOY 19:29
allison excellent
we'll keep that in 1.5
next week we'll break up the 2.0 tasks over the next 6 months
(this is enough roadmap review for one day)
anything else before we close? 19:30
wunderbar, thanks everybody! 19:31
19:31 Util left, PacoLinux left 19:32 NotFound left, chromatic left 19:35 jonathan left 19:36 darbelo left 23:16 Whiteknight joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!