Parrotsketch Every Tuesday @ 18:30 UTC | IRC Log at irclog.perlgeek.de/parrotsketch/today
Set by moderator on 8 December 2008.
04:13 contingencyplan joined 07:24 davidfetter joined 08:31 contingencyplan joined 13:12 Wknight8111 joined 13:37 contingencyplan joined 16:24 contingencyplan joined 17:24 pmichaud joined 17:26 jhorwitz joined 17:36 mberends joined, PacoLinux joined 17:55 allison joined 18:12 Whiteknight joined 18:21 Coke joined 18:23 moritz joined 18:27 Infinoid joined
Infinoid *lurk* 18:27
18:27 chromatic joined
Coke flips chromatic for running the meeting. 18:28
Whiteknight like a cat, chromatic always lands on his feet after a flipping
chromatic Not when I fall down the stairs. 18:29
Then again, I've seen my cats fall down the stairs.
18:29 kj joined, cotto joined
Whiteknight stamps FAIL on that whole mental image 18:29
chromatic 1 minutes 18:30
18:30 jonathan joined
moritz OH HAI 18:31
particle ~~
kj hi
pmichaud present.
jonathan ahoj
jhorwitz alo
chromatic Shall we start?
allison, are you here? 18:32
allison yes
is that a "go ahead and report?"
- Resolved a few issues with the I/O branch after merging it in.
- Created an Ubuntu package archive for Parrot, which allows us to create automated regular builds of Ubuntu packages from the repository (nightly/weekly). 18:33
- Found a few issues while doing that, which I'll check in after the release today.
- Creating a branch for string handles (filehandles that just read/write to a string)
- Starting up the GC branch right after the release today.
EOR
chromatic I fixed a couple of bugs and reviewed the numbers PDD. 18:34
I haven't finished reviewing opcodes for potential deprecations; I probably won't have time until this afternoon.
I also failed to write a journal imploring people to help us review TODO/SKIPs, but I'll do that too.
Coke?
Coke Parrot: 18:35
- doc updates to steer away from RT and to Trac.
- added chromatic++'s 'vgp' tool to the repository
- fix 'examples_tests'
- trac wiki/trac ticket/RT mongering
- remove crufty PLATFORM reports
Partcl:
- trying to get partcl working again against parrot head; many segfaults in the way.
- Need to rethink strategy for targeting versions of parrot. :|
.
chromatic cotto?
cotto * got a Coverity u/p, fixed or ignored a bunch of reports
* ready to remove hash seed from PARROT_EXPORT after the release today
* found a bug in Hash, no luck squashing it, will send to list
.eor
chromatic Infinoid?
allison queues one q 18:36
Infinoid some patch mongering, a bunch of codingstd tweaks.
q1q
^D
chromatic japhb?
jhorwitz? 18:37
jhorwitz mod_parrot 0.5 release is delayed due to problems with the pdd22io merge. wanted to subclass FileHandle but it fails miserably. working with allison to help work around that.
implemented some workarounds for recent rakudo and parrot changes.
various foundation calls to deal with financials.
EOR
chromatic jonathan?
jonathan * Finished up the MMD grant from DeepText and wrote/blogged final report
* Got a Hague grant to Do Stuff in Rakudo - blogged a project plan
* Started working on it; got the type registry pretty much working in a branch (all spectests passing, but need to fix up use - discussing that with pmichaud ATM)
* A few tests we passed for the wrong reasons showed up - did some of those fallout fixes, other casses we'll fudge the tests for now
* Got hypers and various other IO bits done in my Rakudo day last week too.
.end
chromatic kj?
kj == PIRC 18:38
+ various refactoring.
== IMCC
+ removed .namespace <id> / .endnamespace <id> sugar.
== Questions
+ I have 4 quick ones.
.end
chromatic moritz?
moritz * worked on the Perl 6 test suite to get us past 5k passing tests for today's
release
* succeeded.
.end
chromatic particle?
particle ~ got expense form created for pafo, dealt with other foundation activities
~ pds, ppw, and other expenses must be submitted using this form
~ fixed a bunch of broken tests in anticipation of the release
~ looking at a few more on strawberry perl today
~ shepherded some wiki contributors
~ gave up on getting calling_conventions branch working for the release, we'll have to reschedule that roadmap item, no solution for failing tests there, seems a context issue
.end
chromatic PerlJam?
pmichaud? 18:39
pmichaud ** Rakudo spectests (r33949): 249 files, 5139 passing, 0 failing
+334 passing tests since last #ps.
== Parrot stuff
: more subid improvements with tewk++
== PGE stuff
: switched to use :subid
: added <?{{...}}> and <!{{...}}> code assertions (PIR code)
== PCT stuff 18:40
: fixed interactive mode handling post-pdd22io merge
: suggestions on PAST::Control block handling
: updated to make full use of :subid
== Rakudo stuff
: resolved 18 RT tickets
: refactored array, item, list, hash handling for builtins
: fixed list and hash flattening and interpolation
: fixed postcircumfix:<[ ]> to work with non-Rakudo arrays
: added new 0-ary 1-ary block handling to sort
: refactored sort to be stable sort
: added comparison and sorting of Pairs
: added hash slicing
: fixed list assignment of my/our declarators
: generated some articles, statistics on Rakudo progress
: revised test_summary.pl script to simplify graphing, provide more data
: fixed handling of undef/Failure
: added Nil type, handling of "bare return"
EOR
chromatic Tene?
Whiteknight? 18:41
Whiteknight * Prepared for and practiced release
* Lots of testing, various configurations and platforms
* A little bit of nagging on list (sorry!)
* Some tinkering on the book
* Getting my hands on some versions of Borland's compiler for testing
chromatic Did I miss anyone? 18:42
Let's go to question time then. allison? 18:43
allison My q was for Coke, on languages targeting versions of parrot.
The ultimate plan is that languages will target the 6-monthly releases, but the monthly releases will end up being the "release candidates" for those 6-monthly releases. 18:44
Coke: Is there anything we can do to make it easier? You're the test subject paving the way for future languages. 18:45
Coke well, step one, document the plan. =-)
I was unaware about the six-month target release for HLLs. 18:46
but I suspect that it's different pre-1.0
allison that's my task today. At least, the first step of posting something to parrot.org
Coke just added a LeaveTheNest wiki page today.
allison 1.0 is the first of the 6-month releases
cotto q1q
allison so, the next few monthly releases are the "release candidates" for 1.0
Coke ok. well, before then, I'd like to see the monthly releases pass all tests for partcl; 18:47
I'm not sure what I can do other than report all the bugs I find.
pmichaud Rakudo will likely to continue to choose to target monthly releases.
Coke (and fix what I can. but most of them require C-fu I don't have.)
I'm not really in a worse state than I was when I was -in- the repository, fwiw. 18:48
kj although it's mainly Parrot-fu, not C-fu..
allison Coke: it's a tough choice for tcl, because targeting 0.8.2 (for example) would mean that it doesn't work with 0.8.1
Coke allison: yes. I'm probably going to roll partcl trunk back to working with 0.8.1 and hope for 0.8.3
allison the branch idea seemed a good start, creating a branch for the 0.8.2 release 18:49
Coke or perhaps make trunk try to work with parrot svn head, and have a 'stable' branch that works with the last declared version.
(stable branch /in partcl/)
pmichaud stable++
allison expecting any version of the tcl code to pass all tests in both 0.8.1 and 0.8.2 just isn't going to happen
Coke right.
allison and probably won't happen anytime in the next year or two
Coke which won't happen? 18:50
allison (though the changes will certainly slow down)
Coke oh, being able to target 2 consecutive releases? Yah.
I would just like it to be the most recent one.
particle a better name for 'stable' might be 'parrot-0.8.1'
allison I mean, the ability to guarantee that some body of code running on parrot will work without changes in two consecutive releases
yes
particle create maintenance branches for each parrot release
Coke particle: do you know how much work that is? 18:51
allison yeah, that's not worth it
Coke I'm having enough trouble /treading water/
particle you don't need to continue to support old versions
allison 'stable' and 'unstable' (or 'developing') work
particle it's just a better name than 'stable', which doesn't tell you what's considered stable
Coke that's what the wiki is for.
allison particle: true, but then users don't need to understand the guts 18:52
Coke code.google.com/p/partcl/wiki/PartclSource will document what to do.
allison particle: they just need enough info to know "I want the stable version, because it's less likely to have problems"
Coke I'll just split it up into 2 sections: developers and users.
allison ah, very nice
Coke q1q 18:53
allison ultimately, I'd figure "users" would be on 1.0, 1.5, 2.0, 2.5, etc, while developers will be on the monthlies
EOQ
chromatic cotto? 18:54
cotto What's the plan for dealing with Trac spam (both preventative and cleanup)?
allison cotto: are we getting it already?
cotto yes
allison I can turn off anonymous access, that'll help
jonathan But what about people who want to report bugs?
Will that not hinder them?
pmichaud "let 'em register"
allison it takes 2 seconds to create an account
jonathan OK, if registration is easy, that's fine. 18:55
I haven't tried it yet, so I don't know. :-)
pmichaud that's what I've seen for most bug trackers i've worked with -- required registration (but something that isn't onerous)
particle i'm working with osuosl to get trac upgraded, and some plugins/scripts installed
cotto Could Parrot devs get admin privileges so we can delete spam tickets instead of marking them invalid? 18:56
particle 1) allow ticket updates from email, 2) reduce/manage ticket spam
pmichaud queue 1 question
Infinoid (ticket updates from email)++
cotto +1
chromatic Coke?
Tene is here 18:57
allison cotto: certainly, just send me user names and I'll add you to the group
chromatic Coke question going twice.... 18:58
Coke er.
2 parter.
1) can everyone with a bug queue in RT please find a ticket to close today? there are some low hanging fruit still in there; you can probably reject a ticket, especially if you're allison. 18:59
2) can developers please stop creating RTs? I'm looking at you, fperrad.
allison Coke: ok :)
Coke eq
chromatic Tene?
Tene Lots of exceptions work.
Exception handlers in PCT
Exception handlers in Rakudo
nonfatal exceptions automatically resume 19:00
given/when is much nicer
Probably something else? Maybe.
KTHXBAI
cotto exceptional work
chromatic Infinoid?
Infinoid I see a mention of configuring with --optimize in PLATFORMS, suggesting it works on x86, but I wonder how often that's tested.
Do we expect Configure.pl --optimize to work anywhere else? (If not, is that documented anywhere?)
I'm asking because we've had a few tickets crop up due to --optimize causing segfaults on x86-64.
chromatic I build and test with it on x86-32 Linux. 19:01
Infinoid I can reproduce crashes with it on 64 bit ubuntu but not 64-bit gentoo
in a followup to one of those tickets (61242), NWCLARK suggested we put in assert()s to enforce the NONNULL flag. that sounds like a very non-arch-specific problem to me.
think its worth tweaking headerizer to do something like that?
chromatic Yes. 19:02
Infinoid ok, thanks. in the meantime, it might be worth documenting that it's problematic on some platforms. EOQ 19:03
chromatic kj?
kj hi yes
I have 4 short questions
1)
box opcode: 19:04
does it respect .HLL_map, (I think not), and if not, should it?
chromatic It does.
kj iow: would it make sense
oh ok.
2:
chromatic I can tell you exactly where I was when I wrote that part :)
kj ha ha ha
pmichaud that's the whole purpose of 'box'
kj oh ok I just figured it was a shortcut, but makes sense then. 19:05
2:
:-)
2: string index sugar
eh, so $S0[1], it now maps to the substr op
and in pdd19 it says it "should" be deprecated. Just want to do a last-minute check
before I rip it out (after release)
it's not in DEPRECATED.pod 19:06
allison kj: then add it to DEPRECATED.pod now
kj ISTR it was decided earlier it should be removed, but just want to make sure.
chromatic +1
kj ok. will do.
3:
allison (if you get it in before the release, then you can rip it out right after)
kj what *is* in DEPRECATED, is [a..b] syntax 19:07
so, slices. it's deprecated in favor of slice op.
is slice op working correctly? I remember it says *somewhere* that its implementation is experimental/wonky
chromatic I know that IMCC has some weird comments about slicing, but I think that's just IMCC. 19:08
kj any ideas? Thing is, I've tried to run make test after remvoing it, but there's also several test files.
yes, I wasn't planning on cleaning up all the slicing code. Just the grammar
I don't dare to touch the other stuff.
Tene 1 item for pmichaud 19:09
allison kj: as far as I know, 'slice' is still experimental/buggy
pmichaud I think if Parrot is to support slices natively (e.g., via a slice opcode), then we should do it after we have better use cases and not a-priori as it exists now.
allison kj: and that's one I'd be happy to move over to a method instead of an op
pmichaud i.e., I'm in favor of ripping out that op and syntax.
allison yeah, it's very dependent in the implementation of the particular PMC 19:10
s/in/on/
chromatic and it makes Key even uglier
kj so, basically we remove /all/ slicing implementations, both imcc grammar and slice op?
pmichaud +1
allison kj: yup
pmichaud also, note that rakudo is doing slicing just fine via methods at the moment 19:11
kj oki. so I should put in a deprecation for slice op as well?
eh, in DEPRECATED
allison kj: do it 19:12
pmichaud +1
kj okido.
Q4:
on the roadmap it says something like *pdd19_pir, user docs.
Whiteknight is doing a good job on his book 19:13
iow, I hate duplication.
allison yes, the book can be considered user-documentation for the project 19:14
chromatic +1
Whiteknight thanks!
allison (as long as it's under a license we can use, etc)
kj so, basically, that task has been half finished. only pdd19 item stays 19:15
allison kj: which is a quick review for how well it matches reality
kj yes. it shouldn't take too long 19:16
particle q1q
kj I think that wraps it up for me. Thanks!
chromatic particle?
particle the roadmap is not up to date. 19:17
at least, we don't have 'landed', 'missed', 'dropped' on all items
trac.parrot.org/parrot/wiki/ParrotRoadmap
Whiteknight: how long until you start the release? 19:18
in the meantime, will any of these get done?
Whiteknight 5 hours, rough estimae
estimate*
will probably happen after dinner, EST
Coke (parrotroadmap) the entire wiki page should be replaced by real tickets.
at which point the status listed on that page won't matter so much. 19:19
pmichaud (real tickets)++
particle coke: agreed, but that's for another day
chromatic I like having the entire view in one page, but I like tickets too.
Coke which will hopefully be here in the nesxt few days.
allison Coke: not replaced, just add links
pmichaud we can at least make tickets for the next release.
Coke chromatic: with trac's actual roadmap you get both.
allison: yes, replaced.
see:
chromatic Both++
Coke trac.parrot.org/parrot/roadmap
pmichaud trac will create a roadmap from tickets. 19:20
Coke (that's automatically generated based on tickets that have milestones set.)
particle but says nothing about their status
Tene opcode review could potentially take less than five hours, if someone's available to do it.
allison Coke: yes, but there's still no way to easily see the overview list
chromatic I can help, but I have a lot of work to do in the next couple of hours.
allison Coke: or to see, at a glance, all the features that are ahead
chromatic I'll work on the numbers PDD instead though. 19:21
That's very low risk.
pmichaud tickets by milestone: trac.parrot.org/parrot/report/3
particle seems numbers and opcode review will affect the repo/tarball
pmichaud (that's the overview list)
Tene pct exception loops would be approximately five hours, but significant change right before release is potentially dangerous.
Also, I'm at $realjob
pmichaud exception loops moves to jan release.
(but will occur much sooner than that.) 19:22
allison pmichaud: okay, that's tolerable, if we maintain it
Tene Speaking of "moved to january release", when will you update TT #25, pmichaud?
Coke we have to maintain something. might as well just be one thing. 19:23
allison we'll need to add a status that says "this is one of the critical tasks for this milestone"
Coke priority: release
we have blocker.
pmichaud the tickets already have a "critical" flag on them -- we just need to edit the report to add that into the output. 19:24
Coke That's probably sufficient.
pmichaud (and color indicates "priority" or something like that.)
Coke pmichaud: added priority column to the report. 19:25
allison I added a 'release' severity, so we can have a "critical release" ticket
Coke and it'll show on the report. whee.
alright. chromatic has to leave soon. I have lost trac of whose question we're on. 19:26
(I know who'se next, though.)
pmichaud Tene: I've moved TT #25 to the jan release.
Tene pmichaud: I know. I'm just wondering when you'll update it with prerequisites.
pmichaud probably tomorrow, at the current rate I'm going. 19:27
there aren't many prereqs in PCT itself. 19:28
it's also related to getting the .include specifications to work.
Tene Which is no longer a blocker.
... Oh, I should add CONTROL blocks to rakudo before the release. 19:29
pmichaud no, I mean that whatever handles the HLL stuff will also handle .include specificatoins
the fact that PAST::Control doesn't need those is ancillary
Tene Right.
pmichaud Tene: don't add CONTROL blocks until we have an answer back from p6l about how they work.
Tene Okay. 19:30
Coke parrot roadmap: particle, you all set?
particle i'm not happy that we're missing so many items this month
but i can't do anything about it atm, and don't think we should hold the release
Coke You guys set a very aggressive schedule at the conference, it seems.
no, we definitely should not hold the relase. 19:31
particle yes, and we're already slipping.
pmichaud it has to be aggressive to make March.
Coke I'm still wondering why March. =-)
pmichaud the opcode review doesn't have to be done in any particular month, it's not critical or a blocker for any task. That one doesn't bother me.
particle it has to be sometime.
Coke so, next month might be 0.8.3 unless we make up some ground.
particle pmichaud: agreed, opcode review can happen anytime, *except* it has ramifications wrt deprecations
pmichaud loop+exceptions is important to HLLs but also not a blocker. 19:32
Tene I didn't have much scheduled for this coming month, so my work on the bumped items won't be taking away from other items scheduled for Jan.
allison it might help if we broke the opcode review into smaller steps
kj particle: 'so many items': I see 2 'missed'
chromatic calling_conventions is the only one that bothers me.
allison like, reviewing the 'experimental' opcodes, cleaning up the ones that aren't in ops.num anyway
particle kj: and 2 'on track' that will be 'missed'
pmichaud calling conventions is the one I'm concerned about as well.
particle kj: and one that's already been moved to jan 19:33
allison we only put it in this month because it was pretty much done
pmichaud same for pct honors hll, fwiw.
particle pmichaud: we could use your help with calling_conventions branch, seems the failures are context related
nobody else understands that well enough to have spotted the bugs yet
pmichaud particle: I need test cases that fail. Other than that I can't really help much.
they don't fail on my boxes.
allison particle: it's also possible that it's just a bad merge
particle i have them, and we have crossloop 19:34
pmichaud then we need a time to look at them.
particle pmichaud: you should get failures on windows
ok, i'm here all week
ok, next week's meeting?
allison I have a clean diff of the changes in calling_conventions, that we can apply to a fresh branch
particle cancelled, i suppose.
pmichaud I'll be here next week. 19:35
Coke I would recommend we skip.
particle ok, i won't. i forgot, it's only tuesday, not wednesday
Coke but if folks are here and wish to chat, by all means. =-)
particle allison: let's do that asap, so i can test
Coke we still have 2 questions.
pmichaud (also, when I made the schedule for dec release, I _wasn't_ expecting that lexicals would take me yet another full-time week to complete, which is why I'm about a week behind on other tasks.)
allison particle: okay, creating now 19:36
Whiteknight help on that branch would be hugely appreciated, because there is a lot more calling-conventions work that we would like to tackle next
not to mention the GC work that I want to participate in
Coke alright. We've done as good as we're going to for this release modulo some last minute fixups, I think. 19:37
particle i *may* get the strawberry -0 bugs fixed
Coke pmichaud; you still have a pending q?
pmichaud who has admin privileges to parrot-tickets mailing list? That person can disable submissions to that list very quickly. Or if I get the password I can do it. 19:38
allison pmichaud: sure, and I'm happy for any other moderator volunteers too 19:39
(on any of the lists)
Coke let's disable it for now and re-enable it when our hosts get trac2email working.
pmichaud +1
Coke (or just just before they do.)
ok. That leaves Tene, I think. 19:40
Tene What about me?
Oh, already asked.
pmichaud (I think his question was about my updating TT #25) 19:41
Coke k.
We have someone to push this last release to cpan?
chromatic In theory, that's Whiteknight. 19:42
Coke and Whiteknight, any last minute blocks, I'm sure someone on #parrot or the list can help.
chromatic In practice, is it you, Whiteknight?
Coke chromatic: you're gone, aren't you? =-)
chromatic No, our conference system is full so I can't get to the meeting and had to reschedule it for tomorrow, when I only had two or three existing meetings.
Coke ok. I thought I recalled a snafu there. Going forward (after this release), we'll cut releases via parrot.org
Anyone else? 19:43
allison chromatic: no, Whiteknight can't do the CPAN upload, needs someone else to do it for him
particle including this release, we'll go to parrot.org
chromatic CPAN uploads are easy. I can do it if necessary.
Coke particle: yup, thanks for clarifying.
chromatic: danke.
Anyone else?
Bueller?
Alright. Catch as catch can for a meeting next week; definitely see everyone in 2. 19:44
*gong*
19:44 chromatic left, pmichaud left
allison thanks c and c 19:44
19:44 allison left 19:45 moritz left 19:46 Coke left, kj left 19:48 cotto left 19:49 Infinoid left 20:14 mberends left 20:29 jhorwitz left 20:48 jonathan left 22:24 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!