"Tuesday at 20:30 UTC"
Set by moderator on 20 April 2010.
00:16 eternaleye joined 01:07 eternaleye joined 05:24 eternaleye joined 06:45 eternaleye joined 07:33 eternaleye joined 08:02 eternaleye joined 08:30 wagle joined 12:00 bluescreen joined 14:01 plobsing joined
plobsing won't make #ps, pre-posting 14:02
What I Did:
* discussed PBC strings optimization options
* implemented and merged quick solution (pbc_frozen_strings1)
* PBC smaller now, faster to read, but takes much longer to compile
What I Plan:
* Mitigate PBC compilation costs
* Maybe clean up IMCC a little along the way. Maybe unicorns too.
* Get more benefits out of the increased coupling between freezing and PBC 14:03
EOR
14:51 cotto joined 16:27 cotto_work joined 17:05 bluescreen joined 17:34 chromatic joined 17:49 darbelo joined 17:51 mikehh joined
darbelo done { 17:57
Started a wiki page to help GSoCers get started.
Started blogging about NFG.
Got bound to the community ;)
Updated decnum-dynpmcs to be explicit about automatic or manual attrs.
Jumped on git-svn bandwagon, not sure I like it.
}
todo {
Submit GSoC documentation to Google. Waiting on the 2010 forms getting posted.
More blogging. This time of a more concrete nature.
Maybe do some exploratory coding.
}
eor
18:04 NotFound joined 18:09 dukeleto joined
NotFound What I did: 18:17
-parrot
* Created PackfileDebug PMC.
* Implemented vtable init_int in Integer and PMCProxy, using it
wherever possible to avoid temporary objects and shorten up code.
* Update some examples to stop using deprecated or gone features.
* Improved some tests.
* Minor fixes and cleaning.
-winxed
* Fixed usages of parrot deprecated out features.
* Predefied constant __STAGE__.
* Prdefined functions upcase, downcase and titlecase.
* Class scope const now works in stage 1.
* Minor fixes and improvements.
What I will do:
No plan
EOR
dukeleto What I did: 18:24
* Made the open opcode delegate to FileHandle.open to allow intercepting by PL/Parrot
* Added security features to PL/Parrot, specifically, intercepting the open opcode and the open method on the File and FileHandle PMC's
* Added a test to t/src/embed.t for Parrot_compile_string trac.parrot.org/parrot/changeset/46208
* Found lots of bugs in Parrot_compile_string
What I will do:
* Create TT's for each bug I found in Parrot_compile_string and write tests.
* Hopefully fix the implementation of Parrot_compile_string (currently syntax errors in IMCC makes Parrot_compile_string throw an error, instead of populating the error string. chromatic++ gave me some advice on how to fix this)
* Do some stuff related to GSoC 18:25
Blockers:
* Stuff and things
EOR
mikehh What I did since my last report: 19:00
* building and testing parrot on amd64/i386, with gcc/g++
* fixed examples_tests to work with in_place string ops
* some other fixes
* closed TT #1132, #1368 and #1600
What I intend to do in the next week:
* testing and fixing
* document what I needed to get parrot building and testing with new install of Ubuntu 10.04 and Perl 5.12.0
.eor
19:08 Coke joined
Coke parrot: - discussion about migration to git. - apply patches 19:08
- cage cleanups, ticket wrangling.
- make OS and File Dynpmcs
- rename Parrot_string_cstring to Parrot_str_cstring
- rename STRING_is_null to Parrot_str_is_null
- remove pushaction, pushmark, popmark opcodes
- made auto/manual_attrs required
- create branches/codestring to prototype a stringbuilder-like CodeString 19:09
PMC that avoids string concat for 'emit'()
- found segfault fixed by NotFound++
partcl-nqp:
- make it work with parrot-latest
.
19:23 cotto_work2 joined 19:30 moritz joined
cotto_work #did: 19:33
* figured out how to get trac to support multiple vcs backends
- getting trac to support links to multiple vcs backends is a smop at this point
- kicked off discussion about scm migration 19:34
#will do:
* no plan
#closed TTs:
* #1592 (t/op/io.t fails in non optimized build, closed for real)
#eor
q2q (also, I may be unresponsive starting at #ps+30 for 30m +/- due to a meeting)
moritz # did: kept rakudo up to date with current parrot 19:35
rakudo has two problems with current parrot
1) pbc_to_exe needs >1GB memory on 64bit linux platforms
2) on the same platform spectests randomly segfault 19:36
# EOF
dukeleto hello
dukeleto cheers for free wifi at Ithaca airport
cotto_work cheers for dukeleto being early 19:37
Coke "parrotsketch is at the next half hour.
dukeleto may miss #ps due to boarding :) 19:43
Util # Done: 20:14
* Fixed typos in book.
# Plan to do:
* No time; No plan.
# Blockers:
* $WORK
.end
chromatic * worked on CodeString, in trunk and on branch, and it's much faster now 20:15
* worked on speedups of plobsing's constant STRING in PBC branch, but it needs more
* no progress on the GC, unless you count talking about it
* will continue to profile and encourage other people to help optimize
* need to become an awesome mentor to tcurtis, but blocking on time
dukeleto will be missing the meeting, talk to y'all later 20:17
20:25 allison joined
allison What I did: 20:28
- Finished off the lightweight language.
Blockers:
- Got very little time on Parrot this week because of 3 big end-of-year university projects due this week. The last one goes in on Thursday, so I'll have time again after that.
EOR
chromatic #ps in 2
cotto_work hello 20:30
darbelo Hola.
mikehh hi there
chromatic Hello.
NotFound Hola 20:31
Coke ~~
20:31 bubaflub joined
chromatic Let's review last week's goals. 20:31
How are we doing on removing deprecated items?
mikehh lots gone 20:32
chromatic More to go? 20:33
darbelo If there's anything left it's probably small.
chromatic What branch merges do we have outstanding? I know of codestring 20:34
Any progress on the exceptions_refactor branch? Tene? 20:36
Tene chromatic: None.
chromatic Any blockers for you we can help remove?
Tene Sleep?
chromatic Possibly. 20:37
Tene I was attending a convention the past week and dealing with personal drama. That's mostly gone, and I don't have any other blockers to getting anything done.
chromatic Let's talk about priorities for the coming week. Any suggestions?
mikehh none of the other branches have been worked on in the last week
chromatic I'd like to see the codestring branch merge. 20:39
allison I'd say we can move on from the deprecations
chromatic I'd also like to see us fix -o PBC performance.
allison do we have volunteers willing to work on merging the codestring branch?
mikehh I'll do any testing needed 20:40
chromatic Rakudo testing is necessary too.
mikehh I've just been build/make smolder testing 20:41
still takes too long]
chromatic If plobsing et al can fix the PBC generation time, it'll build and test faster soon.
Other suggestions for this week? 20:42
mikehh I'll try and look at that
20:42 smash joined
allison that sounds like plenty for one week 20:42
chromatic Let's move to roadmap review. allison? 20:43
allison our supported release priority for this quarter is GC
we have a break down of smaller tasks 20:44
some of those could become weekly priorities
on the bigger tasks, we really need to pick one or two out of the set of possibilities 20:45
chromatic Are there GC tasks of the proper sizes?
allison some are small, some haven't been broken down yet
the best target for a weekly priority is the GC-related tickets 20:46
(not RFCs, but bug reports)
could we make it a priority to close one GC bug every week?
chromatic We can try. 20:47
mikehh could we get a list of GC related tickets
allison some are easy, like #987 to remove _synchronize
mikehh: already done
trac.parrot.org/parrot/wiki/GCTasklist 20:48
mikehh 'k - i'll look
allison my current thinking is that enabling copying/compacting may be the biggest win
in opening up doors for a variety of high-performance GC implementions 20:49
that's something we can talk about on the list
chromatic Other roadmap items?
allison that's it
(we cleared the roadmap in the dev summit)
chromatic Question time. cotto_work had two. Are you here? 20:50
cotto_work yup, for a few minutes
There's all kinds of discussion about moving to git, why we should, why we shouldn't, etc.
I don't intend these questions to be a continuation of that debate.
Under the assumption that we may eventually switch:
Would it be preferable to host our *blessed* git repo ourselves (git.parrot.org or similar) or externally (github, gitorious, etc)? Note that having a blessed repository doesn't exclude having mirrors elsewhere.
chromatic My preference is hosting with OSU OSL. 20:51
allison We don't really have enough information yet
chromatic That's not a blocking preference.
allison We may have to host with OSU OSL to get Trac integration
I don't know yet
And, it's pretty much guaranteed we will have mirrors elsewhere
cotto_work We could easily set up a regular sync of our trac site with an external host if we decide to go with that. 20:52
allison A purely-internal copy on OSU OSL? Yes that's another possibility 20:53
Coke (deprecated left probably small) not true. 20:54
(move on from deprecations) I disagree. I want all of these deprecations closed or rejected.
allison Hmmm... I've been viewing deprecations as an ongoing thing 20:55
is there time pressure?
or is it more that we've got a backlog of old ones?
I have noticed some as old as 1.0
Coke I can't imagine there's time pressure. 20:56
I just think it's a black eye.
it's certainly a maint. win to remove code we know we're going to throw out eventually. 20:57
cotto_work eoq1, I guess. It isn't a huge problem to work with either solution.
What would be the preferred way to support backwards compatibility (i.e. so we don't have to rewrite all links to specific revisions in tickets, source, the wiki, etc)? (more)
mikehh we probably need to prioritize - but deprecations do not HAVE to happen
allison Coke: could we focus on one or two deprecations a week? 20:58
Coke: to make steady progress toward eliminating all of them?
Coke <shrug> people will work on what they want to; setting goals like that is contraindicated, no?
mikehh they are just a warning not to use this stuff in new code
NotFound mikehh: not true, there are deprecated things with no working alternative. 20:59
allison Coke: if I had one deprecation a week, I would work on it
Coke mikehh: yes. but in the meantime, people DO use them, and we still have to maintain them, maintain the test, coding standards, etc.
allison Coke: I mostly just don't know where to start
mikehh it might just *nudge * somebody to work on it
cotto_work (asking because I have to take off, will backscroll) 21:00
Coke allison: there are deprecation tickets out there that have questions waiting for feedback from you now.
cotto_work 1) Keep svn around in read-only mode. The only potential wrinkle from an implementation perspective is link support, which should be doable.
2) Drop svn completely, using mod_rewrite or similar trickery to keep compatibility with old links.
3) something else
eoq
Coke (I apologize for interrupting cotto's question.)
cotto_work ditto
allison cotto: 2 or 3
cotto: 3 if we can figure out what it is :)
Coke: I know. It's one of those lingering guilty feelings. 21:01
Coke: I can put that top on my list for this month. 21:02
Coke: it's more satisfying to close them than to comment and hope someone else does, but hopefully I can unblock others
chromatic I can remove a couple of deprecated features. 21:03
Coke allison: I of course will not complain if you close them. Just trying to reduce the amount of work you need to do so someone else can move on it.
allison Coke: yes, unblocking first, then more detail in a few 21:04
chromatic Other questions? 21:06
Coke When are board elections? 21:07
(I should know this)
allison beginning of June
which means we need to update the members roster 21:08
mikehh do we have a specific policy on what changes can be made in trunk and what needs a branch? 21:10
darbelo It's ussually up to the developer doing it. "Small" in trunk, "Big" in branch. 21:11
chromatic Stable in trunk, potentially unstable in branch.
mikehh yes but how small is small 21:12
darbelo mikehh: chromatic's explanation is best.
mikehh ?
NotFound mikehh: if you have any dout, it isn't.
chromatic Any other questions, comments, or concerns? 21:13
Okay, let's wrap up this week. Let's close some tickets. 21:16
Coke chromatic++
21:22 bubaflub left 21:30 Whiteknight joined 21:43 darbelo left 22:32 NotFound left 23:22 tcurtis 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!