Parrotsketch Every Tuesday @ 18:30 UTC | IRC Log at irclog.perlgeek.de/parrotsketch/today
Set by moderator on 18 February 2009.
04:46 Aamir joined 06:07 Tene_ joined 09:37 dukeleto joined 10:07 tuxdna joined 12:03 Gerd joined 12:22 ujwalic joined 12:58 wknight8111 joined 15:10 Tene joined 15:38 particle joined 16:14 Util joined 17:13 rdice joined 17:33 rurban_ joined 17:35 NotFound joined
rurban I'm out, so posted before: 17:44
- Preparing cygwin release for 1.0.0
- Weird perl5 File::Temp errors suddenly on one machine,
so missed to implement one File::Temp problem for 1.0 TT #326:
widechars in $TEMP, as in chinese usernames (jimmy)
Workaround: SET TEMP=C:\\temp
- Testing 1.0.0, updated native_pbc's for intel
- openbsd keeps crashing on me, no test coverage. All other platforms are ok.
- sorry for my post-1.0 pbc policy outrage, seems to be settled now.
EOR 17:45
17:45 shoe joined, PacoLinux joined 18:20 allison joined 18:28 masak joined 18:32 barney joined
allison no Coke or chromatic, I see 18:33
who's here?
masak
Util
cotto hi
NotFound hi
barney hello 18:34
allison let's get started, in reverse alphabetical order
Whiteknight?
Util? 18:35
Util Done: Mostly cage cleaning; codingstd and typos
Working on:
Probe for __attribute__format is broken on GCC. (Fixed here; will ticket and commit by tommorow. This will turn on a few more warnings in --cage mode.)
Ovid's email from March 16 - "most links are broken on: docs.parrot.org/parrot/latest/html/info.html" (q1q on site's gatekeeper)
Win32 - Porting Parrot to MinGW-TDM (gcc 4.3.0) (So far, just a 5 line patch enables the Parrot-1.0.0 tarball to pass all `make test*` variants, except 3 nci.t bombs)
New document - Some form of "how to play with Parrot-1.0 on Win32 with GCC" 18:36
EOR
allison ujwalic?
ujwalic yep
allison would you like to report anything? 18:37
ujwalic Hi All and congrats :)
Win32 XP fine
allison pmichaud? 18:38
particle?
NotFound? 18:39
NotFound * Document current practice in pdd13
* Fixing and improving pbc_checker
* Miscellaneous fixes and cleaning
eor
allison masak? 18:40
masak * much fruitful work on some Perl 6 projects: proto, Web.pm, Druid.
* reported a few bugs, not as many as usual.
* overall very happy with Rakudo right now.
* planning to work on Rakudo's &unpack tonight.
* things I'm curious how to do in Parrot: utf-8 decoding, port I/O
.eor
18:41 chromatic joined
allison Gej? 18:41
cotto?
cotto * tried to convert Hash-like PMCs to ATTRS, realized that src/hash.c needs to make fewer assumptions about PMC internals
* currently have code for the following changes ready to commit post-1.0:
- refactor src/hash.c functions to return Hash* instead of using ARGOUT Hash** args
- remove an unused function with a questionable comment (parrot_new_hash_x)
* wrote a fuzzer for Parrot using fusil
EOR
q1q
allison chromatic? 18:42
chromatic Fixed some test failures, diagnosed some weird bugs.
EOR
allison barney?
barney Parrot:
added fetch_languages.pl
Pipp:
Put *.pbc into library/pipp_library
worked on 'make install'
.eor
allison allison: 18:43
- All languages/ migrated out of repository.
- Ticket maintenance, applied patches, fixed bugs, answered questions.
- Test failure review, fix/TODO/skip.
- Removed some ancient examples
- Documentation cleanup. 18:44
- Release process for 1.0.
- Generated test 1.0 packages for Debian/Ubuntu.
EOR
did I miss anyone?
okay, questions queued from Util and cotto 18:45
Util, go ahead
Util Who is the gatekeeper for docs.parrot.org? How to access to update? 18:46
allison docs.parrot.org is generated from the monthly release tarball
currently, Coke is the maintainer 18:47
so, the 0.9.1 docs won't be changed, but the 1.0.0 docs will be added and "latest" will point to them
Util Shouldn't there be a branch with true latest (last nightly build) docs?
allison that's 'make html' 18:48
the target audience for docs.parrot.org isn't developers working on the repository, it's release users
Util So any updates needed after release must wait for next release? 18:49
allison yes
4 weeks is a pretty short cycle 18:50
Util Hmmm. OK for now. Thx. I will follow-up with Coke for access if needed, but your answer removes my immediate need. EOQ.
chromatic Our documentation rides the short cycle.
allison cotto, your question? 18:51
cotto After the release, should I put my fuzzer somewhere in svn, e.g. tools/dev?
It's written in Python with some external library dependencies, but it could be useful for other devs.
i.e. if we start running out of bugs
allison cotto: sure, we won't make the dependencies a Parrot build dependency, but can include it anyway 18:52
cotto ok. I'll make sure it's adequately documented.
eoq
barney what does it do?
cotto it generates somewhat random pir
barney cotto++ 18:53
chromatic I had a Perl program to do that a couple of years ago. It's definitely worthwhile.
allison next question, any release blockers? 18:55
sounds like no 18:56
any other questions?
NotFound BTW, verified C++ buildability right now with r37527
allison excellent, thanks! 18:57
roadmap review: trac.parrot.org/parrot/wiki/ParrotRoadmap
verify Parrot release build against external HLLs and tools? 18:58
landed
implement pdd14-numbers (rescheduled from 0.9.0)?
missed/rescheduled, but where should we reschedule it to? 18:59
1.4? 2.0?
chromatic 1.4 is okay.
allison mmd & subclasses (resolve mmd tickets) (rescheduled from 0.9.1)? 19:00
I did review/fix/resolve a number of mmd tickets, but still more remaining
calling it partial/reschedule, and adding to 1.3 19:01
parrot user documentation, objects, pmcs, dynops, etc (rescheduled from 0.9.0)?
landed
(there's always more documentation work to do, but the big push is done, and I'm satisfied with what's going into 1.0) 19:02
adding attributes to existing classes, remove existing exceptions?
chromatic As far as I know, that's been done for a while. 19:03
If there's more to do there, I don't know what it is.
allison okay, landed
pdd29-pct user doc?
landed 19:04
web site updates?
landed
(again, ongoing task, but push is done)
excellent, we're looking good there 19:05
any other questions/comments?
chromatic Retrospective?
NotFound Who pays the beers for St Patrick's day?
allison NotFound: have a virtual beer, it's free 19:06
19:06 moritz joined
NotFound Thanks 19:06
barney Today I installed Cygwin @work. It was nice to see Parrot in the repository.
allison chromatic: do you have some leading questions for a Retrospective?
chromatic What went right? 19:07
What went wrong?
Why why why why why?
That's also the first verse of the new chromatic and the agilists single, "I Will Retrospect You in the Morning, Baby".
Util Moving to a regular release schedule was a big plus. 19:08
allison Adding milestone tasks for each release has worked well. 19:09
chromatic Milestone tasks did seem to help.
allison Our estimation skills still need work (guessing how much can be completed in a month), but are improving. 19:10
chromatic I think (but can't prove) our ticket handling skills have improved too.
allison And, at least for critical tasks, we hit it bang on the nose for what needed to happen when.
chromatic: yes, I think so too
chromatic Other things that went right? 19:11
Weekly milestone review was a plus.
allison yes, good process addition
wknight8111 (weekly milestone review)++
Util Ditto the bus-number-raising rotation of release managers. Quality of the RM's doc went up. 19:12
Tene YAY MILESTONES!
Util Also, we are no longer burning out key people (AFAIK)
chromatic I'm not so sure about that.
allison Util: more importantly, people burning out isn't a project blocker
Util True 19:13
allison volunteer projects always have a "guard-changing" cycle
people come, people go, people return and go again
chromatic Other successes?
ujwalic (welcoming community)++ 19:14
allison Yes, I've been pleased at the increasing community size.
NotFound Portability of bytecode is going well
Tene The prolific parrot bloggers were what got me to start contributing.
ujwalic me too 19:15
chromatic Okay. What didn't work? 19:16
Util Humor++ ; pmichaud's LOLCODE video has gotten at least 4 PerlMongers to play with Parrot.
chromatic I'll start: our estimate on amount of work we can do in a milestone seems to be optimistic by at least a factor of two.
Util Moving to svn.parrot.org without a definite plan for Rakudo's new home. Caused much chaos, and delays in commits. 19:17
allison chromatic: I think that's a snowball effect
chromatic: that is, it started off as 1-2 tasks off, and collected 1-2 more each month 19:18
NotFound Yeah, lots of things depends on other things being done.
chromatic Sure, we slipped half of the things in the first month.
allison chromatic: so, we need to get smarter about rescheduling missed tasks, and not bump them all to the next month
chromatic Maybe we need to change the way we schedule. Pick the most critical tasks and schedule them throughout the six month milestone period. 19:19
Then every month, pick a couple of "would be nice" tasks.
I think (but can't prove) we're better off having more contributors on a task than one or two.
allison the distribution of critical tasks we did at PDS
the "would be nice" is a good idea 19:20
or, specifically adding a "non-critical" status
Tene I remember several times (mostly before PDS) where I had available tuits, but had trouble finding work to do. I don't know if that's a Parrot failure or a Tene failure.
chromatic We're still not great at breaking milestones into concrete task lists. 19:21
allison Tene: yes, I remember you asked for documentation work, and I didn't have specific tasks to hand you (though I know there's plenty to be done)
Tene I remember trying to work on some of the PDD milestone tasks, asking "Okay, what does this actually mean for me to do?" and had trouble finding an answer. 19:22
allison adding specific tickets to the roadmap items has helped, that may be a way to continue improving 19:23
NotFound Sometimes the problem is to figure if the docs need updating or the code must change to meet the docs.
chromatic There's no substitute for a list of the next discrete and concrete tasks. 19:24
particle self-organizing-project-management++
NotFound Open a ticket for each item, to at least have a site to ask for clarifications.
moritz well, sometimes people want to managed by others 19:25
allison aye, even with the addition of roadmap and tickets, I still keep a Hiveminder list of what I personally should/could be working on
moritz we should offer them guidance, and concerte tasks to do
Tene I've had the feeling a few times that too much of our architecture and plan exist only in Allison's head.
chromatic I find it really hard to work with Trac.
allison Tene: yes, I could use help with documentation there
I plan to spend much of the next several years writing 19:26
first up, publish the parrot book in the next couple of weeks
chromatic We still have code cleanliness issues.
allison chromatic: I find Trac enormously easier than RT 19:27
chromatic I don't.
allison I don't know about other people, but it is primarily responsible for my increased productivity in ticket processing
chromatic Every time I want to do something, I have to navigate to a web page, log in, and then type in a little box.
I get four or five times the mail than I did with RT.
I can't respond to tickets through email.
Tene We've had several problems with branches not being finished and not getting merged and falling out of date. 19:28
particle chromatic: we'll always have cleanliness issues, but i think our cage cleaners have done excellent work, and i expect that will only get better.
chromatic Tene, that's why I think having only one or two people on a branch may be a problem.
particle lack of email responses causes me not to use trac.
allison Tene: I take that as evolution, the dying branches were generally not the right solution
particle we need to smoke branches
Tene Is the GC stuff salvageable, or even still around? 19:29
allison Tene: a good branch is small (or substantial) changes merged in quickly
chromatic We need to smoke branches, and I wouldn't mind pruning all branches older than a month.
Re: Trac and email, I agree with particle. I've not responded to several tickets because Trac was too much of a hassle on the web.
allison Tene: the gc stuff that was merged in was good, the other wasn't the right approach
wknight8111 lots of work to do on the GC still, but the cleaner interface is a big help 19:30
Util particle: add yourself to the 'Cc' field of the ticket
chromatic Speaking of the GC, I think some of the closed tickets were redefined right before closing to "The scope of this ticket was whatever we managed to get done, so I'm closing the ticket now."
barney low performance is a big issue 19:31
allison chromatic: sounds odd, feel free to do a ticket review
wknight8111 those tickets were mostly vague, we need to open and manage more-specific tickets in the future for that kind of stuff
chromatic The GC refactor ticket in particular. 19:32
wknight8111 the "implement a better GC eventually" kind of ticket is a poor one
allison wknight: agreed
chromatic That's why I'd like to see a list of concrete, discrete tasks for milestone items.
allison chromatic: for the big refactors, we had that on a wiki page
chromatic It's easy to argue back and forth "Did we meet the expectations of this vague description?"
I mean for milestone tasks.
I think that'll help our estimates too. 19:33
allison perhaps big/critical milestones should have a wiki page too
wknight8111 Things like the GC should be the milestones, and list subtasks under that
spinclad part of ticket work could be refining the task, cloning/depending new subtasks 19:34
chromatic Is that duplicating information between Trac and the Wiki?
spinclad (looking at debian's process) 19:35
19:35 diakopter joined
NotFound If the ticket is really too bad, better reject it and open a new one. 19:36
allison chromatic: well, some things are tickets, some things are working tasklists
chromatic I don't see a lot of value in making every task into a ticket. Maybe I'm missing something.
allison chromatic: for new development, it's far easier to set out a page of "this is what we're doing for this task" than to navigate a hundred tickets
chromatic: yes, we're on the same page 19:37
chromatic I suppose I can write up a list of milestone handling recommendations.
NotFound A ticket with a link to the wiki page might be helpful in some cases. 19:38
chromatic Sure; a ticket for the milestone seems appropriate.
Anything else that went wrong? 19:40
NotFound We still have to maintain imcc 19:41
chromatic Any suggestions for improvement? 19:43
allison seems like we've mainly been talking through those as we talked about "what went wrong?" 19:44
I am looking forward to not changing the infrastructure for several years 19:46
it was important, and glad we made the changes, but also glad it's done 19:47
chromatic Improvement should also discuss "How can we measure the effect of any improvement?"
Branches land more quickly?
We finish milestones more frequently?
Util Is anyone working on a Parrot+Languages LiveCD, that we can use as crack^Wteasers at conferences to attract more volunteers? 19:48
particle "contributors stay on longer, and don't burn out"
chromatic wknight8111 gets less bored?
wknight8111 has a disorder :(
particle util, i started, and stalled (ENOTIME)
allison particle: that's likely impossible (even paid employees burn out and leave) 19:49
Util particle++
particle allison: sure, but burning out is easier when there's more pain and less fun
how do you measure -Ofun?
NotFound Let's have more fun!
chromatic I can measure the absence of -Ofun
Util Can we measure out-of-nest interest from tarball download count? 19:50
particle i have that now
moritz in WTFs per miinute.
NotFound Util: maybe today a virtual machine is easier to test than a live cd. 19:52
Util NotFound: perhaps so; my mind is stuck in the ancient past. 19:53
NotFound And cheaper to distribute. Just plug a usbpen and copy.
chromatic Other improvements or measurements/ 19:54
Util New language count?
NotFound Well, at least a linux one. A Windows image may need to pay licenses.
Util Retracted - number too low to be a helpful measure 19:56
NotFound Uh, no, people writing languages just to have a feeling of contributing is not a good thing.
moritz uhm. rakudo started to segfault on some tests with parrot HEAD 19:57
allison moritz: which tests?
moritz but I didn't do a realclean before rebuilding, so I'll try that next 19:58
NotFound Just picking brainfuck-inspired ones from the esotheric languages ring can fill a long list :D
Util NotFound: I meant it as: the better we are doing, the more language authors will be attracted to us.
moritz allison: t/spec/S32-str/substr.t for example
chromatic Anything else we need to discuss in #ps today?
Let's call it a week then. 20:00
20:02 Util left, NotFound left
allison thanks all! 20:03
moritz ciao, everyone++
20:04 diakopter left 20:05 PacoLinux left 20:07 particle left 20:12 chromatic left 20:58 davidfetter joined 21:17 davidfetter joined 21:44 Whiteknight joined 23:58 dduncan 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!