Parrot 2.9.1 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | remove deprecations | volunteer for embedding or Lorito teams |
Set by moderator on 26 October 2010.
dalek rrot: r49719 | nwellnhof++ | trunk (2 files):
[str] Fix a bug in index op

work. Rework and reenable the multibyte index tests in t/op/string.t.		 00:04
00:29 dngor left, dngor joined 00:40 whiteknight joined, kid51_at-dinner is now known as kid51
kid51 dukeleto ping 00:45
00:49 Hunger joined
kid51 plobsing ping 00:53
01:08 dngor_ joined 01:09 dngor left
dukeleto kid51: pong (i am going to eat dinner in a few minutes) 01:14
01:18 dngor joined, dngor_ left
dalek rrot: r49720 | nwellnhof++ | trunk/src/string/encoding (2 files):
[str] Remove unused function
 01:21
rrot: r49721 | nwellnhof++ | trunk/t/op (2 files):
[t] Add some string tests
kid51 dukeleto Check your email re message from me re RTEMS. Get back to me when you've got food (and possibly sleep). 01:23
01:27 dngor left
kid51 No wonder the tests weren't passing. 01:39
They were never written!
01:49 preflex left
dukeleto kid51: you should join #rtems on freenode 01:51
kid51: which OS do you want to install RTEMS?
kid51: OS X? They have virtual machine images
01:52 preflex joined
dalek rrot: r49722 | jkeenan++ | branches/gsoc_nci/config/auto/libffi.pm:
Since there's no verbose output possible inside _evaluate_cc_run(), we don't need to pass $verbose to that sub.
 02:07
rrot: r49723 | jkeenan++ | branches/gsoc_nci/config/gen/libffi.pm:
Correct spelling error in description. Eliminate some unneeded code.
rrot: r49724 | jkeenan++ | branches/gsoc_nci/t/steps (2 files):
Refine two configuration step test files.
kid51 dukeleto: I was planning to install it on Linux (Debian) but not on Darwin 02:16
dukeleto: Understand that for now I'm only interested in RTEMS to the extent of giving you the configuration step you need. 02:17
dukeleto kid51: you can use rpm2cpi on the Centos 5 debs to create tarballs and then compile them. Ask on #rtems. Tell them dukeleto sent you and that they need it on their wiki for parrot devs to read. 02:19
rpm2cpio, rather.
kid51: and i meant centos rpms
dukeleto is still sleep-deprived
kid51 Okay, but as I said in my email, that is a *lot* of work (a) just for a config step; (b) in general for anyone. 02:20
02:35 whiteknight left
plobsing kid51: pong 02:43
kid51 plobsing: While looking at some step tests in gsoc_nci branch, I had occasion to study config/gen/libffi.pm. 02:49
IIUC, once you know whether the machine has or does not have libffi, you copy one of two template files into position as src/pmc/nci.pmc. 02:50
Am I understanding that correctly?
dukeleto kid51: if you don't want to do it, you don't have to. 02:51
kid51 dukeleto: As i said in email, if I could get a shell with it installed somewhere, that would suffice. 02:53
dukeleto kid51: ask on #rtems
kid51: hmmm
kid51: you should get access to the gcc compile farm
kid51: it has it on them
kid51: all you need to do is ask the mailing list and give your ssh public key, and tell them how you are involved with parrot 02:54
kid51: the gcc-farm mailing list
kid51 dukeleto: I will consider that
dukeleto kid51: but you can ask #rtems as well
dalek rrot: r49725 | nwellnhof++ | trunk/src/string/api.c:
[pod] Add missing =cuts
rrot: r49726 | nwellnhof++ | trunk/t/op/string_cs.t:
[t] More string tests
dukeleto kid51: i can run code that you commit to a branch 02:55
kid51: but it is better if you have a interactive shell
kid51: all it does is return 1 if __rtems__ is defined, 0 otherwise
kid51 defined ... where? 02:56
dukeleto kid51: you can write and test on your machine. It will always return false.
kid51: magically by the compiler
kid51: just plain C
kid51: in rtems-flavored gcc
kid51 Would this be something that would be detectable by config/auto/attributes.pm or config/auto/headers.pm? 02:57
dukeleto kid51: you are the expert there, have fun :)
kid51 What sort of thingee is '__rtems__' ? A header? An attribute? 02:58
dukeleto kid51: whatever you want it to be 02:59
kid51 Let's suppose my Linux had RTEMS DE installed. Where would I look for __rtems__?
dukeleto kid51: compile a C program and see if it is defined. Does that make sense? 03:00
kid51 dukeleto: Again I ask, "Defined *where*?"
I will try to go on #rtems within the next few days and seek help there. In the meantime ... 03:02
kid51 must sleep
plobsing kid51: your assessment is correct. I dislike that part though, and am working to eliminate it.
kid51 plobsing: Yes, it means that 'src/pmc/nci.pmc' becomes a file generated by Configure.pl ... which means it shouldn't be in MANIFEST ... it should have svn-ignore and .gitignore properties, etc. And I wonder how it would interact with config/auto/pmc.pm. 03:08
plobsing: But I will just monitor what you're doing in that branch and hold my questions till things are more clear. Thanks.
plobsing those are exactly some of the reasons I dislike the copy into place approach. I intend to have a canonical version of nci.pmc which calls functions in src/nci/* which vary based on Configure.pl defined C #defines 03:10
03:25 kid51 left
cotto where's our test coverage report at? 03:28
aloha, coverage?
aloha cotto: coverage is cv.perl6.cz or tapir2.ro.vutbr.cz/cover/cover-results/
cotto didn't expect aloha to be useful
03:29 patspam joined 03:31 dngor joined
cotto anyone want me to put them down as a Code-In mentor for increasing PMC test coverage? 03:36
dukeleto, does the list of Code-In tasks need to be fixed or can we have several related tasks (e.g. increase PMC test coverage) without documenting what they all are? 03:44
03:46 hercynium joined 03:48 bluescreen left 03:58 nwellnhof left 04:08 patspam left 04:15 hercynium left, theory left
dukeleto aloha, msg cotto recurring tasks are fine, as long as they are well-defined 05:18
aloha dukeleto: OK. I'll deliver the message.
dukeleto: Okay.
dukeleto aloha, you have an off-by-one error. You should get that looked at. 05:19
cotto dukeleto, thank 05:31
bacek_at_work dukeleto, "known bug" with Infobot plugin... 05:43
aloha, msg cotto recurring tasks
aloha bacek_at_work: OK. I'll deliver the message.
bacek_at_work aloha, msg cotto recurring tasks?
aloha bacek_at_work: OK. I'll deliver the message.
bacek_at_work: msg cotto recurring tasks are fine, as long as they are well-defined
bacek_at_work this one...
cotto That was odd. 05:48
In soviet Russia, bot tells you to deliver messages. 05:49
bacek_at_work cotto, I didn't have time to fix Bot::BasicBot::Pluggable logic... 05:57
cotto bacek_at_work, sure. It's a tolerable quirk. 05:58
06:31 fperrad joined 07:54 s1n joined 07:59 cosimo left
sorear aloha: msg kid51 try "gcc -x c -E -dM /dev/null" on your favorite gcc system 08:05
aloha sorear: OK. I'll deliver the message.
sorear aloha: msg kid51 __rtems__ is in good company
aloha sorear: OK. I'll deliver the message.
sorear: Okay.
08:19 cosimo joined
cotto sorear, cool trick 08:21
dalek rrot: r49727 | cotto++ | branches/opmap_aware_pmcs/t/pmc/packfileopmap.t:
[t] add tests for PackfileOpMap
 08:36
08:39 lucian_ left
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#844) fulltest) at r49727 - Ubuntu 10.10 amd64 (g++-4.5) 09:21
10:14 tadzik joined 10:32 contingencyplan left 10:34 PerlPilot joined 10:38 PerlJam left 11:08 smash joined
smash hello everyone 11:09
11:16 tadzik1 joined 11:21 tadzik left, tadzik1 is now known as tadzik 12:14 kid51 joined
kid51 msg sorear Thanks. Overnight I surmised that what duke was referring to was some kind of environmental data, and your post confirms. 12:16
aloha OK. I'll deliver the message.
kid51 msg dukeleto On the machine where you have RTEMS DE, what shows up for /__rtems__/i if you run the command described above by sorear? 12:17
aloha OK. I'll deliver the message.
12:23 whiteknight joined
kid51 ponders whether it would make sense to have all the output of "gcc -x c -E -dM" available via Configure.pl. 12:25
12:25 kid51 left 12:28 tadzik left
whiteknight I have no idea what that little incantation does 12:28
12:28 GodFather joined 12:32 tadzik joined 12:51 jsut_ joined 12:55 jsut left 13:09 GodFather left 14:00 s1n left 14:01 nwellnhof joined 14:21 Drossel is now known as Kulag 14:36 GodFather joined 14:40 silug left 14:46 davidfetter joined 15:08 dmalcolm joined 15:09 jan left 15:36 whiteknight left 15:42 theory joined 15:44 mikehh left 15:52 GodFather left 15:56 GodFather joined 15:57 p6eval left 15:59 p6eval joined 16:01 nwellnhof left 16:15 jan joined
dukeleto 'ello 16:23
kid51: why are you not in #rtems/ 16:24
s!/!?! 16:25
kid51: and i have no clue what you are asking me in your msg. I have not backlogged recently. Been busy. 16:26
16:36 cosimo left 16:37 cosimo joined 16:58 tadzik left 17:01 tadzik joined 17:05 whiteknight joined
whiteknight good afternoon, #parrot 17:07
atrodo good afternoon
whiteknight hello atrodo, how are you? 17:18
atrodo Not bad. Working for the man, same old, same old. How about? 17:19
whiteknight same as you. Workin for the man. Can't get no respect, etc. 17:20
17:20 stilgar joined 17:23 tadzik left, tadzik joined, cogno joined 17:25 contingencyplan joined 17:26 patspam joined 17:30 mikehh joined
cotto ~~ 17:30
whiteknight hello cotto 17:31
mikehh hi cotto
cotto so friendly 17:32
whiteknight <3 <3 <3 cotto!!! <3 <3 <3
dukeleto wow. lotta love in the room today. 17:33
cotto feels either welcomed or slightly creeped out. I guess I'll go with welcomed until I notice someone stalking me.
17:33 cogno left
mikehh dukeleto: what's up with GCI? 17:34
dukeleto mikehh: i need to do a lot of work for it today 17:35
mikehh: the page on the parrot wiki for GCI is going to be for Parrot and Perl GCI projects
17:36 stilgar left
dukeleto stalks cotto 17:36
mikehh dukeleto: I spent some time trying to get a list of tests to convert to pir, progressing slowly
dukeleto: we need more stuff there 17:37
dukeleto mikehh: think of more non-code tasks. We need our website translated to other languages, and we need web design work 17:38
mikehh dukeleto: seriously lacking tuits there
cotto Web design work would be amazing.
mikehh still need to work on html_cleanup branch sometime 17:39
dukeleto cotto: if you could give links that define what PMC and PIR mean in task #3, that would be awesome. Don't we have a parrot glossary?
we can also put gsoc ideas that didn't work out on that page, in case we get a few amazing coding students 17:40
mikehh possibly the pir book in docs/website as a ref
dukeleto good idea!
mikehh although it probably need some serious updating
17:40 patspam left 17:42 patspam joined, patspam left 17:43 patspam joined
dukeleto that is another good task :) 17:43
cotto dukeleto, done 17:44
good idea
dukeleto one thing that i've learned, there can only be 1 task per student in the code-in webapp, so if we have tasks that multiple students can do, we will need to create a task for each of them 17:47
that will mostly be my job, i think
17:49 jsut joined
mikehh dukeleto: students can work on more than 1 task, sequentially, 17:54
17:55 jsut_ left
mikehh for example in converting tests to pir, after the first it becomes a lot easier to convert further tests 17:55
cotto: as the architect, what is the situation with pasm, some tests use it, shouldn't we move away from that? 17:58
17:58 smash left 18:09 lucian joined 18:10 integral left
dukeleto mikehh: yes, i know. I am talking about how the tasks are actually stored in Melange. 18:10
mikehh: what is your preferred email/ 18:12
mikehh: i need to cc you some stuff
mikehh: mike.hind@gmail.com
mikehh: ?
mikehh: we need to make each task have an "optional background" and a "recommended background" 18:18
mikehh dukeleto: that's the one I normally use 18:19
dukeleto mikehh: live.gnome.org/GoogleCodeIn/Tasks 18:20
mikehh: community.kde.org/GoogleCodeIn/2010/Ideas
mikehh: those have a huge number of ideas. We need to emulate those. 18:21
whiteknight dukeleto: I 18:25
dukeleto mikehh: www.rtems.org/wiki/index.php/Google...InProjects
whiteknight 'll look over the task list today and add details/ideas
dukeleto whiteknight: the deadline to apply is today, so if you can get some of those ideas in, that would be great 18:32
it really sucks that our bots don't show wiki changes still. It is lessening our productivity. 18:44
mikehh too right
19:07 whiteknight left
dukeleto who is a moderator on parrot-members? 19:12
if you can pass my email thru, that would be awesome. i sent it from a different address than i signed up on 19:13
19:15 cognominal left 19:17 davidfetter left, davidfetter joined
dukeleto mikehh: live.gnome.org/GoogleCodeIn/HowToWriteAGoodTask 19:19
I just gave a task template on the wiki page. We need to update these already existing tasks to give the required info. 19:25
can anybody fix the wikibot? 19:28
dukeleto begs
dukeleto hates wikis that are not based on git. Someone editing while I was, and now I can't submit my changes. 19:30
dukeleto curses the trac wiki 19:31
it has a merge button, but it is useless
19:48 patspam left
mikehh dukeleto: what's your deadline for getting it in? 19:54
and how much can we change before November 5th 19:55
19:55 patspam joined
dukeleto mikehh: well, the deadline is 4pm today, and our task list will be looked at this weekend, when Google decides which organizations to accept 19:55
mikehh: so it is in our best interest to work on the wiki a bunch today and tomorrow. But it needs to be a team effort. 19:56
i've add a template
Please use the new template! 19:57
aloha msg whiteknight could you please use the new task template I added to the wiki? We need way more details in our tasks.
aloha dukeleto: OK. I'll deliver the message.
mikehh dukeleto: so you have about 3 hours to get the proposal in 19:58
cotto mikehh, good question. For the time being, definitely don't write any new tests in it. 19:59
mikehh dukeleto: can we make changes until they look at it?
cotto: was thinking in terms of converting pasm to pir maybe for GIC 20:00
cotto: we have perl based tests thet output pasm code and evaluate the output (also pir based as well) 20:01
20:02 cognominal joined
mikehh cotto: I was trying to define some tasks in terms of converting the perl based tests to convert to pir based ones, but what to do about pasm code? 20:03
plobsing we have pasm code? 20:04
mikehh cotto: was looking at tests that have some form of >>CODE >>OUTPUT in the perl
plobsing: yes in a couple of cases
plobsing oh in test cases.
mikehh plobsing: that's what I was looking at 20:05
plobsing what we really need is a similar interface to what the perl code uses. call down into a *separate* process. Some tests flat out shouldn't be run in the same interpreter because failure usually means crashing out and missing the other tests. 20:06
cotto mikehh, if you think that it'd be a reasonable task, +1 to converting them
plobsing, that'd be nice. 20:07
plobsing it's at the end of a long queue of todo tasks for me. Unfortunately, the queue is weak refs and I have an agressive GC.
dukeleto mikehh: i think you misunderstand something about translating tests 20:08
mikehh dukeleto: elaborate
dukeleto mikehh: some of our tests must be in PASM, because they actually test PASM. Some of our tests are needlessly in PASM, but converting them over doesn't buy us much.
cotto Sure. Translating test from perl to pure pir is valuable for different reasons than moving from PASM to pir. 20:09
20:09 tadzik left
dukeleto mikehh: so while there exist a nonzero number of pasm tests that *could* be translated, what is the benefit? 20:09
cotto actually, I guess it depends on how much pain will be involved in getting PIRATE to support the PASM that's used.
dukeleto mikehh: i understand the benefit of PASM vs PIR, but what is the benefit to change already existing code that works?
cotto: exactly. 20:10
Translating tests now, before Lorito, is not a good investment. Some of them will have to change for Lorito.
cotto If PIRATE supports it, there's no reason to bother.
dukeleto PASM is a Parrot Language, and it is fine to be in our test suite. 20:11
mikehh dukeleto: just getting some ideas together, but in those instances where we are not directly testing pasm, I was woring on the basis that pasm isn't actually going to be around
dukeleto But we shouldn't needlessly write new tests in PASM or anything.
mikehh working
dukeleto mikehh: pasm will be around for a long time.
mikehh dukeleto: I got the impression not 20:12
dukeleto mikehh: i think looking at our code coverage reports, and choosing the top 5 critical subsystems that have the lowest coverage, is a good start.
mikehh: then start writing tests for those subsystems, to bring up coverage.
mikehh: that makes a good meta-plan for the QA team
cotto aloha, clock? 20:13
aloha cotto: cotto: LAX: Fri, 13:13 PDT / CHI: Fri, 15:13 CDT / NYC: Fri, 16:13 EDT / UTC: Fri, 20:13 UTC / LON: Fri, 21:13 BST / BER: Fri, 22:13 CEST / TOK: Sat, 05:13 JST / SYD: Sat, 07:13 EST
cotto bacek, ping
mikehh dukeleto: yes in the long term, I was concentrating on tasks for GCI
dukeleto: mind you getting a list of tasks together is good anyway 20:14
dukeleto mikehh: making students do something that is not useful but easy is still not useful.
mikehh: but i very much appreciate your help. 20:15
mikehh: we need more detailed tasks
mikehh: they can actually say "translate this specific file to X"
mikehh: we are dealing with high school students. They need *very specific" instructions, in digestible sizes.
mikehh: you can copy and paste 10 files to translate and create a task for eafch 20:16
each, even
mikehh: but maybe getting them to increase coverage for the systems you identify would be better
mikehh: i don't think in terms of "the long term". This is the long term. If you identify the subsystems that need code coverage most deperately, you can add each as a task. 20:17
mikehh dukeleto: I was getting a list together of files that could be moved from perl to pir, but will look at the other as well
dukeleto mikehh: and it could even be "increase code coverage of this subsystem by 10%"
mikehh: think small, and create many tasks
mikehh: other orgs have *hundreds* of tasks
we can ask for as much as we want, they just have to be smallish chunks of work
mikehh dukeleto: I can do that, but I don't think I can have it ready in three hours, I need to go somewhere now 20:19
cotto mikehh, are you making a list of which PMCs need better coverage and would be good candidates for the GCI list?
(I want to make sure it gets done. I don't mind doing it myself.)
dukeleto mikehh: any work you can do today or tomorrow would be very useful.
mikehh: Google will be announcing the accepted orgs on Nov 5th
mikehh: and we can continually add to the task list during Code-In. We just want a good enough task list to get accepted. Make sense? 20:20
mikehh dukeleto: yes 20:28
cotto dukeleto, are slots allocated similarly to gsoc or do approved orgs get as many slots as they can get students to fill them? 20:29
dukeleto cotto: i don't know how slots work. I don't know if there *are* slots. All i know about is tasks. 20:30
cotto ok. So the goal is to get lots of tasks on the wiki and hope for the best. 20:31
dukeleto asks people who know
cotto: yes.
cotto eoq
dukeleto 13:33:23 <@carols> dukeleto: yes, you can add tasks and you can even have a queue of tasks that the students can't see so you can release things to work on at your own pace 20:34
13:33:33 <@carols> dukeleto: you can add tasks whenever you want
that is from #gsoc on freenode
dukeleto hopes the logging demons don't come to eat him 20:35
so, we can add tasks all the time! We just have to have enough to get accepted and attract students.
Students can only choose from our tasks. They can't design their own. This is probably good. 20:38
But it means we need to have stuff that people want to do. And to describe it in an exciting way.
20:40 fperrad left
mikehh dukeleto: I have to go to a meeting now, will try to add some stuff later, hope it will be in time to be of use, bbl 20:43
dukeleto mikehh: thanks! 20:47
21:15 tadzik joined 21:36 davidfetter left
dukeleto We should be thinking about tasks that will take a total beginner less than a week to complete, if possible. Some tasks can be bigger, but students are going to want smaller tasks. 22:08
tadzik is Google Codein <18 only? 22:09
dukeleto They get paid a small amount for each task completed, so they are going to look for reasonable stuff
tadzik: 13-18
tadzik: why do you ask?
tadzik dukeleto: I'd like to get involved, yet GSoC always seems to hard to me
s/to/too/
hmm, s:2nd/to/too/
22:10 dngor_ joined, dngor left 22:12 GodFather left 22:17 dngor joined 22:18 dngor_ left 22:20 Andy left 22:21 dmalcolm left
mikehh dukeleto: how's it going? 22:24
dukeleto: I updated the wikiu page with a possible list of tasks (ungraded at the moment) 22:25
22:26 tcurtis joined
mikehh dukeleto: that's tests that could be converted to pir 22:26
dukeleto: possibly, maybe or whatever 22:27
22:33 dngor left
dukeleto mikehh: did you use the template? 22:57
mikehh: we need help converting all the tasks to the new template, which is listed just below the description of Code-In
23:01 kid51 joined
mikehh dukeleto: I'll tidy it up that way after I get some sleep, it's after midnight for me 23:06
dukeleto mikehh: thanks. I appreciate your heroic effort. 23:15
23:18 tcurtis left
sorear hello kid51 23:21
kid51 sorear hello 23:26
sorear I like that command you posted.
23:32 dngor joined
dalek rrot: r49728 | jkeenan++ | branches/gcc_defines:
Creating a branch to explore configuration modifications inspired by sorear++.
 23:47
23:58 bacek_mobile joined
bacek_mobile ~~ 23:59
Cotto, pong
cotto bacek, how does PASM fit in to PIRATE?

Powered by Cro and the Raku® Programming Language.

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

Thank you!