Camelia, the Perl 6 bug

IRC log for #parrot, 2011-04-20

Parrot | source cross referenced

| Channels | #parrot index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:02 wagle left #parrot
00:04 bacek_at_work ~~
00:07 Topic for #parrot is now Topic for #parrot is: Parrot 3.3.0 released | http://parrot.org | Log: irclog.perlgeek.de/parrot/today
00:08 whiteknight there are too many steps in the release process. we need to trim this down
00:09 whiteknight or automate more of it
00:09 whiteknight we should be able to cut a release in less than an hour at most, and without jumping through so many hoops
00:16 whiteknight having to update all these little things on parrot.org or on trac are exactly the kinds of tasks we should be able to automate
00:17 cosimo joined #parrot
00:21 whiteknight none of the milestones since 2.11 have been marked completed on trac
00:21 whiteknight so clearly people aren't following the process too closely now anyway
00:23 dmalcolm left #parrot
00:30 wagle joined #parrot
00:33 dalek parrot: c1bc61d | Whiteknight++ | / (2 files):
00:33 dalek parrot: Parrot 3.3 is 'Fire in the Sky'
00:33 dalek parrot: review: https://github.com/parrot/parrot/commit/c1bc61d28b
00:33 dalek parrot: a9a87f1 | Whiteknight++ | / (10 files):
00:33 dalek parrot: Merge branch 'parrot_3_3'
00:33 dalek parrot: review: https://github.com/parrot/parrot/commit/a9a87f1c7d
00:39 ttbot Parrot 1cfaceab i386-linux-thread-multi make error http://tt.taptinder.org/cmdinfo/69131
00:39 whiteknight wtf
00:40 ttbot Parrot 8a1b743a i386-linux-thread-multi make error http://tt.taptinder.org/cmdinfo/69148
00:41 bubaflub joined #parrot
00:42 ttbot Parrot 8a1b743a i386-linux-thread-multi make error http://tt.taptinder.org/cmdinfo/69165
00:42 whiteknight ...lolwat?
00:42 whiteknight is the bot checking every subsequent commit?
00:45 ttbot Parrot 8a1b743a i386-linux-thread-multi make error http://tt.taptinder.org/cmdinfo/69197
00:49 theory joined #parrot
01:32 whiteknight left #parrot
01:53 jevin left #parrot
01:58 woosley joined #parrot
02:42 bubaflub left #parrot
03:04 dukeleto msg whiteknight i totallly agree about automating more of the "many little steps" in the release process
03:04 aloha OK. I'll deliver the message.
03:14 jevin joined #parrot
03:14 hudnix left #parrot
03:54 cotto ~
04:15 ShaneC left #parrot
04:30 birdwindupbird joined #parrot
04:42 ShaneC joined #parrot
05:02 cognominal left #parrot
05:02 cognominal joined #parrot
05:28 simcop2387 left #parrot
05:29 simcop2387 joined #parrot
05:48 fperrad joined #parrot
05:51 plobsing joined #parrot
05:53 cotto plobsing, ping
06:04 dafrito left #parrot
06:23 plobsing_ joined #parrot
06:27 plobsing left #parrot
06:29 dafrito joined #parrot
06:34 UltraDM joined #parrot
06:36 contingencyplan left #parrot
06:39 simcop2387 left #parrot
07:13 hudnix joined #parrot
07:26 mj41 joined #parrot
07:28 theory left #parrot
08:00 cognominal left #parrot
08:02 cognominal joined #parrot
08:18 TiMBuS joined #parrot
08:23 bacek ~~
08:30 utsl left #parrot
08:41 utsl joined #parrot
09:06 cosimo left #parrot
09:07 jrtayloriv left #parrot
09:20 cotto left #parrot
09:32 simcop2387 joined #parrot
09:47 hudnix left #parrot
09:55 cotto joined #parrot
10:01 woosley left #parrot
10:02 particle left #parrot
11:11 mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#15120) fulltest) at 3_3_0-13-ga9a87f1
11:11 mikehh Ubuntu 11.04 beta amd64 (g++ --optimize)
11:32 woosley joined #parrot
11:56 Psyche^ joined #parrot
11:57 Patterner left #parrot
11:57 Psyche^ is now known as Patterner
12:28 simcop2387_ joined #parrot
12:29 simcop2387 left #parrot
12:29 simcop2387_ is now known as simcop2387
12:32 benabik left #parrot
12:32 smash joined #parrot
12:33 smash hello everyone
12:33 tadzik hello smash
12:38 smash left #parrot
12:38 smash joined #parrot
12:39 whiteknight joined #parrot
12:43 whiteknight good morning, #parrot
12:43 tadzik whiteknight: good morning
12:44 smash whiteknight: mornin'
12:48 mtk joined #parrot
12:50 contingencyplan joined #parrot
12:52 whiteknight good morning tadzik, smash. How are you two doing today?
12:53 tadzik not bad, besides the still-not-functional hand
12:54 whiteknight tadzik: oh, right. what's the schedule for that? How long is it going to be out of commission?
12:54 smash whiteknight++ # release
12:55 whiteknight smash: thanks! I like doing it, but every time I'm reminded about how big a pain in the ass it is to do
12:55 tadzik whiteknight: they're removing the cast and x-raying it once again this friday
12:55 whiteknight tadzik: oh, that's good. therapy after that I assume?
12:55 smash whiteknight: gratz for a job well done then
12:55 tadzik for no one's really sure whether it's broken or just badly twisted
12:55 tadzik whiteknight: no idea, time will tell
12:55 whiteknight tadzik: oh, both are bad enough
12:55 tadzik aye
12:55 hudnix joined #parrot
12:59 mikehh rakudo (e09cb7e) - builds on parrot (3_3_0-13-ga9a87f1) - make test, make spectest_smolder[(#15132), roast (7309150)] PASS
12:59 mikehh Ubuntu 11.04 beta amd64 (g++ --optimize)
12:59 mikehh 27,695 ok, 0 failed, 606 todo, 1,800 skipped and 0 unexpectedly succeeded
13:00 mikehh hi whiteknight, how did the release go for you?
13:02 contingencyplan left #parrot
13:05 Coke left #parrot
13:05 Coke joined #parrot
13:05 contingencyplan joined #parrot
13:05 atrodo whiteknight> Are you going to YAPC::NA this year?
13:10 whiteknight atrodo: no sir. you?
13:11 whiteknight mikehh: release went well, thanks. I'm sorry it had to wait until so late in the evening. I am determined to streamline some parts of the release process
13:11 atrodo whiteknight> Aye.  Just checking the usual suspects
13:11 atrodo aloha, going to YAPC::NA 2011?
13:11 aloha atrodo: I have no idea.
13:11 ambs joined #parrot
13:12 mikehh whiteknight: good - I tend to find releasing a bit stressful, and there is room for improvement in the process
13:14 moritz whiteknight++ # 3.3 release
13:14 whiteknight moritz: just a double-check, the way I released it from a branch works fine with Rakudo, right?
13:15 particle joined #parrot
13:15 moritz whiteknight: yes
13:16 mikehh whiteknight: builds and tests fine for me, but used latest devel version of parrot
13:16 whiteknight okay, awesome
13:16 moritz whiteknight: an easy way to check if the output from   git describe --tags  on master after the rlease starts with RELEASE_3_3_0
13:16 whiteknight moritz: okay, that's good. I'm glad the merge makes that work correctly
13:16 whiteknight and I'm glad git is smart enough to find the most recent tag
13:16 moritz for the actual rakudo release we can use arbitrary git ref specifications
13:17 moritz we could use the SHA1, it's just less pretty - and it makes comparing versions a pain
13:20 mikehh winxed (r966) - make all/test/test1/test2 ok - ran some examples as well - on parrot 3_3_0-13-ga9a87f1
13:20 mikehh Ubuntu 11.04 beta amd64 (g++ --optimize)
13:27 whiteknight mikehh: you're a rockstar
13:28 benabik joined #parrot
13:32 GodFather joined #parrot
13:35 bubaflub joined #parrot
13:36 bluescreen joined #parrot
13:55 whiteknight msg NotFound can I use [flat], [slurpy], and [named] modifiers with multi-returns?
13:55 aloha OK. I'll deliver the message.
14:19 Andy_ joined #parrot
14:20 dalek parrot: 71ed521 | petdance++ | src/embed.c:
14:20 dalek parrot: consting args. Flag unsued args. Add an XXX where we are not setting an outbound pointer.
14:20 dalek parrot: review: https://github.com/parrot/parrot/commit/71ed52129a
14:30 mtk left #parrot
14:40 mtk joined #parrot
14:42 Andy_ Who else besides me is workin' on llvm/clang?
14:42 Andy_ Looks like jkeenan, anyone else?
14:42 Andy_ sorry, kid51, if his client is listening
14:43 mikehh done some testing there
14:43 whiteknight what do you mean working on it?
14:43 whiteknight I use clang as my default compiler
14:43 Andy_ How are you running Configure.pl?
14:43 hercynium joined #parrot
14:43 whiteknight --cc=clang --ld=clang --link=clang
14:44 Andy_ You're not using ccache?
14:44 whiteknight no
14:44 whiteknight I don't think ccache works with clang
14:44 whiteknight at least, it didn't last I looked
14:45 whiteknight although a quick google search seems to have some promising results
14:47 Andy_ I'm working on figuring out what flags to pass to clang, like I did for GC
14:47 Andy_ GCC
14:47 whiteknight okay
14:48 Andy_ and notice config/auto/llvm.pm come out
14:48 UltraDM left #parrot
14:54 Andy_ whiteknight: if I specify --ld, Configure.pl fails.
14:56 whiteknight my exact commandline is: "perl Configure.pl --cc=clang --link=clang --ld=clang"
14:56 whiteknight and that's been working fine for me for a long time
14:58 Andy_ hmmm
15:00 Andy_ sorry, it's if I specify --link=clang
15:00 Andy_ I wonder if your clang is built not to link
15:00 Andy_ rather, MINE is built not to link.
15:00 lucian joined #parrot
15:05 JimmyZ joined #parrot
15:09 theory joined #parrot
15:13 lucian_ joined #parrot
15:17 lucian left #parrot
15:23 dukeleto ~~
15:23 dukeleto whiteknight++ # releasing 3.3.0
15:25 bluescreen left #parrot
15:28 whiteknight good morning, duke
15:29 whiteknight we're coming down to the acceptance deadline for GSoC. I'm starting to get excited about it
15:29 cotto ~~
15:29 whiteknight good morning cotto
15:30 dukeleto whiteknight: indeed. I haven't had enough tuits for it. Where do we stand now and what needs doing?
15:30 dukeleto cotto: top of the mornin'
15:30 whiteknight dukeleto: I sent out an email this morning. If you're happy with your current assignment, you don't need to do anything
15:31 whiteknight dukeleto: We still have one project in limbo that's good enough to accept but for which we don't have a mentor lined up.
15:33 dukeleto whiteknight: which?
15:36 cotto jvm bytecode translator
15:36 whiteknight dukeleto: privmsg
15:40 bluescreen joined #parrot
15:52 jnthn__ Is that for translating jvm bytecode => Parrot?
15:53 dukeleto jnthn__: i believe so
15:54 moritz didn't jnthn__ write something like that ages ago? :-)
15:54 dukeleto jnthn__: did you?
15:55 jnthn__ I did the .Net version of that. :)
15:56 jnthn__ http://www.jnthn.net/papers/200​6-cam-net2pir-dissertation.pdf
16:04 whiteknight jnthn__: are you free this summer and willing/able to do a similar project with JVM?
16:04 whiteknight I haven't been keeping up with it. I have no idea what the GSoC situation in Rakudo-land is looking like
16:06 moritz if all goes well, I'll mentor a POD parser project by tadzik++
16:06 moritz and that's it
16:07 tadzik \o/
16:10 jnthn__ whiteknight: Free-ish. How much are they proposing to do? I spent quite a while on it and was already familiar with Parrot and knew quite a bit of .Net, and only got so far.
16:11 jnthn__ I guess my main interest in mentoring such a project would be to see how a static language on 6model pans out. In theory, I've planned for that. In practice, nobody has done one yet. :)
16:12 whiteknight http://www.google-melange.com/gsoc/prop​osal/review/google/gsoc2011/bbatha/1001
16:13 whiteknight that's the proposal. You can judge for yourself if it looks reasonable
16:13 whiteknight jnthn__: of course, we probably won't be using 6model in parrot during this time
16:16 jnthn__ whiteknight: The proposal is quite ambitious, though it omits various things I took on which were challenging.
16:17 jnthn__ Actually I started out with class and method stubs as my starting point
16:17 jnthn__ And then started filling them out as far as was possible with the translator.
16:17 mj41 left #parrot
16:17 jnthn__ Exceptions were tricky.
16:18 GodFather left #parrot
16:19 jnthn__ I suspect much of what's proposed can be done, but it won't lead to running a great deal of actual interesting Java code.
16:20 jnthn__ I started running into things like, you want to translate the IO stack. But it depends on concurrency to make sure that IO is done thread-safely.
16:20 jnthn__ I spent weeks on it and didn't get hello world to run ;)
16:21 cotto that's instructive
16:21 JimmyZ left #parrot
16:22 jnthn__ It's not because I couldn't translate many instructions. In the end I had decent coverage of those, and the OO features, and exceptions, etc.
16:22 jnthn__ It was all the other stuff that goes around it.
16:22 jnthn__ And the fact that there were 200+ "VM internal" methods, iirc.
16:23 whiteknight is any of that the same with JVM?
16:23 whiteknight we don't even know what we don't know at this point
16:24 whiteknight that's why none of our existing mentors are taking this one
16:24 jnthn__ I know far less about the JVM but my expectation would be to find a similar experience.
16:24 woosley left #parrot
16:25 whiteknight getting something as simple as "hello world" working by the end of the summer would be a necessity
16:25 whiteknight anything less, there's no way to consider that a success
16:25 jnthn__ Well, a class with a method that returned hello world would run fine :)
16:26 jnthn__ My tests were basically, take this bunch of C# and compile it to a DLL, run the DLL through the translator to get a PBC, then load it and write PIR tests that instantiated the classes in it and made method calls.
16:26 jnthn__ If getting "hello world" to work was critical, sure, you could put in some translation hack somewhere. :)
16:26 jnthn__ But doing it "properly" is probably a serious amount of effort.
16:27 whiteknight I wish we had this discussion weeks ago, when this proposal first came in
16:28 cotto Now is the second-best time to have it.
16:28 whiteknight of course, the vagueness of the proposal might suggest that the student hasn't done his due diligence in the first place
16:28 whiteknight maybe the student doesn't even understand what kind of task this is supposed to be
16:30 jnthn__ I guess my overal summary would be, the deliverables are probably do-able by a decent student, but won't allow one to "run previously written JVM applications" as suggested in the abstract.
16:30 jnthn__ *overall
16:31 whiteknight so best-case scenario is that we are looking at a complicated translation engine that only is able to run specially-written and toy applications?
16:33 jnthn__ If all that gets done on it is the work in this proposal, then - depending on how well engineered it is - it could well be a good starting point. But it'll be plenty more work beyond that before it will run anything non-toy.
16:34 whiteknight ...which brings us to the next question: do we have any devs waiting in the environs who are itching to continue such a project?
16:35 jnthn__ It's just too big for a GSoC project, plus a lot of serious Java apps expect threads, which I suspect will become a blocker at some point too.
16:35 whiteknight yeah, we really need to get onto the threads boat soon
16:35 whiteknight that's going to become a big priority for me in the coming months
16:35 whiteknight I suspect a mind-meld with cotto is imminent on that topic
16:36 birdwindupbird left #parrot
16:36 ambs left #parrot
16:37 jnthn__ So yes, it's only interesting if you expect it to seed something that the Parrot project plans to continue focusing on for a while. And IMHO, there's more critical things (e.g. Lorito) that could use the focus.
16:37 whiteknight okay
16:40 davidfetter joined #parrot
16:45 cognominal left #parrot
16:46 cognominal joined #parrot
16:46 Coke left #parrot
16:46 mj41 joined #parrot
16:47 Coke joined #parrot
16:47 yell0_ joined #parrot
16:47 Drossel left #parrot
16:47 Kulag joined #parrot
16:48 cognominal left #parrot
16:51 lucian_ left #parrot
16:51 lucian joined #parrot
16:53 darbelo joined #parrot
16:56 lucian_ joined #parrot
17:00 lucian left #parrot
17:07 Coke left #parrot
17:07 Coke joined #parrot
17:11 dodathome joined #parrot
17:23 cotto_work ~~
17:36 ShaneC left #parrot
17:37 Coke left #parrot
17:37 Coke joined #parrot
17:38 cotto_work Can someone update docs.parrot.org?
17:39 mtk left #parrot
17:44 cotto_work I'd like to see more people get the necessary account to do that.
17:44 mtk joined #parrot
17:49 lucian_ left #parrot
17:50 dalek parrot-linear-algebra: ddedd0b | Whiteknight++ | / (79 files):
17:50 dalek parrot-linear-algebra: rewrite harness. first pass to start updating tests.
17:50 dalek parrot-linear-algebra: review: https://github.com/Whiteknight/parr​ot-linear-algebra/commit/ddedd0bf3f
17:53 lucian joined #parrot
17:53 Eduardow left #parrot
17:54 whiteknight cotto_work: oh right. Who do I contact about that? osl support?
17:56 cotto_work You might be able to get them to give you an account on #osuosl
17:56 whiteknight where's that, irc.freenode.net?
17:56 cotto_work yes
17:58 preflex left #parrot
17:58 Eduardow joined #parrot
18:03 cognominal joined #parrot
18:11 dmalcolm joined #parrot
18:14 cotto_work whiteknight++ for an excellent release quote
18:14 soh_cah_toa hey guys. i was planning on asking this during yesterdays #parrotsketch but a family member needed to use my computer and i missed the meeting...
18:14 soh_cah_toa anyway, i was wondering if anyone had ever considered using flatrr (http://flattr.com) for accepting "micropayment" donations. i'm certainly not the business-type but i thought it may be a cool idea...
18:14 whiteknight I'm actually in the middle of reading that book, and came across that quote just a few days ago
18:14 soh_cah_toa there are a lot of other foss projects that use it (http://flattr.com/browse/software/all/1) and apparently its been quite successful
18:15 soh_cah_toa it's just an idea
18:15 cotto_work soh_cah_toa: that's an interesting idea.  I'm not sure how the legal details would work.
18:15 ShaneC joined #parrot
18:15 whiteknight soh_cah_toa: Parrot Foundation doesn't have non-profit status right now. Money is hard to deal with
18:15 lucian left #parrot
18:15 soh_cah_toa i don't think any of that stuff matters
18:15 whiteknight I'm absolutely certain that it does matter
18:16 cotto_work +1 that it does
18:16 ShaneC left #parrot
18:17 ambs joined #parrot
18:17 soh_cah_toa well, i think they differentiate between non-profits and profits. again though, not much of a business man
18:18 whiteknight we will definitely keep that site in mind
18:18 cotto_work I've been following their blog for a while.
18:19 soh_cah_toa oh, so you've heard about it. cool
18:19 benabik left #parrot
18:20 cotto_work I like the idea and would like to see Parrot take advantage of it, but having the legal stuff ironed out a hard prerequisite.
18:20 soh_cah_toa it's a pretty neat idea. b/c there's a ton of projects i'd love to support but as a student i'm quite low on funds. something like this is great for people like me
18:21 bluescreen left #parrot
18:23 soh_cah_toa that reminds me (i don't know why)...who comes up w/ the names for each release?
18:23 whiteknight the release manager
18:23 cotto_work It's the release manager's job
18:23 soh_cah_toa and that's...?
18:23 soh_cah_toa you?
18:23 whiteknight the person who makes the release
18:23 whiteknight it's different every month. We take volunteers
18:23 soh_cah_toa oh, okay
18:23 cotto_work docs/project/release_manager_guide.pod (at the bottom)
18:24 whiteknight we like to have many release managers, so if some people disappear we can keep functioning like normal
18:24 soh_cah_toa right
18:24 whiteknight soh_cah_toa: of note, we have openings in August and September and November
18:24 cotto_work dukeleto: ping
18:24 whiteknight soh_cah_toa: if you want to volunteer for any of those, you can pick your own name
18:25 whiteknight actually, it would be very good if some of those slots were taken by GSoC students
18:25 dukeleto cotto_work: pong
18:26 soh_cah_toa whiteknight: well, that would be pretty cool but i think i'd need to get involved w/ some "beginners-responsibilities" first
18:26 whiteknight soh_cah_toa: okay, but keep it in mind. I suspect you will be more than ready to do it by the end of the summer
18:26 soh_cah_toa whiteknight: which i hope to accomplish at pds
18:26 cotto_work dukeleto: do you have an outline of your yapc::na talk?  I want to minimize overlap.
18:26 whiteknight it's basically a matter of having a commit bit and following a checklist
18:26 soh_cah_toa whiteknight: sure, what else would it involve?
18:27 whiteknight https://github.com/parrot/parrot/blob/mast​er/docs/project/release_manager_guide.pod is the whole checklist
18:27 dukeleto whiteknight: we need to automate 80% of the current release manager guide
18:27 whiteknight you basically update the version number information in the repo. Run lots of tests, etc
18:27 whiteknight dukeleto: yes, I want to start working on that soon
18:27 atrodo 99%+
18:27 dukeleto cotto_work: i don't have anything more than my abstract right now
18:27 cotto_work dukeleto: ok
18:27 dukeleto atrodo: sure, but if we do 80% a few times, it works out to be the same :)
18:27 cotto_work I wouldn't either if I weren't talking on the 30th
18:28 soh_cah_toa whiteknight: yeah, you're right. i think after a few months i'd be more than capable
18:28 atrodo dukeleto> Point
18:28 dukeleto cotto_work: well, i can watch your talk and make sure that I am not overlapping too much
18:28 dukeleto cotto_work: you are talking about your "state of the vm" talk, right?
18:28 cotto_work dukeleto: or that
18:28 cotto_work yup
18:28 whiteknight what would be the best is if we had some kind of bot or daemon running on parrot.org which could update the drupal bits of the release process
18:29 whiteknight I'm sure it's just a few short update queries to the db
18:29 dukeleto cotto_work: i think my talk will be much more about the basics, and I won't concentrate much on the latest shiny, which I assume you will be doing
18:29 dukeleto whiteknight: sounds pleasant
18:29 whiteknight being able to update the trac milestones should not be too bad either. trac uses sqlite, I think
18:29 cotto_work dukeleto: I'll be doing some basics.  I want a complete newbie not to be too lost.
18:30 dukeleto cotto_work: ok, well i will observe your talk and then design mine to not overlap too much
18:30 dukeleto cotto_work: i submitted a last minute talk about Jitterbug to LFNW. Not sure if I will be chosen to speak, but I figured I should try.
18:31 soh_cah_toa make sure someone records your presentations. i'd love to go but it'd conflict w/ gsoc
18:31 whiteknight I don't think it would be too hard to put up a small password-protected admin site on parrot.org with a "bump version number" button on it
18:32 bluescreen joined #parrot
18:34 whiteknight I'm also wondering if we should start moving away from the osl ftp site and start using something like github to host the tarballs
18:34 whiteknight Especially if it's an undue pain to update the ftp site every month with public keys and ssh and all the best
18:34 whiteknight all the rest
18:34 cotto_work I'm fine either way.
18:35 whiteknight my only worry is that github probably doesn't have unlimited storage for an endless stream of releases
18:35 soh_cah_toa that reminds me, http://www.parrot.org/release/supported is broken  550
18:36 whiteknight damnit
18:36 whiteknight stupid drupal
18:37 cotto_work ftp://ftp.parrot.org/pub/parrot/releas​es/supported/3.0.0/parrot-3.3.0.tar.gz
18:37 cotto_work fixed
18:37 soh_cah_toa fixed. nice and fast. i like
18:37 cotto_work see if you can spot the goof
18:38 dukeleto whiteknight: github also names tarballs in a non-standard way that you can't change
18:38 dukeleto whiteknight: it embeds the sha1 in the tarball name
18:38 dukeleto but cotto_work's worry is valid as well
18:39 whiteknight oh weird
18:39 whiteknight that's a stupid typo on my parto
18:39 dukeleto we just need to automate docs.parrot.org and the ftp links. Those are the most common things that are broken after a release
18:39 dukeleto history tells us that they are mundane annoying tasks that almost everybody forgets to do
18:40 dukeleto soh_cah_toa: you would gain many karma points by helping us automate those 2 things
18:40 cotto_work Automating the tarball upload and docs generation would be *really* nice.
18:41 cotto_work It's easily-automated busy work.  The only reason I haven't automated it is because I only do it once a year (+/-).
18:41 soh_cah_toa well, what do you guys do to upload the tarball now?
18:41 whiteknight soh_cah_toa: scp and a little voodoo
18:41 * dukeleto is the release manager for 3.9 and wants it to be as little work as possible
18:42 soh_cah_toa hmm...i know scp but not much voodoo
18:42 whiteknight the biggest portion of time in the release is running fulltest twice
18:42 soh_cah_toa whiteknight: ouch
18:42 whiteknight although, that's also the easiest part.
18:42 whiteknight updating the version numbers in all the various files is tedious and easy to automate
18:43 whiteknight uploading the tarballs can stay manual. Updating the drupal links should be automated
18:44 cotto_work It's funny how often having a dayjob where I do Drupal hacking all day comes in handy.
18:44 soh_cah_toa well, i'm not too familiar w/ content management systems like drupal. do is have some sort of scripting engine for automation? if so, that'd be easy
18:44 dukeleto whiteknight: for updating strings in many files: https://github.com/leto/Util​/blob/master/bin/replace.pl
18:44 * dukeleto goes on a run
18:45 cotto_work soh_cah_toa: there's a scripting thingy called "drush"
18:45 cotto_work It'd probably be a pain to set up permissions for it though.
18:46 soh_cah_toa well, how do you propose to automate the release process then?
18:47 soh_cah_toa b/c i'd love to help. i just need some guidance
18:50 bluescreen left #parrot
18:51 whiteknight I could probably write a short script in ~20 minutes that would do most of the necessary work.
18:53 jevin left #parrot
18:53 jevin joined #parrot
18:53 whiteknight 1) build the old parrot for bootstrapping 2) update version numbers and other details in various files 3) use old parrot to build the core ops 4) reconfigure + rebuild to get new parrot 5) make fulltest 6) make release 7) make release_check
18:53 benabik joined #parrot
18:53 whiteknight that's all very very easy to automate. Everything besides step #2 there could be a small makefile
18:54 soh_cah_toa whiteknight: do you guys use git tags for version numbers?
18:55 whiteknight we tag releases, yes
18:55 soh_cah_toa whiteknight: lightweight or annotated?
18:55 whiteknight lightweight
18:55 soh_cah_toa okay
18:55 soh_cah_toa actually, it'd be pretty easy w/ a simple perl 5 script
18:55 cognominal left #parrot
18:55 whiteknight although I guess there's no reason we couldn't annotate a release tag
18:56 whiteknight soh_cah_toa: if you think you have the perl5 chops, go for it
18:56 whiteknight I'd probably try to write it in winxed :)
18:56 soh_cah_toa most of the things that need to be done are system commands anyway
18:56 whiteknight A short script could prompt for a few pieces of information: Enter the version number. Enter the release name. Enter a short quote (optional)
18:56 whiteknight exactly
18:57 whiteknight Can you pipe output through a system() call?
18:57 whiteknight I've never tried to do it directly
18:58 soh_cah_toa not sure. you can w/ open()
18:58 whiteknight ah, right
18:58 whiteknight You could use backticks too. Some parts of the release need to extract console output
18:58 whiteknight although, if "make fulltest" passes, there's no real reason to log its entire output
18:59 soh_cah_toa that's find
18:59 soh_cah_toa *fine
19:00 whiteknight an automated solution should probably check if PBC_COMPAT needs a bump, but shouldn't try to do it directly
19:00 whiteknight we can save that for human hands
19:01 soh_cah_toa alright
19:05 bluescreen joined #parrot
19:06 whiteknight I have a general question, why the hell do we have a ChangeLog file if we robotically update it every month to say "nothing useful here, look at NEWS instead LOL"
19:06 whiteknight I suggest we delete ChangeLog, or replace its entire contents with a single redirect to NEWS and never ever update it again?
19:08 soh_cah_toa whiteknight: when building old parrot, should i run a 'make install' or just 'make'
19:08 whiteknight and what exactly is the distro_tests target? It's so short I don't see why we wouldn't be running it more often
19:08 whiteknight just make
19:08 soh_cah_toa ok
19:09 whiteknight and does "make world" do anything different in our makefile than just "make" does?
19:09 whiteknight crow.pir should be able to search for .sha256 files in the repo and include them automatically
19:10 whiteknight crow.pir should *definitely* be able to email a list of recipients
19:12 soh_cah_toa whiteknight: do you want the script to clone the current version or should it be executed after a 'git clone'?
19:13 whiteknight I don't clone. you can do whatever you want there
19:14 soh_cah_toa so i can assume that i'm in the parrot root directory? for now, that can be fixed later
19:14 whiteknight yeah
19:14 soh_cah_toa alright
19:14 whiteknight assume that the person using this tool is a competent Parrot hacker
19:15 dukeleto soh_cah_toa: yes, assume you are in the root of a checkout
19:16 soh_cah_toa alright. i'm working on it now. you can take a look at it once it done. it's just a rought draft
19:16 dukeleto soh_cah_toa++
19:16 dukeleto whiteknight: i agree with all of your above sentiments
19:16 dukeleto +1 to removing ChangeLog
19:16 dukeleto whiteknight: i think ChangeLog is because debian requires a file like that, but I could be wrong
19:17 whiteknight dukeleto: We could generate such a worthless and repetitive file on the fly
19:17 whiteknight dukeleto: we have the technology to put a short text template in a short for-loop
19:18 dukeleto whiteknight: yes. I totally agree.
19:18 dukeleto whiteknight: should we be tracking all of this in a TT?
19:18 whiteknight maybe
19:18 whiteknight These are all things that I'll  be mentioning at #ps in about 5 months when I am next able to attend, if nobody mentions it first
19:19 dukeleto whiteknight: mentioning these in 5 months is not acceptable. You have the memory of what really sucked right now. You won't in 5 months.
19:20 * dukeleto creates a ticket
19:20 whiteknight I am tempted to suggest we should stop using milestones on trac too
19:21 whiteknight I went in last night and closed every track milestone since 3.0, all of which should be getting closed at their respective release
19:22 whiteknight and we haven't really been using them recently anyway. We may want to just reject the whole milestones system on trac
19:23 dukeleto whiteknight: i haven't found it very useful. It seems to be mostly wishful thinking
19:23 whiteknight once we started moving to our new system of more strict milestone goals, we stopped using the mechanism on trac
19:23 whiteknight so maybe we should just disable it entirely if we aren't using it
19:25 dukeleto whiteknight: http://trac.parrot.org/parrot/ticket/2099
19:25 whiteknight dukeleto++
19:25 dukeleto whiteknight: please add stuff that you feel is important to automate that I missed
19:25 whiteknight will do
19:25 dukeleto whiteknight: awesome
19:30 soh_cah_toa whiteknight: before you mentioned to rebuild and reconfigure. would this just be a 'git pull' and then rebuild like usual?
19:31 whiteknight make reconfigure and make. You probably don't need to pull at any time after the release has started
19:32 dukeleto soh_cah_toa: as a general rule, you should run Configure.pl after any git pull or rebase, lest you wake the daemons
19:32 soh_cah_toa okay, that's after 'make bootstrap_ops' for old parrot?
19:32 dukeleto soh_cah_toa: it will work without doing that sometimes. And other times, it will cause subtle bugs that will drive you insane.
19:32 soh_cah_toa the 'make reconfigure' i mean
19:32 soh_cah_toa dukeleto: alright
19:33 dukeleto soh_cah_toa: just general advice :)
19:33 soh_cah_toa dukeleto: i usually do that anyway. just for good measure
19:33 dukeleto soh_cah_toa: there is a good reason for it :)
19:33 benabik What is 'make reconfigure'?
19:34 soh_cah_toa yeah, i can't find it either. whiteknight, where's 'make reconfigure'?
19:34 dukeleto benabik: basically calls Configure.pl again with the same args as last time
19:34 dukeleto it is "make reconfig"
19:34 soh_cah_toa dukeleto: yeah, there it is
19:35 dalek TT #2099 created by dukeleto++: Automate more of the release manager process
19:35 dalek TT #2099: http://trac.parrot.org/parrot/ticket/2099
19:38 soh_cah_toa 'make fulltest' has to be run twice, right?
19:39 whiteknight yes.
19:39 whiteknight make release_check automatically does that, I think
19:39 whiteknight or, does it the second time
19:39 soh_cah_toa alright
19:43 dukeleto someone needs to port BLOOD to Parrot: http://portal.acm.org/citation.cfm?id=130489
19:43 dukeleto so we can say "some parrot hackers write in BLOOD"
19:45 jevin left #parrot
19:50 jevin joined #parrot
19:55 jevin left #parrot
19:59 lucian joined #parrot
19:59 benabik left #parrot
20:00 benabik joined #parrot
20:03 rohit_nsit08 joined #parrot
20:04 rohit_nsit08 hello #parrot
20:04 soh_cah_toa rohit_nsit08: hey
20:04 rohit_nsit08 soh_cah_toa: hi
20:04 rohit_nsit08 how are u ?
20:04 whiteknight hello rohit_nsit08
20:04 rohit_nsit08 whiteknight: hello
20:05 rohit_nsit08 parrot 3.3.0
20:05 tadzik oh, the rhyme
20:05 rohit_nsit08 i liked the name "fire in the sky" :-)
20:06 * davidfetter puts on some deep purple for rohit_nsit08
20:06 tadzik dum, dum dum dum dum duum, dum
20:06 soh_cah_toa smoke on the water
20:06 tadzik bravo!
20:06 rohit_nsit08 soh_cah_toa: one of my fav!!
20:07 benabik_ joined #parrot
20:07 soh_cah_toa i wonder if that's what it's actually named after
20:08 benabik left #parrot
20:08 benabik_ is now known as benabik
20:08 whiteknight I read an interview one time that says it was named after a building fire the band witnessed
20:09 soh_cah_toa the gambling house
20:09 rohit_nsit08 whiteknight:  I'm trying to build 3.3.0 on my system, pulled it from git. says incompatible version of 'core ops'
20:09 rohit_nsit08 pbc_to_exe error
20:09 whiteknight rohit_nsit08: make realclean, reconfigure, and try again
20:10 jevin joined #parrot
20:13 rohit_nsit08 whiteknight: worked and done.
20:14 whiteknight rohit_nsit08: awesome. I wish we could avoid that problem
20:14 soh_cah_toa whiteknight: alright, i've got a rough draft for the release manager script. should i nopaste? dcc it to you?
20:15 whiteknight soh_cah_toa: email it? I'm heading home from work soon
20:15 soh_cah_toa parrot at work? that's dedication
20:15 whiteknight barely
20:16 whiteknight all I do from here is chat and run tests
20:20 * soh_cah_toa will be back in a minute
20:24 whiteknight soh_cah_toa++
20:25 dodathome left #parrot
20:26 mtk left #parrot
20:26 bluescreen left #parrot
20:28 cotto_work soh_cah_toa: you can nopaste it too.  I know whiteknight isn't the only one who cares.
20:33 mtk joined #parrot
20:36 whiteknight left #parrot
20:38 dalek tracwiki: v2 | cotto++ | ParrotTalks
20:38 dalek tracwiki: add OSBridge talk
20:38 dalek tracwiki: http://trac.parrot.org/parrot/wiki/P​arrotTalks?version=2&action=diff
20:42 nopaste "soh_cah_toa" at 192.168.1.3 pasted "Release Management Automation Script (Rough Draft)" (146 lines) at http://nopaste.snit.ch/41132
20:42 * cotto_work looks
20:44 soh_cah_toa don't mind the indentation in the tag() function. i accidentaly had them as tabs instead of spaces and nopaste reformatted it
20:44 soh_cah_toa actually, i see there's a few places. easily fixed
20:45 cotto_work soh_cah_toa: have you looked at the release makefile target?
20:45 cotto_work don't worry too much about that at this point
20:45 soh_cah_toa whiteknight told to
20:45 cotto_work sketching out the idea is the important part
20:45 soh_cah_toa *told me to
20:45 cotto_work (I mean stuff like whitespace fixes)
20:45 soh_cah_toa oh, okay
20:46 soh_cah_toa anyway, yeah i did look at the release target. why?
20:46 cotto_work That seems like a natural place to put the code
20:47 cotto_work maximum laziness for the release manager
20:47 benabik soh_cah_toa: Why chomp $line; print "$line\n" and not print $line ?
20:48 * soh_cah_toa smakes forehead
20:48 soh_cah_toa yeah, that just removes the newline only to add it again in the next line
20:49 soh_cah_toa and i meant to say smack
20:49 benabik soh_cah_toa: That was what I thought, but I'm only glancing at it while doing homework.
20:49 * soh_cah_toa SMACKS again
20:49 soh_cah_toa benabik: good catch
20:50 soh_cah_toa the old way adds a lot of unnessecary overhead
20:53 cotto_work Automation is much more valuable than efficiency.  If the script takes twice as long as it would to do it by hand, but the script can be run unattendedly, it's still an improvement.
20:53 soh_cah_toa cotto_work: true
21:02 fperrad left #parrot
21:12 bubaflub left #parrot
21:13 hercynium left #parrot
21:21 sigue joined #parrot
21:28 ambs left #parrot
21:34 ShaneC1 joined #parrot
21:39 davidfetter_ joined #parrot
21:41 theory left #parrot
21:42 davidfetter left #parrot
21:43 davidfetter_ is now known as davidfetter
21:46 darbelo left #parrot
21:46 darbelo joined #parrot
21:47 benabik left #parrot
21:50 mj41 left #parrot
21:51 darbelo left #parrot
21:52 benabik joined #parrot
21:52 benabik left #parrot
21:54 ShaneC joined #parrot
21:56 ShaneC1 left #parrot
22:00 rohit_nsit08 left #parrot
22:22 Coke left #parrot
22:22 Coke joined #parrot
22:26 bacek_at_work ~~
22:42 bubaflub joined #parrot
22:48 cognominal joined #parrot
22:50 smash left #parrot
22:50 kid51 joined #parrot
22:58 cognominal left #parrot
22:58 whiteknight joined #parrot
22:59 yell0_ left #parrot
23:02 pranq joined #parrot
23:02 pranq left #parrot
23:04 Andy_ left #parrot
23:06 theory joined #parrot
23:06 dalek nqp/ctmo: e199d50 | jonathan++ | src/pmc/stable.pmc:
23:06 dalek nqp/ctmo: Mark WHO.
23:06 dalek nqp/ctmo: review: https://github.com/perl6/nqp/commit/e199d50809
23:06 dalek nqp/ctmo: 30a5cbb | jonathan++ | src/metamodel/how/NQPClassHOW.pm:
23:06 dalek nqp/ctmo: Catch adding a null method - it's almost certainly a sign of a problem.
23:06 dalek nqp/ctmo: review: https://github.com/perl6/nqp/commit/30a5cbb263
23:06 dalek nqp/ctmo: a89b31b | jonathan++ | src/Regex/P6Regex/Actions.pm:
23:06 dalek nqp/ctmo: Actually declare that $REGEXNAME is our scoped rather than never actually declaring it anywhere in NQP.
23:06 dalek nqp/ctmo: review: https://github.com/perl6/nqp/commit/a89b31b2d7
23:06 dalek nqp/ctmo: 592c026 | jonathan++ | src/ (2 files):
23:06 dalek nqp/ctmo: Avoid a double-encoding bug (by us then by the toolkit) that oddly manifested itself during other refactors.
23:06 dalek nqp/ctmo: review: https://github.com/perl6/nqp/commit/592c026989
23:12 dalek nqp/ctmo: c45260d | jonathan++ | src/NQP/Actions.pm:
23:12 dalek nqp/ctmo: First bite into the package refactor. This changes our-scoped variable storage and lookups in packages. It's enough to give us working storage and lookup of package-scoped variables in lexical packages though, which has never worked before. Various unsurprising regressions. Also eliminates a bit more Q:PIR.
23:12 dalek nqp/ctmo: review: https://github.com/perl6/nqp/commit/c45260d0fd
23:16 dalek nqp/ctmo: fe98d9f | jonathan++ | src/stage0/ (7 files):
23:16 dalek nqp/ctmo: Rebootstrap with the first set of package changes. No especial need, it's mostly just comforting that it still passes the same tests as pre-rebootstrap. :-)
23:16 dalek nqp/ctmo: review: https://github.com/perl6/nqp/commit/fe98d9fbe4
23:26 lucian left #parrot
23:46 kid51 Hmm, 'master' is now building successfully on Darwin/PPC.
23:46 * kid51 keeps fingers crossed re tests.
23:47 cotto_work I didn't know anyone had tried to fix it.
23:47 kid51 But still getting 4 test failures in t/src/extend_vtable.t
23:48 kid51 linux/i386: make fulltest PASS
23:48 whiteknight I'm starting to sincerely dislike that test
23:48 kid51 discuss w/ dukeleto!
23:51 GeJ Good morning everyone.
23:51 kid51 Bonjour, G�raud!
23:54 whiteknight section 10.f of the release manager guide is *extremely* light on details
23:54 cosimo joined #parrot
23:57 whiteknight so I'm supposed to take docs/resources and docs/html, and expand them both into a single directory in the web root?
23:59 whiteknight and maybe somebody can explain to me why this is a manual process at all? Why don't we have a post-commit hook or something here to automatically pull changes from the repo and build the html files in place?

| Channels | #parrot index | Today | | Search | Google Search | Plain-Text | summary

Parrot | source cross referenced