Parrot 2.9.1 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | svn is dead, long live git! github.com/parrot/parrot
Set by moderator on 8 November 2010.
00:24 whiteknight joined 00:25 dmalcolm left
cotto_work hio whiteknight 00:28
whiteknight hello cotto
cotto_work It kinda died out in here after the migration. 00:42
00:44 particle joined
whiteknight that's fine. I'm sure people are hard at work trying to figure out how to use everything 00:48
where is that git documentation page that dukeleto made? 00:52
cotto_work docs/project/git_workflow.pod ? 00:54
whiteknight ah, I hadn't updated my fork to master yet 00:55
mikehh dunno if I screwed that up or not, push to github, see what happens 01:02
dalek rrot: 6151934 | mikehh++ | t/manifest/01-basic.t:
TODOed a test to get it to pass - problem with git migration
 01:04
rrot: 40e8e2c | mikehh++ | MANIFEST (2 files):
ran tools/dev/mk_manifest_and_skip.pl to update MANIFEST and MANIFEST.SKIP
rrot: 64b2954 | mikehh++ | t/configure/018-revision_to_cache.t:
added TODO to get test to pass after git migration
cotto_work nice to see that working 01:05
01:05 Limbic_Region joined
mikehh does it send emails, cause haven't got 'em yet 01:07
cotto_work mikehh: nyi 01:08
high priority though
GeJ Bonjour everyone. 01:09
mikehh hi GeJ 01:10
GeJ Quick question, could someone tell me how big the parrot repository is after the git clone?
heya mikehh.
mikehh GeJ: 107.2 MiB 01:11
GeJ mikehh: thank you. 01:12
whiteknight github does have a post-commit hook that can send emails
mikehh thats what I got anyway
whiteknight we can probably have those sent to parrot-dev, or whatever list it is
cotto_work whiteknight: yes but it doesn't send full diffs
bacek_at_work ~~ 01:19
seen mj41
aloha mj41 was last seen in #perl6 31 days 6 hours ago joining the channel.
bacek_at_work sigh...
We've lost taptinder.
GeJ can't we use the diff from the commit URLs?
mikehh bacek_at_work: it still thinks we are on svn 01:21
cotto_work goes home 01:24
possibly never to check parrot out via svn again 01:25
01:37 dngor_ joined 01:41 dngor left
cotto ~~ 01:47
mikehh you got home :-} (20 minutes commute?) 01:49
mikehh gets to work at home most of the time
mikehh especially as I have no local clients :-} 01:50
whiteknight yay! I merged my first branch 01:58
of course, I did it in my fork, bu it's merged nonetheless 01:59
GeJ still checking out... :/
cotto GeJ, at least it'll be fast once you're over the initial clone 02:00
GeJ cotto: indeed. 02:01
whiteknight I've cloned a few times today, and all of them were relatively quick 02:02
GeJ need MOAR BANDWIZZZ!
cotto GeJ, what's your connection? 02:03
speed
GeJ 512KB yeah, I know. 02:04
dalek rrot: 7c6c829 | Whiteknight++ | / (28 files):
Merge remote branch 'origin/stdhandle_meths'
 02:07
GeJ I meant, 512 Kbps
dalek rrot: 0671540 | Whiteknight++ | / (4 files):
Merge branch 'master' of github.com:parrot/parrot
02:19 Andy left
plobsing is the funny character on the 3rd line of CREDITS (<feff>) supposed to be there? 02:28
cotto plobsing, looks broken 02:29
GeJ isn't that the BOM thingy?
bacek_at_work plobsing, it's unicode BOM. Should be first bytes in file
cotto we can nuke $Id$ now 02:33
bacek_at_work we _should_ nuke $Id$ now 02:36
plobsing I looked through the file for obviously unicody names, but I don't see any that have been mangled. removing.
dalek TT #264 closed by whiteknight++: Add experimental method stdhandle to the interpreter pmc 02:39
TT #264: trac.parrot.org/parrot/ticket/264
rrot: 95508f2 | plobsing++ | CREDITS:
kill stray BOM
cotto \\me will now start referring to plobsing as "the BOM squad" 02:42
GeJ Hum, looks like gerd and myself merged.
cotto Hmmm. You're definitely not him. 02:43
GeJ I know that. But github doesn't seem to think so. 02:44
cotto are commits by you and him attributed to the same person? 02:45
dukeleto, ping
GeJ At least two of my commits (on CREDITS) are attributed to him. 02:46
02:46 dngor_ is now known as dngor
GeJ back to work know. I'll have a look at it later. 02:46
GeJ &
02:56 cognominal left 02:57 rurban_ joined
dukeleto yay, dalek is reporting git commits! 02:57
sorear++ # was that you?
02:59 rurban left, rurban_ is now known as rurban
dukeleto whiteknight: ping 02:59
msg whiteknight can you please use "git pull --rebase" to update your git repo? that will avoid " Merge branch 'master' of github.com:parrot/parrot 03:00
aloha OK. I'll deliver the message.
dukeleto should Parrot::Revision even exist anymore? 03:02
plobsing I had to 'git add -f' to add files from include/parrot (they are somehow getting filtered by .gitignore)
that's not supposed to happen right?
whiteknight dukeleto: aye aye 03:03
but now I am off to bed
03:03 whiteknight left
cotto dukeleto, did gej and gerd get merged by your branch filtering? 03:03
03:05 bluescreen left
GeJ me being 'geraud' in trac. Just saying. That'd be a Levenshtein distance of 2. 03:07
bacek_at_work plobsing, which files in parrot/include? 03:09
plobsing bacek_at_work: include/parrot/{encoding,hash,list,nci,oplib/core_ops,...}.h but it just says include/parrot is ignored 03:12
dukeleto cotto: i hope not
bacek_at_work plobsing, yes. core_pmcs.h should be removed from .gitignore now 03:13
cotto dukeleto, I think they did
dukeleto cotto: gist.github.com/668635 03:14
bacek_at_work plobsing, hang on. May be not.
dukeleto cotto: that is my script. gej and gerd have different entries
cotto: oh shit.
"geraud") n='Gerd Pokorra' ; m="pokorra@uni-siegen.de" ;;
cotto trac.parrot.org/parrot/changeset/49717 should be geraud
dukeleto that is not correct, i am guessing?
damnit. 03:15
that was a jet-lag induced error, i tell you.
what do we do?
bacek_at_work plobsing, hmm. These files are not ignored in my checkout. 03:16
dukeleto how many commits are wrong?
bacek_at_work 47000? 03:18
bacek_at_work hides
dukeleto chases bacek 03:20
GeJ dukeleto: I haven't had the chance to make a lot of commits. I'd say half a dozen tops. 03:21
cotto I have 5 commit messages from gej 03:22
dukeleto GeJ: i am really sorry about this. I am trying to figure out what is best to try to do
GeJ cotto: confirmed.
bacek_at_work dukeleto, rewrite history...
cotto dukeleto, +1 to what bacek_at_work said. better now than later 03:23
dukeleto fun thing is that we have to know which sha1's they are and do it manually, because there is nothing to rewrite now
cotto dukeleto, that's easy 03:24
just follow the urls in the commit messages and they'll redirect to the right place
dukeleto This was caused because i want hand-editing the filter script, because the data was not put into CREDITS properly 03:25
cotto: no, this is nontrivial
GeJ dukeleto: There's nothing to be sorry about. Given the 5 commits I did, I'd say "do nothing". 03:26
dukeleto cotto: we need a script that stops at those sha1's and commits them with a different author
GeJ: i feel really bad that your commits got lost in the shuttle. It isn't right. 03:27
cotto: can you give me a list of sha1's that should be GeJ's ?
cotto sure. just a sec
GeJ dukeleto: well don't! :)
dukeleto i can write the script, i just need the sha1's
I really hope this is the only mistake. I don't want to do this *yet* again. 03:29
plobsing what is the optimal wait period such that we expect to find all similar errors, but not too many people will need to rebase after? 03:31
nopaste "cotto" at 192.168.1.3 pasted "sha1sums for most of geraud's commits" (5 lines) at nopaste.snit.ch/25409
cotto the middle one didn't appear in the output of git log
nopaste "GeJ" at 192.168.1.3 pasted "GeJ's commits (4 out of 5)" (6 lines) at nopaste.snit.ch/25410
cotto probably because it was just a metadata change
GeJ thanks cotto :) 03:32
03:32 lidden left
dukeleto sha1's are based on content *and* metadata 03:32
"most" of the commits?
mikehh codetest blows up straignt away 03:33
dukeleto mikehh: the tests for Parrot::Revision are just horrible
mikehh: the test about $Parrot::Revision::current fails because it is in a temp dir, but there is odd stuff going on when you require the module, so I couldn't get all the tests to pass at the same time 03:34
Parrot::Revision seems to be a lot more complicated than it needs to be 03:35
cotto: so why is one of GeJ's commits in svn but not git?
cotto dukeleto, that's my best guess
dukeleto cotto: what repo where you running that on? 03:36
cotto: it was perhaps a commit that happened after you last updated it
mikehh dukeleto: I TODOed the test , but it reports something like - fatal: Not a git repository (or any of the parent directories): .git
cotto dukeleto, could be
dukeleto mikehh: yes, it is running it from a temp directory
mikehh dukeleto: I'll have to have a look later - it's approaching 4am for me and I definately need a break - bbl 03:38
dukeleto i am not sure what to do without the last of GeJ's sha1's 03:39
mikehh oh make corevm/make coretest, make world/make test PASSed ok, but as I said make codetest blows up straight away 03:40
GeJ the commit message says "Pet file_metadata.t by adding missing SVN properties." but I can't find such file.
sorry, nevermind 03:41
mikehh GeJ: I think that has gone
GeJ I'm useless today.
dukeleto delete any test that is talking about svn metadata 03:42
GeJ dukeleto: most probably it was a `svn propset` only commit.
not sure if it got logged in git.
dukeleto hates svn properties even more now.
mikehh GeJ: you got to wrap your head around new ways of doin' things :-}
dukeleto $ svn log -r49390
------------------------------------------------------------------------
that is all it prints.
svn-- 03:43
GeJ: ok, then. I will fix your 4 code commits. You deserve it.
cotto dukeleto, that's probably because the svn server is down
nm
dukeleto cotto: nope. run "svn log", it works fine
cotto it works for some commits
dukeleto cotto: svn log is read-only 03:44
why doesn't svn show the svnpropset commit? I don't care.
cotto Moving away from it was a good idea.
mikehh anyway I am starting to miss keys so I better get some sleep, cu 03:45
cotto mikehh, g'night
GeJ good night mikehh 03:46
dukeleto fixes parrot.git 03:48
03:51 moritz_ joined, moritz left
GeJ dukeleto: you really don't have to. But thanks. 03:54
cotto GeJ, if you committed something it's important that our repo reflect that. It wouldn't be proper to let your commits fall through the cracks. 03:59
dalek rrot: 2ee2d42 | plobsing++ | / (2 files):
cleanup t/distro/file_metadata.t references

t/distro/file_metadata.t is gone with svn->git migration. harnesses should no longer be trying to run it.		 04:03
dukeleto cotto: can you look through gist.github.com/668635 and see if you can spot any more errors? 04:04
cotto: i am writing the new filter-branch command now
cotto looking now 04:05
dukeleto, I've looked through it and don't see anything amiss, apart from the lack of an umlaut in masak's full name. 04:11
plobsing cotto: that error appears to have been propagated from the CREDITS file 04:13
04:14 Limbic_Region left
GeJ dukeleto: in your gist, line 56, I guess the information for "geraud" would be the same as "GeJ" 04:21
I don't know if that matters, but there's a "geraud" on github, and he's not me either.
(I probably should unify my accounts and put myself as GeJ on trac) 04:22
cotto GeJ, github will go by email address
as long as he's not using yours, you'll be linked to your commit 04:23
dukeleto gist.github.com/668696 <-- my proposed fix
cotto: code review please ^^^
cotto: please verify the commits mentioned there are the ones you told me were GeJ's 04:24
GeJ: stuff on the left are svn usernames, that get translated into names and emails 04:25
cotto dukeleto, looks good
dukeleto there is a subtle bug there. rev-parse HEAD doesn't change inside of a filter-branch, but $GIT_COMMIT is there for that 04:27
jhelwig++
jhelwig: code review again please : gist.github.com/668696 04:28
jhelwig That should do what you're looking for, yeah. 04:29
I normally quote both sides of the test, but other than that... ( [ "$foo" = "bar"] )
Writing anything non-trivial in shell sucks. :-\\ 04:30
GeJ dukeleto: I added a comment on gist.github.com/668635
dukeleto GeJ: thanks :) I am writing a different script to fix that error. That was the original script that I ran, with the error in it 04:31
GeJ: fixing the script does us no good, because all your commits have be rewritten
jhelwig dukeleto: You could re-run the original script (with the bugfix) on a clean repo. 04:33
The original refs should all be under refs/original/*
dukeleto jhelwig: i did the filter-branch many times. Not even sure which was the original. And there have been commits to git, so I need to use the newest repo 04:34
jhelwig Fair enough.
dukeleto jhelwig: so i am just going to run that smaller fixer
GeJ: does your info look good in gist.github.com/668696 ? 04:36
sorear dukeleto: yes, I added the dalek git parrot thing
dukeleto sorear++ 04:37
sorear: i am going to force push a new repo soon. Hopefully that doesn't kill dalek
sorear I'll restart the poller 04:38
GeJ dukeleto: looks good, thanks. 04:39
sorear actually it won't kill dalek
but the last 30 or so commits will be reported twice
dukeleto has found the magic commitish 22840928853892a1b5ed020799702956d2f5ee76^..498de4d5554f9d3e5dba6634e4e7c3c20c5ef28a^ 04:44
which is small and contains the wrong commits
dukeleto fixes stuff
jhelwig: if i only filter part of history, that makes the history of that disjoint, yes? 04:46
s/of that/after that/
jhelwig: gonna do a : git filter-branch --env-filter '. /home/leto/fix_gej.sh' -- --all 04:47
jhelwig From that point forward (anything that has that as a parent), yeah.
Where "that point" is the first "changed" commit. 04:48
dukeleto jhelwig: as i thought, my magic commitish is for naught
gist.github.com/668718 <-- what dukeleto is using to fix bug with GeJ++'s commits 04:56
dukeleto crosses fingers
cotto crosses dukeleto's fingers 04:59
dukeleto crosses his crosses with double crosses
05:01 eternaleye left, eternaleye joined 05:26 davidfetter joined
dukeleto cotto: ping 05:28
cotto: gist.github.com/668749 <-- that is the new output of "git shortlog -se", which shows 4 commits from GeJ
GeJ: ^^^ 05:29
someone tell me it looks sane, and I will do a force push and let people know
Is anybody still awake? 05:31
dukeleto does a while(1) { $house.ring_door_bell } 05:32
dukeleto doesn't feel like waiting much more 05:33
bacek_at_work dukeleto, afternoon
dukeleto bacek_at_work: !
bacek_at_work dukeleto, ?? !!
dukeleto bacek_at_work: i am about to do another force push, so GeJ++ gets his commits back
bacek_at_work: you will need to re-clone
bacek_at_work dukeleto, yes
actually git pull --force should work 05:34
cotto dukeleto, I see gej people 05:35
dukeleto cotto: does it look good?
bacek_at_work: you wil have disjoint history that takes up lots of space
bacek_at_work: but a gc could fix that
bacek_at_work: but it is easier to tell people to re-clone 05:36
bacek_at_work dukeleto, who cares about space???
dukeleto bacek_at_work: will "git pull --force && git gc" work ?
bacek_at_work: some people :)
bacek_at_work it's 100MB for 10 years of development
cotto dukeleto, +1
dukeleto stdhandle_methods got merged, correct? 05:37
bacek_at_work dukeleto, git gc should work too
dukeleto does a whole bunch of force pushes
only some branches were effected 05:38
only a small slice of history has their sha1's changed by this
basically, all branches and tags after GeJ's commits, which were all in the last few months
all branches that contain a commit in the last few months, that is
It is done. 05:40
bacek_at_work: let me know if "pull --force" works for you. I am trying it myself. 05:41
bacek_at_work: nope. it creates conflicts
bacek_at_work dukeleto, did you push it?
dukeleto bacek_at_work: pull is a merge
bacek_at_work: yes, it is there.
bacek_at_work: but a "pull --force" creates merge conflicts 05:42
cotto dukeleto, git pull --force worked, if noisily
dukeleto cotto: it worked for you? Perhaps i was doing it on a wonky repo
cotto: can you nopaste the output it gave you?
cotto np
dukeleto has about 17 parrot repos laying around 05:43
nopaste "cotto" at 192.168.1.3 pasted "git pull --force output" (19 lines) at nopaste.snit.ch/25417
dukeleto jhelwig: thanks for your help!
jhelwig: do you have any wisdom about "git pull --force" ?
cotto: cool! I shall notify the masses. 05:44
cotto I'll need to work with osuosl to update the trac db
shouldn't cause much interruption though 05:45
dukeleto cotto: yes, please and thank you. I need to sleep soon. I am running on 3 hours of sleep from last night, which is why I made stupid mistakes.
cotto and only during the few minutes when the upgrade is being done
dukeleto, night
aloha, clock?
aloha cotto: cotto: LAX: Mon, 21:45 PST / CHI: Mon, 23:45 CST / NYC: Tue, 00:45 EST / UTC: Tue, 05:45 UTC / LON: Tue, 05:45 GMT / BER: Tue, 06:45 CET / TOK: Tue, 14:45 JST / SYD: Tue, 16:45 EST
cotto dukeleto, I hope not to hear from you until morning, then 05:46
Hmmm. aloha seems to be addressing me twice 05:49
Hmmm. it looks like my local repo isn't happy 06:02
nm. I probably messed it up with an over-eager sed command 06:03
dukeleto cotto: don't scare me.
Yes, i am not asleep yet.
I just fixed another buglet.
dukeleto waits for the commits to show up in here before attempting to go back to sleep 06:04
plobsing: your last commit accidently didn't go into the newest parrot.git, but I committed myself and gave you a ++ 06:05
s/committed myself/committed it myself/ 06:06
but the first is not far from the truth at this point
perhaps dalek hates me?
msg GeJ you have your commits back in parrot.git :)
aloha OK. I'll deliver the message.
plobsing dukeleto: so long as its still there, I'm happy. did you get the gsoc_nci master merge as well?
that one is more important (it was a PITA) 06:07
dukeleto plobsing: when did it happen?
plobsing: github.com/parrot/parrot/commits/gsoc_nci 06:08
plobsing: is that out of date?
plobsing: i got everything before your BOM commit
plobsing: but the one commit after that wasn't in my force push. but everything before that should be there. But i don't see a gsoc_nci merge 06:09
plobsing: did you push it?
plobsing I thought I did
dukeleto plobsing: if you have a local copy, feel fresh to make a new git branch for it
plobsing: i don't think i ever saw a gsoc_nci merge happen, but I could be insane.
plobsing gitk shows my copy being at the same point as remotes/origin/gsoc_nci
(haven't force pulled yet for fear of losing work) 06:10
(yes, github's gsoc_nci is out of date)
wtf is going on?
cotto removing $Id$ isn't as simple as I'd hoped 06:12
dukeleto plobsing: save your work as patches with format-patch, and create a new branch with it
cotto: why not?
cotto: tests?
cotto: you should probably only remove that string if it is in the first 5 or 10 lines of the file
plobsing: you can do a "git apply *.patch" from the patches that format-patch spits out 06:13
plobsing: in a new branch
cotto dukeleto, go to sleep. I can figure it out. 06:14
;)
dukeleto sleeps
cotto just a healthy number of edge cases that prevent a naive sed command from dtrt
06:21 moritz_ is now known as moritz
GeJ msg dukeleto Thank you, and sorry for the trouble. :) 06:22
aloha OK. I'll deliver the message.
06:29 davidfetter left 06:34 sECuRE left, sECuRE joined
plobsing gsoc_nci fixed. brute force copy ftw. 06:50
cotto $Id$ is mostly purged 07:24
there are three remaining instances that I don't feel like figuring out 07:26
trac.parrot.org/parrot/wiki/GitMigration has a nice tasklist for anyone interested 07:50
cotto sleeps
GeJ Good night cotto. 07:51
Anyone witnessing build failures? 07:52
I think I have a fix, running fulltest to be sure. 07:54
If I'm the only one, then I'm probably going to break something. 07:55
bacek GeJ, on which platform? 07:56
GeJ FreeBSD 8.x
amd64
But I don't think it's platform related. The way I get it, it has to do with : github.com/parrot/parrot/commit/ff...16#diff-69 07:57
bacek looks suspicious 07:58
which tests are failing?
GeJ build failure.
bacek ./parrot -o parrot_nci_thunk_gen.pbc tools/dev/nci_thunk_gen.pir 07:59
error:imcc:The opcode 'concat_s_s_ic' (concat<3>) was not found. Check the type and number of the arguments
in file 'tools/dev/nci_thunk_gen.pir' line 354
make: *** [parrot_nci_thunk_gen.pbc] Error 1
this one?
GeJ precisely
I think I got it fixed.
just running tests to be sure.
bacek ok than
push it 08:00
GeJ will do in a few minutes. Thanks
08:05 jsut_ joined
GeJ msg dukeleto I just did my first commit under the git era. Everything's shiny. Not to fret. thanks! 08:09
aloha OK. I'll deliver the message.
08:10 jsut left 08:26 theory left 09:11 bacek left 10:13 bacek joined 10:29 masonkramer__ joined, masonkramer___ left 10:39 contingencyplan left 10:57 rurban_ joined 10:59 rurban left, rurban_ is now known as rurban 11:14 fperrad joined 11:29 kid51 joined
kid51 Who is supposed to post to parrot-dev re conversion to git? 11:29
Is there someone who has already performed Parrot checkouts from git who could pick up trac.parrot.org/parrot/ticket/1844, that would be appreciated. 11:42
People who participated in the git transition, please pre-post about that for today's #parrotsketch. Thanks. 11:44
11:45 gerd joined 12:10 particle1 joined 12:13 particle left 12:16 jsut joined 12:20 jsut_ left
moderator Parrot 2.9.1 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | svn is dead, long live git! github.com/parrot/parrot | Parrot mailing lists are down 12:28
12:29 bluescreen joined
kid51 gerd: Am cc-ing you on the OSUOSL bug ticket re mailing list being down 12:37
gerd kid51: Thank you, I can read the email with the web interface 12:40
but the mail server seems not to send them out 12:42
12:45 gerd left 13:00 patspam joined, patspam left 13:24 kid51 left 13:26 whiteknight joined
whiteknight good morning, #parrot 13:31
mikehh hi whiteknight - I am having great difficulty keeping track of what is happening at the moment 13:48
mailing list is down, dalek is not keeping up with git 13:49
dukeleto waves 13:50
mikehh: yeah, OSUOSL peeps aren't awake yet
mikehh: but they will probably wake up within an hour or two
mikehh dukeleto: I did a git pull a while back and loads of conflicts, re cloned 13:52
dukeleto mikehh: yes, sorry about that. There was an email to parrot-dev, but it didn't get there
whiteknight mikehh: that stinks. I'm sorry to hear that. This mailing list thing is becoming a little bit aggrevating
dukeleto mikehh: "git pull --force" would have worked, but a new clone is even better
whiteknight: indeed it is
whiteknight it is hard to complain about a service we are getting for free, but at the same time we are pretty dependent on that service 13:53
13:53 darbelo joined
mikehh I have a sort of master clone, which I am cloning various local copies (different ones for testing on different Ubuntu versions 13:54
I use that to pull from the local copy and then push to master 13:55
dukeleto mikehh: that is smart 13:56
mikehh: you are saving lots of bandwidth that way
mikehh: but you should read the email to parrot-dev, because I had to modify the repo last night
mikehh: clones before that require a "git pull --force"
mikehh: lists.parrot.org/pipermail/parrot-d...04992.html 13:57
ARG
the stupid mailing list corrupted a URL in the email 13:58
git clone git at github.com:parrot/parrot.git should be: git clone git@github.com:parrot/parrot.git
mikehh: if you have any questions, please ask me. 13:59
gsoc_nci got merged! 14:02
yay.
oh wait, i am wrong
but it seems a lot closer.
hmmmm. i think it was merged, but the remote branch did not get deleted 14:03
plobsing: ping
for people that did a "git pull --force" : I think you will also need to run "git pull --force --tags" to get the new tags 14:13
For people that did a new clone: You are all good. 14:14
how long until #ps? 14:26
sorear: ping
whiteknight dukeleto: what did you modify about the repo last night?
dukeleto whiteknight: GeJ++'s commits were incorrectly given to gerd++ 14:27
whiteknight: which was not acceptable to me.
whiteknight: so i had to filter-branch again to fix that
whiteknight: you may need a "git pull --force --tags" if you have the old clone
14:31 bluescreen left, bluescreen joined 14:32 bluescreen left, bluescreen joined
whiteknight gotcha 14:34
why were the commits being misattributed? 14:43
also, how do I fix this situation in my fork? Do I need to re-clone or can I update the history in-place? 14:44
dukeleto whiteknight: "git pull --force --tags" 14:46
whiteknight: i was jet lagged and Gerd and Geraud are very close. It was a mistake on my part when hand-editing the filter-branch command to fix stuff that was not in CREDITS 14:48
whiteknight okay, that's cool. Those kinds of things do happen. So long as it's fixed any everybody is happy about it, everything is good 14:49
and everybody expects some stumbling blocks the first day of the migration, so I doubt anybody is upset about it 14:50
dukeleto whiteknight: yep. 14:56
whiteknight dukeleto: now that all the hard stuff is done, I think you've earned yourself a little vacation 14:57
I don't know anybody who'se worked harder in the past few weeks than you 14:58
who's*
15:05 jsut_ joined 15:10 jsut left 15:15 AzureSto_ joined 15:17 AzureStone left 15:19 plobsing_ joined
jhelwig dukeleto: Seeing as Parrot is in git now, I'm shutting down the auto-updated git-svn clone. 15:26
whiteknight good idea 15:29
jhelwig++
plobsing_ dukeleto: pong 15:34
whiteknight plobsing_: dukeleto wanted to know whether the gsoc_nci branch was merged in last night, and if not, what the status is 15:37
plobsing_ It isn't merged. It got synced w/ master (conflicts, ugly) which was the only large barrier remaining against the merge (AFAIK). There's a couple codingstd tests failing, but I can squash those. 15:38
As indicated in my #ps pre-post, I do intend to have it merged sometime this week. 15:40
speaking of #ps, can somebody ask about abandoning pkg-config (as I'm suggesting with TT #1842)? once again, I cannot attend (except possibly if it runs long). 15:41
whiteknight I actually might be able to attend today, though I will likely be a little late 15:46
atrodo I'm excited. I forget that the time change means #ps starts sooner 15:48
plobsing_ oh. right, time change. Now I *definitely* cannot attend.
cotto This is one of those times I'm glad my phone lets me set events in utc. 15:50
15:53 cognominal joined
whiteknight plobsing_: question. How hard is it to convert a packfile structure into a PackFile PMC? 15:54
is it a VTABLE_set_pointer, or is it more nefarious?
plobsing_ that's it. it's just a VTABLE_set_pointer.
whiteknight perfect. That makes me and the new embedding API very happy
Because the new API is only going to expose PackFile PMCs to the embedding code, never the packfile structures 15:55
plobsing_ please also make it possible to operate without having to touch packfiles if we don't need that level of control 15:56
having to provide a dummy packfile always seemed like such an arbitrary chore. 15:57
whiteknight plobsing_: I do agree with you there. But there are plenty of cases where we might want to have the packfile objects available, so we need to cover those cases 15:59
plobsing_ true. that level of control should be available. but there's a large set of cases where embeders shouldn't have to care that much.
whiteknight From the embedder's perspective, most programs will have two parts: obtain packfiles, execute packfiles 16:00
whether those packfiles come from loading a .pbc file, or are output by a compiler is up to the embedder 16:01
or even from compiling a string
dukeleto jhelwig: sounds good. Git-svn is dead to me. 16:02
plobsing_ OK. the compiling/executing a string case is the one I'm most interested in.
but I can see the case for consistency
dukeleto plobsing_: i agree. dummy packfiles => we are doing it wrong.
whiteknight: i would say that, from the embedders point of view, packfiles don't exist. Only PBCs 16:03
whiteknight dukeleto: whatever you want to call it is inconsequential to me. It's an object representation of a slug of bytecode 16:04
whether that comes out of a compiler, from loading a .pbc file, or elsewhere doesn't matter.
dukeleto whiteknight: congrats on baby teeth :) 16:05
whiteknight You get the bytecode object, you execute the bytecode object
dukeleto whiteknight: sure. But no one but core parrot devs should even hear the word "packfile"
whiteknight: i am agreeing with you :)
whiteknight dukeleto: thanks. The first 8 teeth came in one-at-a-time and were manageable. This next set are all coming in at once and the kid is miserable 16:06
and he's miserable while I'm trying to sleep, which makes me miserable
plobsing_ whiteknight: one more request. If we can kill Parrot_run_native and PARROT_OP_enternative, I will be very happy. They are both crufty and non-reentrant.
dukeleto plobsing_: thanks for you work on gsoc_nci ! 16:07
atrodo whiteknight> I know the feeling. My kid got his first 8 teeth in just over a month
whiteknight plobsing_: is that an embedding API goal? I would need to read over the contents of those things first. I'm not familiar with them
dukeleto what is up with dalek?
whiteknight atrodo: what's funny is that it's such a pain in the ass to grow them, and my kid has already chipped one of his front teeth
plobsing_ whiteknight: let me know if/when/how I can help executing your master plan. 16:08
whiteknight didn't even take the time to appreciate them before breaking one
atrodo kids are so thoughtless
whiteknight plobsing_: right now it's all design and basic prototyping. This kind of feedback is invaluable
Most of it should be straightforward. In many cases I'm just creating thin wrappers around existing functions
thin wrappers with a consistant aesthetic and proper error-handling 16:09
plobsing_ Parrot_run_native (the only user of PARROT_OP_enternative) is used to execute C extn code while not under a runloop by setting up a dummy runloop and then calling a callback.
whiteknight plobsing_: let me abbreviate that for you: "it's stupid" 16:10
that sounds horrible
what's the possible reason for that?
plobsing_ oh, it also stores the callback in a static global variable
dukeleto This is relevant to getting Javascript working on parrot: matt.might.net/articles/by-example-...ing-style/
plobsing_ have fun with threads on that one
dukeleto plobsing_: please kill that with fire
whiteknight is that just a misguided way to capture exceptions from extension code? 16:11
plobsing_ whiteknight: reason? to this madness?
whiteknight plobsing_: I'm sure the original author had a reason, even if it was a bad one
who calls Parrot_run_native?
plobsing_ it looks almost like it's trying to setup an inferior runloops problem 16:12
maybe that's useful somehow
whiteknight ha
i need to go grab an early lunch. I'll be back in a bit
dukeleto mikehh: please read the "Keeping branches updated" section in docs/project/git_workflow.pod 16:14
mikehh: basically, use "git pull --rebase" instead of "git pull"
mikehh: this will make it so we don't see lots of merge commits like "Merge branch 'master' of /home/mhu/parrot"
mikehh: not a huge deal, but going forward, please use "git pull --rebase" 16:15
16:15 sECuRE left 16:26 sECuRE joined, plobsing_ left 16:28 dmalcolm joined 16:29 fperrad left
mikehh dukeleto: should I always include --rebase when I do a git pull (both directions?) 16:37
dukeleto mikehh: what do you mean "both directions" ? 16:38
mikehh: --rebase is only for "git pull", not "git push"
mikehh: is that what you are asking?
mikehh dukeleto: I have a git clone as a master and the I have local clones off that that I use eg in /home/mhu/parrot (i386) or /home/mhc/parrot (amd64) 16:39
dukeleto: I have been doing a git pull on the main clone and the pull from that for the local clone, when I have updated the local I pull it into the main clone and push from there 16:41
dukeleto mikehh: always use "git pull --rebase". It should Just Work.
mikehh: please let me know if it doesn't
mikehh: for your use case, i am pretty sure you can always do a "git pull --rebase" 16:42
mikehh dukeleto: 'k, I'll try on all the git pulls
dukeleto mikehh: thanks! 16:43
mikehh: it will get rid of "useless merge commits"
mikehh: "git pull" = "git fetch" + "git merge" 16:44
mikehh: "git pull --rebase" = "git fetch" + "git rebase"
mikehh: just so you know what is happening underneath
mikehh: git fetch copies the remote index to your local index. merge and rebase work on your "working copy"
mikehh: this should be explained in our git docs, but please let me know if there are some holes. I am sure some exist. 16:45
mikehh: "working copy" = "the actual files in the directory structure"
mikehh: make sense?
particle1 am i missing the friendly email to all parrot users that lists how to clone the new parrot git repo?
16:45 particle1 is now known as particle
particle the messages that mention the repo location are titled 'a slightly newer...' 16:46
darbelo particle: You are missing all emails to the list. parrot-dev is down.
particle aha.
dukeleto particle: i am talking to OSUOSL about it 16:48
particle: the archives work, tho
particle: git clone git@github.com:parrot/parrot.git
particle: that is for core devs that have a commit bit
moderator Parrot 2.9.1 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Long live Git! github.com/parrot/parrot | git clone git://github.com/parrot/parrot.git | parrot-dev is down 16:50
dukeleto the "git clone" command in the topic should work for everybody. Core parrot devs should use "git clone git@github.com:parrot/parrot.git" 16:52
16:52 nwellnhof joined 16:54 theory joined 16:55 contingencyplan joined
nwellnhof What's the best way to move local branches from my old git-svn clone to my new git clone? 16:57
git format-patch and git am?
darbelo Sounds safe enough.
17:03 dukeleto left
cotto_work ~~ 17:03
Is parrot-dev up? I just got some subscription confirmation messages. 17:04
17:05 dukeleto joined
nwellnhof I also got some mails from parrot-dev about half an hour ago. 17:06
17:11 dukeleto left 17:13 dukeleto joined
jhelwig nwellnhof: You could add your old git-svn clone as a remote then cherry pick, but format-patch && am is probably the most bullet-proof way. 17:16
nwellnhof ok, i'll go with the format-patch and am combo. 17:23
17:23 nwellnhof left 17:24 dukeleto left 17:36 dukeleto joined, dukeleto left 17:37 dukeleto joined 17:46 Andy joined
Andy OK, so the Parrot github repo is THE repo now. 17:47
Not a copy of the svn repo.
darbelo Yep. 17:48
Andy What's work flow then? 17:49
Who's pushing up to parrot/parrot from our forks?
Mail says "You can now safely push to parrot/parrot on github" but who is "you"
17:56 M_o_C joined
darbelo I read that as you = committers 17:57
There's a s git_workflow.pod file somewhere under docs/
moritz currently ./parrot_config revision returns a sha1 17:58
could it instead return the output from git describe?
that way we could compare revisions without having a parrot repo available 17:59
darbelo That'll probably involve a rewrite of the relevan config step.
moritz which one is it?
darbelo auto::revision IIRC 18:00
moritz thanks
PerlJam moritz: what do you mean by "compare revisions"? 18:01
moritz PerlJam: answer the question "is this parrot as least as new as $reference?"
PerlJam how does the output of "git describe" help with that?
moritz by providing a sequence number + origin (which is the last tag) 18:02
darbelo $ git describe --tags 18:03
RELEASE_2_9_1-197-g882b414
moritz $ git describe --tags
RELEASE_2_9_1-204-gc762e42
right, darbelo++ was faster
darbelo Yeah, but I'm 6 commits behind :) 18:05
moritz lib/Parrot/Revision.pm line 87
so, should I change it? 18:06
darbelo I have no idea. To be honest. I'd float it at the next #ps 18:10
PerlJam good thing that's later today :) 18:11
How does parrot use this revision interally? If it's just for exact matching (and I think that's all it can be with the SHA1) then it should be fine. 18:12
darbelo It doesn't really. It's mostly used by HLLs to determine wether a given parrot is 'too old' for the build. 18:13
PerlJam With just a SHA1, you can't determine too-old unless you have a parrot repo around (and even then, you can only answer "is this revision in my history") 18:14
darbelo PerlJam: Exactly. That's why moritz++ is suggesting the change.
My guess is that the config step was changed to the nearest git analog of the svn rev. 18:15
That's certainly good enough for rakudo's —gen-parrot stuff 18:16
moritz no.
darbelo Doesn't —gen-parrot fetch a specific rev?
I was sure it did.
moritz I'm a bit disappointed, because I've discussed this with dukeleto, and he seemd to agreed that git-describe was a good way to do it 18:17
darbelo: it does, but only if $available_rev < $required_rev
whiteknight moritz's suggestion here does seem pretty worthwhile 18:21
cotto_work I'm working with osuosl to update the trac db table that stores svn revisions, git commit ids and commit messages. In order to ensure that trac knows about all commits, lease hold off pushing to github until we're done. It shouldn't be more than a few minutes.
whiteknight Is there any reason we can't capture both: the get-describe info and the SHA1 of the specific commit?
cotto_work *please
darbelo whiteknight: Keys only hold one value? 18:22
moritz whiteknight: I'm sure we can capture both; I just don't know which knobs to turn
darbelo: there can be two keys
PerlJam there could be a "revision" key and a "tag" key (or something)
moritz let's say revision and git_hash
whiteknight darbelo: then add a second key
PerlJam moritz: works for me. 18:23
to be clear, git_hash is the SHA1, revision is the output of "git describe --tags" 18:24
moritz right
but I'm not keen on duplicating the capture-cache-and-store mechanism that we currently use for the revision 18:25
whiteknight do we want the github issue tracker and github wiki to be active? I'm inclined to hide those things and redirect traffic to Trac
moritz so +1 from me, if somebody else does it :-)
whiteknight: +1 18:26
darbelo whiteknight: Let's keep Trac. One migration is enough for this week. 18:27
cotto_work whiteknight: +1 18:28
whiteknight darbelo: I'm not talking about migration. There were links on github for a wiki and an issue tracker. I just disabled them 18:29
dukeleto: can you add me as an "owner" of the parrot org? 18:31
I figure all metacommitters need to be listed as "owners" to have the same capabilities as they did in svn
I don't think Andy (petdance) is listed as a committer there yet, hence his confusion 18:32
Andy I guess I was figuring that there might be a change in workflow going to github, but apparently not.
Committers is committers. 18:33
whiteknight Andy: yeah, no serious change in workflow
cotto_work Andy: not for now.
Andy Not complaining.
whiteknight there's room for some experimentation, but the meat and potatoes is the same
moritz I guess people will feel less averse against branching 18:36
cotto_work trac has been updated to know about the rewritten history. Resume pushing to github. 18:37
It looks like no commits were dropped on the floor either. 18:38
good times 18:39
moritz \\o/
cotto_work I hope we don't have to do that again, but the trac end of it will be easy enough to deal with if we do. 18:40
Andy Bravo for the work you guys did pulling this off. 18:46
I was only vaguely aware of it and I expected something longer and more agonizing.
cotto_work Andy: thanks 18:51
18:57 rurban_ joined 18:59 rurban left, rurban_ is now known as rurban
cotto_work matt.might.net/articles/by-example-...ing-style/ 18:59
dukeleto cotto_work: i already posted that in here :) 19:02
cotto_work dukeleto: when?
clearly I need to backscroll more often 19:03
dukeleto cotto_work: indeed you do. Earlier today.
cotto_work: irclog.perlgeek.de is full of win
moritz: i am looking at your patch now
moritz: i like your patch, but we need to keep the "revision" config key around as a plain int, for all the HLL's and projects that already exist 19:04
moritz: so, i would apply your patch, if it was stored in a different config key 19:05
moritz dukeleto: it's not an int right now.
dukeleto moritz: yeah, that is probably a bug too. It should be. That is my mistake.
moritz dukeleto: I don't understand the configure system enough to know how to install a new config key
dukeleto moritz: every HLL in existence now looks at the revision key and expects it to be an int 19:06
moritz so I'd happily delegate that, rather focus on the Rakudo side once git-describe output is available
dukeleto moritz: yeah, it is not exactly intuitive 19:07
moritz: yep. Can you create a TT and include your patch, assign it to me, and maybe put a note that the patch needs to effect a differnent config key
moritz: then you can go work on the rakudo side, and i will take care of it 19:08
moritz dukeleto: well, I need something to test the rakudo side with
dukeleto moritz: i think i will make the new key "sha1", for lack of a better name.
moritz wfm
wait
for the sha1, or for git-describe?
dukeleto moritz: oh, yes. we need both. 19:09
moritz: the "version" key is already taken
moritz: what should we use for the git-describe string?
moritz has no idea... git_rev maybe?
or maybe just git_describe
dukeleto pontificates 19:10
moritz anyway, I'll open a ticket
dukeleto moritz++
moritz: so, for your work, assume there will be 3 keys: 'revision', 'sha1', and $foo, where we haven't figured out $foo yet. But $foo will contain a git-describe string 19:11
mikehh dukeleto: has anybody chacked if make smoke works - about to try
dukeleto how long until #ps?
mikehh: good idea. It should. It worked in git-svn clones
mikehh just over an hour 19:12
dukeleto mikehh++
cotto_work sorear: what happened to dalek? 19:16
php-- 19:20
stupid case-insensitive function names
whiteknight Andy: I just added you as a committer 19:21
(you should have been previously. Sorry about the oversight)
dukeleto who is a metacommitter? 19:22
whiteknight I assume most of the old board is 19:23
"admins" on Trac include NotFound, allison, basic, chromatic, dukeleto, jhorwitz, jkeenan, particle, and myself 19:24
cotto_work basic?
whiteknight whether the Trac "admin" role is synonymous with our notion of "metacommitter" is another thing entirely. Those are just the people with the technical access to add committers 19:25
particle smash was a trac admin for a time, too
admin !=== metacommitter
whiteknight particle: yeah, that's true. But you need to be an admin on Trac to add committers
so at the very least, it's necessary but not sufficient 19:26
particle correct.
allison particle: admin has equalled metacommitter since we switched to Trac, since the committer rights were administered through Trac
particle basic works at ousosl, btw
allison: basic is not a metacommitter.
cotto_work particle: good to know. thanks
particle we added smash as a trac committer through the voting process 19:27
he was not a metacommitter.
allison particle: technically, basic was a meta committer, even though he wasn't a committer
particle we *could* create a trac privilege set called metacommitter, just for bookkeeping 19:28
but i think what we have now works.
whiteknight now that we're on github, Trac permissions involving metacommitters are moot
particle the confusion is limited to a handful of people who can quickly figure it out.
whiteknight but yes, for bookkeeping, we don't have a readily-available list of metacommitters
allison why on earth did wayland have ticket modification permissions but not "developer" permissions?
and atrodo had wiki modification permissions but not "developer" permissions? 19:29
mikehh dukeleto: smoke worked - see #998 - we are still getting a lot of svn smokes at r49810
allison seems silly to grant wiki but not ticket, or ticket without wiki
particle allison: then join the privsets into one
allison particle: they are joined, it's called "developer" permissions 19:30
particle as long as they are separate and we don't have someone tasked with documenting and how maintaining the permissions are granted...
allison particle: (which doesn't grant commit access)
whiteknight we also use "developer" as a marker for committers
particle could we call them 'contributor' permissions, then?
whiteknight you need to be in the developer group to become a committer
particle and use 'committer' for committers?
or... call them 'member' rather than contributor or developer? 19:31
should all members have access to modify tickets and wiki items? 19:32
i think yes.
whiteknight yes
particle should this privilige be limited to members?
i think yes. 19:33
whiteknight becoming a member is sufficiently easy
so, yes
sorear cotto_work: dalek is fine
particle so then they're the same, and the privilege set named 'developer' should be renamed to 'member' to reduce confusion.
and any doc for adding a member should mention that privilege 19:34
allison editing the wiki and tickets isn't restricted to members
the "member" permission only exists on www.parrot.org, not on Trac, and it's for voting
cotto_work sorear: why isn't he reporting any commits? 19:35
dukeleto mikehh: awesome. thanks
so, again. who is a metacommitter?
particle hrmm... i suppose editing a wiki is a gateway to membership
allison whiteknight: you have to have "developer" permissions before you can get commit rights, but there's an extra step to grant commit rights
sorear cotto_work: oh...
allison dukeleto: a metacommitter is anyone who has access to add a new committer to the git repository
whiteknight allison: yes, Only developers can be committers. Necessary but not sufficient
allison: he's asking for a list of current metacommitters 19:36
we don't have a list handy. I suspect each metacommitters knows whether she is one or not
dukeleto allison: not "what",
allison: "who"
allison we do have a list handy, it's in Trac
dukeleto i could argue about the "handy"-ness of that 19:37
whiteknight allison: okay, so all "admin"s on trac are metacommitters?
dukeleto just wants to allow others to add github committers 19:38
allison NotFound, allison, chromatic, coke, dukeleto, jkeenan, particle, whiteknight
dukeleto allison: thanks!
allison whiteknight: all of them except the OSU OSL admins, yes
whiteknight okay, that's what I was thinking
dukeleto allison: all of those people can now add people to our github repo 19:39
allison: as well as pmichaud, since he created the parrot org on github 19:40
19:46 davidfetter_vmw joined
davidfetter_vmw dukeleto, kudos on the SCM move! 19:46
moritz dukeleto: trac.parrot.org/parrot/ticket/1845 btw 19:47
dukeleto davidfetter_vmw: danke 19:48
moritz++
19:48 chromatic joined
davidfetter_vmw dukeleto, got any observations, or are you just trying to recover right now? 19:49
davidfetter_vmw contemplating a blog post
dukeleto davidfetter_vmw: observations? 19:51
davidfetter_vmw: it is sunny outside. That is one of my observations 19:52
davidfetter_vmw on the process, how it affected you, whether there's anything other projects might learn from it...
dukeleto davidfetter_vmw: well, i learned a bunch. i should probably write a blog post. But i have at least 4 other blog posts that are higher priority, related to Google Code-IN 19:53
davidfetter_vmw heh
dukeleto is wondering where to commit the scripts that he used to convert the parrot repo 19:54
moritz separate repo 19:55
GeJ Bonjour everyone. 19:56
dukeleto GeJ: welcome
GeJ: your commits are alive and well in parrot.git
darbelo aloha: clock? 19:57
aloha darbelo: darbelo: LAX: Tue, 11:57 PST / CHI: Tue, 13:57 CST / NYC: Tue, 14:57 EST / UTC: Tue, 19:57 UTC / LON: Tue, 19:57 GMT / BER: Tue, 20:57 CET / TOK: Wed, 04:57 JST / SYD: Wed, 06:57 EST
whiteknight chromatic: what is lacking in the GC MS 2 branch, just testing? 20:00
dukeleto mikehh: just to let you know, Parrot::Revision is wrong right now. I am going to fix it. Hold off on changing any tests related to it.
cotto_work #ps in 30 20:01
chromatic Just testing. 20:03
dukeleto chromatic: howdy
whiteknight chromatic: do you need timings and things too? 20:04
chromatic Timings would be very nice.
whiteknight okay. No promises for today, but starting tomorrow I will see what I can do
chromatic My guess is somewhere between 3-10%.
whiteknight over trunk, or over the generational_gc branch?
chromatic Over trunk.
whiteknight okay
what is your branch's relationship to the generational_gc branch? 20:05
chromatic No relationship.
dukeleto Andy: i just replied to your email. let me know if you have other questions 20:07
whiteknight chromatic: ah, okay. So are the two going to have a combined beneficial effect when they're both merged?
chromatic I don't know. 20:08
dukeleto Andy: you can commit to parrot.git on github, so you don't need really need a fork, but you are free to have one
Andy: it will be more work for you to keep your fork in sync, than to just have branches in the main repo
NotFound Please don't put me first in metacommiters lists, I'm possibly the less experienced with git and github, 20:10
20:11 hudnix left
mikehh chromatic: if you didn't get my messages - gc_ms2_sf branch gives about 10% improvement in rakudo build times, only 2-3% in stresstest 20:12
whiteknight NotFound: metacommitting is just an interface on github. No knowledge of git required to add committers 20:13
chromatic What's the Rakudo stresstest? 20:14
mikehh spectest with long running tests 20:15
NotFound whiteknight: people usually assume that the first on a list is the appropiate to answer related (or unrelated) questions.
dukeleto NotFound: i can take you off the list if you would like. Do you want to not be a metacommitter on github? 20:18
20:18 tcurtis joined
dukeleto NotFound: I just copy and pasted allison's list of metacommitters 20:18
NotFound: did not mean to put you first
cotto_work hi tcurtis
dukeleto tcurtis: werd up
allison NotFound: and my list was sorted by Trac order, which puts all capital letters before lowercase letters, sorry about that 20:20
NotFound No problem, but if someone is making a wiki page or something please don't. 20:21
dukeleto NotFound: that is what you get for capitalizing ;)
NotFound: 10-4
20:22 hudnix joined 20:28 kid51 joined
mikehh chromatic: I ran the full rakudo stresstest -> spectest + 2 extra tests - integration/advent2009-day17.t and S03-operators/range-int.t 20:29
chromatic: on both amd64 and i386 - same results as trunk - a bit faster 20:30
dukeleto misses dalek 20:32
mikehh++
mikehh ain't it past time for #ps 20:33
dalek rrot: 0abb5b0 | leto++ | lib/Parrot/Revision.pm:
[lib] Make Parrot::Revision not break old HLLs
rrot: 0058ed8 | leto++ | t/steps/auto/revision-01.t:
Revert "Modify test for git migration"

This reverts commit 29ccd1d401d59c405cc59dc3947f27cf6b3bfec4.
rrot: 693bb4e | leto++ | / (2 files):
[t] TODO out some config tests relating to revisions
kid51 parrotsketch has begun
dukeleto yay, dalek is back! 20:34
sorear++
tcurtis Hi, dukeleto, cotto_work. Sorry about the slow reply. Building gcc in a vm appears to work my laptop quite hard. 20:36
dukeleto tcurtis: indeed. 20:37
kid51 tcurtis: I've found building GCC *anywhere* works a box hard ;-) 20:38
20:39 lucian joined 20:43 perlite_ joined 20:46 perlite left
rurban dukeleto++ for the git conversion 20:49
sorear people are having dalek issues?
dalek kudo: 4d33bae | masak++ | src/core/Cool-str.pm:
[Cool-str] fixed .trans case where @changes is []

This un-breaks one of the tests in trans.rakudo.
dukeleto rurban: thanks. 20:51
sorear: it seems to be back now.
whiteknight where does dalek live? 20:52
feather?
sorear appflux.net
diakopter's VPS
rurban dukeleto++ git is a massive improvements for working on branches and everything else. gitk forever 20:53
cotto_work dukeleto++
cotto_work wonders how long we'll keep referring to "trunk" 20:58
sorear forever, if we want
master is renamable
perl5 git has all development on the "blead" branch
whiteknight dukeleto: what's the problem with forks? 21:01
dukeleto master is master
whiteknight: for non-committers, they are fine. For core committers, they are redudant 21:02
whiteknight how are they any more redundant than a branch? 21:03
chromatic Seems like a matter of preference to me.
dukeleto chromatic: it is a matter of preference. The preference to do and create more work for others, or less. 21:04
forks require pull requests
forks will have divergent histories
forks will create more merge conflicts
do you want me to go no?
go on, even.
if someone does it correctly, a fork is basically an epsilon more of work. 21:05
allison parrotsketch time changed this week?
dukeleto If they don't do it correctly, it creates a bunch more work.
allison (I guess because of daylight saving time)
cotto_work allison: yes
whiteknight forks don't require pull requests. I cloned my fork, added parrot/parrot as a remote, and pushed to it directly
dukeleto Core committers should commit to parrot.git branches, but I am not going to demand it. It is "strongly recommended"
whiteknight: you are a core committer. yes, you know how to use them correctly. You know which remotes to push to. 21:06
allison rather than derail the meeting, this was my report, a prototype sandboxing vm: people.canonical.com/~allison/src/l...0.1.tar.gz
dukeleto I don't care if people fork, unless it starts to create more work for me.
whiteknight gotcha
chromatic Sandboxing how?
allison demonstrating the value of interpreter-level sandboxing to linux developers who only know about OS-level sandboxing 21:07
dukeleto allison: is there a repo for that? or only a tarball?
chromatic CAPs?
allison also, a permission model that Ubuntu wants to use
dukeleto: it's a local bzr repository, just a prototype
whiteknight is this the kid of idea that Parrot will want to adopt?
kid -> kind 21:08
allison whiteknight: I'm not sure
dukeleto allison: i am reading the README. Looks very cool.
allison whiteknight: the sandboxing is similar, and the permissions model would certainly have benefits for Parrot
whiteknight obviously we do need more security/sandboxing stuff. It's a matter of finding a good model to follow
allison whiteknight: as in, Parrot could be the avenue for implementing this for Ubuntu
chromatic And yet again I regret never typing my notes on Lorito and sandboxing.
allison chromatic: well, you and I talked about it, so I know what you were going to write 21:09
whiteknight allison: anything parrot can do to get more adopted by more people is something that we shoudl always seriously consider
chromatic The only tricky part is the graph walking if you have capabilities as composed op metadata. 21:10
whiteknight dreams about being to Ubuntu what Dalvik is to Android
chromatic ... but I suppose you could tag basic blocks in bytecode with their required caps....
allison whiteknight: this permissions model was interesting, because it a) didn't assume that permissions would be fixed at the install/config level, but instead assumes that users will be determining what they want an app to have access to
chromatic ... and if they need runtime checks.
allison whiteknight: and also, doesn't fall into the Android trap of asking for everything when you install the app, which is annoying, and not particularly useful 21:11
whiteknight: so, it's a dynamic, runtime, user-controlled permission model
whiteknight: seems quite... parroty
dukeleto allison: /me likes
whiteknight you had me at "dynamic" :)
allison remembers to set parrotsketch and sixperl back by an hour in calendar 21:14
whiteknight attending #ps makes me so happy 21:18
I've missed it these last 6 months
tcurtis: make sure you do a dry-run between now and then. I'll give it a try myself too 21:19
dukeleto whiteknight: good to see you there 21:20
allison: if you subscribe to the google calendar, it is in UTC
allison: should make your life easier
allison: it made my life easier
whiteknight just whitelisted the github commit emailer to parrot-commits
allison dukeleto: yeah, but it doesn't give me alerts, so not useful 21:21
dalek rrot: 9456558 | leto++ | lib/Parrot/Revision.pm:
[lib] Make Parrot::Revision lie correctly, darbelo++
dukeleto allison: you can set up google calendar to send you SMS's. That is what I have it do. 21:22
whiteknight I had it sending me emails for a while
dukeleto allison: you have to verify your phone # in the google calendar config first, then it will show up as a type of reminder
allison dukeleto/whiteknight: I do it all the time on events in my calendar, usually popup and email reminders, but not on external calendars 21:23
whiteknight #parrotsketch++ 21:26
still has that new-chatroom smell 21:27
21:27 darbelo left
whiteknight I think we can get gc_ms_2 and gsoc_nci merged this week. That's a good goal 21:29
dukeleto squeals with delight
cotto_work It's a good day. 21:30
21:30 kid51 left
whiteknight a good day indeed 21:30
but on that note, I have to leave. Later
21:30 whiteknight left
cotto_work sorear: is dalek getting user aliases from CREDITS on github now? 21:34
21:36 bluescreen left 21:37 perlite_ is now known as perlite 21:42 hudnix left
dukeleto allison: which language do you propose "larch policies" would be written in? 21:43
21:45 bacek left 21:46 hudnix joined
allison dukeleto: they aren't designed to be extensible modules (because that would be a security risk), so the policies would be baked into the core VM 21:46
dukeleto: they could ultimately be written as C modules, but they should never be HLL modules 21:47
(HLLs are contained within the sandbox, and so aren't allowed to modify the boundaries of the sandbox)
21:47 hudnix left
allison dukeleto: think of it as similar to a browser, where HLLs are javascript 21:48
dukeleto: you could be running any arbitrary code downloaded off the internet, and need to be exceedingly paranoid about what that code might do 21:49
dukeleto allison: sure. so you are saying that policies can only be set at compile-time ? 21:50
NotFound First push to parrot git repo done! Back in business :) 21:51
cotto_work NotFound: great 21:52
dalek rrot: a0f3af4 | NotFound++ | t/pmc/string.t:
test String set_string_keyed vtable
 21:53
dukeleto NotFound++
NotFound What about tapir2.ro.vutbr.cz/cover/cover-results/ ? Will it be updated from the git repo? 21:54
allison dukeleto: sort of. The list of available permission sets can only be set at compile time, but users can grant and remove individual permissions for individual apps at runtime 21:56
dukeleto: once an app is granted a permission by a user (like "has the ability to access my address book") that permission is always granted to that app, everytime it runs 21:57
dukeleto: kind of like Facebook, but less viral and permission hungry
dukeleto: (so, apps don't get to request permission to the kitchen sink)
chromatic Are permissions parametric? 21:58
allison chromatic: meaning, can they act on more information than the Application and the User? 21:59
chromatic: what did you have in mind?
sorear cotto_work: yes
cotto_work nice work. sorear++ 22:00
sorear What are larch policies?
chromatic can_read_files[in_home_directory]
That's more specific than can_read_files.
allison chromatic: I would see those as two different permissions, but in more of a hierarchy
sorear can_read[$HOME/.ssh/id_rsa] subsumes quite a lot of other permissions
allison chromatic: another example might be the ability to read from the address book, but only for a specific user or set of users 22:01
chromatic You can view sandboxed op dispatch as a degenerate form of multiple dispatch with all the concomitant narrowifications.
allison chromatic: at which point, you might be looking for email address parameter 22:02
chromatic: that's what's interesting about this sandboxing model, is there is no selective disabling of certain ops 22:03
chromatic Right.
22:03 tcurtis left
allison chromatic: instead there's "ops that don't require advanced permissions" and "ops that do require advanced permissions" 22:03
sorear What are you talking about? Is it on the wiki?
allison chromatic: and then checking against a permissions et
sorear: sorry, it's a prototype I wrote this week
chromatic I drew the line between "ops with permissions checkable at runtime" and "ops with permissions checkable at compile/load/verification time". 22:04
allison sorear: people.canonical.com/~allison/src/l...0.1.tar.gz
chromatic: the first are useful for embedding in postgres, etc, but not so much for the application sandbox 22:05
chromatic: really, not at all for the application sandbox, because it has to be a "one-size-fits-all" install from a single standard package
chromatic: like a browser
sorear is larch connected to lorito? 22:06
chromatic I might be comfortable letting Google Gears for GMail write to a specific directory, but I'm not letting Facebook write to my local hard drive.
allison sorear: not particularly, but exploring other ideas we've had for Parrot
chromatic: exactly, your choices will be for a specific app 22:07
chromatic: it has to ask for your permission before it can do anything 22:08
22:08 nwellnhof joined
allison (anything outside of the sandbox, that is) 22:08
dalek rrot: 7161db0 | nwellnhof++ | compilers/opsc/src/Ops/Emitter.pm:
[opsc] Don't output $Id$
 22:14
22:15 hudnix joined
cotto_work nwellnhof++ for noticing that. My brain refused to do much more work after reviewing the commits that removed the bulk of those. 22:16
nwellnhof cotto: I noticed it when I ran ops2c while testing the remove_charset_ops branch. 22:19
I'm about to merge that branch now, btw 22:20
cotto_work merge away
nwellnhof I'm still running some tests...
It's basically only one commit, but I'm doing a branch merge to see how it works with git. 22:22
dukeleto nwellnhof++
cotto_work Our karma system favors such a workflow.
22:29 masonkramer_ joined, masonkramer__ left 22:31 hudnix left 22:33 hudnix joined
dukeleto This may be relevant to larch policies: download.oracle.com/javase/1.4.2/do...sions.html 22:34
at least to compare and contrast.
dalek rrot: a9e814c | nwellnhof++ | / (8 files):
Merge remote branch 'origin/remove_charset_ops'
 22:35
dukeleto nwellnhof++ 22:36
nwellnhof dukeleto++ # for all your git work
dukeleto nwellnhof: thanks! 22:37
nwellnhof: make sure to delete your merged branch
nwellnhof: git push origin :branch_name_to_delete
nwellnhof: don't forget the ":" :)
nwellnhof dukeleto: Yes, I've done this before. It's a bit weird. 22:38
There are also some of my other old branches that are still around on github. I'll delete them too. 22:39
dukeleto nwellnhof: thanks for cleaning up :)
nwellnhof: that syntax comes from the general form of : git push remote local_branch:remote_branch 22:40
nwellnhof: when you don't give a local_branch, it means "delete the remote branch" instead of pushing local_branch to remote_branch
nwellnhof: that is how it makes sense to me
nwellnhof: i think this is described in the git_workflow.pod 22:42
nwellnhof OK, 34 branches left.
dukeleto Some of our branches don't seem to ever have a change of merging.
Some haven't been touched in over a year. 22:43
22:45 PerlJam left 22:50 whiteknight joined
bacek_at_work tilda tilda 22:54
22:56 M_o_C left
cotto_work ~~ 22:58
23:07 bluescreen joined
dukeleto bacek_at_work: i expect your productivity to go up 100000000% now that you don't have to use git-svn. I fear the number of commits that will flow through your fingers. 23:12
plobsing oh noes. it's the bacek singularity
dukeleto 1/bacek = OH MY GOD, IT'S MADE OF COMMITS 23:14
whiteknight is anybody seeing emails from github on parrot-commits? 23:16
chromatic yes 23:17
whiteknight okay 23:18
chromatic They don't include diffs, but it's a start.
whiteknight I haven't seen any yet. I don't think I get sent a copy of emails I moderate 23:21
23:32 Kulag left, Kulag joined 23:45 plobsing left 23:46 plobsing joined
dalek TT #1778 closed by nwellnhof++: Charset opcodes are deprecated 23:49
TT #1778: trac.parrot.org/parrot/ticket/1778
rakudo: e6486e5 | masak++ | src/core/Any-list.pm: 23:59
rakudo: [Any-list] added &elems sub

Powered by Cro and the Raku® Programming Language.

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

Thank you!