couchdb-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From ASF IRC Bot <asf...@urd.zones.apache.org>
Subject Summary of IRC Meeting in #couchdb-meeting
Date Wed, 14 May 2014 19:52:44 GMT
Summary of IRC Meeting in #couchdb-meeting at Wed May 14 19:01:33 2014:

Attendees: jan____, rnewson, awenkhh, tilgovi, Kxepal

- Preface
- bigcouch merge
  - Action: everyone to checkout the 1843-feature-bigcouch branch of couchdb.git and help
with testing. report all errors you found on dev@ ML or #couchdb-dev
  - Info: to run the bigcouch from merge branch: clone couchdb.git, checkout 1843-feature-bigcouch,
./configure && make && dev/run
- 1.6.0 release
  - Link: http://www.mail-archive.com/dev@couchdb.apache.org/msg32610.html
  - Link: https://gist.github.com/chewbranca/07d9a6eed3da7b490b47
- couchdb meetings
  - Link: https://gist.github.com/andywenk/3168d403ea62f204c383
  - Link: https://plus.google.com/u/0/communities/103656875572948787245


IRC log follows:

## Preface ##
## bigcouch merge ##
[Wed May 14 19:02:08 2014] <rnewson>: so I made some significant progress on the merge
this week (and last).
[Wed May 14 19:02:28 2014] <awenkhh>: YAY :)
[Wed May 14 19:02:31 2014] <rnewson>: the contribution from cloudant is now merged with
couchdb as it was in March of this year.
[Wed May 14 19:02:37 2014] <rnewson>: with the exception of "ibrowse".
[Wed May 14 19:02:55 2014] <rnewson>: I will complete that tomorrow, now that I (finally!)
discovered that it was the source of my couch_replicator bug.
[Wed May 14 19:03:09 2014] <rnewson>: I have spent days tracking it down, the whole
time looking in the wrong place.
[Wed May 14 19:03:39 2014] <rnewson>: after ibrowse is fixed tomorrow, I will then update
all repos with couchdb updates since March.
[Wed May 14 19:04:07 2014] <rnewson>: any subproject with a 'master' branch with the
exception of 'ibrowse' should now be considered 'merged' as of March.
[Wed May 14 19:04:54 2014] <Kxepal>: hooray! that's the great news(:
[Wed May 14 19:04:55 2014] <rnewson>: since I cannot rewrite the master branch of any
repo hosted by the ASF for policy reasons, there is some ugly "history" in some repos which
will make review a bit harder.
[Wed May 14 19:05:11 2014] <rnewson>: hopefully no one is going to peer too deeply there,
and I can show the true diffs if needed. 
[Wed May 14 19:05:24 2014] <rnewson>: essentially it's a few commits that should be
one.
[Wed May 14 19:05:44 2014] <awenkhh>: really great 
[Wed May 14 19:05:49 2014] <rnewson>: There's another topic, though, the pre-receive
hook prohibition, while well-intentioned, is a hindrance.
[Wed May 14 19:06:33 2014] <rnewson>: I'm sure it comes from a desire to prevent the
kind of deletion I'm attempting, but it doesn't account for the nuance of this situation,
we're I'm building a new history. No one but me should have seen the mistaken middle step.
[Wed May 14 19:06:46 2014] <rnewson>: there's also a veiled threat to extend the hook's
prohibition to *all* branches
[Wed May 14 19:06:53 2014] <rnewson>: if that happens, we might as well use Subversion.
[Wed May 14 19:07:02 2014] <rnewson>: back to the merge topic.
[Wed May 14 19:07:15 2014] <rnewson>: I encourage everyone to checkout the 1843-feature-bigcouch
branch of couchdb.git
[Wed May 14 19:07:37 2014] <rnewson>: this is now a shell, containing no code, but just
the scripts needed to fetch and build couchdb.
[Wed May 14 19:07:50 2014] <rnewson>: any errors found with that should be reported.
[Wed May 14 19:08:05 2014] <awenkhh>: is there a preferred ticket for reports?
[Wed May 14 19:08:06 2014] <rnewson>: the master branches of each subproject will be
brought up to current.
[Wed May 14 19:08:18 2014] <rnewson>: awenkhh: good question.
[Wed May 14 19:08:30 2014] <rnewson>: I suggest we keep things light and use #couchdb-dev
[Wed May 14 19:08:45 2014] <rnewson>: the resultant commits will hit the mailing lists,
so we're not off-the-record
[Wed May 14 19:09:07 2014] <awenkhh>: fine with me - I also thought of you creating
a Jira with some basic info and we report there but dev@ is fine 
[Wed May 14 19:09:19 2014] <rnewson>: we need to discuss the rcouch merge, and then
when active development shifts to the multi-repository model.
[Wed May 14 19:09:24 2014] <Kxepal>: #action everyone to checkout the 1843-feature-bigcouch
branch of couchdb.git and help with testing. report all errors you found on dev@ ML or #couchdb-dev
[Wed May 14 19:09:36 2014] <rnewson>: #couchdb-dev is the IRC channel :P
[Wed May 14 19:09:46 2014] <awenkhh>: so do I understand correctly that after finishing
the merge, the next release will be 2.0 and not 1.7?
[Wed May 14 19:09:55 2014] <rnewson>: I am always connected to IRC via irccloud
[Wed May 14 19:10:05 2014] <Kxepal>: oh..I hope everyone will guess that there is no
#couchdb-dev email address (;
[Wed May 14 19:10:18 2014] <rnewson>: awenkhh: that's a different topic.
[Wed May 14 19:10:24 2014] <awenkhh>: ok ... 
[Wed May 14 19:10:26 2014] <rnewson>: the version that includes clustering will certainly
bump the major.
[Wed May 14 19:10:44 2014] <rnewson>: whether a version later than 1.6.0 but earlier
that bigcouch is released is another topic.
[Wed May 14 19:10:49 2014] <rnewson>: probably not, but who knows?
[Wed May 14 19:10:55 2014] <rnewson>: certainly I don't preclude it.
[Wed May 14 19:10:59 2014] <Kxepal>: rnewson: as about the merge. what's the next steps?
testing is the one of, I feel the second one is updating the docs, right?
[Wed May 14 19:11:04 2014] <awenkhh>: I should have asked it in this way ...
[Wed May 14 19:11:07 2014] <awenkhh>: thx 
[Wed May 14 19:11:22 2014] <awenkhh>: Kxepal ah good point
[Wed May 14 19:11:27 2014] <Kxepal>: or it's all depends from rcouch merge too?
[Wed May 14 19:11:33 2014] <rnewson>: Kxepal: testing (from everyone other than me),
merging updates since March and fixing ibrowse (from everyone that is me)
[Wed May 14 19:12:02 2014] <rnewson>: we do not have to wait for that, but we said in
couchdb-1999 that we would.
[Wed May 14 19:12:32 2014] <awenkhh>: I think we should start with the docs early ...
I guess it will be a lot of work 
[Wed May 14 19:12:50 2014] <awenkhh>: dumb question: can we "lend" some docs from cloudant
?
[Wed May 14 19:12:53 2014] <rnewson>: I don't know what bandwidth/time benoitc has to
merge the pending 1994-merge-rcouch branches.
[Wed May 14 19:13:24 2014] <rnewson>: yes, cloudant docs should flow back to couchdb,
it's only fair.
[Wed May 14 19:13:31 2014] <Kxepal>: awenkhh: we can start, but if there will be also
rcouch merge and some bits will be changed (and I feel the will) I think how to avoid doing
the same job twice
[Wed May 14 19:13:45 2014] <awenkhh>: rnewson cool!
[Wed May 14 19:13:55 2014] <rnewson>: I don't believe the release with bigcouch is blocked
by the rcouch merge, I don't mind if that happens in a release afterward.
[Wed May 14 19:14:00 2014] <rnewson>: if they happen together, fine.
[Wed May 14 19:14:05 2014] <awenkhh>: Kxepal yes true but don't you think it's still
good to start at all 
[Wed May 14 19:14:13 2014] <rnewson>: but testing what I've made today is important
regardless.
[Wed May 14 19:14:15 2014] <Kxepal>: sure, it is
[Wed May 14 19:14:24 2014] <awenkhh>: yes prio 1
[Wed May 14 19:14:26 2014] <rnewson>: Cloudant folks know how bigcouch *should* behave.
[Wed May 14 19:14:34 2014] <Kxepal>: rnewson: agree. testing is the first checkpoint
[Wed May 14 19:14:38 2014] <rnewson>: so testing what I made today is easier in that
context.
[Wed May 14 19:14:48 2014] <rnewson>: where the release points occur is a detail
[Wed May 14 19:14:56 2014] <awenkhh>: as stated today I will start after the meeting
[Wed May 14 19:15:06 2014] <rnewson>: and, with semver, the number we give is derivable
directly from the diff between versions
[Wed May 14 19:15:25 2014] <rnewson>: adding rcouch is not a compatibility break, but
adds features, so it would, on its own, be a minor version bump.
[Wed May 14 19:15:37 2014] <rnewson>: for example, bigcouch release is 2.0.0, bigcouch+rcouch
is 2.1.0
[Wed May 14 19:15:46 2014] <awenkhh>: sounds feasible
[Wed May 14 19:15:48 2014] <rnewson>: that feels the likeliest outcome, but it's not
the only one.
[Wed May 14 19:16:27 2014] <rnewson>: I will note that there's a list of JIRA tickets
for post-merge work that also need to happen before the bigcouch release.
[Wed May 14 19:16:43 2014] <rnewson>: and those will take time, enough time for rcouch
to land if benoitc has the time and inclination.
[Wed May 14 19:17:00 2014] <rnewson>: I will need a few days away from hardcore merging,
but after that I think I will be able to help him.
[Wed May 14 19:17:35 2014] <rnewson>: I will probably have a look anyway, it's possible
I can merge it myself. I do understand some erlang, after all.
[Wed May 14 19:17:48 2014] <awenkhh>: ah really ? :)
[Wed May 14 19:17:56 2014] <rnewson>: my focus is finishing bigcouch and responding
to the horrible bugs you are bound to find.
[Wed May 14 19:18:22 2014] <rnewson>: I should note that cloudant did not stand still
since the 1843-feature-bigcouch branches were made last year.
[Wed May 14 19:18:34 2014] <rnewson>: so there's a whole ton of fixes to come from there
too.
[Wed May 14 19:18:44 2014] <rnewson>: I have *not* taken those yet.
[Wed May 14 19:19:07 2014] <rnewson>: Cloudant/IBM also intend to move their day-to-day
development to the asf git repositories.
[Wed May 14 19:19:23 2014] <awenkhh>: oh wow ... 
[Wed May 14 19:19:24 2014] <rnewson>: that is happening in parallel to this effort (different
branch, presumably, I'm not directly involved).
[Wed May 14 19:19:37 2014] <rnewson>: so that work will appear in our repos soon enough.
[Wed May 14 19:19:46 2014] <rnewson>: and we'll merge those into master too.
[Wed May 14 19:20:01 2014] <awenkhh>: taking the fixes is after the so called BigCOuch
merge right?
[Wed May 14 19:20:15 2014] <rnewson>: the model for Cloudant's development of couchdb
will follow the successful model we've taken with Fauxton
[Wed May 14 19:21:07 2014] <rnewson>: chronology: 1) testing (everyone) 2) fix ibrowse
(me) 3) import latest couchdb master (me) 4) import latest cloudant work (me, probably)
[Wed May 14 19:21:19 2014] <rnewson>: rcouch can happen at any point during that I think
[Wed May 14 19:21:27 2014] <awenkhh>: ok cool !
[Wed May 14 19:21:51 2014] <Kxepal>: sounds like a plan (: great news, great notes,
rnewson!
[Wed May 14 19:21:56 2014] <rnewson>: I would like to see you all running this and finding
brokenness this week, if anyone has cycles.
[Wed May 14 19:22:13 2014] <Kxepal>: I hope benoitc will join to merge party soon
[Wed May 14 19:22:14 2014] <rnewson>: the sooner I get reports, the sooner I can fix
while the merge (and the conflicts!!) are fresh in my mind.
[Wed May 14 19:22:38 2014] <rnewson>: the best plan for benoitc right now would be to
make a branch cloned from origin/master and merge 1994-merge-rcouch into it.
[Wed May 14 19:22:58 2014] <rnewson>: if those turn out easy/sane/etc, we can then do
the same but to 'real' maser.
[Wed May 14 19:23:01 2014] <rnewson>: master
[Wed May 14 19:23:09 2014] <rnewson>: I suggest this so as not to confuse the testing
effort.
[Wed May 14 19:23:22 2014] <rnewson>: but I also don't want to hold benoitc up unnecessarily.
[Wed May 14 19:23:42 2014] <tilgovi>: rnewson: running 1843-feature-bigcouch and finding
brokenness?
[Wed May 14 19:23:44 2014] <rnewson>: the merge effort can be done now, I'd just rather
it not hit the branch called 'master' until we've lived with the merge I've done for a little
bit.
[Wed May 14 19:23:49 2014] <rnewson>: tilgovi: yes pls
[Wed May 14 19:23:54 2014] <tilgovi>: will do
[Wed May 14 19:24:08 2014] <rnewson>: couchdb.git, checkout 1843-feature-bigcouch, ./configure
&& make && dev/run
[Wed May 14 19:24:12 2014] <rnewson>: and you will have a 3 node cluster
[Wed May 14 19:24:27 2014] <rnewson>: ports 15984, 25984, 35984 for each nodes clustered
interface.
[Wed May 14 19:24:37 2014] <rnewson>: ports 15986, 25986, 35986 for each nodes local
interface.
[Wed May 14 19:24:44 2014] <rnewson>: each node *is* a couchdb instance.
[Wed May 14 19:24:56 2014] <rnewson>: everything should work on both interfaces with
some obvious caveats.
[Wed May 14 19:25:14 2014] <rnewson>: I will note that cloudant never implemented vhosts
in the clustered interface, since we use haproxy for that.
[Wed May 14 19:25:28 2014] <rnewson>: a decision has to be made about whether we do
so or not. 
[Wed May 14 19:26:06 2014] <rnewson>: oh, and clustered compaction is not a thing, you'll
get a 400 message.
[Wed May 14 19:26:09 2014] <Kxepal>: #info to run the bigcouch from merge branch: clone
couchdb.git, checkout 1843-feature-bigcouch, ./configure && make && dev/run
[Wed May 14 19:26:17 2014] <rnewson>: again, because Cloudant automate this for their
users.
[Wed May 14 19:26:29 2014] <rnewson>: or, if you like, we don't let you pound our disks
whenever you feel like it ;)
[Wed May 14 19:26:57 2014] <tilgovi>: heh
[Wed May 14 19:27:04 2014] <rnewson>: compaction is initiated at each node, on the ***6
interface, by shard.
[Wed May 14 19:27:05 2014] <tilgovi>: I assume I can run compaction against the nodes
individually?
[Wed May 14 19:27:09 2014] <tilgovi>: yep
[Wed May 14 19:27:10 2014] <rnewson>: yup.
[Wed May 14 19:27:37 2014] <rnewson>: separate to the merge, but maybe not the release,
we should finally add a background priority for compaction and automate this for everyone
forever.
[Wed May 14 19:27:55 2014] <rnewson>: and ditch that awkward-to-configure compaction
daemon.
[Wed May 14 19:28:06 2014] <rnewson>: when compaction can run without impact, the scheduler
becomes much simpler.
[Wed May 14 19:28:45 2014] <rnewson>: That's everything I can think of, does anyone
have questions for me?
[Wed May 14 19:28:58 2014] <jan____>: great report rnewson, thank you very much!
[Wed May 14 19:29:10 2014] <awenkhh>: yes awesome thank you!
[Wed May 14 19:29:12 2014] <rnewson>: you're very welcome, now test my stuff.
[Wed May 14 19:29:35 2014] <awenkhh>: ACTION cloning fresh couchdb.git
[Wed May 14 19:29:57 2014] <Kxepal>: rnewson: we'll do it(: thanks a lot for the report!
[Wed May 14 19:30:09 2014] <jan____>: rnewson: checkout out as we speak
[Wed May 14 19:30:18 2014] <rnewson>: :D
[Wed May 14 19:30:35 2014] <rnewson>: ACTION hopes he didn't forget a push....
[Wed May 14 19:30:53 2014] <Kxepal>: ((:
## 1.6.0 release ##
[Wed May 14 19:32:42 2014] <Kxepal>: ok, 1.6.0-rc5 is still open for testing and voting.
please everyone who didn't test and +1 for it - you have your last chance to do that (;
[Wed May 14 19:32:44 2014] <jan____>: ship it
[Wed May 14 19:33:00 2014] <awenkhh>: +10^2
[Wed May 14 19:33:24 2014] <Kxepal>: by the same time Nick provided windows binaries
against Erlang 17 - works like a charm!
[Wed May 14 19:33:43 2014] <Kxepal>: if there is any windows users - please feel free
to test them (:
[Wed May 14 19:34:24 2014] <Kxepal>: #link http://www.mail-archive.com/dev@couchdb.apache.org/msg32610.html
[Wed May 14 19:34:44 2014] <Kxepal>: and yes...SHIP IT (:
[Wed May 14 19:35:09 2014] <rnewson>: cool
[Wed May 14 19:35:30 2014] <rnewson>: I'll note the issue with 17.0 is not merely compiling
and running, it's scheduler stuff for BIF's and NIF's
[Wed May 14 19:35:37 2014] <rnewson>: I'm certain that we have not solved all of those.
[Wed May 14 19:35:57 2014] <rnewson>: chewbranca outlined what it will take to use 17.0
in production.
[Wed May 14 19:37:36 2014] <Kxepal>: #link https://gist.github.com/chewbranca/07d9a6eed3da7b490b47
[Wed May 14 19:38:00 2014] <Kxepal>: ^^^ about CouchDB and Erlang releases by chewbranca
[Wed May 14 19:38:56 2014] <Kxepal>: so, moving on...
## couchdb meetings ##
[Wed May 14 19:40:00 2014] <awenkhh>: yay
[Wed May 14 19:40:04 2014] <Kxepal>: recently there were two couchdb meeting. so how
it was, nslater, awenkhh?
[Wed May 14 19:40:22 2014] <awenkhh>: I start then 
[Wed May 14 19:40:37 2014] <awenkhh>: yea yesterday we had the first CouchDB Meetup
Hamburg 
[Wed May 14 19:40:45 2014] <awenkhh>: I wrote a recap here
[Wed May 14 19:41:01 2014] <awenkhh>: #link https://gist.github.com/andywenk/3168d403ea62f204c383
[Wed May 14 19:41:20 2014] <awenkhh>: and I created a G+ community where already 6 members
are :) 
[Wed May 14 19:41:34 2014] <awenkhh>: #link https://plus.google.com/u/0/communities/103656875572948787245
[Wed May 14 19:42:12 2014] <awenkhh>: we are now thinkg of putting together the Hamburg
Beamers with the CouchDB Meetup - both brand new groups 
[Wed May 14 19:42:29 2014] <awenkhh>: was a lot of fun and yeah ... looking forward
for the next one 
[Wed May 14 19:42:40 2014] <jan____>: sounds great!
[Wed May 14 19:42:45 2014] <awenkhh>: I would like to post my recap on the CouchDB blog
[Wed May 14 19:43:44 2014] <jan____>: +1
[Wed May 14 19:44:03 2014] <Kxepal>: +1 that's good idea
[Wed May 14 19:44:43 2014] <awenkhh>: ok - before or after the weekly news post tomorrow?
[Wed May 14 19:45:31 2014] <Kxepal>: I feel when you think it'll be ready to
[Wed May 14 19:45:47 2014] <awenkhh>: hm it is written so I can post it any time ...

[Wed May 14 19:45:58 2014] <jan____>: awenkhh: whenever is fine
[Wed May 14 19:46:12 2014] <awenkhh>: ok ... that's it from my side ... 
[Wed May 14 19:46:50 2014] <Kxepal>: jan____: have you been on CouchDB Hack Night 8th?
[Wed May 14 19:47:40 2014] <jan____>: have not, but an oral report from nslater said,
lots of talking an good discussions :)
[Wed May 14 19:49:38 2014] <Kxepal>: oh, it's a pity..hope we'll read the post from
him about too (:
[Wed May 14 19:50:53 2014] <Kxepal>: I think that's all for today unless anyone has
any other topic to discuss
[Wed May 14 19:51:17 2014] <awenkhh>: fine - I have no additions to make 
[Wed May 14 19:52:31 2014] <Kxepal>: good! thank you awenkhh for users meeting and news
about (:
[Wed May 14 19:52:35 2014] <Kxepal>: ASFBot: meeting end


Meeting ended at Wed May 14 19:52:35 2014

Mime
  • Unnamed multipart/mixed (inline, None, 0 bytes)
View raw message