couchdb-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF IRC Services" <asf...@wilderness.apache.org>
Subject Summary of IRC meeting in #couchdb-meeting, Wed Feb 27 20:00:55 2013
Date Wed, 27 Feb 2013 20:58:12 GMT
Members present: eckoit, davisp, nslater, Wohali, vmx, benoitc, drsm79, garren, rnewson, wendall911,
chewbranca, mikewallace

----------------
Meeting summary:
----------------

1. Preface

2. fauxton / ip clearence
  a. fauxton ip clearance is done, and code on fauxton branch in asf git repo (nslater, 2)
  b. drsm79 wants to work on tests before merging in with master (nslater, 2)
  c. fauxton will live beside futon until we have feature parity (or there abouts) (nslater,
2)

3. bigcouch status
  a. bigcouch merge to start in next few weeks (rnewson and davisp) and will take a few weeks
to finish (nslater, 3)

4. rcouch status

5. autotools
  a. nslater still wants to ditch autotools and move to riak model of packaging (nslater,
5)
  b. rcouch merge timeline doesn't block bigcouch merge timeline (nslater, 5)
  c. get benoitc and davisp/rnewson talking about rcouch + bigcouch merge timelines (nslater,
5)

6. release
  a. AWS based build system for releases was shut down by infra (nslater, 6)
  b. continued problems with test suite showing up in travis - details to be posted later
(nslater, 6)
  c. NEWS and CHANGES being brought up to speed, then release will proceed (nslater, 6)

7. rcouch
  a. benoit hopes to have rcouch build work on asf branch very soon (pending ip clearence)
(nslater, 7)
  b. nslater to help benoit with ip clearence (nslater, 7)
  c. rebar will allow us to change source releases to tgz copies of git and will allow our
binary packages to bundle deps (nslater, 7)


--------
Actions:
--------
- get benoitc and davisp/rnewson talking about rcouch + bigcouch merge timelines (nslater,
20:23:51)
- nslater to help benoit with ip clearence (nslater, 20:40:03)

IRC log follows:


# 1. Preface #
20:01:19 [nslater]: is anybody there? can we have a show of hands please?
20:02:40 [chewbranca]: o/
20:02:40 [eckoit]: here
20:02:47 [mikewallace]: hello
20:03:27 [davisp]: present
20:03:32 [drsm79]: ola
20:03:40 [nslater]: any topics to add?
20:04:10 [nslater]: i will just press on and topics can be added as we go


# 2. fauxton / ip clearence #
20:04:25 [nslater]: chewbranca: would you mind providing a status update?
20:04:32 [nslater]: or drsm79...
20:04:40 [chewbranca]: nslater: Fauxton updates are in the ASF repo
20:04:48 [chewbranca]: as far as I know all ip clearance is done
20:04:55 [nslater]: on a topic branch i presume? (totally not been following this at all)
20:04:55 [chewbranca]: I pushed up the latest updates over the weekend
20:05:02 [chewbranca]: yeah on the fauxton branch
20:05:02 [nslater]: sweet
20:05:10 [nslater]: what's the plan for merge with master?
20:05:27 [drsm79]: I think it's targetted for 1.4
20:05:27 [chewbranca]: no set in stone plans, but I believe the idea was to have a stand alone
preview in 1.4
20:05:42 [chewbranca]: right now we deploy fauxton statically to /_utils/fauxton/index.html
20:05:50 [chewbranca]: so it can live beside futon without issue
20:05:50 [nslater]: well, let's forget about release numbers. i was more thinking about a
date for merging in with master
20:06:05 [drsm79]: I'd like to get a test suite done for it before we do
20:06:15 [drsm79]: which is the next big thing on the list
20:06:20 [nslater]: ah, cool
20:06:27 [drsm79]: there was talk of fauxton being a separate thing in Berlin
20:06:28 [nslater]: why would it live beside futon? would we not replace one with t'other?
20:06:35 [garren]: sorry I'm late.
20:07:05 [drsm79]: nslater: it's not got feature parity currently
20:07:15 [nslater]: ah okay. cool. thanks guys
20:07:20 [drsm79]: I think it's valid to have both for a little while
20:07:28 [rnewson]: hello
20:07:35 [chewbranca]: I'm still working on the view editor, and we don't have things like
replication support yet, etc etc
20:07:42 [nslater]: #info fauxton ip clearance is done, and code on fauxton branch in asf
git repo
20:07:42 [chewbranca]: its coming along but not a full replacement yet
20:07:57 [nslater]: #info drsm79 wants to work on tests before merging in with master
20:08:14 [eckoit]: drsm79: a separate thing, but only in berlin? :)
20:08:20 [drsm79]: ;)
20:08:21 [nslater]: #info fauxton will live beside futon until we have feature parity (or
there abouts)
20:08:35 [nslater]: let's move on!
20:08:42 [drsm79]: ok


# 3. bigcouch status #
20:08:52 [nslater]: over to you rnewson 
20:09:05 [nslater]: (or davisp)
20:09:12 [rnewson]: I'm just finishing up some tasks and then davisp and I are getting together
to do the merge.
20:09:13 [davisp]: drsm79: test suite for fauxton?
20:09:27 [davisp]: hooray mergey mergey
20:09:35 [nslater]: rnewson: oooooh very exciting. when do you expect to start, or perhaps,
when do you expect to finish?
20:09:37 [chewbranca]: davisp: testing fauxton functionality itself, not CouchDB
20:09:50 [rnewson]: well, we have travel to figure out, but I hope in the next few weeks
20:09:50 [davisp]: nslater: Sometime before April 8 me and rnewson are getting together to
spend a solid week on the merge
20:09:53 [rnewson]: and it'll take a few weeks
20:10:12 [rnewson]: a solid week of together time for the hardest bits, and then followup

20:10:12 [davisp]: nslater: goals are basically to get the code merged and into a branch read
for ASF without changing single node operation
20:10:27 [nslater]: #info bigcouch merge to start in next few weeks (rnewson and davisp) and
will take a few weeks to finish
20:10:34 [rnewson]: yes.
20:10:34 [eckoit]: chewbranca: would be nice in travis to push the couchapp fauxton to a location
people can replicate from
20:10:42 [nslater]: davisp: okay cool
20:11:09 [nslater]: i think we can move on, unless anyone has any questions?
20:11:09 [chewbranca]: eckoit: yeah that would be slick


# 4. rcouch status #
20:11:38 [nslater]: we sort of need benoitc here for this
20:11:45 [davisp]: I'm still wondering what midwest adventure I should force rnewson to endure
20:12:00 [nslater]: the more time that passes the more i want to throw autotools out of the
window
20:12:01 [Wohali]: butter sculpture
20:12:07 [nslater]: i would merge in rcouch today if i could
20:12:15 [nslater]: i dont think i understand what the steps are between here and there
20:12:30 [nslater]: rnewson / davisp - how does that work intersect with the bigcouch stuff?
20:12:30 [rnewson]: davisp: the trick will be building a special portable sensory deprivation
system for the journey to/from the airport.
20:12:45 [rnewson]: nslater: dunno
20:13:00 [nslater]: does bigcouch use rebar?
20:13:01 [davisp]: rnewson: Not overly familiar with rcouch
20:13:09 [davisp]: nslater: it do but I was planning on using autotools
20:13:15 [davisp]: cause you all the things
20:13:15 [Wohali]: nslater: rebar and sconscript
20:13:25 [nslater]: davisp: well... we want to move to rebar and drop autotools completely
20:13:37 [davisp]: nslater: we do?
20:13:52 [davisp]: no more autotools?
20:13:52 [nslater]: yep. by "we" i mean me, and jan and benoit - and i think you and other
people too?
20:14:00 [nslater]: yeah, **** dat ****
20:14:01 [rnewson]: davisp: nslater was recently replaced by a pod person so we can ditch
autotools.
20:14:08 [davisp]: nslater: I was under the impression autotools was non-negotiable
20:14:22 [Wohali]: ACTION is surprised as well
20:14:25 [nslater]: ah, i think you must've missed the thread where i threw my toys out of
the pram and decided i wanted to ditch it
20:14:37 [davisp]: nslater: Guess so
20:14:37 [nslater]: i  think autotools is hindering uptake of couchdb - both users and devs
20:14:45 [davisp]: nslater: for building Erlang code itself, rebar is fine
20:14:46 [wendall911]: +1 ditch autotools, makes for dependency hell 


# 5. autotools #
20:15:01 [davisp]: nslater: but that doesn't address the places we actually use it
20:15:01 [wendall911]: currently can't build on el5 without serious patching
20:15:15 [nslater]: davisp: yeah, i basically convinced myself that it's not necessary
20:15:16 [Wohali]: So right now, bigcouch does the couchjs build using sconscript
20:15:22 [nslater]: davisp: i looked at how riak and whatever do things
20:15:30 [davisp]: nslater: did you have any thoughts for things like docs, installation,
etc etc
20:15:30 [davisp]: Oh?
20:15:30 [Wohali]: I recently did some work to patch that up for js185 on various OSes
20:15:37 [nslater]: davisp: yeah.. my thoughts are, punt on it
20:16:00 [nslater]: i.e. when you download the source, it is set up to run out of the local
dir you unpacked it in
20:16:07 [Wohali]: is dch around to toss in a windows build chain comment?
20:16:07 [nslater]: just like riak and mongo and every other damn thing
20:16:15 [davisp]: nslater: Oh?
20:16:18 [nslater]: and then if people want to put it on to the operating system - they do
it manually
20:16:30 [nslater]: or, as will actually be the case, install a package
20:16:37 [nslater]: davisp: yeah - take a look at how other people do it
20:16:52 [davisp]: nslater: I have. You convinced me a long time ago it was a dumb way of
doing things :P
20:17:15 [davisp]: nslater: I'm ok with that approach
20:17:16 [nslater]: yeah, well i have been convinced that what we do actually makes it harder
to develop couchdb, harder to install, and offers practically zero benefit
20:17:30 [nslater]: i mean, in theory, it would seem like the right choice - but i dont think
the pain is worth it any more
20:17:45 [davisp]: nslater: I would say the biggest downfall is the steep learning curve to
autotools
20:18:00 [nslater]: there is that too. it's also brittle and overly-complex
20:18:07 [nslater]: okay, so... moving on from this revelation...
20:18:09 [davisp]: cool, so if you're convinced I'm good to go on that
20:18:15 [nslater]: how does this news impact your ideas for bigcouch?
20:18:23 [davisp]: Makes one of the easy parts easier
20:18:30 [nslater]: do we do the rcouch merge first? after? not important?
20:18:52 [nslater]: #info nslater still wants to ditch autotools and move to riak model of
packaging
20:18:53 [davisp]: The only issue I see is building ocuchjs
20:19:07 [davisp]: other than that it should be straight forward
20:19:15 [nslater]: this might go away if/when we switch to V8? or is it the compilation of
C in general?
20:19:15 [davisp]: as to rcouch, I have no idea on what code that contains
20:19:31 [nslater]: no me either. in my head "rcouch" == "benoit's rebarification of couchdb"
20:19:33 [davisp]: nslater: compilation of anything not Erlang C
20:19:37 [nslater]: i am just operating at a conceptual level here
20:19:48 [nslater]: well, riak manage it
20:19:53 [nslater]: so it's gotta be possible :)
20:20:00 [davisp]: nslater: rebar will handle building our NIFs and drivers, but a standalone
executable with links and what not is a bit wonky
20:20:07 [davisp]: nslater: They handle what?
20:20:15 [davisp]: Or where?
20:20:18 [nslater]: do they not handle v8 stuff? maybe not...
20:20:38 [nslater]: oh, i guess if they don't need a wrapper there's no need. hmm...
20:20:45 [davisp]: Last I saw they were still using erlang_js which downloads a tarball, scripts
the build, and then statically links against that
20:20:45 [nslater]: feh. we'll figure it out
20:20:51 [davisp]: nslater: +1, will get to it
20:20:53 [Wohali]: nslater: riak doesn't ship a windows version, do they?
20:21:00 [nslater]: Wohali: not sure
20:21:05 [Wohali]: We can certainly make it work though with effort.
20:21:21 [nslater]: davisp: you're saying the autotools stuff for bigcouch is gonna be easy?
20:21:27 [Wohali]: If we go scons we make python a build-time dependency.
20:21:42 [davisp]: nslater: I'm saying that if we remove autotools it makes some of the build
system stuff easier
20:21:50 [davisp]: nslater: Ie, we can drop rebar in pretty easily
20:22:15 [nslater]: davisp: i doubt we're gonna get to that before you start the merge. so
i am guessing the merge will have to be autotools, and then we strip it all out later. i am
not sure of benoit's timeline
20:22:50 [davisp]: nslater: either way is fine. If I know we're skipping autotools I can rewrite
the src/Makefile.am to just call rebar as well
20:22:50 [nslater]: okay
20:23:05 [davisp]: obviously that breaks make dist immediately, but if the plan is to ditch
we can start moving that direction
20:23:05 [nslater]: #info rcouch merge timeline doesn't block bigcouch merge timeline
20:23:28 [davisp]: well, I'm only saying I have no idea what's in rcouch
20:23:30 [nslater]: davisp: problem is we wouldn't be able to make a release unilt it was
fixed
20:23:50 [davisp]: nslater: fair enough
20:23:51 [nslater]: #action get benoitc and davisp/rnewson talking about rcouch + bigcouch
merge timelines
20:23:57 [nslater]: let's move on


# 6. release #
20:24:20 [nslater]: okay... *sigh*
20:24:28 [nslater]: the test suite has gone haywire
20:24:36 [nslater]: take a look at the jenkins build logs if you don't believe me
20:24:50 [nslater]: there seems to be about a 50% chance that any commit is going to send
me a "build broken" email
20:24:58 [nslater]: fortunately, jenkins also has lots of logs, so you can poke through them
20:25:08 [nslater]: i am going to send an email about this once i've got the actual release
off my plate
20:25:12 [davisp]: what's our jenkins project URL again?
20:25:45 [nslater]: sorry travis
20:25:51 [nslater]: https://travis-ci.org/apache/couchdb/builds/5113588
20:25:57 [nslater]: okay, in the mean time i am coding around it
20:26:07 [nslater]: i wrote a whole thing that could build couchdb on a freshly spun up AWS
image
20:26:12 [davisp]: ah
20:26:20 [nslater]: and it would just keep building and building until it passed
20:26:27 [nslater]: so i could go to bed or whatever you know have a life that isn't just
watching the terminal for four hours
20:26:36 [nslater]: then infra shut me down basically (check the lists) because of policy
20:26:50 [nslater]: they dont want release signing on aws
20:26:57 [nslater]: and they're a little sketchy about release building on aw
20:26:58 [nslater]: ... s
20:27:05 [Wohali]: nslater: That looks fixable, it doesn't know how to start the build so
it presumes it's ruby
20:27:12 [nslater]: so i am repurposing it to run locally. BIG SIGH
20:27:35 [nslater]: Wohali: oh wow i have no idea what the **** that is
20:27:38 [Wohali]: nslater: If we can put a .travis.yml into the repo root I can resolve that.
Last I recall there were issues with asfinfra being unhappy with that.
20:27:42 [nslater]: take a look at some of the older build failures
20:27:57 [nslater]: we have one in master i think i dunno
20:28:07 [nslater]: i think it might work on some branches and not others
20:28:12 [Wohali]: ah, that's attempting to build on 1.0.x, if you see
20:28:15 [Wohali]: We'd have to push the .travis.yml everywhere to fix that.
20:28:20 [nslater]: thats fine i think
20:28:26 [nslater]: but look at the other ones
20:28:33 [vmx]: sorry for chiming in late: dch did a lot of work to make rebar work nice on
windows, even for nifs
20:28:49 [Wohali]: vmx: *nod* still need to sort out couchjs, then.
20:28:49 [nslater]: Wohali: https://travis-ci.org/apache/couchdb/builds
20:29:03 [Wohali]: nslater: Yes, I see, all of the failed ones are older branches missing
.travis.yml.
20:29:10 [Wohali]: 0.11, 0.8, 1.1, etc
20:29:10 [nslater]: nope
20:29:17 [nslater]: look for ones on master and 1.3 too
20:29:26 [Wohali]: ok, i'll look into it.
20:29:33 [nslater]: https://travis-ci.org/apache/couchdb/builds/4967867
20:29:56 [nslater]: i dunno. the whole thing is a mess
20:30:03 [nslater]: anyway. i want to punt on that for now
20:30:11 [Wohali]: Aha. Looks like some of the dependencies required to build documentation
are missing.
20:30:11 [nslater]: so, i am repurposing the script to run locally
20:30:18 [nslater]: Wohali: that is yet another issue
20:30:27 [nslater]: there are other logs there that show the test suite is just randomly falling
over
20:30:33 [nslater]: you have to click around a bit
20:30:41 [Wohali]: ok, i take your word for it
20:30:49 [nslater]: well, i will send out a summary email once i get to this :)
20:31:11 [nslater]: if i think about it too hard right now it will just stress me out again
haha
20:31:26 [nslater]: at the moment, i have written a bunch of scripts which automate the maintenance
of NEWS and CHANGEs
20:31:33 [nslater]: they work very well, and i am going through the laborious admin work of
fixing it all now
20:31:48 [nslater]: once that's done, hopefully tonight, i should be able to get back to actually
building the release
20:32:04 [nslater]: so... real soon now. i am on 50% of my evenings at the moment
20:32:11 [benoitc]: i'm just arriving 
20:32:11 [eckoit]: the whole build thing makes me think fauxton should have a separate build
20:32:20 [benoitc]: a meeting at this hour on the middle of the week is definitely not for
me
20:32:33 [nslater]: #info AWS based build system for releases was shut down by infra
20:32:48 [nslater]: #info continued problems with test suite showing up in travis - details
to be posted later
20:32:57 [nslater]: #info NEWS and CHANGES being brought up to speed, then release will proceed
20:33:11 [nslater]: benoitc: let's circle back around to rcouch


# 7. rcouch #
20:33:26 [nslater]: benoitc: see above... some questions for you:
20:33:41 [nslater]: 1) given that we want to jettison autotools, is rcouch itself the thing
we want to merge in?
20:33:48 [nslater]: 2) what are the steps between here and there? and what is the timeline?
20:33:56 [nslater]: 3) would you see us doing this before or after bigcouch?
20:33:57 [benoitc]: jettison?
20:34:03 [nslater]: get rid off
20:34:18 [benoitc]: so status: i' i
20:34:26 [benoitc]: m in the middle of a patch 
20:34:43 [benoitc]: i said sometimes this week before a branch i think i'm still good on that
20:35:11 [benoitc]: so it's full rebar, a rebar plugin to detect icu
20:35:11 [nslater]: a branch in the asf git repos?
20:35:13 [benoitc]: yes
20:35:20 [nslater]: we need to do ip clearance first
20:35:28 [nslater]: are you happy running point on that, or do you want/need help?
20:36:03 [benoitc]: ip clearance for what? the patch is only giving a part of rcouch and is
a source code rework
20:36:06 [benoitc]: i don't think you need any
20:36:18 [benoitc]: especially since i've the right on it
20:36:18 [benoitc]: s
20:36:26 [nslater]: any considerable work done away from the project's lists or repos needs
ip clearence
20:36:33 [nslater]: it doens't matter whether you have an ICLA on file or not
20:36:41 [nslater]: the work was done away from the project, so we need to put it through
this
20:36:41 [benoitc]: mmm ok
20:36:48 [benoitc]: then i will publish a branch somewhere
20:36:58 [benoitc]: i'm still unsure how to manage this damn spidermonkey
20:37:10 [nslater]: no no, let's get your work in to the asf repos!
20:37:10 [benoitc]: i only handle 1..8.5
20:37:25 [nslater]: when you say it's only a part of rcouch, what do you mean?
20:37:49 [benoitc]: it won't include other changes for now. only the build work
20:38:04 [benoitc]: view changes and co may come after a while but as different patches
20:38:10 [nslater]: is this a fair representation of your plan:
20:38:25 [nslater]: 1) put build work on asf branch
20:38:26 [nslater]: 2) merge in work
20:38:40 [nslater]: 3) put other work on a separate branch
20:38:47 [nslater]: 4) merge in work
20:38:48 [nslater]: etc, etc
20:38:55 [benoitc]: yes
20:39:09 [nslater]: cool. okay, let's start this ip clearence stuff. would you like help with
that?
20:39:38 [benoitc]: totally i don't even know how it works. 
20:39:46 [nslater]: okay, i will help
20:39:47 [benoitc]: also let start this point next week
20:39:53 [nslater]: #info benoit hopes to have rcouch build work on asf branch very soon (pending
ip clearence)
20:39:53 [benoitc]: i prefer to release the code first
20:40:03 [nslater]: #action nslater to help benoit with ip clearence
20:40:16 [nslater]: benoitc: when you say "release", what do you mean?
20:40:46 [benoitc]: finishing and testing the full patch
20:40:53 [nslater]: okay cool
20:41:08 [nslater]: the way a typical ip clearence works is that we fill in the paper work,
and then you donate a patch or tarball to the project and we commit it to the offical repos.
it doesn't matter if that patch is published elsewhere first or whatever
20:41:31 [benoitc]: ok
20:41:38 [nslater]: benoitc: do you think bigcouch should wait until the rcouch build work
is done?
20:41:53 [nslater]: davisp said it might make things a little more simple - but it doesn't
appear to be a critical issue
20:42:03 [benoitc]: since bigcouch is using rebar i think the build change will help
20:42:16 [benoitc]: and should come first imo
20:42:18 [nslater]: my only concern here is timelines
20:42:31 [nslater]: we shouldn't delay bigcouch any more than we have to
20:42:49 [benoitc]: not sure how things can work in parallel, only cloundant guys have idea
on that
20:43:01 [benoitc]: rnewson: davisp kocolosk ?
20:43:01 [nslater]: benoitc: davisp / rnewson think they can start on this in the next few
weeks. you think we can get the rcouch build work done by then?
20:43:01 [davisp]: nslater: benoitc: It makes no difference to me on the build work. I've
just been looking through rcouch realizing its mostly build differences at this point
20:43:16 [benoitc]: davisp: not totally
20:43:31 [benoitc]: also view changes, and lot of other things
20:43:38 [davisp]: benoitc: well, I was equating refuge_couch with rcouch before I started
reading
20:43:46 [davisp]: benoitc: ah
20:43:53 [benoitc]: rcouch.org
20:44:01 [davisp]: I don't know anything about those changes so I can't comment
20:44:02 [benoitc]: but we are only speaking about the build here for now
20:44:08 [nslater]: benoitc: right - but for this, we're *just* talking about the build changes
20:44:16 [davisp]: right, +1 to build work only for now
20:44:16 [benoitc]: yes
20:44:31 [nslater]: cool - looks like it's fine for these two merges to be happening in parallel
then?
20:44:38 [davisp]: fine by me
20:44:47 [benoitc]: should probably work
20:44:49 [davisp]: there's lots of stuff in the rcouch build system that I'm a bit uncertain
of in asf land
20:44:53 [benoitc]: what few weeks mean ?
20:45:01 [nslater]: cool - can't wait for couchdb 2.0 :)
20:45:03 [benoitc]: 1, 2, 3?
20:45:19 [davisp]: ie, fetching github repos is a no-no. ASF infra is gonna have to give us
a decision on whether we can start creating a large number of git repos
20:45:23 [nslater]: davisp: we're fine to ship binaries that have all the deps pre-fetched
20:45:31 [davisp]: no
20:45:38 [davisp]: err
20:45:38 [benoitc]: davisp: so you don't need to 
20:45:53 [davisp]: benoitc: don't need to what?
20:46:01 [benoitc]: couch_core is now pretty much integrated
20:46:01 [nslater]: davisp: what are you saying "no" at?
20:46:08 [benoitc]: only deps are mochi & co
20:46:08 [davisp]: nslater: binaries aren't releases though
20:46:24 [nslater]: davisp: who cares? our releases would literally just be tar.gz copies
of the source tree
20:46:26 [davisp]: benoitc: I just looked through and saw gproc and lager
20:46:31 [benoitc]: davisp: well make dist create a release will all deps
20:46:33 [nslater]: davisp: with every release we'd distribute the package produced by rebar
20:46:46 [benoitc]: davisp: that won't be part for now from the new build
20:46:53 [davisp]: benoitc: ah, k
20:46:54 [nslater]: davisp: this is pretty much how ever other asf project does it
20:47:01 [benoitc]: i ditched couchdb logs by lager caus eit' sbetter
20:47:10 [benoitc]: also gproc is here for some stuffs around
20:47:16 [nslater]: releases are just zipped up source trees - and the useful stuff are the
jars or the whatever
20:48:09 [davisp]: Hrm
20:48:11 [nslater]: benoitc: we're on the same page here, right? it's totally fine for us
to fetch and bundle any software, as long as the license is free software
20:48:16 [davisp]: no
20:48:23 [davisp]: oh wait
20:48:23 [davisp]: binaries
20:48:25 [davisp]: hrm
20:48:31 [nslater]: yes!
20:48:46 [benoitc]: https://github.com/refuge/rcouch/blob/master/Makefile#L56 is building
the source archive
20:48:53 [benoitc]: ok perfect then
20:49:01 [nslater]: the actual release is *just* a tgz'd copy of the source tree - and with
ever release we provide the rebar package, which has all the goodies. and that's what we expect
people to download and use
20:49:16 [benoitc]: ok fine
20:49:38 [nslater]: again - this is how practically every asf does their releases. it makes
preparing the release very simple and it allows us to bundle whatever we want, as long as
the license is compatible
20:49:40 [Wohali]: and we can redist rebar?
20:50:01 [nslater]: Wohali: clarify?
20:50:01 [davisp]: Wohali: Its ASF 2.0, so we should be fine
20:50:08 [davisp]: nslater: most projects check a rebar binary into the source tree
20:50:16 [Wohali]: ok cool
20:50:16 [benoitc]: Wohali: we can build it 
20:50:16 [Wohali]: but what davisp said
20:50:23 [nslater]: ah cool
20:50:24 [davisp]: we'd be redistributing rebar in a release which is a thing to clear (but
should be fine)
20:50:25 [benoitc]: i have it in support/
20:50:38 [Wohali]: cool cool, just wanted to be sure
20:50:46 [benoitc]: but yeah license is fine
20:50:54 [nslater]: davisp: nope we wouldn't!
20:51:08 [davisp]: nslater: nope we wouldn't what?
20:51:16 [nslater]: be redistributing anything
20:51:31 [davisp]: nslater: how so?
20:51:33 [davisp]: unless you're not planning on checking it into git
20:51:46 [benoitc]: rebar script is like the ./configure 
20:51:54 [benoitc]: except it's gzipped
20:52:04 [nslater]: ah. well, you *could* fetch it at binary package building time if you
want. or you could include the source in our source
20:52:16 [benoitc]: so it worth to ask if we can provide it in the archive as is
20:52:17 [nslater]: but our source releases would just be a zipped up copy of our git tree,
with no additions
20:52:23 [davisp]: right
20:52:53 [davisp]: nslater: that seems right to me
20:52:54 [benoitc]: in any case we can build it from sources and membed it
20:53:01 [nslater]: so yes, either we include the source to rebar in git, and build it as
part of the binary package preparation - or we have a script which fetches it. either is fine
20:53:16 [benoitc]: the binaries don't need rebar at the end
20:53:23 [benoitc]: it's just a build script
20:53:25 [nslater]: ah cool
20:53:31 [nslater]: well, anyway. let's move on
20:53:31 [nslater]: i strongly suspect we're going to be completely fine on all fronts
20:53:32 [Wohali]: so no checking rebar binray into the tree.
20:53:38 [nslater]: which is emensely good news
20:53:38 [benoitc]: totally a bunch of erlang script put in the same file
20:53:38 [Wohali]: otherwise we're all good.
20:53:48 [benoitc]: ok
20:54:31 [nslater]: #info rebar will allow us to change source releases to tgz copies of git
and will allow our binary packages to bundle deps
20:54:39 [nslater]: that's it from me. i don't think we have any other topics
20:54:54 [nslater]: unless anyone has anything else, i think we're done?
20:55:31 [nslater]: okay, i'm calling it. thanks everyone! :D


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