couchdb-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF IRC Services" <>
Subject Summary of IRC meeting in #couchdb-meeting, Wed Oct 2 13:01:16 2013
Date Wed, 02 Oct 2013 14:01:07 GMT
Members present: djc, deathbear, nslater, JasonSmith, garren, benoitc, dch, jan____, strmpnk

Meeting summary:

1. Preface

2. Fauxton deploy for 1.5
  a. garren and deathbear to prepare compiled Fauxton in _utils/next (garren, 2)

3. plugins 1.5

4. node view server

5. Couchdb Conf Vancouver
  a. everyone tell everyone about (jan____, 5)
  b. put a banner on the website for couchdb conf (nslater, 5)
  c. jan____ has spoken to Yuriy about the banner (nslater, 5)

- garren and deathbear to prepare compiled Fauxton in _utils/next (garren, 13:27:15)
- everyone tell everyone about (jan____, 13:56:22)
- put a banner on the website for couchdb conf (nslater, 13:59:29)

IRC log follows:

# 1. Preface #
13:01:45 [garren]: ASFBot: #topic Fauxton deploy for 1.5

# 2. Fauxton deploy for 1.5 #
13:02:24 [garren]: Ok for Fauxton release. 
13:02:46 [garren]: We can either release Fauxton as a couchapp or we can do a compile and
deploy it to share/www
13:03:10 [JasonSmith]: IMO I like how Futon builds as part of CouchDB
13:03:10 [garren]: I like the idea of the share/www and make it similar to the _util for futon.
13:03:24 [JasonSmith]: I personally, and for CouchDB hosting, I like to compile once and it
is installed for all users
13:03:32 [JasonSmith]: garren: +1
13:03:40 [jan____]: garren: yeah, could we make it /_utils/next/ or something?
13:03:41 [garren]: JasonSmith: great, to compile Fauxton we would need node integrated into
the couchdb build process.
13:04:20 [garren]: Or should one of the fauxton team members rather just compile beforehand
and commit the compiled Fauxton into git?
13:04:51 [strmpnk]: +1 maybe we should have a _contrib/
13:05:37 [garren]: jan____: which is easier at this stage, compiling Fauxton in the build
process or compile it beforehand?
13:05:39 [JasonSmith]: The way I always imagined it would work was ./configure would detect
if you could build it (you have Node and maybe Grunt)
13:05:52 [JasonSmith]: Checking for Node.js.....ok
13:06:25 [garren]: JasonSmith: yeah that makes a lot of sense.
13:06:40 [garren]: who is the most capable bash-fu expert to help us with that?
13:06:47 [strmpnk]: I don't see why the artifacts should be committed.
13:06:47 [JasonSmith]: To me the thing I am not sure about is, npm installs at build time?
I kind of don't personally like that
13:06:57 [jan____]: garren: either is possible, I think doing it as part of the build (like
Jason says) makes the most sense, but it would mean some autofoo work needing to be done
13:07:02 [JasonSmith]: to me your npm packages should be a build-time dependency, like libicu
or libmozjs
13:07:43 [JasonSmith]: Correct me if I'm wrong, but I really want to roll it out on Iris Couch
and then overnight anybody can just try it out on their account
13:07:50 [jan____]: nslater to the rescue!
13:07:57 [garren]: JasonSmith: that would be great.
13:08:29 [jan____]: garren: can you sum up the question for nslater ?
13:09:01 [garren]: nslater: we trying to find the best way of deploying Fauxton for the next
release. Should we commit the compiled artifacts of Fauxton into git or..
13:09:31 [garren]: should the build process do all that for us. Bearing in mind that we would
then require node plus some npm packages as dependancies of Couchdb.
13:10:09 [garren]: It might be easier for this first release just to commit the compiled artifacts
and later we look to integrate Fauxton into the build process.
13:10:49 [nslater]: "compiled"?
13:11:40 [garren]: nslater: We compile all the css, html templates and javascript and compile
it into 3 files. Instead of the 50 or so files we have in development.
13:12:09 [nslater]: i think we should do it in make
13:12:18 [nslater]: note that this doesn't introduce a user dependancy on node
13:12:42 [nslater]: we could do it so that we ship the pre-compiled files along with the source
13:12:47 [garren]: nslater: could you explain a little more.
13:12:56 [nslater]: sure
13:13:28 [nslater]: in autotools it is possible to set it up so that during the "make dist"
step of preparing the release tarball, some targets are built from the sources, and shipped
along with the rest of the source, so that end users don't need to remake them
13:13:38 [nslater]: the manpages are one example of this (we can't expect users to have help2man
13:13:44 [nslater]: the entire docs are another example
13:13:51 [deathbear]: hi :)
13:14:00 [garren]: hey deathbear.
13:14:38 [dch]: hey, late but made it
13:14:38 [garren]: nslater: that sounds good. Would that mean that the release manager would
then be the only one needing the node dependancies?
13:14:45 [garren]: hey dch 
13:14:54 [nslater]: garren: yes. or anyone trying to built from a git checkout
13:15:07 [nslater]: but if you're trying to build from a git checkout then you need everything
else too
13:15:17 [garren]: ok great.
13:15:19 [nslater]: i am talking autotools, autoconf, help2man, sphinx, etc, etc
13:15:33 [garren]: nslater: would you have time to help us set this up in make?
13:15:34 [nslater]: we call these developer dependancies in the root doc files
13:15:55 [nslater]: garren: how much help do you folks need? are you comfortable setting up
the default targets?
13:16:11 [nslater]: (which i could then tweak)
13:16:19 [nslater]: or do you need help with the whole thing?
13:16:35 [garren]: nslater: to be honest I'm rubbish with make. I could take a look. Our whole
build process is run via grunt which is similar to make.
13:16:51 [jan____]: I’m happy to land a hand, too, but I will need some guidance
13:17:02 [nslater]: don't worry about it. i am happy to pair with garren on it. it sounds
simple enough
13:17:07 [garren]: So all we would need to do is possibly do a check that node is installed,
install npm dependancies and run grunt release task.
13:17:07 [nslater]: but thanks for the offer
13:17:18 [nslater]: garren, we wouldn't install npm stuff from make
13:17:29 [nslater]: we'd just bail out of configure unless the modules were detected
13:17:38 [garren]: ok great.
13:17:53 [nslater]: jan____ corrects me on this
13:17:54 [JasonSmith]: To me the most complex and error-prone part is the rather large npm
13:17:55 [garren]: nslater can I follow up with you after this meeting and we can set a time
we both around in the next 24 hrs our so to do this.
13:17:55 [nslater]: but we'll see when we get to it
13:18:07 [nslater]: next 24 hours? woo jeez
13:18:07 [deathbear]: hi, we have a make file for deploying
13:18:07 [deathbear]: if that helps
13:18:30 [nslater]: what is "grunt" and do we need to continue using it?
13:18:39 [jan____]: garren: deathbear: a broader question before we set times, is Fauxton
ready to be shown off in 1.5.0?
13:18:48 [garren]: deathbear: that could help. Maybe you and I can take a stab at this today
and then get nslater and jan____ to check it and apply any tweaks.
13:18:55 [deathbear]: I think it is, as experimental 
13:19:03 [garren]: yeah definitely as experimental.
13:19:10 [nslater]: when are we pulling the trigger on 1.5.0?
13:19:11 [deathbear]: we've been working really hard on it since the redesign
13:19:33 [jan____]: nslater: grunt is a make for JS projects. we definitely want to keep using
13:19:34 [JasonSmith]: nslater: grunt is a build tool, to a first approximation it is Node.js's
13:19:41 [jan____]: nslater: djc wants to cut in ~24 hrs.
13:19:49 [nslater]: i am wary of including a build tool inside a build tool
13:19:50 [garren]: nslater: djc sent an email saying he is cutting the release in the next
24 hrs or so.
13:19:55 [nslater]: but i guess we did it for sphinx
13:20:03 [nslater]: jan____: okay then fauxton isn't going to go in
13:20:17 [jan____]: nslater: why not?
13:20:24 [nslater]: because i can't commit to getting this working in 24 hours
13:20:42 [jan____]: tomorrow is a holiday in .de ;)
13:20:49 [garren]: nslater: can we not rather do a precompiled artifact for this release?
And then after 1.5 integrat into the build procedure
13:20:57 [jan____]: I can give it a shot.
13:20:58 [JasonSmith]: nslater: This is what I was saying earlier. I would say call Grunt
a build dpeendnecy
13:21:06 [jan____]: garren: that’s a decent shortcut
13:21:08 [JasonSmith]: so you need Grunt the same way you need libicu-dev
13:21:13 [nslater]: jan____: disagree
13:21:19 [jan____]: JasonSmith: nslater wasn’t around then
13:21:28 [nslater]: i think it would run afoul of asf principals
13:21:44 [jan____]: which one?
13:22:07 [djc]: (sorry I'm late)
13:22:15 [nslater]: our source releases should be source releases. they should contain everything
you need to modify and run the product. i don't think we're allowed, and i don't think we
should, ship compiled files
13:22:53 [garren]: nslater: when we talk compiled files its just a html file, a javascript
file, a css file, a image file and 1 font file.
13:23:00 [jan____]: our docs are compiled
13:23:08 [nslater]: jan____: but we also ship the source for them with the compilation
13:23:24 [jan____]: we can also ship the fauxton source
13:23:33 [garren]: yes
13:23:42 [nslater]: this seems very 11th hour
13:23:49 [jan____]: plus, it is a preview, not a final release, so I think we have some leeway
for shortcuts
13:23:56 [nslater]: part of the rolling release schedule is that we don't panic about this
13:23:59 [djc]: I think shipping source + "compiled" is fine
13:24:05 [jan____]: nslater: I think this is a simple issue.
13:24:05 [nslater]: they'll be another release in a month
13:24:06 [jan____]: nobody panics
13:24:08 [nslater]: jan____: i don't feel comfortable doing this
13:24:20 [nslater]: and i would prefer about 1w to work on the build system for it
13:24:44 [nslater]: and even then, i would prefer a bit longer, because i would have to stop
working on everything else couchdb related
13:24:46 [nslater]: and i have some other pressing items that need my attention
13:24:54 [jan____]: src/fauxton/ explains how to build from source, and we also ship
share/www/next/ (or whatever) with the compiled version of fauxton. End of story
13:25:08 [jan____]: then let’s not integrate that into hte build today but ship a compiled
13:25:33 [nslater]: src/fauxton/ with instructions, and a compiled share/www/next is not ideal,
but i agree it doesn't violate our principals
13:25:40 [garren]: awesome.
13:25:48 [deathbear]: :) 
13:26:03 [jan____]: I agree it is not ideal, but it would allow us to ship a fauxton preview
tomorrow :)
13:26:05 [nslater]: i would like to work on it so that in the next release we have it properly
wired up to the build system
13:26:11 [jan____]: plus, it is not a dirty hack
13:26:18 [garren]: deathbear and I can work on getting fauxton working on _utils/next url.
13:26:26 [nslater]: (With garren or deathbear or whomever)
13:26:34 [deathbear]: in the next release fauxton will be even more awesome.
13:26:43 [jan____]: yeah, I consider it a blocker for a proper fauxton release that it is
integrated into the build system
13:26:45 [garren]: nslater: I would definintely be happy to help integrate into the build
tools. I do think thats the best option long term.
13:26:50 [nslater]: okay cool
13:27:07 [djc]: consensus \o/
13:27:15 [garren]: #action garren and deathbear to prepare compiled Fauxton in _utils/next
13:27:29 [garren]: everyone happy ready for the next topic?
13:27:29 [deathbear]: woo
13:27:38 [djc]: bikeshedding: does _utils-ng/ or something make more sense?
13:27:38 [jan____]: ^5
13:27:54 [deathbear]: ^5
13:27:56 [jan____]: djc: literally don’t care

# 3. plugins 1.5 #
13:28:24 [nslater]: wait wait
13:28:33 [nslater]: is _utils-nx the new X-feature?
13:28:35 [nslater]: *_utils-ng
13:28:48 [nslater]: i.e. we're not going to be lumbered with this url choice are we?
13:29:02 [djc]: nslater: no
13:29:19 [djc]: just while it's experimental
13:29:34 [garren]: djc yea I agree.
13:30:05 [dch]: ACTION then call it _experimental and make it crystal clear. But as djc said
"don't care" lets pick one.
13:30:29 [garren]: dch: I can send an email with suggestions and people can vote on one?
13:30:52 [JasonSmith]: yeah xylophone
13:30:52 [jan____]: hell no, just pick one
13:30:55 [djc]: xylophone
13:30:59 [dch]: lets not, pick one now.
13:31:08 [dch]: ding ding
13:31:15 [garren]: ok we will pick one.
13:31:23 [garren]: Okay next topic?
13:31:24 [djc]: yeah
13:31:32 [jan____]: aye
13:31:32 [garren]: JasonSmith: can you start up off on plugins for 1.5
13:31:32 [djc]: I think jan____ killed plugins for 1.5
13:31:47 [JasonSmith]: really?
13:32:03 [jan____]: soo, I wanted to spend a minute today to see what’s missing for plugins
13:32:05 [jan____]: I haven’t managed to do that yet
13:32:19 [djc]: JasonSmith: if you have round tuits to make it happen, I'm all for it!
13:32:22 [jan____]: I think I can frame it shippable, but I only will know later today
13:32:27 [djc]: just that jan____ didn't have time, I think
13:32:42 [jan____]: yeah, I found some time under the carpet
13:32:49 [jan____]: but not before the meeting
13:32:59 [JasonSmith]: jan____: I think I am happy with plugins, but just my definition of
"plugins" may be different from yours
13:32:59 [jan____]: I will update the 1.5 thread on dev@ later today
13:33:38 [jan____]: JasonSmith: well, you rip out /_plugins and /_utils/plugins.html
13:33:40 [jan____]: which is all that I did
13:34:10 [jan____]: e.g the binary installer & one-click-install-registry page
13:34:33 [jan____]: the rest were just minimal changes in how CouchDB operates, but that’s
all that you neede
13:34:33 [jan____]: d
13:34:57 [benoitc]: one thing to consider is how the pluging would work in a system where
you do live upgrade of a node
13:35:12 [jan____]: so far the things I am not too happy about are the barren look of /_utils/plugins.html
/ same for the fauxton version
13:35:13 [benoitc]: they won't be part of the release which may be problematic
13:35:28 [jan____]: benoitc: yeah, totally, but for this release that is out of scope and
won’t work
13:35:53 [jan____]: e.g. if you do live upgrades you will want to make the plugins you need
part of your source distribution
13:36:15 [benoitc]: well if we ship it , we make a promise to the user that it will stay for
a long time
13:36:38 [benoitc]: so we need to make sure it can really exists with a release system
13:36:40 [jan____]: benoitc: no, we ship it as “this will change a lot”
13:36:53 [jan____]: as a preview, this isn’t a commitment yet
13:37:44 [benoitc]: in the head of people it can be
13:38:00 [nslater]: well then we need to properly set expectations
13:38:14 [jan____]: yeah, but we can’t win that. we need to embrace experimntal features
13:38:15 [JasonSmith]: I am happy regrading plugins
13:38:17 [JasonSmith]: regarding plugins
13:38:22 [benoitc]: why would they care to try a feature that could change a lot tomorrow
13:38:30 [JasonSmith]: 1.5 as-is, I myself am happy
13:38:30 [benoitc]: this is not that i am happy or not
13:38:38 [benoitc]:  i like the idea of having plugins
13:38:54 [djc]: benoitc: if we document it clearly as changing, then there's nothing else
we can do
13:38:59 [benoitc]: but i wonder how it can really work with an erlang program
13:39:08 [djc]: if it changes out from under people, it's their fault for relying on it
13:39:24 [djc]: and experimental stuff leads to experimentation, which is good
13:39:31 [jan____]: benoitc: I don’t feel comfortable shipping plugins as a long-term feature
without having run a test version through a umber of users
13:39:31 [djc]: we need more ideas about what/how to work this stuff
13:39:40 [jan____]: regardless of how long we work on it unreleased
13:39:58 [strmpnk]: benoitc: I agree on the worry about how it might cause problems but this
is why it should be released as experimental, so those his are explored.
13:40:55 [benoitc]: i would expect they land in master for a release or two before going directly
in a release anyway
13:41:09 [benoitc]: at least I wouldn't expect them for 1.5
13:41:18 [djc]: benoitc: that's not how we work anymore, master means releasable
13:41:33 [benoitc]: yes
13:41:33 [benoitc]: so it is not in master
13:41:55 [djc]: yes, but there's also no "bake-time" on master
13:42:09 [jan____]: benoitc: I can merge it into master within the next 24 hours and then
it is relesable
13:42:09 [djc]: they bake in releases, need users to mature anyway
13:42:18 [jan____]: the question is are we happy with the current state
13:42:43 [jan____]: I think I can be happy with the current + minor fixes state
13:42:49 [djc]: I have no answer to that, but am happy to defer to Jan & Jason
13:43:35 [jan____]: right, again, will look into this after the meeting, but before tonight
13:43:55 [benoitc]: i don't see the point to release a thing in 24h that have never landed
in master for a while. 
13:44:18 [benoitc]: this is not how a quality software should work imo but this is out of
13:44:33 [jan____]: benoitc: we don’t do baking in master anymore. branches are ready to
ship or they are not
13:44:47 [djc]: okay, I'm ready for the next topic
13:44:55 [JasonSmith]: Ready
13:44:55 [garren]: great
13:44:56 [jan____]: benoitc: and especially with new stuff, we mark it as experimental, so
people can play with it without expecting it to all work

# 4. node view server #
13:45:18 [garren]: jan____: Can you start us off or is it JasonSmith?
13:45:18 [jan____]: I also don’t quite see why we have to discuss the how of releases that
we had agreed upon again
13:45:20 [jan____]: ready.
13:45:28 [JasonSmith]: I am done, plugins look good
13:45:33 [jan____]: I got it
13:45:34 [JasonSmith]: fdmanana says hi
13:46:16 [jan____]: as far as I am concerned the nodejs viewserver works well enough to ship
as an experimental release. There are some obvious todos, but they can be done later.
13:46:29 [benoitc]: i don't remember to agree on such thing but anyway it wasn't the point
13:46:36 [jan____]: the goal of the release is to get this into people’s hands so they can
play and try to break it
13:46:36 [benoitc]: i was speaking on a technical level.
13:46:52 [benoitc]: let's go to the other topic
13:47:00 [garren]: jan____: is there documentation on how to get it up and running. Eg I install
1.5 how do I activate the nodejs view server?
13:47:30 [djc]: the nodejs view server doesn't add any dependencies for non-users, right?
13:48:14 [jan____]: djc: correct
13:48:39 [jan____]: garren: one sec
13:48:46 [garren]: sure
13:49:24 [JasonSmith]: jan____: +1
13:49:40 [jan____]: garren:
13:49:49 [jan____]: e.g. this would show up on
13:50:32 [djc]: I like the looks of this
13:50:42 [garren]: jan____: perfect thanks.
13:51:02 [jan____]: :D I made a poinnt of adding docs to get djc approval :D
13:51:33 [jan____]: it is a little bare-bones, but we can update the doc on the go
13:51:33 [djc]: and you shall have it
13:51:40 [jan____]: (love the new docs immediate publish setup)
13:51:55 [garren]: Excellent.
13:52:10 [jan____]: nice
13:52:17 [jan____]: djc: I’ll get that merged in time
13:52:40 [garren]: Great any other topics someone wants to add?
13:53:02 [jan____]: CouchDB Conf Vancouver
13:53:17 [deathbear]: who is going?

# 5. Couchdb Conf Vancouver #
13:53:24 [jan____]: <
13:53:26 [deathbear]: I am still thinking about it
13:53:54 [jan____]: deathbear: would be nice to meet you finally :)
13:53:54 [nslater]: Yuriy should post an announcement about the tickets to both public couchdb
13:53:54 [garren]: Unfortunately I can't make it.
13:54:25 [nslater]: ACTION wished he flew
13:54:27 [jan____]: It would also be cool if everyone here could use their social media outreach
to get people aware of it
13:54:32 [benoitc]: i probably can't make it happen. my assistant forgot to book the flight
13:54:32 [jan____]: s/get/make
13:54:41 [deathbear]: nslater are you scared of planes? cause ME TOO. 
13:54:41 [benoitc]: and cascadiajs don't interrest me at all
13:54:48 [jan____]: benoitc: Doh :(
13:54:55 [nslater]: idea: hook up skype to a projector that covers one of the walls. and i
can have an omni-tele-presence 
13:55:04 [nslater]: silently watching and judging you all
13:55:05 [jan____]: heh, cascadia is optional :)
13:55:20 [jan____]: nslater: sounds good :)
13:55:27 [nslater]: benoitc: time to get a new assistant! ;)
13:55:36 [benoitc]: yup but was trying myself to still come even if a flight is 1700 euros
13:55:37 [nslater]: deathbear: yes
13:55:49 [jan____]: #task everyone tell everyone about
13:55:58 [nslater]: is it not #action ?
13:56:07 [jan____]: benoitc: get in touch if price becomes an issue, we might be able to help
13:56:22 [jan____]: #action everyone tell everyone about
13:56:37 [benoitc]: yuup was about to do it thx
13:56:45 [benoitc]: can we have an ad on the site ?
13:56:54 [benoitc]: like banner or sort of ?
13:56:59 [jan____]: yeah great idea
13:57:09 [nslater]: +1
13:57:15 [jan____]: benoitc: excellent idea, I’mma look after that
13:57:30 [garren]: Any idea who could do that for us?
13:57:52 [jan____]: garren: I pinged Yuriy in email
13:58:00 [nslater]: do we have any designers in da house?
13:58:14 [garren]: cool.
13:58:59 [nslater]: ait
13:59:00 [nslater]: wait
13:59:08 [nslater]: you've pinged yuriy about the website banner?
13:59:22 [jan____]: yes
13:59:29 [nslater]: #action put a banner on the website for couchdb conf
13:59:38 [nslater]: #info jan____ has spoken to Yuriy about the banner
13:59:38 [nslater]: okie dokie
13:59:39 [nslater]: thx
13:59:45 [jan____]: :)
13:59:56 [garren]: ok great.
14:00:01 [nslater]: me and jan____ are in the same room together. this adds a new dimension
to the meeting
14:00:02 [garren]: ASFBot: meeting end

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