couchdb-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Paul Davis <>
Subject Re: dependancy tracking
Date Tue, 03 Jul 2012 07:07:07 GMT
On Wed, Jun 20, 2012 at 11:16 AM, Benoit Chesneau <> wrote:
> In my last attempt to make the core more erlangish, I'm trying to find
> a way to manage dependencies like webmachine, oauth, snappy, ... . I
> thought one way would be to get them while preparing the release like
> I do in rcouch, but maybe some still want to have them in the
> repository.
> So:
> 1. What do you prefer ?
>    a. keep the dependency in the source
>    b. getting them at the release (with possibly rebar), they will be
> distributed with the archive

ASF policy says that all source code in a release *must* be in ASF
source control. So to not include a dependency in a release we have to
force users to get it on their own. For things like snappy, we may be
getting to a point where that's possible, but for the Erlang apps that
doesn't seem like a viable option at this point in time.

> 2. In case of a), is there a good pattern to maintain/track changes
> upstreams ? We for now do big merge, without keeping upstream history.
> It may be a problem or not. Thought?
> - benoit

I know that Dave Smith has a tool to keep leveldb sources up to date
in eleveldb. We might take a look at that. Although I'm perfectly fine
with just using a big commit to bring our deps up to date (although we
may want to mark them with a specific URL/commit/version thing in the

Also, we patch a number of our Erlang deps, so that's another vote in
favor of having them in the repo.

View raw message