Hi, While I'm a huge fan of having API client libraries galore, I don't think it's a very good idea to maintain them in the actual ESME repository. There are several reasons for this:
1. Most client library authors will not be commiters, creating an unnecessary bottle-neck around library checkins. 2. There may be more than one version of a client in a given language/platform, and we probably shouldn't be playing favorites. 3. Checking them in as part of the project gives a sort of "Apache stamp of approval", which we may indeed want to give, but should probably do by listing the clients on the wiki rather than checking them into the repository. Overall, I'd say that the general approach for API clients to projects like this is for authors to maintain them in public repositories (usually Github or Google Code, depending on the choice of version control system). They can then be managed outside of the Apache process. Often, an implementation will provide a reference client in one or more languages, not meant for production use. If that is what we are trying to do here, then I guess I could come around to supporting that position, but we should be aware that we are signing up for more work :-) Wow, I'm such a contrary person aren't I? Richard (and others) what do you think? Were you going for "reference client" or were you planning for this to happen for all clients? Ethan On Sun, Dec 6, 2009 at 4:56 PM, Richard Hirsch <[email protected]> wrote: > FYI > > I just added Daniel's initial code drop > (https://issues.apache.org/jira/browse/ESME-14) for a Java client for > the streaming API to SVN. > > D. >
