On Thu, Aug 16, 2012 at 10:10 AM, C. Michael Pilato <cmpil...@collab.net> wrote:
> So, incorporating some of the great suggestions on this thread so far, but
> adding some of my own, I'd like to reset this design discussion with the
> following proposal:
>
> # Reports the general name of the client.
> #
> # Examples: "svn", "tortoisesvn", "svnkit"
> #
> # Default value: "svn".
> #
> client-name=[-_a-zA-Z0-9]+
>
> # Reports the specific version of the client.
> #
> # Examples: "1.6.0a2", "1.5.4dfsg1-1ubuntu2.1"
> #
> # Default value: ${SVN_VER_NUMBER}
> #
> client-version=[-_a-zA-Z0-9\.\+]+
>
> # Reports the version of the Apache Subversion libraries with which
> # this client is most compatible.
> #
> # Examples: "1.6.0", "1.8.0-dev", "1.7.0-alpha1"
> #
> # Default value: ${SVN_VER_NUMBER}
> #
> client-compat-version=[0-9]+\.[0-9]+\.[0-9]+(-[a-z0-9]+)?
>
> Here are some examples:
>
> client-name=svn
> client=version=1.8.0-dev
> client-compat-version=1.8.0-dev
>
> client-name=svnkit
> client-version=1.7.5-v1
> client-compat-version=1.7.5
>
> client-name=tortoisesvn
> client-version=1.7.8
> client-compat-version=1.7.6
>
> client-name=svn
> client=version=1.5.4dfsg1-1ubuntu2.1
> client-compat-version=1.5.4
>
> I think this will allow the best of both worlds for hook authors: the
> deeper insight needed to rule in/rule out specific clients and client
> versions, plus a way to make more general rulings based on the set of
> features expected to be present in a given Apache Subversion release.
>
> Thoughts?
It probably does not hurt anything to "do more" as your proposal does.
But personally, I think that your "client-compat-version" value is
the only one that is needed and I would be fine if that was the only
thing we added.
--
Thanks
Mark Phippard
http://markphip.blogspot.com/
