David Glasser wrote:
> On Feb 1, 2008 9:00 AM, Karl Fogel <kfogel_at_red-bean.com> wrote:
>> "C. Michael Pilato" <cmpilato_at_collab.net> writes:
>>> IMO, a 1.5 client should never set svn:mergeinfo when talking to a 1.4
>>> repository (note that I said "repository", not server). To accomplish
>>> this, we need to grow FS-level capabilities reporting and integrate
>>> that into our RA capabilities checks.
>> This is an interesting point.
>>
>> Up till now, we haven't carefully distinguished between server
>> capabilities and repository capabilities. Here are tentative
>> classifications of all our current svn_ra capabilities:
>>
>> SVN_RA_CAPABILITY_DEPTH ============> really about server
>> SVN_RA_CAPABILITY_MERGEINFO ========> really about repository
>> SVN_RA_CAPABILITY_LOG_REVPROPS =====> really about server
>> SVN_RA_CAPABILITY_PARTIAL_REPLAY ===> really about server
>>
>> Since every RA session is anchored on a particular URL -- that is, a
>> particular repository -- one way to handle SVN_RA_CAPABILITY_MERGEINFO
>> is to have it report negative if the repository in question does not
>> support mergeinfo, *even* if the server itself is 1.5 or higher.
>>
>> Does this sound sane?
>
> +1
>
> I think Mike was looking into how to implement that. Mark's point
> about helpful error messages is definitely also valid.
I filed issue #3089 to track this.
--
C. Michael Pilato <cmpilato_at_collab.net>
CollabNet <> www.collab.net <> Distributed Development On Demand
Received on 2008-02-02 05:06:05 CET