[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: svn commit: r1214677 - in /subversion/trunk/subversion: libsvn_client/client.h libsvn_client/merge.c libsvn_client/ra.c libsvn_client/switch.c tests/libsvn_client/client-test.c

From: Julian Foad <julianfoad_at_btopenworld.com>
Date: Fri, 16 Dec 2011 14:23:58 +0000 (GMT)

C. Michael Pilato wrote:

> On 12/15/2011 04:53 AM, julianfoad_at_apache.org wrote:
>> Teach svn_client__get_youngest_common_ancestor() to output the ancestor as
>> a URL as well as a relpath, because some existing callers want that and it
>> will also simplify writing a higher level version of this API which I
>> intend to do next.
>
> May I suggest that if you plan to publicize this functionality, you let the
> server do this YCA calculation with a new RA function, only falling back to
> the client-side calculation when communicating with older servers?

That sounds like a good optimization to make if and when we find this operation becoming frequently used and inefficient.  I'm certainly not opposed to it but at the moment it doesn't appear to be either very frequent (assuming once or twice or so per merge is OK) or too inefficient.

While I'm first putting an extra call to it in the 'svn' client, the same code path goes on to call one of the libsvn_client merge functions and those also call get_youngest_common_ancestor in some cases, so by refactoring I may be able to eliminate the duplication.

- Julian
Received on 2011-12-16 15:24:36 CET

This is an archived mail posted to the Subversion Dev mailing list.