Jim Blandy wrote:
> On 11/14/05, Hyrum K. Wright <hyrum_wright@byu.edu> wrote:
>
>>To help me understand better, why would it be better for only a single
>>call to the client library from the client application for a move or
>>copy command, instead of looping like the add command? In my
>>understanding, we don't have to have atomicity because we aren't doing
>>something like a commit, which does send all of its arguments to the
>>client library at once. I don't have a position on which is better or
>>not, just wondering why.
>
> If the sources and dest are URL's, then it makes sense to want the
> multiple moves to appear as a single transaction. By contrast, 'add'
> never operates on the repository; it only makes sense for a WC.
Yes, that's right (we ARE doing something like a commit in some versions of the
copy and move commands). Only the copies and moves where the destination is
the repository need to be implemented as one multiple-source operation; those
where the destination is the working copy can be implemented as a series of
many single-source operations.
> (Caveat lector: the client library is terra incognita to me, if you
> catch my lingua franca. So take this with a grain of tortellini.)
Heh! :-)
- Julian
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Nov 15 14:24:03 2005