Daniel Shahaf wrote:
> Stefan Sperling wrote on Fri, Jan 27, 2017 at 10:18:05 +0100:
>> As far as I recall, the concern here has always been about cases where
>> the user is passing specific arguments, e.g. should 'svn up dir', where
>> 'dir/child' was moved to 'otherdir/child', update just 'dir' or both 'dir'
>> and 'otherdir/child'? SVN asks users to make this decision at the conflict
>> prompt.
>>
>> I think such questions simply boil down to whether we expect path arguments
>> to represent working copy "paths" or working copy "nodes" addressed by path.
>
> This gets a little less obvious when multiple layers are involved, e.g.,
> .
> svn rm kappa
> svn mv iota kappa
>
> Should 'svn commit kappa' commit only the rm, or both the rm and the mv?
I'm willing to draw up a proposal that satisfactorily addresses these
sorts of combination cases. A rule that's simple and easy to understand.
No time to write a full version just now, but along the lines of: if a
move target path within the requested commit has a corresponding move
source path that is not within the requested commit, then automatically
include the source path *iff* the only scheduled change to that source
path is the move-away. (So in your example, commit both.)
- Julian
Received on 2017-01-28 20:47:11 CET