I'm not up to speed on the structure of WCs and such, but I've wondered why
a WC doesn't allow an arbitrary set of add's, delete's, and copy's before
commit. It seem that there is only one rigid constraint, that the WC must
record the contents of the working copy of each file, and (implicitly or
explicitly) for each file, the revision and the file name within that
revision that is the immediate ancestor of this file. In practice, the
ancestors of most files/directories are not recorded, and if not, they are
implicitly the corresponding subordinate of the parent directory's ancestor.
Within this concept, there is no difficulty, say, copying (or renaming) an
added file -- the contents of the file are in the WC, and the copied file
has no ancestor, so there is no difficulty constructing a new element in the
WC with the correct contents and ancestry (none).
Similarly for other situations -- there are probably a number of cases to
enumerate, but none of them should present any problem to represent in the
WC.
Dale
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Mon Aug 8 21:10:06 2005