Issue #1213.

Although I am still having problems with SWIG and Python, I thought I
would make a quick stab at issue #1213.

I'm not quite sure of the desired behavior. One option would be for PREV5,
for example, to walk back 5 _changes_ to a particular file (which could be
many more than 5 revisions to the repository).

The other approach would be for PREV5 to walk back 5 revisions from the
commited version of a particular WC entry. This is the approach I have
taken. The other approach is likely to be quite a bit harder.

For syntax, I chose something that I thought was tolerable yet extensible:

   prev => Exactly 1 revision backwards from committed.
   prev/1 => Synonym for plain "prev"
   prev/2 => Two revisions backwards...
   prev/3 => Three revisions backwards...

if the previous commits changed the file or not). Searching back through
the nodes table would be a lot harder patch and probably outside the scope
of my current understanding of the code.

I was also unsure as to this routine:

  svn_boolean_t
  svn_client__revision_is_local (const svn_opt_revision_t *revision)
  {
    if ((revision->kind == svn_opt_revision_unspecified)
        || (revision->kind == svn_opt_revision_head)
        || (revision->kind == svn_opt_revision_number)
        || (revision->kind == svn_opt_revision_date))
      return FALSE;
    else
      return TRUE;
  }

Shouldn't PREV always be considered non-local. It's not like it's in the
text base, right? But my patch leaves the behavior for "PREV" as it was
previously (but says non-local for PREVX where X>1).

Comments welcome. I can't run "make check" yet, but its a start.

L8r,
Mark G.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org