Philip Martin wrote:
>
> "deleted" with "schedule=replace" would be a bug, it should not occur.
OK; that makes sense.
> But you are right, I was probably a bit hasty. "absent" is often
> treated very much like "deleted" in the existing code, but as far as I
> know "absent" can't be deleted or added and so can never be
> "schedule=anything". Your code probably has redundant checks.
Right, as does the code from which I copied it:
subversion/libsvn_wc/lock.c:prune_deleted()
(This presently has two slightly different variants which are equivalent in
practice but unsettling to read.)
subversion/libsvn_wc/entries.c:handle_start_tag()
This looks suspect:
subversion/libsvn_wc/adm_ops.c:svn_wc_add2()
> /* You can only add something that is not in revision control, or
> that is slated for deletion from revision control, or has been
> previously 'deleted', unless, of course, you're specifying an
> addition with -history-; then it's okay for the object to be
> under version control already; it's not really new. */
I can't make sense of this "unless" part. What's "the object" - the one you're
adding or one that already existed at the path? Obviously the object you're
adding (with history) is already be under VC; but surely there must not be a VC
object already at the path? Does it mean,
"At this level of operation (the WC), it doesn't matter if you overwrite
a versioned object, because you will have taken care of it at a higher
level."
?
> if (orig_entry)
> {
> if ((! copyfrom_url)
> && (orig_entry->schedule != svn_wc_schedule_delete)
> && (! orig_entry->deleted))
> {
> [error: "'%s' is already under version control"]
- Julian
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Oct 12 21:57:12 2005