Re: Symmetric merge and deleted subtrees
From: Julian Foad <julianfoad_at_btopenworld.com>
Date: Mon, 13 Aug 2012 20:30:04 +0100 (BST)
I (Julian Foad) wrote:
> I have only had a chance to skim through your reply so far. I'm glad that
OK, I follow your explanation and agree; thanks. I have closed the isse #4217.
In investigating this, I tried your suggestion of doing the 'remove_noop_subtree_ranges' after the 'fix_deleted_subtree_ranges'. It triggered an assertion, one that I added recently, checking that the session URL was unchanged; this led me to find a bug I introduced (I think) whereby in several functions we reparent an RA session 'temporarily' but we don't take care to restore the original session URL if an error occurs.
On fixing that, I find that moving this code makes no difference to the editor drives made by this test. I haven't looked further into eliminating the no-op drive.
I do wonder, though, if instead of trying to detect and eliminate revision ranges here that would cause an 'invalid editor drive' later, it might be simpler instead to change the 'do an editor drive' subtroutine to handle such cases (where the start and/or end revision of the range doesn't exist). I will look into that some time.
> I have assigned Issue #4217 to track this, as I want to be able to refer to it
This is an archived mail posted to the Subversion Dev mailing list.