Philip Martin <philip.mar...@wandisco.com> writes: > What I propose is that copy only produces the incomplete state for nodes > that are copies (or copies of copies) of repository nodes. Copy would > not produce the incomplete state for copies of added nodes. When an > added node is copied the copy is an added note completely separate from > the source and if an interrupted copy causes it to have fewer children > this is not the same as the copy of a repository node having missing > children.
It appears that wc-to-wc copy already works this way as far as setting incomplete. That makes me more confident that r1177732 is the correct approach: - read_info converts working/incomplete to status=added, - scan_addition returns status=incomplete - callers of scan_addition are responsible for handling status=incomplete (in lots of cases like status=copied) -- uberSVN: Apache Subversion Made Easy http://www.uberSVN.com