Branko Čibej <br...@wandisco.com> writes: > What we /should/ do is create a tempfile and keep changing and stat'ing > it until its mtime changes. We could do this with an exponential backoff > sleep, too. Then there would be no guesswork about the timestamp > resolution of the WC file system. > > The problem is that we don't always get a path into > sleep_for_timestamps; but we already have a fallback for that.
There are other things we could try. When we do get a path we could identify the wc root and then run a SQLite query to get the most recent last_mod_time. We could use that to reduce the sleep. -- Certified & Supported Apache Subversion Downloads: http://www.wandisco.com/subversion/download
