Alex Pankratov :
Nov 11, 2013
1. Fixed an issue with copying files that are being shrunk - this is that single serious issue that's been outstanding for a couple of weeks now. Kudos to Terry for the most excellent log file that put everything in place.
The cause was a duplicate re-initialization of internal statistics in the delta copying module. In one particular case it would see more I/O buffers at hand that it has (supposedly) allocated and it would panic. So it wasn't leaking anything after all, it was an accounting problem.
2. Fixed an issue with the timestamp granularity detection - in certain cases the planning module failed to properly detect timestamp granularity, fell back to using a default that was too little and that cause some files to be updated when they should've not been updated.
3. Fixed a UI issue in case of losing connection to the service - this is an exotic problem. If Bvckup2 service was manually stopped, the UI would display a message to that effect, but the message box didn't have a parent window and was free-floating. So it could end up sitting under the main window, allowing for jobs to be stopped and started ... and that led to all sorts of wonderful things as jobs were no longer there.
4. Fixed an issue with drive information not getting refreshed - as of previous release every backup maintains a little cache of drive/volume information for both source and destination. This includes things like (my favourite) timestamp resolutions, file system type, etc. Regrettably, this information wasn't purge when the location was changed, so, again, that might've led to some interesting behavior.
Added detection of actively modified source files - this is a freebie from the first bugfix. This is the case that caused that accounting problem and so, while at it, I added a bit of code to detect *source* files getting changed while being copied. When this happens, the UI spit out a little warning into the log and suggest switching to Shadow Copying (though this doesn't really resolve the root of the issue - if an app fails to properly lock their files when updating them, the chances of it properly supporting Volume Shadow Copying are extra ultra slim).
Destination folder going missing now forces a re-scan - This compliments existing safeguards when running off the destination snapshot, which force destination re-scan whenever the app discovers something unexpected during the backup (e.g. a subfolder going missing or a folder suddenly appearing on its own accord). Same idea here - if the destination should exist and it doesn't, then the snapshot is discarded and rebuilt.