AutoRecover seems to be unreliable. I am constantly losing many hours of work because autorecover simply does not recover recent changes after an enforced windows re-boot (which we can no longer prevent). It seems to either fail completely or simply "recovers" the last saved version.

Similar problems with AutoRecover not working have been reported and apparently fixed with earlier versions so I am surprised that this problem is still happening.

I don't yell at waiters or kick dogs (ref previous responses to AutoRecover issues!) and so consequently I also do not write angry, rude, patronising bug reports to software engineers who are good enough to support free open source software. However I would like to point out that frequently losing many hours of work is very frustrating so I think this issue should receive a high priority.

There have been previous issue reports about the recovery dialog being unnecessarily complex and confusing and it still seems to suffer from this. Could this be why something that should be simple to implement is unreliable? I am a programmer myself and have implemented auto recover functionality before so I am aware of how complex the whole issue can get if you are not careful.

My recommendation would be to replace the current strategy with something along the lines of how Microsoft Office implement it. So, for each auto recovered file, display a dialog telling the user that the file was auto recovered and then create (and open) a renamed version of the file in the same directory as the original file. The name could be the original file name with the word "recovered" and a time stamp added. Microsoft do not add the time stamp but I think this would be useful for people (like me) who constantly forget to save their work and so could possibly end up over-writing the recovered file. Using this strategy, the user is fully informed of what happened and can manage the complexities of having different versions of the same file in their own way and in their own time because the files are clearly named and easily accessed. So no need to confuse them with questions during the recovery process, just inform them that the file was either recovered or that the recovery failed. Microsoft do not create the new recovered file in the directory (it will only be created if the recovered version is saved) but in my opinion they should because the recovered file can be lost if the operating system again re-boots before the user has saved the recovered file (this has happened to me).

I may sound like a dis-organised, absentminded idiot that deserves everything he gets and so shouldn't really be catered for. However an alternative interpretation could be that I am overworked, constantly multi-tasking and distracted and plagued by an ever increasing number of operating system re-boots that I have no control over! And I am guessing that there are many other people using OpenOffice under the same circumstances.