Alan McDonald wrote:

>>Still, there is a theorical potential for corruption isn't there? I know
>>I've had corruption before, due to having to kill the ibserver.exe
>>process, because of a recursive SP. I think the same thing could happen,
>>theoretically, if the machine gets shut off abruptly while a database
>>write is in progress.
>>
>>Am I wrong?
>>
>>Jonathan Neve.
>>
>>
>
>That shouldn't leave the db in a bad state - the transaction which was
>running at the time of the 3 finger salute is just never committed. If you
>have had coruption from this vefore, then I would have to be suspicious of
>other things.
>Alan
>
>

Well, possibly... But I am fairly sure that this is what caused it
(there really were no other factors). Perhaps it was just a bit of bad
luck. However, personnaly, I would expect it to be corrupted. What if
it's in the middle of writing a record to disk, and it suddenly gets cut
of, with only part of the record written? Wouldn't that cause it to get
corrupted? (On the other hand, you might be right, as I can't pretend to
understand the internal workings of FireBird very well!).

Jonathan Neve.


[Non-text portions of this message have been removed]