Subject RE: [firebird-support] More on database corruption
Author Alan McDonald
> Hi!
>
> From what I gathered from my previous thread, mostly with Ann Harrison,
> it seems that FireBird is actually designed to be 100% corruption-proof
> in case of abrupt shutdown. Is this true, or am I pushing things a bit
> too far? I always used to think that FireBird did its best to avoid
> corruption in such cases, but that it was still theoretically possible;
> if, as it seems, FireBird is actually designed to make corruption due to
> abrupt shutdown _impossible_ (provided ForcedWrites is true of course),
> then should not any corruption be considered a bug, and get reported?
>
> As for logicial corruption, shouldn't we make that as impossible as can
> as well? And therefore, if I get such corruption one day, shouldn't I
> report it?
>
> Thanks!
>
> Jonathan Neve.

What about the corruption caused by a faulty NIC which provides just enough
bytes of a datastream such that the server writes "garbage" to the file. The
next read may not be valid but I can't say that this is any bug in FB.
Alan