> While I appreciate that this problem is a total show stopper for you,
> I'd personally and humbly, as a reader of this forum, suggest you keep
> your voice down a bit until someone has been able to confirm that it is
> actually a bug in FB.

hmmm, i don't raise my voice (sorry if it's look like this) ... i simply say that their is a problem, i just insist on the fact that it's a serious problem. everyone in the firebird project do a spectacular job, and if i can give a little help to them by become an alarm to them when their is a bug, i will off course don't be silent too much :)

> You claim that it cannot be a bug in your code, since you communicate
> via fbclient.dll. But wouldn't it be possible that your code passes some
> invalid data into fbclient calls, which in turn might cause the FB
> process to run havoc? Or possibly there's some faulty hardware in your
> system? Or some other process running on your system that e.g. locks the
> databse file from time to time, disrupting FB:s updates? Just to mention
> a few possible alternatives to FB being buggy in this case.
>
> Anyway, it's not very productive to go shouting. Keep calm and go on
> providing extra info to help narrow down where the problem lies, and
> then find out how to resolve it.

thanks, you are right, i will try to ask one developper to do a sample project that you will be able to run on your own computer and you will see by yourself how the database become to be corrupted !

for what i know, the key element of the bug are :

1/ one table with more than 10 000 000 rows
2/ around 25 index on this table
3/ lot of simultaneous read only Select on this table
4/ Lot of Insert / delete / update of this table
5/ eventually some sleep operation at the same time (don't know yet if it's connected)

soon as the demo project will be ready i will come back here !

thanks for all
stephane