mason wrote:

> We are getting blob corruption in users notes though I also suspect
> users maybe cutting and pasting from word, unsure really.

One possibility is that you're using read-committed, read-only
transactions on volatile data. There is a window between reading
a record and reading the blob data. If, in that window, someone
else deletes the record, commits the delete, and a third transaction
garbage collects the record, the read-only, read-committed
transaction will get a "blob not found" error that is reported as
a corruption.

If that same transaction tried to read the record again, it wouldn't
see it either. That's the "beauty" of read-committed transactions -
you see the world changing under your feet.

We've discussed ways of fixing the problem in the code, but the
changes are complicated and in a delicate part of Firebird. So,
for the moment, the best suggestion I have is to trap that error
and ignore it.


Regards,


Ann