Re: problems with transaction blocks

>>>>> "BW" == Bruno Wolff, <Bruno> writes:

BW> Currently there is no provision for recovery from error by the application
BW> inside a transaction. What you can do is have the application check for
BW> problems before trying the insert. Depending on the problems you expect,

I have an application that does this. First it attempts to do a mass
number of inserts based on some external events all inside a
transaction. If we abort due to referential integrity constraint
violation, we restart the whole process but before each insert a check
is done to see if the required FK's are satisfied.

This gives us the benefit of 99% of the time when the FKs are ok we
zip along pretty darned fast, and for the 1% of the time when some
stale data is re-injected into the stream, we just restart that batch
and pay the penalty for it.

The benefit of being able to do all the inserts within a single
begin/end cannot be understated.

--
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Vivek Khera, Ph.D. Khera Communications, Inc.
Internet: khera(at)kciLink(dot)com Rockville, MD +1-301-869-4449 x806
AIM: vivekkhera Y!: vivek_khera http://www.khera.org/~vivek/