The enormous s->childXids problem

I've tracked the problem I mentioned earlier with my 4.5 million node  
linked list of s->childXids down.  We use plperl to connect to Oracle  
over DBI.  The select function is dbi-link's remote_select.   
remote_select will perform the query and then for each row  
return_next which calls the SPI.xs stuff to do plperl_return_next  
which is wrapped in a PG_TRY block.  I see the value of the try block  
to kick back sensible errors to perl, but creating childXids for  
every row of a setof seems wildly excessive.  What's the harm in  
simply not TRY'ing around there?

I ask with respect to the suitability as general solution and as the  
suitability for my acute issue (of a 5 million row setof returned  
from that).  Will it break anything?

Best regards,
Theo

// Theo Schlossnagle
// CTO -- http://www.omniti.com/~jesus/
// OmniTI Computer Consulting, Inc. -- http://www.omniti.com/