Reflecting on this thread:
http://archives.postgresql.org/pgsql-general/2008-06/msg00344.php
it strikes me that the elog messages in tuptoaster.c would be
significantly more useful if they gave the name of the toast table
containing the problem, which is readily available at the sites of
the elog calls. Any objections? Should I back-patch that, or
just do it in HEAD?
(Since these are elog's not ereport's, there's no translation impact
from choosing to back-patch. I agree with their being elog's because
they should be can't-happen cases, but when they do happen ...)
regards, tom lane