"Frank D. Engel, Jr." <fde101@fjrhome.net> writes:
> Given that this issue is a violation of SQL compatibility, shouldn't
> there really be an option to turn off interpretation of backslash
> characters in string literals as escapes? Maybe as a session variable
> of some kind, with a default being set in postgresql.conf?
That has about as much chance of getting in as a session variable to
change the identifier-case-folding behavior, and for the same reason:
any such variable is certain to break tons of existing client-side code
that doesn't know about it. We already learned this lesson with respect
to autocommit :-(.
It is irritating that we can't easily support exactly-spec-compliant
applications, and I'd like to find a solution. But a GUC variable
ain't it. See past discussions about case-folding behavior for some
of the ramifications.
regards, tom lane