Bruce Momjian <pgman@candle.pha.pa.us> writes:
> Does anyone know the ramifications of allowing SET to work in an aborted
> transaction?
This is not an option.
The case that will definitely Not Work is SET variables whose setting
or checking requires database accesses. The new search_path variable
certainly works that way; not sure if there are any other cases at the
moment, but I'd not like to say that there can never be any such
variables.
regards, tom lane