Using ( or trying to ) the JDBC interface, one finds the my porting
effort is just not complete enough.
Postgres seems to like 2^32-1 byte/file as a rule. On the ALPHA there
is the ability to go beyond 2^32-1 bytes ( in theory ~17gb, due to
limitations of the ext2 filesystem ). u do not get an EOF/error at the
(32bit) 2gig intel file limit.
And What of loread/lowrite(loseek/lotell) ? it is theoretical to give a
len >2gig ( but the mem/architecture is just not there yet for mem that
large ). Are offsets passed back to the JDBC/client as 32bit ints?
Can any of ur DB file architectures handle the >32bit offsets? If so how
far? ( ie ext2 is limited to 17gig ).
Where are u folks going with this "offset" inconsistency?
gat