[postgis-users] jdbc, bytea for the write side

[David Blasby]

> Funny, since the same "read 20000 geometries and convert to JTS" sintetic test
> went down from 19 secs, using the WKB format, to 12, using
> bytea(asBinary(...)). I did not check, I kind of remember from the manual
> that effectively special chars such as the 0 gets turned into 4 bytes, but
> the others remain pretty the same. Hum, I guess I should try with more
> variations in the data read for the test...

This is a very funny result - did you do this for many runs?

bytea(<wkb>) is a 'null' function - it just copies the <wkb> argument
and postgresql re-types it.

The only way I can account for this would be that libpq/JDBC optomizes
itself for transfering BYTEA results.  I havent seen anything in the
manual about this, but it is reasonable...

dave