Taravudh Tipdecho <taravudh@hpcc.nectec.or.th> writes:
> Our problem is with the insert command when we insert a large polygon
> (~200 points). With version 6.3 we have no problem inserting the
> polygon. In fact all the polygons were inserted, verifying that they
> were less than the 8K limit. However with version 6.4.2, the backend
> closes the connection during the insert.
> The first time we noticed this, we had a lot of trailing zeros on the
> values of the points and eliminating the zeros allowed a smaller polygon
> (that previously failed with ~ 170 points) to be inserted in version
> 6.4.2. This seems to indicate that there is some kind of limit on the
> length of the query string. Isolating the query string that failed
> indicates that the string is about 4500 bytes.
That's really odd; it's hard to believe that the maximum query length
got shorter.
I presume the backend dropped a corefile when it crashed; can you use
gdb on the corefile to provide a backtrace? Alternatively, can you
provide a fairly short psql script that demonstrates the problem?
(If you're right about the problem, just a CREATE TABLE and INSERT
oughta do it...) I will look into it if I can reproduce it here.
regards, tom lane