JT Kirkpatrick wrote:
> I am connecting to postgres 6.4.2 on rh linux 5.2, using new kernel 2.2.2,
> from access97 on a win98 machine through an obdc driver (somehow, that was
> a mouthful!). I am converting some access files into postgres. One file
> in particular has a NOTES field in it (access memo datatype, postgres text
> datatype) that may have large amounts of data in various records. When I
> run an insert query to insert into the postgres tables from the access
> table, it bombs when it hits a record with"too much data" - I get the error
> message "tuple size too big, size 8360". In the obdc driver, I tried
> increasing the "text as longvarchar" amount to 15000 - no good. I then
> tried just unchecking the "text as longvarchar" option - no good. I think
> the message is coming from postgres, not the obdc driver. I have found a
> record in particular that causes the problem. When I remove a few lines
> from the data, it pulls in just fine. The few lines I removed had no
> special characters, etc. what to do?? I need to be able to store this
> data in its entirety. How to increase the allowable "tuple" size? What is
> a tuple anyway??
>
The data for that notes field is probably making the overall size of the query
too big for the backend to handle. It has a limit of around 8k. The real
solution is to have the driver use a large object instead of in-line the data.
I know I have said this before, but I will try to add that option next since it
seems to be really necessary now, alot of people need to store large amounts of
character data.
Byron
