Humm. Thanks for that Tom. I was expecting you to suggest the array 
method might be the best because if I fully understand what is going on 
it will match same field names from different tables ..... as long as 
they are identical field names and compatible types.

Another alternative then is to use a Python Script to directly add one 
to the other and call that from Paradox. No intermediate step as Python 
reads .db files directly. And Paradox runs Python Scripts really easily.

The current method works just fine but but is just slow. Maybe none of 
the other methods are going to be any faster as the speed problem is 
possibly related to the insert - commit at the AWS end.


On 10/05/2020 6:11 PM, Tom Krieg wrote:
> Every one of those is going to give you headaches. Why don't you just
> export Paradox data to a csv, and bulk add it into PostgrsSQL? You can
> create a Paradox script to do that.
>
> On 10/05/2020 5:38 pm, Robert MacMillan wrote:
>> So adding works just fine using a standard add. Just like in Liz's
>> question from years ago. It is pretty slow though so I am rapidly
>> going to run into performance problems.
>>
>> I am wondering whether a scan reading each record into a dynamic array
>> then inserting a blank record at the other end and then copyfromarray
>> will work better and faster?
>>
>> Also if one uses this method does it also mean that as long as the
>> field names are identical then the order of fields in the two tables
>> can be different?
>>
>> And supplementary question. Does that also mean that as long as every
>> field is in the AWS table then that table could contain additional
>> fields that the rows will fill in correctly and fields that are not in
>> the Paradox Table but are in the AWS Table will be left blank? I
>> presume if the AWS Postgresql Table has fields that are not allowed to
>> be null that it wont work?
>