Franz,
TYPE SDF doesn't explain what rule it uses in determing field sizes, all it does is output a given file with each field a fixed size.
For most practical purposes it's a fairly useless option. If you want a particular file structure either use a SQL-select to create character fields for each column and dump that or use TEXTMERGE to output the file. I have a process that moves millions of rows each weekend between a transactional SQL database and a OLAP SQL database and it uses BCP on the SQL sides and textmerge in VFP for the output to go into the OLAP. It performs quite acceptably.
>Well, I would rather the result were predictable - as I said, the pad amount varies with different numbers of fields, etc... Moreover, I'm bulk loading into SQL SERVER using format files and I don't know how big (in advance) the integer column will be written from VFP, judging from this example.
>
>Here's another interesting example:
>
>
>
>CREATE CURSOR Mewp (int1 I, int2 I, char1 c(10), char2 c(10))
>INSERT INTO mewp (char1, char2, int1, int2) VALUES ('asdf','qwer',1,2)
>INSERT INTO mewp (char1, char2, int1, int2) VALUES ('zxcv','zxcv',3,4)
>INSERT INTO mewp (char1, char2, int1, int2) VALUES ('tyui','ghjk',5,6)
>COPY TO c:\temp\temp1.txt FIELDS char1, char2, int1, int2 sdf
>
>The integers are LEFT justified, 10 characters long
>