Hi all:

We are running two servers: ASE 11.9.2 and ASE 12.0 on the same NT 4.0
machine. A C program inserts rows into one table, called ms_orig.

Running the same program with the same input data versus ASE 11.9.2 and ASE
12.0 we obtained two ms_orig tables, one for each server. Let's call them
'(1192)ms_orig' and '(12)ms_orig'.

The first strange result was that applying bcp out in character form for
both ms_orig tables we found out that rows generated for '1192(ms_orig)'
are delimited by 0x0a and rows generated for '(12)ms_orig' were delimited
by 0x0d 0x0a. The command line was the same:
bcp product..ms_orig out <file_name> -S<server> -U<user> -P<pass> -c

The second, and most annoying result was: when bcp in '(1192)ms_orig' the
file generated by bcp out from '(12)ms_orig', we expected to have the same
set of rows twice in '(1192)ms_orig', because the two tables were
containing the same rows.

But, running a clustered index with ignore_dup_row, in order to remove
duplicate rows from 'concatenated''(1192)ms_orig', there were still
duplicate rows un-removed. If the 'concatenated''(1192)ms)orig' table is
transferred into another table:
select * into test from ms_orig
and the same clustered index is applied to 'test', all duplicate rows were
removed and 'test' table was an exact copy of 'original'(1192)ms_orig.

The same result occurs if we 'concatenate' (12)ms_orig in the same fashion
and apply the same steps as above.

Any help will be greately appreciated.

stefan
stefan.giuros@connex.ro