Hi Jan,
I don't know that it makes any difference, but ;
The request was for the numbering sequence to begin anew every day hence
today = 20080114001
tomorrow = 20080115001
conflicts with
20080114063030 (YYYYMMDDHHMMSS).
Unless the time value is more meaningful somehow.
fwiw,
Ben
jan johansen wrote:
Bob, Alastair, David,
Thank you for your suggestions.
The only reason for the YYYYMMDD sequence was to allow a form of
secondary identification
to a lot i.e. visual using human readable.
I had initially proposed just using an autonumber but the client
requested a form of secondary
identification. The request was for the numbering sequence to begin
anew every day
hence
today = 20080114001
tomorrow = 20080115001
so concatenating an autonumber to the date doesn't quite work
Another goal would be to keep the length the same
since a 1 character 128code is narrower than a 6 character.
So just using an autonumber starting at one would not quite work
unless it could be
formatted as 0000000001 (formatting an autonumber integer?) or just using
10000000001 and go with that. But again what does that mean to someone
looking at it.
I'm begining to lean towards 20080114063030 (YYYYMMDDHHMMSS).
This format remains consistent (as long as someone doesn't screw with
the DATETIME settings
and allows it to be stored as INTEGER vs TEXT.
Enjoyed the discussion.
Jan
