From: "Doug Chamberlin" <dchamberlin@...>
Sent: Wednesday, August 08, 2001 3:55 PM

> My last post detailing GetFullPathName was in response to Nando who seems
> to be trivializing the problem of using it. For the most part
> GetFullPathName is worthless, especially for this problem.

I think this is why, IMHO, that it would be beneficial to be able to use the
alias name you give a database when registering it with the server. It may
not stop someone from using things incorrectly (who can?) but it would at
least provide a uniform way of accessing databases.

How about an idea like this: Is there already or could something be added to
the database structure which would be an identity code of some type? When
the server is asked to connect to a database, the internal code number for
the database would be read and compared with connections already
established. It could be a code created with the database whether you create
a new one through DDL or by restoring a backup, it shouldn't matter. That
would allow the server to connect using any of the possible connection
strings and still understand that it's the same database file. Can anyone
see problems with this other than the obvious which means a file structure
change in the next version. (But, is that really a problem since you have to
convert each time anyway?)

Woody