Hello fellow coders,
I have successfully subclassed DBI with the intent of overridding the
connect method so that I can retrieve the password from my password
server. One of my primary considerations in subclassing DBI was to not
change the interface. However, one of my co-workers is insisting that
we modify the interface so that a RDBMS's name be passed separately. My
point back to him was that there are RDBMS's that do not have a name but
rather an address (host:port).
My question is: Do I attempt to parse the DSN for the name of the
server, or force my users to pass it separately ? I am heavily leaning
to the former as this allows a very seamless transitition to the new
module. However, I am loathe to investigate and write/lift parsing code
for each DBD I want to support.
Here's where you come in and say, "Good idea Chuck, but...."
TIA,
Chuck