[Xastir-Dev] Xastir-2

[Howard, Chris]

Thoughts from a lurker...

I think one place to start would be to develop a list
of the types of queries you will want to be doing.

If you are going client-server,  your software will
consist not only of the client-side presentation code,
but also of the database schema, stored procedures/functions,
and possibly the communications protocol between the two.
Install, configuration, upgrades, etc. will all have to be
smarter.

I think it would be very convenient to have a protocol
layer that allows replacing the full-blown SQL backend
with something less sophisticated, as necessary.

Would you want to do the ODBC/JDBC kind of thing? or
do you want to create your own protocol?  As I see
it the benefit to the ODBC model would be portability.
But the downside can be performance, and another
external piece of software that the user has to
find, install, and configure.  If the queries are
relatively simple, performance isn't an issue.

And I think you will want to, early on, think about what
data goes into the database and what does not.  For
example, user session info, screen colors, window sizes,
etc.  I would suggest those should be maintained
in config files on the client end.  Anything that would
be particular to that client workstation/environment
should be maintained on the client _unless_ you want
to go for the whole user-access-same-from-any-workstation
kind of viewpoint, which would require user-config maintenance
on the backend, with appropriate forms/screens for adding
users, changing passwords, etc. etc. and a raft of
info about workstation/environment capabilities.

Are you going to allow multiple backends?  I mean, how
big do you want this thing to be?  One backend for 
each user/friendly-group/1,000 people/state?  
  like a distributed findu?  I can see why
it might be convenient to have a heirarchy of databases
to query for out-of-local-area information.



Chris Howard
chris at yipyap.com