Re: turn the tables

On Tue, 29 Jul 2003 23:53:01 +0100
"Balog Pal" <[email protected]> wrote:

> > To come back to my question: Is there a useful application to client side
> checks ?
> 
> Sure, quite plenty. :)
> 
> Server is far away, and accessing it requires time and network resources.
> Also, server response time may be high or unpredictable.
> Also, error reports from the server may be designed to be short and > not meaningful. [eg. to fight some attacks]   Especially if the server > is suposed to work with its own client doing checks. Then server > response may be kust a success/failure, nothing more.
> 

Ok, in the real world resources (e.g. network capacity, machine power) are limited but in an ideal world with unlimited good network connection and super fast servers these resource arguments do not count.
I think software design should orientate in the very first step by this ideal and later of course pay attention to the limited resources.

> Doing check on client side you can interact with the user and lead him to
> fill everything correctly consistently and conveniently.  (Marking required
> fields, graying out irrelevant controls and options is common.)  The
> 'Submit' button is only accessible when all checks passed, all needed > data collected.
> If you send a few megs of scanned pages with your request you'll not > want to repeat that due to a mistyped date.
> 

These are non-resource arguments but they have nothing to do with keyholes. These checks are done to make the client more pleasant for the
user. These checks increase the functionality of the client. 
Possibly one can say: Checks at client side should only be done with the aim to improve the surface (e.g. disable some edit fields) for the user.
But the client should avoid any checks which have nothing to do with the
interface to the user.