Re: design issue: query type checking via DTD

looking over -- much data

at first glance I feel it is important to stress how it is almost never
favorable to *restrict* anything :-)

lisa rein

Jim Davis wrote:
> 
> Alan's responses make me realize that the true force of the Con arguments
> was left implicit in the arguments as I summarized them, or perhaps the
> issue should be once again reframed.  Let me try to make that explicit.
> 
> DASL servers themselves must independently validate any query.  They cannot
> rely on clients submitting only valid queries.  (While not everyone agrees
> with every argument advanced for this point, I think everyone agrees with
> at least one of them.)  If this is true, then it calls into question the
> value of client side error checking.
> 
> Alan argues that such checking is very desirable in UIs. (P2 makes this
> point).  Someone else might argue that such checking is redundant, since
> the server will repeat it regardless, and that redundancy, per se, is a bad
> thing, and since the only proposed means to support such client side
> checking is so unappealing.
> 
> So perhaps the argument is really over whether there ought to be a
> requirement that DASL enable agents (client side apps) to assist users to
> construct valid queries, e.g. by preventing them from constructing invalid
> ones.
> 
> I hope someone besides Alan (whose position is clear already) or I (who is
> simply trying to keep this discussion clear) will speak up.

Received on Monday, 8 June 1998 13:17:28 UTC