> > > Is the DTD wrong - should NAME be required?
> > It's not required for TYPE=SUBMIT or TYPE=RESET,
> > but it is for all other TYPEs. This an application
> > convention (see 7.1.2, under "NAME") that can't
> > be enforced by SGML.
>
> Fair enough - so we're at the second case.
>
> The RFC needs to give some indication of what an application should do
> with a naked <INPUT>, as the DTD says such are legal. Even if it's
> merely to note that this isn't really legal.
What I said was that NAME should be #REQUIRED, so that users with proper
editors get prompted for a name, even at the expense of putting "submit"
or "reset" as the value for NAME as well as TYPE...
///Peter