David M. Lloyd wrote:
> On Thu, 17 May 2007 14:15:46 -0400
> jacob_at_hookom.net wrote:
>
>> I agree with Adam here... but this is one specific use case.. are
>> there others?
>
> I don't know. By Adam's logic, the whole request/session/attributeMap
> ought to have key validation to prevent keys from being created
> non-identifier characters like ".".
No, that's not an outcome of my logic.
> This is certainly not the intent
> of the equivalent structures within the Servlet spec; why should there
> be additional artificial restrictions imposed just because the thing in
> that part of the map happens to be a managed bean?
>
> The fact is that EL is not the only possible way to access a managed
> bean.
And yet EL is a key and critical way of accessing managed beans,
and in fact the *primary* way of accessing them. The mappings onto
the Servlet spec are important, but secondary.
-- Adam