David Worms wrote:
Great, that was the response I was expecting.
1. can you avoid using a selector by declaring a dependency on a selection service?
you're right, this is the most respected solution.
2. if you are absolutely locked into selector semantics, how urgent it is?
I am not.
Good - then we don't have a problem. *steve will sleep peasefully during the night*
But I can think of a couple of scenarios in which such a functionality would be required (SEDA architecture for example).
That just means that the SEDA architecture is broken because if SEDA if assuming semantics based on the specified semantics of framework selector - its a no-show. This issue is all about the specification of the selector. Under the current specification the behaviour is undefined. If SEDA has a specification of what selection means, then there may be a fighting chance.
3. workarounds to the problem are possible via cascading containers
maybe this has been discussed before. Do you mean that a component behaving as a container can lookup all of its child components? Can u explore more or redirect me to some prior discussions?
What I mean is this - both ECM and Fortress make assumptions about what selection means that is unrelated to the framework specification. You can embed Fortess in Melin, and deploy a selector dependent component inside Fortress - and basically this is a workaround to a problem of insuffiently defined specifications. So long as the specifications remain as they are, you and I will have a problem. Becuase your expecting something and I have now way of resolving what I need to deliver. You get pissed off because your not getting what you want, and I get pissed off because your not telling me what you want.
Who was it that said that we don't need specifications?
4. think about option (1) 5. consider option (4) 6. your best bet is to go with any option less than 2 or greater than 3.
thks.
No problem (so long as you stick with opion 4, 5, or 6).
Cheers, Steve.
--
Stephen J. McConnell mailto:[EMAIL PROTECTED]
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
