Udo Richter wrote:
> ...
> There's a minor taste issue that I want to bring up before its too late.
> The current interface suggests (but not enforces) that
> Service("someid",NULL) asks whether this protocol is supported or not,
> and doesn't cause any action. If a plugin just wants to signal something
> and has no additional data, enforcing this rule would require some dummy
> pointer to be sent. However, the plugin could just take action even with
> Data=NULL, violating the rules.
>
> Variants for this:
> 1. Keep it as-is
> 2. Drop the supported check from docs - each plugin can offer a check if
> it likes
> 3. Introduce a constant handle, so Service("someid",QUERY_SERVICE) is
> the supported check
4. Service("?someid"), maybe...
(Not "someid?", which would be the "natural" way to form a question, since
"?someid" makes it simple to check for the '?').
Klaus
