Feedback and comments on ARM proposal

Joshua Bloch jjb at google.com
Sat Mar 21 11:06:08 PDT 2009


On Sat, Mar 21, 2009 at 9:54 AM, Tim Peierls <tim at peierls.net> wrote:

> On Sat, Mar 21, 2009 at 12:08 PM, Joshua Bloch <jjb at google.com> wrote:
>
>> What about the magic marker interface (if you'll pardon my wordplay)? This
>> may be difficult to specify.
>>
>
> There are spec issues to resolve with the magic package approach, too: What
> do you do if an interface type extends two or more interfaces from
> java.lang.auto? Probably not hard to find reasonable answers, but you could
> say the same about the magic marker approach.
>

I do think it's (a bit) harder with the magic marker approach, as you can
have multiple paths to one "root" magic marker interface, as per previous
mail.  The package would contain a bunch of interfaces each of which has
exactly one parameterless method.  (Each such method would probably have a
name distinct form all of the others, but that's not a hard requirement.)
 If you implement two or more such interfaces, you lose.  That's a pretty
simple rule.

      Happy weekend,

      Josh



More information about the coin-dev mailing list