Re: Ada advocacy

["Randy Brukardt" <randy@rrsoftware.com>]

"Jeffrey R. Carter" <spam.jrcarter.not@spam.acm.org> wrote in message 
news:kvofa4$dnc$1@adenine.netfront.net...
> On 08/29/2013 08:28 AM, Robert A Duff wrote:
>>
>>   Also, you can do
>> multi-way waits with accept statements but not entry calls.
>
> You can do
>
> select
>    P1.E1; -- An entry call.
> then abort
>    P2.E4; -- Another entry call.
> end select;
>
> to wait for 1 of 2 entry calls, and you can nest another ATC in there to 
> do 3, and so on to whatever depth you like, so I don't really see a reason 
> for the language not to provide a multi-way select for entry calls.

The idea was extensively studied during the Ada 9x process, and the 
conclusion was that it would be very expensive to implement. (I forget why, 
but there should be papers on the topic somewhere in the Ada 9x documents.) 
Moreover, the hard real-time users were the ones doing the lobbying for the 
feature, and it was abundantly clear that they would not be able to use it 
because of latency issues. (It certainly would have been excluded from 
Ravenscar.)

Your workaround is also insanely expensive, but at least that is obvious 
syntactically ("abort" should "feel" expensive to an Ada programmer in a way 
that an entry call should not.)

                                    Randy.