Re: Extensible Enummerated types FW: When will next Ada revision be? (83, 95, ?)

["Nick Roberts" <nickroberts@callnetuk.com>]

Robert Leif's example seems to lead to a slightly hairier alternative (to
the simple proposal I made, unwittingly on behalf of Jeff Carter
apparently). Here, an enumerated type declaration, either a standard one,
e.g.:

   type Prefix_Type is (None, Mr, Ms, Miss, Mrs, Dr, Prof, Rev);

or a special one, e.g.:

   type Prefix_Type is (None, Mr, Ms, Miss, Mrs, Dr, Prof, Rev) with others;

declares a type which actually has more values (e.g. 256 or 65536) than are
specified. A size representation clause could be used to ensure a big enough
number are available. The first named subtype Prefix_Type denotes only the
declared values, as if the following had been declared:

   type {anon} is (None, Mr, Ms, Miss, Mrs, Dr, Prof, Rev);
   subtype Prefix_Type is {anon} range None..Rev;

so most things work as for current Ada. The other values have no designators
(but, I suppose, can still be referenced by using Val). A special kind of
declaration can then declare identifiers (or character literals) for these
values (starting at the first vacant position), e.g.:

   subtype New_Prefix_Type is Prefix_Type with (Guru, Ustad);

The new subtype New_Prefix_Type denotes the full new range None..Ustad. Note
how this allows multiple names to be declared for the same values:

  subtype My_Prefix_Type is Prefix_Type with (Master, Maestro);

It may be that an enumeration representation clause must be disallowed for
this kind of enumeration type. It may also be that the [Wide_]Image and
[Wide_]Value attributes must also be disallowed for such a type and any of
its subtypes.

Anyone saluting this alternative?

--
Nick Roberts
http://www.AdaOS.org