[Cfp-interest] format - type names issue
Jim Thomas
jwthomas at cup.hp.com
Thu Feb 4 08:26:27 PST 2010
Maybe we're making an unnecessary problem for ourselves. C99 has
specification to define complex and imaginary types in terms of real
types without the need of additional tokens. For example, following the
C model, if _Float256 is a real type, then _Float256 _Complex would be
the complex types whose associated real type is _Float256. (As Fred
notes, _Complex _Float256 would be allowed too.) I believe we got on the
track of having a single token for complex and imaginary types because
of a belief that C++ was unhappy with the C complex and imaginary types
not being a single token. But if that's an issue, it's already an issue
and implementations have dealt with it. Also, it would be better to
follow the existing C model because the specification would fit more
easily into the C standard and be easier to maintain.
-Jim
Jim Thomas wrote:
> Fred J. Tydeman wrote:
>
>> On Wed, 03 Feb 2010 11:30:45 -0800, Jim Thomas wrote:
>>
>>
>>
>>> Given that the existing names for imaginary and complex types have
>>> the associated real type first (e.g., float complex), it seems more
>>> consistent with C style to place _Complex and _Imaginary last.
>>>
>>
>>
>> Both
>> float complex
>> complex float
>> are valid C99.
>>
>> As per 6.7.2 Type specifiers may occur in any order.
>>
>>
> All instances in the C standard have _Complex, _Imaginary, complex, or
> imaginary last.
>
>> ---
>> Fred J. Tydeman Tydeman Consulting
>> tydeman at tybor.com Testing, numerics, programming
>> +1 (775) 358-9748 Vice-chair of PL22.11 (ANSI "C")
>> Sample C99+FPCE tests: http://www.tybor.com
>> Savers sleep well, investors eat well, spenders work forever.
>>
>> _______________________________________________
>> Cfp-interest mailing list
>> Cfp-interest at oakapple.net
>> http://mailman.oakapple.net/mailman/listinfo/cfp-interest
>>
>>
>>
>>
>
>
More information about the Cfp-interest
mailing list