RTEMS patch sweep: *types.h

[Paul Schlie]

> Corinna Vinschen wrote:
>> On Feb 22 05:50, Paul Schlie wrote:
>> (don't know where it would belong, but do honestly believe, particularly
>>  for small targets with sparse data memory resources, there needs to be
>>  some way easy way for the programmer to determine that data structures
>>  need to ideally be declared as frugally as possible, while not needlessly
>>  compromising more ideal sizing for larger targets; something basically
>>  along the lines of where a target may designate themselves as having
>>  either a small <16KB, medium ~128KB, or large >1MB total data memory;
>>  thereby a target size aware function implementation may then at least
>>  attempt to size their allocations proportionally to available resources;
>>  but again it just provides the opportunity to do so, utilizing the
>>  information effectively is the harder part.)
>
> That's the whole purpose of defining abstract datatypes like off_t,
> size_t, uid_t, pid_t, isn't it?

Yes, they enable the type to be flexibly aliased; but to what, and under
what criteria/goals determines it's utility for any particular purpose.

One still needs the information which type/size decisions may be based on
itself to be formalized and reasonably universally abided by, as otherwise
all one's got is a bunch of knobs, not connected to anything, it would seem?