Refactor fileptr_type_node handling

[Marc Glisse]

On Thu, 22 Jun 2017, Richard Biener wrote:

> On Thu, Jun 22, 2017 at 10:46 AM, Marc Glisse <marc.glisse@inria.fr> wrote:
>> Hello,
>>
>> I was asked to handle (const) fenv_t and fexcept_t the same way as FILE and
>> const tm. Since these have special handling in quite a few places, it seems
>> necessary to make their support a bit more generic first. If I didn't mess
>> up, this patch should not change anything.
>
> It looks like a good thing to do before extending the set of types
> handled this way.
>
> The only thing I dislike is the name 'predeclared', isn't it actually
> forward declared or simply declared (as opposed to defined)?
> So - declared_ptr_types maybe?
>
> Sorry for bikeshedding... - the patch itself looks good to me.

I had "predefined_types" at some point, which turned into 
"predeclared_types", I am fine with a different name.

This kind of declaration does not seem quite equivalent to a declaration 
in user code, but the difference is minimal. IIRC the comments do describe 
this as some kind of temporary declaration until we see a real one.

I am ok with declared_ptr_types, but it may not convey the idea that this 
is not a list of all the pointer types in the program, just a few standard 
ones.

standard_ptr_types? standard_structptr_types?

-- 
Marc Glisse