Re: [Devel] Re: Strange code in cff/t2parse.c?

[Toby J Sargeant]

On Thu, Dec 21, 2000 at 10:44:05AM -0500, Tom Kacvinsky wrote:
> Try replacing the code with this:
> 
> #define T2_REF( s, f) (FT_UInt)( offsetof( s, f ) )

My reading of the T2_REF macro is that it isn't the same as offsetof; it's
trying to facilitate some simple introspection on structs. the result of
T2_REF is dereferenced structure member which you can use as an argument to
sizeof/typeof etc.

I can't see any way of making this macro palatable to an AS400 compiler.
replacing 0 with NULL might help, as NULL needs to be an already cast
(to void *) pointer, if the compiler is going to whinge badly about casting
an integer to a pointer.

If, on the other hand, it triggers a compiler bug, then you might be a little
stuck. The only thing I can think of (and it's ugly, believe me...)
is to define something like:

#define BEFORE_T2_REF(s) static s _fake_ ## s ;

#define T2_REF(s,f) (_fake_ ## s ## . ## f)

Which will work as follows:

address@hidden $ cpp
#define BEFORE_T2_REF(s) static s _fake_ ## s ;
#define T2_REF(s,f) (_fake_ ## s ## . ## f)
BEFORE_T2_REF(Foo)
T2_REF(Foo,bar)
# 1 ""


static  Foo  _fake_Foo  ; 
(_fake_Foo.bar ) 

... The big problem is that now you need to use the BEFORE_T2_REF macro before
you use the T2_REF macro, and you have to do it in a location where you can
declare variables. YUCK. And doubly yuck when you use T2_REF inside another
macro.

Toby.