bitmask lookup table creation

[Kaleb Keithley]

Don't like to follow up my own article, but:

In article I wrote:
>I want to create the following bitmask lookup table:
>    unsigned short masks[] = {
>        0x0000, 0x0001, 0x0003, 0x0007,
>        0x000F, 0x001F, 0x003F, 0x007F,
>        0x00FF, 0x01FF, 0x03FF, 0x07FF,
>        0x0FFF, 0x1FFF, 0x3FFF, 0x7FFF 
>    };
>
>in a machine independent way.
>
>In /usr/include/values.h, the following are #defined:
>#define BITS(type)  (BITSPERBYTE * (int)sizeof(type)
>#define HIBITS  ((short)(1 << BITS(short) - 1))
>#define MAXSHORT    ((short)~HIBITS)
>
>which are 0x8000 and 0x7FFF respectively on my machine, which happens to
>be a Sun Sparc.
>
>It is nearly trivial to do it at run-time, and the above definitely works
>on my Sun.  Does anyone have any neat trick to generate this table at 
>compile time in such a way as to be completely machine independent?

Well, so far I got two answers, one said the above was correct, the other
told me to look at X Window Server code.

I don't want to belittle the person who told me the above was correct, but...
I looked at the MIT server code, and found that it makes two distinctions: 
If you have a VAX or a 386 then BITMAP_BIT_ORDER is defined as LSBFirst, 
otherwise it's MSBFirst.  Then there are two versions of the table, much 
like the one I have above, one for MSBFirst, one for LSBFirst.  I could do 
this, and in fact, probably will, but, the question still remains; Is there 
a way, using the three defined macros, to generate the correct table at 
compile time?


kaleb at thyme.jpl.nasa.gov            Jet Propeller Labs
Kaleb Keithley

"So that's what an invisible barrier looks like"