int8, etc.

[Karl Heuer]

In article <27.266bddfd at spanky.mgi.com> cepek at spanky.mgi.com writes:
>[What's wrong with a typedef `int8'?]

People who've used PL/I can probably give first-hand stories about why
explicit sizes are a bad idea.

If you were on a machine with 36-bit ints, 18-bit shorts, and 9-bit bytes,
what typedefs would you have?  If you define only `int9', `int18', and
`int36', then none of your code using `int8' will compile.  So I presume you
actually maintain a file of bitsize-typedefs `intNN' for all reasonable values
of NN.  But then why did you specify `int8' for the boolean type instead of
the more logical `int1'?  (Probable answer: since C requires that `char' be at
least 8 bits anyway, there's no need to distinguish anything smaller.  This
isn't entirely true, since an implementation on a bit-addressible machine
might provide a `short char' extension, and one should be able to take
advantage of that.)

Karl W. Z. Heuer (karl at ima.ima.isc.com or harvard!ima!karl), The Walking Lint