Re: size_t & unsigned long

[Michael Livshin]

Dirk Herrmann <address@hidden> writes:

> IMO, we should at least _try_ to be type safe, and also
> use the most restrictive possible types, like prefering unsigned types for
> values that can't be negative.

total agreement here.

> Another point:  We have a type scm_sizet (which, IMO, should better be
> renamed to scm_size_t whenver we fell masochist enough for such a change),
> the purpose of which does not seem to be too clearly defined.  In which
> situations should we use this type?  What is the difference to
> size_t?

it seems to be a compatibility wrapper: on systems that have `size_t'
it is defined as such, on other systems it is defined as `unsigned
long'.

since Guile requires ANSI C for quite some time now, `scm_sizet'
should just be deprecated.  as far as Guile cares, `size_t' is
available everywhere.

so how about this: I'll comb the source and change it to use `size_t'
and `ptrdiff_t' in all the appropriate places.  I'll also add
`scm_sizet2num', `scm_ptrdifft2num', `scm_num2sizet' &
`scm_num2ptrdiff'.

comments?

do we still want to have `scm_size_t' and `scm_ptrdiff_t' typedefs,
just in case?

-- 
Hit the philistines three times over the head with the Elisp reference manual.
                -- Michael A. Petonic