Re: [Hamlib-developer] Year 2038 issues

[Greg T. <gd...@le...>]

"C. van Wüllen" <van...@ch...> writes:

> I believe that the assumption
>
> sizeof(time_t) = sizeof(long) = sizeof(void *)
>
> will hold for the forseeable future just for supporting
> code of lazy programmers (there is no *need* to have
> the same size for these), so I guess there will be 
> *something* done for 32-bit systems, e.g. internally
> referencing this to 1. Jan. 2000, but this remains to
> be seen.

That assumption already is not true and has been not true for well over
10 years.  C99/POSIX does not guarantee anything close to that,
especially time_t, which is only required to be an integral type:
  https://pubs.opengroup.org/onlinepubs/9699919799/basedefs/sys_types.h.html

Therefore any code which assumes anything about which integral type has
been buggy from the day it was written.