Re: [PATCH] Fix diff build on arm

[Bruno Haible]

Paul Eggert wrote:
> Is there any possibility that FreeBSD/arm can be fixed to have 64-bit
> long int?  This is what nearly every other platform does.

This is unrealistic. On all modern platforms, 'long int' has the size
of a machine word, i.e. of a integer register in the CPU. [*1]

If on a 32-bit platform, you use a 64-bit type as elementary type, you can
expect a slow-down of 20% to 30% for many applications. [*2]

The 64-bit time_t obviously has the purpose to avoid year-2038 problems.[3]
I guess that more platforms will follow the lead of FreeBSD, similarly to
the switch to a 64-bit off_t for which FreeBSD was also ahead of other
systems.

Bruno


[*1] The only known exceptions are m68k, where the machine word has 16 bit
     and a register has 32 bit, and MIPS in n32 ABI, where a machine word
     is 32 bit and a register has 64 bit.

[*2] This estimation is based on experiments with GNU clisp's -DWIDE flag.

[3] http://en.wikipedia.org/wiki/Year_2038_problem