Re: New function xpalloc in module xalloc

[Paul Eggert]

On 4/6/21 12:23 PM, Marc Nieper-Wißkirchen wrote:
> Where is the flaw in my reasoning?

Oh, you're right; any nonnegative signed integer value will fit into 
uintmax_t. (Perhaps this wasn't always true in older standards, but it's 
true of the recent C standard.)

So that cast should work. Still, I wouldn't use a cast; I'd assign the 
integer to an uintmax_t local value, and use that.

And anyway, for this particular case widening to uintmax_t is not 
necessary; it complicates the code and can make it less efficient on 
some platforms by forcing an unnecessary conversion at runtime. So let's 
not do that.