[postgis-devel] size_t: width and signedness of counters

[Sandro Santilli]

On Thu, Dec 28, 2017 at 09:18:05AM -0500, Daniel Baston wrote:

> I understand that the size of the platform-dependent size_t
> should always be >= uint32_t, but I can't see a reason to prefer it.

Unless you have profiling results showing use of size_t being
significantly faster than uint32_t on a 64bit architecture
(I'm assuming that was your concern) then I'd go with uint32_t too,
for the specific case in which the source is an uint32_t...

--strk;