Reasons for 64-bTT & DHCP import

[Marcel Moolenaar]

On Sun, Mar 14, 2004 at 09:20:37PM +0100, Roderick van Domburg wrote:
> 
> Probably a Real Stupid question, but why hasn't __time_t been a __int64_t
> from the beginning? I'm asking out of sheer curiosity and insight in
> development.

I don't know the real reasons, if there were any reasons at all, but
in general things like this are most of the time the result of chance.
Some developer has to have the right amount of awareness in order to
raise the issue. The right amount of awareness in this case includes:
o  The fact that __time_t is 32-bits. We very likely know the answer
   when it matters, but at other times we tend to forget about it.
o  The fact that with 32-bits, Y2K resurfaces as Y2038. Again, when
   explicitly confronted with it, we can figure it out but we won't
   wory about it every day.
o  The fact that 40 years is soon. This typically is not a concern at
   all for the developers who are working on the initial port. Getting
   it to work is priority 1.

And so on. The result is that before you get to a point where you can
worry about it, you have snapshots, ports and users. Even if you then
realize that a 64-bit time_t would be ideal, you very likely don't
bother. Especially if no other platforms have it.

> Also, I am curious why we're not importing ISC DHCP 3 in the base system.
> Skimming over the ISC license, I don't see any direct issues. Once again:
> don't mean to bash, just curious.

Someone has to feel the need. If it's the maintainer who feels the need,
you don't anything else. Otherwise you need enough voices to make the
maintainer feel the need to do it. Sometimes there are good reasons to
not upgrade. I don't know if that applies here. I don't think so.

FYI,

-- 
 Marcel Moolenaar	  USPA: A-39004		 marcel at xcllnt.net