Re: Removing SunOS/68k compatibility before 8.0

On Fri, 29 Dec 2017, Frank Wille wrote:
>> Indeed. The Sun-3 MMU was custom made by Sun and honors only 28 bits of
>> VA from a given address (the "context" is global).
> Ahh! Thanks. That's what I tried to remember.

Wow, only a team of uber geeks could dig up a fact like that and elucidate 
it as the cause. Nicely done.

> Indeed. Having more memory has become more important during the last 
> years. I'm not even sure if gcc can still be built on NetBSD/68k 
> hardware.

I don't know about GCC, but I know the last time I tried to compile 
Blender on a PC with 2GB of RAM (and 10G of swap) it kept hanging. At one 
point I let it run on an i7 box for four days. It still never worked until 
I built the package on a box with 16G of RAM. I was watching vmstat the 
whole time and at one point the VSZ was around 12G for the participating 
compiler processes. That was the first time I ever saw an example that 
dramatic.

It's cool to have newer compilers, but I'm sad that they are generally so 
unfriendly to older hardware (I know I know, march of progress). Same with 
C99 vs C11. C99 was a bunch of awesome long-time-coming changes I thought 
were well thought out. C11 uhh, well, Mom said if I couldn't say something 
nice....

-Swift