> From kenh@cmf.nrl.navy.mil Thu May 21 21:48:29 1998
> To: Ross Harvey <ross@teraflop.com>
> cc: port-alpha@netbsd.org
> Subject: Re: Fixing pointer-to-int casts 
> X-Face: "Evs"_GpJ]],xS)b$T2#V&{KfP_i2`TlPrY$Iv9+TQ!6+`~+l)#7I)0xr1>4hfd{#0B4
> 	WIn3jU;bql;{2Uq%zw5bF4?%F&&j8@KaT?#vBGk}u07<+6/`.F-3_GA@6Bq5gN9\+s;_d
> 	gD\SW #]iN_U0 KUmOR.P<|um5yP<ea#^"SJK;C*}fMI;Mv(aiO2z~9n.w?@\>kEpSD@*e`
> Date: Fri, 22 May 1998 00:48:07 -0400
> From: Ken Hornstein <kenh@cmf.nrl.navy.mil>
>
> >usually better to use void * or char * or some_other_type *. Remember
> >that int is always 32 bits on unix, but long is 32 or 64. There is
>
> Obviously you've never programmed on a Cray :-)  (int is 64 bits, short is
> 32 bits of integer precision but takes 64 bits of memory).  Kerberos
> was hell to port to it.

Right, good points. I am actually aware of several rare implementations
of IPL64, another one was the Hal workstations. The many problems
coupled with the fact that "short" must be assigned either 16 or 32,
thus leaving the _other_ one with no standard type is what made LP64
universally accepted in the mainstream for migration from the standard
32-bit architectures.

Ross