[Info-vax] Process memory settings

[John Reagan]

"Howard S Shubs" <howard at shubs.net> wrote in message 
news:howard-5D24CA.01314714082009 at news.newsguy.com...
> In article <_KqdnZnGt6U9QhnXnZ2dnUVZ_jmdnZ2d at earthlink.com>,
> "John Reagan" <johnrreagan at earthlink.net> wrote:
>
>> Well, it won't be stable after we take a roto-tiller to the code.   We'd
>> have to expand fields in the compiler when we build it (and hope we get 
>> the
>> sign-extensions correct).  You can't be suggesting we now make all our
>> fields in internal data structures be run-time sized with run-time
>> offsets?!?  Yeah, that'll be stable... not...
>
> When are y'all planning to make the 64-bit change, then?  Never?  I
> tried to use 64-bit memory a week ago in C, and it was all good, until I
> needed the Fortran I/O library, at which point Strange Things happened.
> They stopped happening when I went back to 32-bit pointers.
>

That's a totally different question.  I'm talking about the compilers 
themselves being 32-bit applications.

User programs compiled BY those 32-bit compilers can certainly use 64-bit 
pointers if the language supports it.  You should be able to use 64-bit 
pointers in C, C++, Fortran, Macro-32, Pascal, and COBOL (for as much as 
COBOL supports ANY pointer).

If you had an F90 program that used 64-bit pointers but got strange results, 
then somethings broken (either in your code, the F90 compiler, or the F90 
RTL).  Read the Fortran release notes to see if you bumped into some known 
restriction.

John