unusual storage representations - looking for info

[seas.gwu.edu!mfeldman@uunet.uu.net (Michael Feldman)]

Perhaps some of you out there can help me out. As we know, the LRM
does not "second-guess" storage representations, in the main. Integer
ranges, Float precisions, array mappings, etc., are all implementation
defined.

I'm looking for examples of Ada compilers and targets that use unusual 
storage mappings. Specifically:

1. are there any compilers for which the predefined numeric types do _not_
occupy multiple-of-8 storage? For example, the CDC Cybers have a 60-bit
word. Does predefined Integer use the full range of the 60 bits? Also,
as I recall, the old Honeywell architecture used a 36-bit word. So
what is the range of Integer on these? Which other targets don't use 
multiple-of-8-bit predefined ranges?

2. are there any compilers in which multidimensional arrays are _not_
stored in row-major form? Clearly column-major is not only legal but might
be useful in a compiler optimized for pragma-interfacing to Fortran.
Wilder still, it is certainly not gospel that an array even has to be
stored contiguously.

Anything else worthy of note or unusual?

Please reply by e-mail; I'll post a summary if I get good stuff.

Thanks!

Mike Feldman
------------------------------------------------------------------------
Michael B. Feldman -  co-chair, SIGAda Education Committee
Professor, Dept. of Electrical Engineering and Computer Science
The George Washington University -  Washington, DC 20052 USA
202-994-5253 (voice) - 202-994-5296 (fax) - mfeldman@seas.gwu.edu (Internet)
"Pork is what those other guys get from the Government."
------------------------------------------------------------------------