Message-ID: <1032@hou5e.UUCP>
Date: Thu, 15-Nov-84 16:53:41 EST
Article-I.D.: hou5e.1032
Posted: Thu Nov 15 16:53:41 1984
Date-Received: Fri, 16-Nov-84 06:54:36 EST
Organization: AT&T Information Systems Laboratories, Holmdel, NJ
Lines: 46
[...]
Can anyone out there help me with the following situation?
Mission Impossible:
Machine A is 68000-based, running on a non-standard OS (i.e.,
not publicly available).
Machine B is either an IBM-PC or an AT&T 6300.
Program X is compiled in C to run on both A and B. Obviously,
different compilers are being used.
A data area created by X will be copied byte-by-byte from A to B.
Because of the differences in the way compilers arrange data
variables, it is unlikey that X on B will be able to use the data
area correctly without some sort of conversion, even though X
uses identical structures on A and B.
Concerns:
Some of the conversion problems I foresee are-
- length compatibality for data types (char and short, sighned and
unsigned are used almost exclusively)
- alignment and consequent padding (for individual items and
structures, especially unions)
- byte ordering
- embedded pointers (pointing to within the data area)
The conversion program will run on B when it receives the data area from
A. This program can have access to both A's and B's symbol tables as
well as the source structure definitions.
Help requested:
Is this possible????
Does such, or similar, program exist - even if for different machines?
Are there C compilers for IBM-PC that give me control of alignment?
Have I missed worrying about some conversion detail?
Any help at all to guide me in my folly.
Don't ask me why I want to do this. (I won't do it if I can't.)
Thanks for any responses. Send mail please.
Irwin Walkenfeld
AT&T Information Systems
Holmdel, NJ 07733
....!hou5e!ihw