-
The internal data representations
are different.
Cross-architecture
access is used when the client session and the server session are
running on machines that internally represent data differently due
to hardware differences between two machines. For example, IBM mainframe
machines represent floating-point numbers differently than computers
that use Intel CPUs. The code pages that are used to represent character
data also vary. For example, EBCDIC and ASCII are two major character-encoding
methods.
-
The C-language compilers that are
used are different.
Different operating
environments and C-language compilers also cause differences in data
representation due to the varied alignment requirements of aggregate
data types, such as the inter-element padding in a specific C structure.
Also, two compilers for the same type of CPU might implement simple
data types that have different lengths.
-
Operating environments are the
same, but the machine architectures are different.
It might not always
be obvious when the cross-architecture feature of
SAS/SHARE is required.
z/OS to CMS access
is not cross-architecture because the underlying representation of
data in the two operating environments is the same. However, sharing
data between OpenVMS for VAX and OpenVMS for AXP uses cross-architecture
access because data is represented differently on the Digital VAX
and Alpha AXP architectures, even when the same operating environment
is used.
For complete details
about architectural compatibility, see Identical Architectural Groups.
Note: Although neither CMS nor
OpenVMS VAX are supported in later versions of SAS, they are included
in the preceding example for backward compatibility.