The distinction is between permissive standards and restrictive standards.
The former specifies what must be done and the rest is optional; the latter
specifies what must be done and the rest is prohibited. One classic example
of this distinction is FORTRAN vs. Ada.
With respect to permissive standards, Microsoft has a reputation for
adding undocumented functions, but it is possible to document extensions.
(I recall that CDC did a particularly good job in their FORTRAN reference
manual: the extensions were typeset in a different font.)
I think the goal in writing SANE modules would be to adhere to the
"principle of least surprise" long known to the UNIX community, i.e.,
writing them so that they contain the fewest violations to popular
assumptions about their behavior and then clearly documenting those
violations.
Randolph Bentson
-- Source code, list archive, and docs: http://www.mostang.com/sane/ To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com