Subj : Re: Non-strictly-conforming and unspecified versus undefined behavior
To   : comp.programming.threads,comp.std.c
From : Douglas A. Gwyn
Date : Tue Feb 22 2005 02:32 pm

James Kuyper wrote:
> Douglas A. Gwyn wrote:
> > not all programs need be s.c.  There is also the
> > useful category of a (plain) "conforming" program,
> I'll agree that some (but not all) programs in that category are useful,
> but calling the category itself useful seems quite a stretch. What's the
> category useful for? Useful categories have both a significant number of
> members, and a significant number of non-members; the category of
> "conforming C programs" has too few non-members to be very useful.

I didn't mean that it was useful from a *conformance* point of view.
In fact I've said that it wasn't, if not nearby then certainly in
previous postings.  I mean that many useful programs are not
strictly conforming.  Strict conformance is important for ensuring
a high degree of portability, but obviously that isn't relevant
for intrinsically nonportable access to platform-specific features.
It is wise to isolate such nonportable code, however, in order to
minimize the work and expense of porting the app.

.