In article <t53d8qcbrvb.fsf@rover.cygnus.com> marc@cygnus.com (Marc Horowitz) writes:
>
>The code in NetBSD will not drop into any other unix derivative
>cleanly.  It will require some work, some of which may be nontrivial.
>Deansification is relatively easy; there are tools to do it for you.

And then you cannot share your cvs tree. I am currently able to build
all the NetBSD applications I use using a set of custom headers and a
library derived from libc to add the missing functionality.

>gcc is a better compiler than then SunOS native one.  And, as above,
>you can use unprotoize if you have to use the vendor compiler.

Ahem... Better when it produces correct code and it does not always.

>No, it simply means you need to have an ANSI compiler.  ANSI is 8
>years old, which is approximately forever in computer time.  The
>chance you will not have an ANSI compiler is very small.  IMHO, the
>risk is worth the benefit.  gcc has already been ported to practically
>everything which it can run on, and new hardware comes with ANSI
>compilers.
>
>I think we should optimize our style for NetBSD's use.  If this makes
>using the code elsewhere substantially more difficult in a lot of
>cases, we should consider that, too.  However, making our code better
>at the cost of a little more work for a few people seems like a good
>tradeoff to make.

I am fine with that if it will buy us something... I am not convinced that
using __P is such a difficult thing.

christos