> Just noticed this source commit. 

[ one of my changes that uses gcc's inline feature for the signal mask 
  manipulation functions. ]

> Doesn't this end up sacrificing the ability of users to drop in any
> GCC they pick up whenever they like, or use other
> compilers/languages, and doesn't it also lower portability? Was
> signal mask initialization time a real problem?

Using gcc's inline extension is conditionalized on __GNUC__, so it
won't effect other compilers.  There is similar code in stdio.h that
worked when we were using gcc 1.37, so there shouldn't be any problems
with old versions of gcc either.

When/if the next ANSI C standard includes inline (many people on
comp.std.c seem to think it's one of the more likely changes), we can
change the conditional to something like:

	#if __STDC__ == 2 || defined(__GNUC__) || ....

I'm not worried about other languages, they'll likely have their own
header files (or whatever is equivalent in that language).


With today's cpu's, this change doesn't make that impressive of a
change.  But why make 1 + n function calls when it can be replaced 
by a single assignment?

	--jtc