>>>> "SJT" == Stephen J Turnbull
<turnbull(a)sk.tsukuba.ac.jp> writes:
SJT> I am not able to assess the likelihood of that so I went with slightly
SJT> safer code that will increase its advantage as NAS >= 1.4 becomes
SJT> prevalent. If you judge that that risk is very low, you can munge the
SJT> *_ENDIAN macros freely. But such a mistake cannot possibly be caught
SJT> by the compiler. My code can't catch the mistake either for NAS <=
SJT> 1.2p5, but code using *_ENDIAN will work as expected for NAS >= 1.4.
In practice, I think this is a non-risk, since the first compilation
on a non-glibc platform would get a trivial-to-fix compile error. The
really nasty bugs are the ones the compiler can't find.
SJT> I agree that your code is simpler, and should work. If you say you
SJT> want it that way, I'll write it and test it (I can test XEmacs 21.1
SJT> and 21.2, NAS 1.2p5 and 1.4 without too much trouble), then submit
SJT> patch(es).
Naturally I prefer my code, but I will accept either patch, as long as
it works.
Stephen, thanks for the doing the hard work behind the scenes, like
testing.
Martin