>I had a terrible time compiling xemacs. The standard netbsd distribution did
>not build. 

FSF Emacs 20.3 tries to put a backtick-expansion (ALL_LDFLAGS) inside
another back-ticked expression. Is that what you saw?  There's a
trivial fix: patch the inner backquote with what we know its going to
expand to.


>I had to hack unexecelf.c (?) to look for .sbss instead of .bss,
>I believe, and it built a working non-stripped executable. However,
>stripping the executable destroyed it, so now I am running the non-stripped
>version. 

Hm.  The comments in unexelf.c say exactly why that would break if
you're doing a `strip'.  I really wonder what happeins if you replace
all those tests for sony-news with a HAVE_SBBS, and turn on HAVE_SBSS
for both netbsd/mips and sony news 5.x.  That might evne work....

>I am not impressed by emacs' brittle "unexec" build scheme, either. Blech.

Yeah, but the alternatives all run into the exact same ELF problems,
or are derived from the VMS hack, which has its own lossage.