NAT_FILE always required?

[Andrew Cagney]

> It looks as if I'm ready to get rid of NAT_FILE for *BSD/amd64.
> Except there's this funny bit of code in configure.in:
> 
> if test "${nativefile}" = ""; then
>         < Makefile \
>         sed -e '/^NATDEPFILES[[ 	]]*=.*\\$/,/[[^\\]]$/s/^/# /' \
>             -e '/^NATDEPFILES[[ 	]]*=/s/^/# /' \
>         | sed -e '/^\(NATDEPFILES[[ 	]]*[[+]]=[[ 	]]*\)/s//# \1/' \
> 	> Makefile.tem
> mv -f Makefile.tem Makefile
> fi
> 
> This code comments out the definition of NATDEPFILES if NAT_FILE isn't
> there.  What's the purpose of this code?

As you imply -> it was assuming that nativefile=="" implied no native 
support.   Grubbing around, there was code like:
	if host==target
		native=`sed magic ...`
so native= for cross targets was empty and the above would have disabled 
native support.  The predicate was moved between 5.2 and 5.3 and my 
fuzzy memory is because:

- it never really worked

- (possibly) it was realized that "one day soon" GDB would support both 
native and true cross debuggers in a single binary

> Can anybody see a reason why
> this code can't be deleted?

As far as I can tell, ever since 5.3 nativefile!="" has held, and hence 
the code was dead.

Andrew