>> 	which change between 1.4.x and 1.4[A-Z] affects the behavior?
>> 	the (very simple) patch is attached.
>Signal handling has changed, for one...
>Unless I'm really confused.

	The thing puzzles me is that the configure.in script allows 1.4.x
	for certain architectures, while disallowing it for i386.

	pattern match rule against $host_cpu-$host_os says like this:
	was signal handling different across platforms when 1.4 gets branched?

alpha-*-netbsd1.4*) 
arm32-*-netbsd1.4*)
sparc-*-netbsd1.4*)
i386-*-netbsd1.4[A-Z])
m68*-*-netbsd1.4*)

itojun