> Your test cases work fine on the i386.  Anyone with another big endian 
> machine care to test this out?  I want to know if this is a hp, or big
> endian problem, as it is probably in the regular expression routines 
> themselves.

Peter G. confirmed that the bug is present on Sparc's too.

At first glance, It seems that this may be a problem with the regex code 
not handling 64bit off_t's correctly (One of it's fundemental types, 
regoff_t, is an off_t).

A quick solution may be to define regoff_t to be a 32 bit int, perhaps a 
ssize_t (Actually, that seems to be a more correct definition anyway, 
we're talking about memory offsets, not disk offsets).

I don't know if it will work, but it's worth a try.  If someone want's to 
give me an account on a big endian machine, I can try it myself.

	--jtc

------------------------------------------------------------------------------