🚀 go-pugleaf

In article <1704@prlhp1.prl.philips.co.uk> glenn@prl.philips.co.uk
(Glenn Coombs) writes:

> I have just compiled ispell-2.0.02 and added ispell.el to my 
> GNU Emacs 18.57 and everything works fine.  However, I would 
> like to be able to use ispell on both Sun3 and Sun4 (Sparc)
> machines.  What I would like to know is can I use just one
> ispell.hash file in the ispell/lib directory ?

The short answer is "it depends on the hardware involved."  The medium
answer is "try it;  if ispell doesn't bomb on startup, then it
should work just fine."

BTW, a machine-independent hash file is one thing on my wish list for
future ispells.  You won't see it in Ispell V3, though.

> The ispell.hash file that I get when compiling on a Sparc is
> the same length as the one I get when compiling on a Sun3.  
> The two files are different though.  They are the same for
> the first ~280K which is just the list of words.  After that
> they appear to differ quite regularly up to the end of file.

The things that can affect ispell's hash file format are byte ordering
and structure padding.  If ispell can read the hash file header, then
the byte ordering isn't a problem.

The differences you are seeing are probably due to different fill
values in the padded part of the structure.  If you can ispell a
reasonably large document without a crash, then you're exercising the
whole hash file, and I think you can safely continue operating with
just one hash file.
-- 
	Geoff Kuenning   geoff@ITcorp.com   uunet!desint!geoff