Re: ediff failure if diff doesn't exist

[Jesper Harder]

Richard Stallman <address@hidden> writes:

>     I wasn't really proposing a solution, just pointing to some places
>     that might fail to work as intended, because the code doesn't take the
>     possibility of signalling an error into account.
>
> Would you like to look at those places to see if there are real
> problems?  For each place, the question is: is there a real case where
> the program would not be found, and where signalling a Lisp error
> leads to bad results?
>
> I would expect that in most of these cases a Lisp error and its error
> message would be an ok outcome.  But you may find some where it is
> desirable to arrange some other behavior.

I looked through all the uses of `call-process', and found that
`ediff-test-utility' was the only place which I thought important
enough to change now.