Re: [Nmh-workers] Weird behavior with non-ascii code in headers

[Ken Hornstein]

>I agree with Ralph:  ripple the error back up (with returns,
>not longjmps :-) and don't even open the editor.  In this
>case, it looks like that might not be too difficult:
>fmt_scan() currently doesn't return anything useful.  Fix
>that, and have its callers check the return value.

One minor nit ... so, what fmt_scan() returns right now is a pointer
to the last format instruction executed, which is currently always NULL.
This is a vestige of something John Romine was working on, which you can
get more info about here here:

  http://lists.nongnu.org/archive/html/nmh-workers/2010-12/msg00050.html
  http://lists.nongnu.org/archive/html/nmh-workers/2012-01/msg00144.html

This never got completely implemented, and thinking more about it, I
think if I was going to implement per-format column headings I'd just
have a separate format string deal with it, rather than put it as part
of the normal scan format string.  So I think it's safe to return something
more useful if we decide that's the route to go.

--Ken