Re: [Nmh-workers] Loop support in mh-format?

[Ken Hornstein]

>I think the %[...%] is by “JLR”?  do_loop() sets up an FT_DONE that's 1
>rather than the normal 0 and then the end of fmt_scan() at finished:
>checks if it's true and if so returns the next format instead of NULL,
>but only #ifdef JLR.  Without that defined it appears to be ignored so
>formatting stops at the first %].  Just from reading, not running, so
>could be wrong.  Didn't spot a caller that bothered to check the return
>for non-NULL and call again...

Hm, I think I'm following you ... the idea was that if fmt_scan()
returned non-NULL, you'd call fmt_scan() again and you'd pick up
after the FT_DONE instruction?  And that would be another round of
the loop.  But how would you exit the loop, that's what I don't
understand.  It occurs to me that you could construct a loop with
the existing instruction set, so I'm not sure why the idea was to
overload FT_DONE.  I'm not going to bother to fully implement the
loop construct unless someone has a credible reason for wanting it
(and comes up with a workable proposal to the syntax of it).

As a side note ... this was my first note that I replied to where I used
my new replcomps that does "From" address selection, and it worked great!
But Ralph's note came in as base64-encoded UTF-8 which repl deals poorly
with ... we've got more work to do, sadly.

--Ken