#: 20256 S12/OS9/68000 (OSK)
    27-Aug-94  01:33:05
Sb: #screen & ispell
Fm: Ken Scales 74646,2237
To: Bob van der Poel 76510,2203 (X)

         <... slowly raising head above the edge of the abyss,
                  and cautiously peering around...>

Is it safe to come out yet?

Seriously though, sorry guys -- I'm in the middle of major crises at work
(today my boss and I continued our discussions of the possibility of disbanding
the group I have spent the last year building up... in spite of the major
successes we have had...)

 > Okay...I have ispell working under screen. It appears that the problem is
 > as follows:
 >
 > 1. I believe that ispell changes the EOR setting in the path desc.
 > Normally this is a $0d. I don't know that ispell changes it to, but I
 > assume a 0.

I don't think that 'ispell' monkeys with the EOR value, but wouldn't discount
that possibility.  It uses a lot of calls to 'ioctl()' in blarslib.l, and it
would take a while to re-trace through them all.

 > 2. ptyman's output routine checks characters against the EOF setting in
 > the path desc. If it matches and autolf is needed, a LF is output.
 >
 > What I have done is to modify the test to see if the character==EOR or if
 > it is a $0d. However, this is probably not the way to do it properly.
 > Before I update my posting...anyone know how scf handles this? I see a few
 > ways....
 > 1. Test for EOR or a CR,
 >
 > 2. Test for a CR only if EOR is set to 0,

From subsequent postings, I see that you found in The OS-9 Guru book that
PD_ALF is specific to CR, and doesn't relate to the PD_EOR value (p330-331).
Whew!  Never would have guessed that!  That book is certainly a "must-have"!
Lots of gems in there.

Are you comfortable with this fix?  Or do you feel that 'ispell' still needs
investigation?

 > 3. Force Ken to recompile ispell.

  No problem there...  In fact, 'ispell' is in serious need of a recompile
(actually, re-link) because the distributed version used an early version of
the 'cgfx.l' library, and therefore won't work with the 68340-based MM/1s.  (I
have already sent a re-linked version out to a couple of users, for trial.)

 > BTW, Steve, please cross post this over to Ken's arena.

Got it.  Thanks, Steve.
 --------------------------------------------------------------------------
 Ken Scales   Delphi:KSCALES   Internet:kscales@delphi.com   CIS:74646,2237

There are 2 Replies.