In a msg on , Victor Loberger of 1:154/731
writes to Matthew Laird:
 ML>> do not work.  Unless Scott wants to rewrite NT, there's 
 ML>> nothing he can do about it.
 ML>> The only partial solution is to use a mailer called Argus, 
 ML>> then at  ML> least your mailer can be 32-bit even if your 
 ML>> board can not.
 VL>         I wonder how complicated the routine is that allows 
 VL> Argus/NT to pass an open com port to Maximus for DOS?  I would 
 VL> think that the same principle that allows Argus, a 32-bit 
 VL> program, to call Maximus/DOS, a 16-bit program, would work for 
 VL> Maximus/NT calling DOS doors.  Perhaps it'd be just a "simple" 
 VL> re-write.  8-)
The problem is that Win95 is COM-port stupid.  The TCP/IP protocols in Win95 
are screwed up enough that the passing of a COM port within the kernel is all 
but impossible...the only reason I can't say it _is_ impossible is more out 
of hope than any tangible materials.
The compiled kernel for Win95 simply isn't sophisticated enough to run 
between 16/32-bit software combinations *and* handle TCP/IP manipulations.
WinNT is _closer_, and the server version of it is closer _still_...but it 
isn't completely adequate and still requires you (as operator) to sacrifice 
what you want for what NT can handle.
There is a possibility, I am told, that the Win98 rewrite *may* have partly 
addressed this, and opened the door for third-party software to allow it to 
pass 'hot' COM ports much more easily, but I won't believe it until I can see 
for myself.
If you know _anyone_ on the Win98 Beta teams, push them to ask MS to address 
the issue.  Force the issue, and be public about it.
-zf-
--- msgedsq 2.1
---------------