Columbia Kermit/kermit.zip/newsgroups/misc.19960715-19961006/[e-mail] _Thu Oct 3 23:35:55 1996.msg

Internet Message Format | 2020-01-01 | 4KB

Return-Path: [e-mail]
Received: from newsmaster.cc.columbia.edu (newsmaster.cc.columbia.edu [128.59.35.30]) by watsun.cc.columbia.edu (8.7.5/8.7.3) with ESMTP id XAA10360 for <[e-mail]>; Thu, 3 Oct 1996 23:35:54 -0400 (EDT)
Received: (from news@localhost) by newsmaster.cc.columbia.edu (8.7.5/8.7.3) id XAA10948 for kermit.misc@watsun; Thu, 3 Oct 1996 23:35:53 -0400 (EDT)
Path: news.columbia.edu!sol.ctr.columbia.edu!news.msfc.nasa.gov!news.sgi.com!news.mathworks.com!howland.erols.net!cs.utexas.edu!news.cs.utah.edu!cc.usu.edu!jrd
From: [e-mail] (Joe Doupnik)
Newsgroups: comp.protocols.kermit.misc
Subject: Re: 2nd post: Chars lost by DOS Ker under OS/2 during macros
Message-ID: <[e-mail]>
Date: 3 Oct 96 20:41:42 MDT
References: <[e-mail]>
Organization: Utah State University
Lines: 53

In article <[e-mail]>, "Space Computer Corp." <[e-mail]> writes:
> No response to the first post of this, but I'm still looking for
> suggestions for a solution.  Can anyone relate to this? --
>
> I am running DOS Kermit 3.14, Patch 9, under OS/2 Warp with a FOSSIL
> connection using Ray Gwinn's SIO programs.  This works fine for
> "normal" usage, but while Kermit is executing the "Product" macro it
> appears that the input buffer overflows, and characters are lost.
> (This procedure worked perfectly when DOS Kermit was run under actual
> DOS, rather than a DOS window under OS/2.)  More specifically, the
> host is a VAX/VMS system, and both the host and Kermit are set for
> xon/xoff flow control.  The host sends the escape sequence to cause
> Kermit to execute the "Product" macro, which in this case is just "log
> session file.dat", and follows this immediately with a data stream.
> The beginning of the data stream is captured in file.dat, then there
> is a big gap, then the rest of the data stream is captured.  A similar
> thing happens when the second "Product" macro is executed.  In this
> case, the host sends another escape sequence to cause Kermit to
> execute the second "Product" macro, which is just "close session; run
> myprogram.exe", followed by more data for the Kermit screen.  The
> logging is correctly ended, myprogram.exe is correctly executed, and
> the first of the data for the Kermit screen is correctly buffered (it
> is received while myprogram.exe is running).  When myprogram.exe ends
> and control is returned to Kermit, Kermit shows the screen data it
> correctly buffered, but everything after that (about half a screen
> full) sent while myprogram.exe was executing is lost.
>
> One more observation which may be relevant:  When a long stream of
> data is being sent to Kermit from the host, I can hit ctrl-S on the
> keyboard to suspend the sending of data, but it keeps coming for
> several seconds, which seems like an inordinately long time, and makes
> me suspect that Kermit's xoff signals are not being sent to the host
> quickly enough to prevent buffer overflow.  When I type ctrl-Q to
> resume sending, the response is pretty quick, but not really
> instantaneous (to the eye).

    No, the XOFF goes out instantly, but there is a huge amount of
data in transmission buffers etc. The proper way to stop on a dime locally
is to use Digital's HoldScreen button, keyboard verb \Kholdscrn.

> Has anyone else encountered similar problems, and is there a setting
> of Kermit's parameters which will prevent this?

    The overflow isn't Kermit's fault, as you amply demonstrated with
the DOS test. I think the fundamental problem is you are using XON/XOFF
flow control, and Kermit is provding that support on the local end, and
there is that large amount of buffering in transit, cited above. Better
to let modems use hardware flow control, RTS/CTS, which is a local-only
effect and then the consequences are propagated upstream by the other
modem, etc.
    Joe D.
>
> Bill Kendall