Re: RFCI vendor-specific info

["Robert G. Moskowitz" <[e-mail]>]

>>Also we should review these vendor 'add ins' and design RFCS so they will
>>not be needed.  That is to say, if a vendor saw fit to add a feature in,
>>then we should have a recommended implementation that addresses that
>>feature.  Comments?

>We could do this, but I'm not sure anyone but the particular vendor might
>understand the reason behind some of the 'add-ins'. Perhaps a better
>approach is to have a vendor explain his add-ins and propose something
>for RFCS if he wants to.

I had specific items in mind.

One that I am constantly dealing with is how the server finds out that the
client disappeared and cleans up the LU.  IBM's MVS/TCP uses the TELNET TM
option.  OCSII relies on TCP keepalives.  Both have configuration and
network concequences.

Another is how to handle a reconnect on a dedicated LU.  We have had a
couple of configuration problems with OCS's handling of that.

So I envision an implementation section that makes recommendations on how
best to handle things like this.  They are not TN3270 protocol per say, but
they are important to the operation of a TN3270 Client/Server setup.


Robert Moskowitz
Chrysler Corp
TN3270E WG Chair