Re: [LINUX5250] device naming

[Scott Klement <klemscot@xxxxxxxxxxxx>]

How hard?   Not sure how to measure that. :)

Right now, we load up all of the env.XXX variables when the emulator
starts, and assume them to be valid variables for the AS/400 to receive.

When the AS/400 requests "NEW_ENVIRON" (see RFC 1572), we simply dump all
of these variables...  we pay no special attention to DEVNAME or any other
variable, just dump em all...   (That's not the "correct" way to handle
it, but it hasn't caused problems yet.  It's listed as a bug in the
BUGS file in the distribution)

If you wanted to make it do "device name collision processing" (as it's
called in RFC2877) you'd need to change this behavior to actually look
at which variables the AS/400 is requesting, and only sending those.

Then, keep track of the state of the DEVNAME variable, if you get back
another request for it after the first one, the device is in use, and you
need to try again with a different value for DEVNAME...

This would repeat until you found an available name, or you sent the
AS/400 the same device twice in a row...  once it gets the same device
twice, it gives up :)   (which is what happens now, except that we don't
just send DEVNAME, we send everything...)

See telnetstr.c, sslstream.c and RFC 2877...


On Tue, 22 Jan 2002, Sean Porterfield wrote:
>
> Anyone have any idea how hard it would be to allow other dynamic naming from
> tn5250?  What I mean is in Client Access I can set my device name to MYPC=
> and it will automatically give me MYPC1 followed by MYPC2, MYPC3, etc.
> There are other options besides = that will allow alphanumberic name
> suffixes.  I lost the link that explained the different options.
>
> The only one I use is = because I normally only run 1 or 2 sessions.  It's a
> bit inconvenient to have multiple config options, especially if I give it to
> an end user with a desktop icon.  I was going to have a look myself, but I
> didn't have time yesterday, not that I'd be likely to figure it out...
>
> TIA
>