Re: Telnet 5250 in Linux or Browser -- LINUXDESKTOP4I

Right, for me it's not that critical because from Fedora I can always yum -y install tn5250 and that product works through the HMC. It would be nice to just use the TN5250j only whenever I need a GS on the HMC, and not have to "remember" and switch.

On 11/11/2009 05:27 PM, Pete Helgren wrote:

OK. Got it. I didn't realize you could "pass through" the HMC to get to the i with 5250.

As far as I know, tn5250j was based on an early release of tn5250 for linux (2003) so there might be some missing functionality. My exploration of the 5250 data stream, which has no recent/useful documentation that I know of, left me trying to reverse engineer the encodings based on what I would see coming into the buffers in tn5250j and it wasn't easy. Since I didn't "grow up" on either programming or 5250 data streams, the lack of documentation on 5250 data streams makes it devilishly difficult to write code to handle it. Short of tracing I can't think of any way to implement changes to handle the advanced functions since I have been unable to find a manual that defines it.

The only thing at my disposal is the "5494 Remote Control Unit" publication by IBM from 1995 and that requires requires me to put my "propeller hat" on (REAL programming - bits, bytes, hex and that "stuff"). If there is a better source of documentation I'd love to get my hands on it since there some to be some interest in getting tn5250j whipped into shape.

Pete

Simon Coulter wrote:

On 12/11/2009, at 2:46 AM, Pete Helgren wrote:

Could be I am not understanding how the HMC plays here. My understanding would be that tn5250j is installed ON the HMC, just like in Linux, and so the JVM installed on the HMC would be an issue to consider. But, I don't have an HMC and haven't used one for months so I may not be grasping the issue.

I think that's not allowed. HMC is unsupported with anything but the shipped software. Connectivity is from client to host via HMC simply to connect to service ports. I'm not sure of all the details either not having an HMC to experiment with. I do know that console connectivity via PC5250 to HMC is effectively passed through to the target host. The HMC is simply a connexion conduit.

I also know that tn5250j does not implement the full 5250 data stream and even when it implements some of the so-called advanced DDS it does so incorrectly. By that I mean it informs OS/400 that it can handle groups of advanced DDS functions via the 5250 Query Response but then chooses to implement only a subset of those groups (from memory it handles push buttons but not the other features that are implied by setting the bit for that group). Some while ago I did trace the data stream from a tn5250j connexion and it is apparent that the Query Response is simply cloned from a trace of some other emulator-- probably Mochasoft--rather than being set for the capabilities of tn5250j.

I have a bunch of 5250 test-cases that can easily confuse tn5250j and most of the other emulators. PC5250 is the only emulator I know that supports a full implementation and even that does the odd weird thing.

It is this background that makes me suspect the parsing of the 5250 data stream rather than any other cause.

Regards,
Simon Coulter.
--------------------------------------------------------------------
FlyByNight Software OS/400, i5/OS Technical Specialists

http://www.flybynight.com.au/
Phone: +61 2 6657 8251 Mobile: +61 0411 091 400 /"\
Fax: +61 2 6657 8251 \ /
X
ASCII Ribbon campaign against HTML E-Mail / \
--------------------------------------------------------------------