<Comments on 3270 stuff snipped> > > Has anybody considered running "something" which uses subsystem SVC > > screening to "trap" the TSO terminal I/O SVCs, such as TGET, TPUT, and > > TPG and then converting this to "something else" such as XML or HTML to > > ship to a different client? I think that BMC has something like this > > because they have an interface other than IKJEFT01 to run some of their > > TSO based Mainview displays. Or they did in the past. I remember doing > > this so that I would eat up my TSO region running Mainview.
The thing you are referring to is Mainview Alternate Access. It is still available and it comes in two flavors. One allows you to run a locally attached 3270 without either TSO or VTAM being up and if you're heroic enough it will even run without JES. The intent is to allow you to have a (dedicated and secure) terminal that gives you access to all of your management software in the situation where VTAM and/or TSO are not available. In theory you can edit parmlib, drive your monitors etc. and it does work, but I am not sure how often, or how successfully it is used in those scenarios. The other version (it's the same code base) is a VTAM application that runs the TMP in a STC. You logon just like you would to TSO and it starts an STC to run a standalone copy of the TMP connected to your terminal. There are some subtle to significant differences from "real TSO(tm)", e.g. being a STC gives it wildly different default performance characteristics than real TSO - even though it is running identical code and workloads. The rationale for this option was that you could "logon" even when TSO was down, i.e. when the TSO terminal controller was not running. How rational is that in reality? Well not very, but it was good check-list marketing back in the day :-) This function dates from the mid-1990s and it was in turn a complete rewrite of an earlier but similar function that ran all of the users in a single address space. That had hideous reliability problems as you might imagine. MVA has been fairly reliable and stable. So to address your question about doing a general web-style interface to TSO using this sort of technique... it could be done and for some TSO applications it might even work. I suspect you would run into some profound strangeness with applications like ISPF that have a fetish about the shape of the display window. ISPF does not want the shape of the viewport to change once it has obtained the bind information, but even that might be possible to overcome if you planted enough smarts in the underlying intercept code. However, given that you can run a plain old 3270 emulator (with no risk) in either a desktop window, or a servlet container like HOD, why would you bother? CC ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
