RE: XML to 5250

["Joe Pluta" <joepluta@xxxxxxxxxxxxxxxxx>]

> From: Leif Svalgaard
>
> Jon, I have worked on and off on this (not done yet). The problem
> (and confusion) is that XML in itself does not describe anything.
> To do what you want you have to first *define* your data structure
> in XML (like you have to describe the syntax of DDS), then populate
> the structure with data (giving the values of the labels and fields
> defined in the DDS), then finally from that generate the 5250. This
> last step is trivial and the first step has no common meaning
> (unless you choose to mimic DDS). To talk about a tool that
> accept XML and generates a 5250 data steam is like talking
> about a tool that accepts English and generates 5250. What
> is missing a definition of the *structure* of the data. This structure
> you can make an XML description of, but the structure itself is
> empty.

Yes, it would be a two-step process, Leif, especially if the idea was to
talk to a program using a buffer approach like OS/400 does.  The program
would be communicating using buffers that contain just the varying data of
the fields (and possibly some sort of "indicator area" to condition the
display).  The static information (literals and so on) would have to come
either from the program (which would mean a lot of extra formatting for
every buffer) or from an intermediate representation which would, as you
say, mimic the DDS.  Then you could spit out the 5250.

The trick is not the formatting of the data stream, as you've alluded to
(although it's not exactly trivial, IMHO - you still have to generate the
field tables and all that good stuff).  The trick is combining the static
information with the dynamic information from the program into a pre-5250
form.  You either have a DDS lookalike (and if you're truly mimicking DDS,
you're talking about things like message fields and the like) or you spit
out the literals yourself over and over again, more like a CGI type of
approach.

This is a huge amount of work.

Joe