Reply to thread

Message

<blockquote data-quote="Davin" data-source="post: 1254237" data-attributes="member: 1183">You seem to have some strong feelings on this subject, but I think your general complaints are valid ones.&nbsp; Most people doing this sort of work aren't doing (IMO) a terribly good job of generalized design.&nbsp; Things I've seen tend to be too task-specific, and as such they suffer badly when it comes time to communicate between disparate systems.&nbsp; You're going to need &quot;indians&quot; as well as &quot;chiefs&quot; in something like this, but your top-level design people will probably need to have strong coding skills (any language) and strong database design skills (any database) in addition to good technical XML-related knowledge &amp; experience.&nbsp; Otherwise the design will suffer, perhaps badly.Many people tout the brilliance of using XSL[T] to convert between formats, but they fail to take into account that most of the data structures they're converting between don't have the same data in them.&nbsp; To reiterate, they're each usually to specific to the task for which they were designed, so they don't tend to have the right kind of data for the application they need to be translated to.&nbsp; For a simple example, one application may output only &quot;totals&quot; for something, but the receiving application needs it broken down into details.&nbsp; In such a case, the translation is impossible.I'm not sure that all these problems even have solutions, much less feasible ones.&nbsp; And I'm far too busy myself to do more than get my fingers in the way of the people that are really doing the work, even if I had the necessary skills to contribute profitably.&nbsp; But if you're serious about doing this kind of thing, you'd best do some meta-design to make sure the organization works and can produce the kind of results you're looking for.</blockquote>

[QUOTE="Davin, post: 1254237, member: 1183"] You seem to have some strong feelings on this subject, but I think your general complaints are valid ones. Most people doing this sort of work aren't doing (IMO) a terribly good job of generalized design. Things I've seen tend to be too task-specific, and as such they suffer badly when it comes time to communicate between disparate systems. You're going to need "indians" as well as "chiefs" in something like this, but your top-level design people will probably need to have strong coding skills (any language) and strong database design skills (any database) in addition to good technical XML-related knowledge & experience. Otherwise the design will suffer, perhaps badly. Many people tout the brilliance of using XSL[T] to convert between formats, but they fail to take into account that most of the data structures they're converting between don't have the same data in them. To reiterate, they're each usually to specific to the task for which they were designed, so they don't tend to have the right kind of data for the application they need to be translated to. For a simple example, one application may output only "totals" for something, but the receiving application needs it broken down into details. In such a case, the translation is impossible. I'm not sure that all these problems even have solutions, much less feasible ones. And I'm far too busy myself to do more than get my fingers in the way of the people that are really doing the work, even if I had the necessary skills to contribute profitably. But if you're serious about doing this kind of thing, you'd best do some meta-design to make sure the organization works and can produce the kind of results you're looking for. [/QUOTE]

Verification