On Sep 10, 2013, at 8:57 AM, Kingsley Idehen wrote:
- COBOL
- Assembler
- CICS
- JCL
- IMS (how many sub languages in & around a 40 year-old IMS application?)
- ETC (an HTML-like document composition language from Applied Data Research)
- ??? report writer.
To what degree (if any) are the items above loosely coupled with data?
Huh?
I do not understand the question.
Slightly switching metaphors here... I would somewhat loosely couple the machine tools (e.g. software languages) with the materials (data) the tools process. Rather like taking a sheet of metal, running it through multiple steps that drill holes & bend shapes & output a car door.
How many languages are there in a SW application? RDF, OWL, SPARQL, ???
With regards to the items above, loose coupling with data has been generally overlooked in most associated narratives. Net effect, a lot of totally avoidable confusion and many unnecessary wars, at a variety of levels.
We just need logic, data, and applications to be loosely coupled. Remember, history has shown that conflating the aforementioned items always fails :-)
Theory like that is all nice & good, but often doesn't play well in the real world.
What happens when the logic, data, and applications are NOT loosely coupled?
In the process of producing & delivering software it is totally typically specifications to be lacking & the product delivered to not well serve the now changed, or better understood business needs.
End users tend to be devilishly clever... how about if we put "TTE" in the title of the customer line to indicate "trustee" and then run extracts pulling out TTE records for special processing.
Is that what is called fixed coupling... the direct opposite of "loose coupling?"
On that note, I think I'll try to remember to inject the "MVC"—why isn't that VCM?—jargon into conversations & see what the response is.
I have to assume the MVC paradigm is absent from the vast majority of Systems of Record systems in use in the Fortune 500.