In article <55v2a0$ohd@romeo.logica.co.uk> gregorie@see.sig.for.address (Martin Gregorie) writes:
>>I have had several preliminary meetings with prospective clients who have serious
>>Y2K problems, most of them on mainframes running COBOL or RPG apps. So far all my
>>analyses indicate that it would be far less expensive to re-write the entire apps
>>in a modern 4GL like Progress and run the app in a modern client-server
>>environment, than to try to fix the present code. Few of the apps really require a
>>mainframe, although in those few cases the mainframe could continue to be used as
>>a data server.
>I'd agree but for one small point.
>Many sites have out of date documentation if they haven't lost it
>entirely, and often have no written record of changes since the system
>was implemented. As a result a rewrite would first involve
>recreating/updating the spec from the code and *this* is can be
>sufficiently expensive to make the rewrite quite unattractive.
Okay, let's say that the client is American Express and the program in
question is a COBOL application that processes several hundred thousand
transactions an hour from all over the globe and it's working fine but we're
worried about the Year-2000 date-handling and ... and your "solution" is to do
what?!?!
I don't know where your analysis came from, friend, but I'm afraid it smacks
of "this is what I'm in business to do and I want to persuade you that I'm an
authority so you'll agree that what I'm in business to do is what you need...
or at least I have this report to sell you." Sorry, but that's exactly how it
reads.