Robert,
It is not normal however it happens every so often.
I don’t worry too much about making false assumptions since I don’t really start any configuration until we are well along in a very detailed integration specification.
At the point where I am satisfied the structure is confirmed, I can build an Xlate that will handle most everything (including iterations).
Most of the time when we are that point, the source system can provide messages I can test with – but not always.
So now I use the configuration I built based on the spec and an Xlate to build to that configuration. I save the output of the Xlate tester to a file.
The I build the Xlate that is supposed to take the source messages and create the receiving messages and I test that.
I can vary the data sufficiently to test most of the construct parsing and some of the other logic as well. By the time the source system delivers messages I am well on my way.
I used this techniques successfully recently with the UHC 278 project. UHC in their test environment was not able to provide any 278 Results but I needed to be relatively sure I could handle their specified result sets completely.
By using the Xlate trick I was able to have that so fully tested by the time we rolled into production that there were only a couple of issues.
The key is to have detailed specifications created as a result of anlaysis (not relying on the vendor specifications which we all know is full of half truths or just plain errors).
email: jim.kosloskey@jim-kosloskey.com 29+ years Cloverleaf, 59 years IT - old fart.
.
