Trace Analysis
Having just watchedLaurence Melloul and yourself explain both the trace application as well as your live trace, I see a small (but none the less important) element that is missing (my view only BTW), I will explain:
In my current arsenal of responsibilities, I have a bespoke middleware product (written in .NET 1.1 but migration plans to 2.0 or 3.0 - more than likely 3.0).I have had the solution re-architected twice to have a reasonable level of trace provisioning available.
Now to the issue, during the demo, I noticed that the assumption (correct me if I am wrong), is that we (the customer) enable tracing of both the client and the server.This is great where the platform allows such behaviour, but what if (n this case), I need to connect and provide messaging (that falls out of scope for WCF) to legacy solution that cannot provide said information.
Case in hand: The current implementation provides data and transaction requests to a core banking product (Misys Equation) using a proprietary message structure.In this case, you can trace as far as the request to the host (Equation), but no further.
Now the question: In all cases where failure occurs (not that often BTW), the host provides a generic error message – will the log trace catch the error once returns i.e. can it be made scheme aware?I ask as this in general terms, would not constitute an error or report as such (although we have developed scheme inference – thus we know if an error occurs) and simply pass back up stream.
How is such a scenario managed?
Thanks, David
