http://www.themidnightcoders.com/products/ria-apppuncher/overview.html

I started with Flex about year and half, I work very fine for me help me with the errors I always use CF for backend. I use RemoteObject object and I build a function on fault=”failError(event)” event, this catch everything happens and really makes my life easy. I just simple show inside a Alert box the errror.

The code looks like this.

private function failError(event:FaultEvent):void{
Alert.show(‘Error ‘ + event.toString());
}

In my little experience CF and Flex Rocks…

Logs != disk. SmartClient retrieves server logs from an in-memory revolving buffer attached as a standard Log4j Appender.

Logs != unstructured. Log4j files are sufficiently structured to parse, and viewers already exist.

I think you’re still mystifying most readers as to what you’re getting at, and in my case, it’s certainly not due to lack of experience building complex RIAs (SmartClient shipped as the user interface to major BI products 5 years ago).

Perhaps if you went into more detail on some concrete scenarios that you believe are amenable to standardization.

For example, if you are connecting SmartClient to a non-Java server and you have no pre-existing services, we provide a complete protocol that includes signaling validation errors, session expired errors, stale cache errors, and general errors, with message formats available for JSON, XML or SOAP. Is that the kind of thing you have in mind?

What I am saying is that we can measurably improve RIA developer productivity by (a) establishing a common format/schema for common types of diagnostic output information combined with a common viewer for that type of information and (b) have a way to maintain some of it potentially automatically in memory on the server as opposed to having to persist everything in log files.

I don’t see how a world with no structured logging capabilities, when every app server, framework and tool provides its own way of dealing with diagnostic output leaves developers better off…