> - Protocol Simulator for testing RFC compilance.
> Problem is that it is that tests cannot be easily extended - no
> functions/objects
This is similar to something I am considering, which is a client simulator
that would simulate real-world interactions James has with various mail
clients. Basically, a user-oriented script (not a software script, a
"human" script) is created that covers the majority of client actions, like
check mail, send mail, send mail w/attachments, check mail with incorrect
username, etc.
While a human was running the user script against James with a specific
client app, a mailet would capture the _exact_ stream from the client and
store it, along with the James return stream for each action in the script.
When someone makes changes to James, they would start up james, then run the
simulator, which would simulate the client actions by playing back the
stored stream for each action. The return stream from James would be
checked against expected results, and deviations could be considered bugs.
It would be fairly easy to expand both the set of actions and the clients
recorded. It would also likely be a fast test, so someone could run a dozen
or more clients against their new James to determine the stability of the
change before every checkin.
Matt Bishop
[EMAIL PROTECTED]
"We are all here on earth to help others. What I can't figure out is what
the others are here for."
- W. H. Auden
--
To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
