For a user (or a unit test simulating one), a discovery process goes like this :
start (user creates a new discovery wizard, or a new unit test is launched...)
listens a first time to get a first "master" run, and detects a first detected api using default heuristics, which it displays in order to * help the user decide whether it is OK or not yet
then alternatively improve the heuristics conf (on the detection module), or the master run (by getting a new one in a new recording session),
and after each (drop the existing and) redetects the api and displays it,
finally validates the triplet master run + heuristics conf + api (sets metadata in registry)
LATER
enrich it with GUI specifications and develop it
alternatively to listening to a new run, the user could choose an existing one
TODO 20111201 update according to 73 "Message"
develop the following process as a unit test.
For a user (or a unit test simulating one), a discovery process goes like this :
LATER