--
Async.Generator DEBUG runTestGenerator-0: self.cb generated at test_async_missing_yield.js:28
Async.Generator DEBUG secondGen-1: self.cb generated at test_async_missing_yield.js:20
Async.Generator DEBUG secondGen-1: done() called.
Async.Generator DEBUG runTestGenerator-0: self.cb() called, resuming coroutine.
Async.Generator DEBUG runTestGenerator-0: done() called.
Async.Generator DEBUG secondGen-1: self.cb() called, resuming coroutine.
Async.Generator DEBUG secondGen-1: done() called.
Async.Generator ERROR Async method 'secondGen-1' is missing a 'yield' call (or called done() after being finalized)
--
As you can see, I've added log messages whenever the Generator's 'cb' property is accessed--this is almost guaranteed to be very close to a 'yield' statement, and therefore provides us with a decently accurate idea of where the generator 'stopped'. We also log a message when the generator continues, and by doing so we get an idea of how the coroutines interleave.
Another idea I had was to actually match calls to self.cb with calls to 'yield' to automatically detect e.g. two yields in a row (which will ordinarily result in a generator 'hanging'), a generator exiting while a self.cb still hasn't been called, but I'm not sure what kinds of reprecussions it may have.
* async generators: much better stack traces
* dav: use global identity system rather than login/logout to manage
usernames and passwords.
* dav: there is a checkLogin() instead of login() which can be used to
verify auth at any time.
* dav: make it so that we can (only internally atm) hold multiple
locks for different URLs.
* identity: add an identity manager singleton service to keep
identities globally, referenced by a name, with aliasing support (so
e.g., dav can ask for the 'dav' identity, while something else can
alias 'dav' to another identity).
* service: keep track of logged in status here, rather than in the dav
service. Use the global id manager.
* Logging improvements / default log levels tweaked. Less chatty now.
Ability to tweak log levels of individual loggers via prefs.
* Various crypto module fixes, specially for RSA.
* 'service' lock removed, reuses server lock now. dav module supports
temporarily blocking locks to simulate the 'service' lock (login()
needs this, since dav is not configured at that time).
* PKI support: data encryption uses randomly-generated symmetric keys,
which are then encrypted with RSA public keys and stored on the
server.
