implement readonly attribute boolean isLoggedIn

[swick]

fixes #34

[infinity0]

Your logic can be flipped to argue that we should always return TRUE; there is nothing to distinguish the two cases. What I need to find out is how firefox treats TRUE vs FALSE. You've only had a few days to test "everything works" and I doubt you covered all the corner cases.

Don't be lazy, you can convert a macro into an integer in the Makefile. But it's OK I've found a better way of doing it, by compiling a stub that tests for GetIsLoggedIn directly.

[infinity0]

@dolske: You recently fixed mozilla bug #853308 which resulted in mozilla/releases-mozilla-central@bee13fc8. The explanation was "Otherwise tabbing through (blurring) a field pwmgr thinks is a login's username or password field would trigger a MP prompt, and that would be annoying."

Unfortunately, the GNOME keyring API does not provide equivalent functionality to tell if a keyring is unlocked or not. So we must either always return true, or always return false. What do you advise? I am aware of two dangers in each approach:

• always return true: will be annoying for the user in the way stated in the comment in the bug
• always return false: may cause incorrect behaviour that cancels some action based on the (potentially false) belief that the keyring is locked.

(I am aware that mozilla is trying to integrate this functionality into firefox/thunderbird itself using libsecret, but until that happens I need to keep maintaining this extension.)

[infinity0]

@dolske got back to me via email and pointed out a similar issue on mozilla bug 839961. The recommended implementation is actually to always return TRUE.

I will merge this request tomorrow (too tired right now) and make the necessary adjustments (or feel free to make them yourself). Please remind me if I don't do this within a few days.