[webkit-dev] Improving the flexibility of the cookie manager

[Drew DeVault]

I'm looking into making a few patches that make the cookie manager more
flexible, and I'm stopping by the ML to see if anyone has
concerns/advice about this first. I'm using WebKitGTK, and in webkit1,
it was possible to have the UA manage the cookie store for more subtle
behaviors that aren't in the scope of webkit. In webkit2, the cookie
manager was introduced and is much less flexible with respect to cookie
policies and manipulating the cookie store. There's also issues with
sharing cookies between processes in a multiprocess browser.

I have a two ideas around strategies to improve this:

1. Make WebKitCookieManager subclassable and have the UA provide their
own implementation if they need more flexibility.
2. Add sufficient signals and APIs to WebKitCookieManager to allow the
UA to effectively manage cookies themselves.

I'm leaning towards the second option. Would like to hear your thoughts.

--
Drew DeVault