Closed lopuhin closed 8 years ago
The most non-obvious issue is library support: right now we use scrapy to fetch pages, find login and registration forms, and log in. It is possible to hide it all behind a synchronous API, the only caveat is that AFAIK there should be only one twisted event loop in a thread, so we'll have to offload all autologin operation into a different thread. Does it sound like a sane approach?
Done in #13
This branch brings some improvements:
Still, there are some unresolved issues: