Paul Colomiets

Well, I don't understand how 2dfe91a fixes the issue.

The proposed strategy is: 1. Sync all inactivity timers across all the replicating nodes. Probably by grouping them in bulks with 100ms - 1s latency. 2. Split session namespace into...

Well, under some circumstances nginx forwards zero-length chunk but strips `Transfer-Encoding`: ``` 13648 recvfrom(10, "HTTP/1.1 304 Not Modified\r\nTransfer-Encoding: chunked\r\nContent-Type: application/octet-stream\r\nDate: Wed, 02 Aug 2017 14:43:38 GMT\r\nServer: Python/3.6 aiohttp/2.2.0\r\n\r\n0\r\n\r\n", 4096, 0,...

All in all, I think we have two options: 1. Accept zero-length chunk if exists 2. Reject as if some garbage was sent after the response (like we do now,...

Currently looks more like [this is a bug in aiohttp](https://github.com/aio-libs/aiohttp/commit/cc614ca3ecbcb3d49854a2a7411cf64eb724923d), and nothing to do in swindon.

>> So the idea is to create small DSL > > Ouch, it looks creepy at the first sight... So what are the alternatives? All ones mentioned in "motivation" section...

Updated

Well, after some thinking, I don't want any special tags on `addresses:`, I think we should implement configurable name service, with some `.verwalter` suffix resolved via verwalter's API. Keeping this...

Well, I'm not sure about ORSWOT, because this requires enumerating all the actors. I'm not sure how we should treat backend processes. But on the other hand, I'm pretty fine...

Yes, this bothers me too. In practice, however, we delete items as needed from the database, and keep them in swindon as long as user session is active (remove after...