Changes between Initial Version and Version 1 of Ticket #30350, comment 11


Ignore:
Timestamp:
May 3, 2019, 5:43:25 PM (4 months ago)
Author:
dcf
Comment:

Legend:

Unmodified
Added
Removed
Modified
  • Ticket #30350, comment 11

    initial v1  
    22> This makes sense to me, I think we do need to do something about item 2 as well though. I wasn't able to test the behaviour of tor browser at the VPS but just the Tor client was having trouble quickly deciding whether or not the proxy was being blocked. I suppose this won't be as big an issues when we have more diverse proxies though.
    33
    4 No disagreement there. Snowflake can do better about detecting a lack of connectivity. Actually it's kind of a general thing that affects more than just Snowflake. See #24640 for meek for example. The PT interface was designed with a mindset of obfs4-like transport that make a single TCP connection to a single bridge and always have a well-defined connectedness state--it doesn't map perfectly onto transports that don't work like that. I think there are bugs on the tor side as well--see comment:3:ticket:26891 where deleting a state file made the connection start working, which looks like #11301 even though that is supposed to be fixed.
     4No disagreement there. Snowflake can do better about detecting a lack of connectivity. Actually it's kind of a general thing that affects more than just Snowflake. See #24640 for meek for example. The PT interface was designed with a mindset of obfs4-like transport that make a single TCP connection to a single bridge and always have a well-defined connectedness state--it doesn't map perfectly onto transports that don't work like that. I think there are bugs on the tor side as well--see comment:3:ticket:26891 where deleting a state file made the connection start working, which looks like #11301 even though that is supposed to be fixed. Edit: also #29245.