10:27 <Candle> Is there some documentation on the first parameter 'auth' in net/TCPConnection.create(...)? i.e. why it is required, and what the various different types (Ambient/Net/TCP/TCPConnect) allow/disallow?

18:22 <jonas-l> Candle, there's a little bit at http://www.ponylang.org/ponyc/builtin-AmbientAuth/

18:25 <jonas-l> as I understand it, auth parameter requires particular module to have the right to do TCP connection

18:26 <jonas-l> and to prove it, you have to provide TCPConnectAuth value

18:32 <jonas-l> https://github.com/ponylang/ponyc/issues/293 explains why it is required

18:55 <ponysaurus> @sylvanc: you promised to send some notes regarding Issue 380. whenever you get a chance ...

21:02 <ponysaurus> @jemc: Just watched the video .. awesome!

21:04 <ponysaurus> I am curious to know how atomicity is maintained in distributed pony

21:04 <ponysaurus> actually i am curious about a million things in distributed pony

21:09 <jemc> ponysaurus: me too :)

21:10 <jemc> there are some public materials online, some of which you may or may not have seen:

21:10 <jemc> http://www.doc.ic.ac.uk/~scd/Disributed_Pony_Sept_13.pdf

21:10 <jemc> http://www.doc.ic.ac.uk/teaching/distinguished-projects/2013/s.blessing.pdf

21:14 <jemc> as far as atomicity goes, I don't think much will need to change in distributed pony

21:14 <jemc> each actor will still follow the single-turn model of executing one uninterrupted behavior when they receive message

21:18 <jemc> as far as I can guess, the main interesting areas of problems in distributed pony are (1) maintaining causal message order in a fault-tolerant way (2) giving the user some sort of interface/pattern for dealing with faults at the application level, and (3) giving the user some way to pin certain actors to a machine because they are tied to local data access

21:18 <jemc> I'm probably missing some things though

22:03 <SeanTAllen> you hit the big ones jemc.