Vložit
- čas přidán 18. 02. 2019
- Let's #TalkConcurrency Panel Discussion with Sir Tony Hoare, Joe Armstrong, and Carl Hewitt with host Francesco Cesarini.
When considering the panel to discuss concurrency, you’d be pushed to find a higher calibre than Sir Tony Hoare, Joe Armstrong, and Carl Hewitt. All greats within the industry and beyond, they give an amazing insight into the lifeline of concurrency and actor models over the past few decades, their bountiful experiences within the concurrency field, and where they see concurrency heading in the future.
Full transcript: www.erlang-solutions.com/blog... - Věda a technologie
What an interview! We stand on these peoples' shoulders.
i know it is quite off topic but does anybody know of a good place to watch newly released series online ?
@Jordan Elliot Flixportal :P
@Leo Amari Thanks, I went there and it seems like a nice service :D Appreciate it !
@Jordan Elliot you are welcome xD
We miss you Joe
Great men!
Joe looked so intimidating but it's a great discussion anyway!
He confessed to me (before the interview) that he was intimidated by Tony Hoare and Carl Hewitt.
12:00 There is nice disagreement about synchronous (Sir Tony Hoare) vs asynchronous (Joe Armstrong and Carl Hewitt) messaging communication. What's strikes me, that neither Joe nor Carl especially which has significant knowledge about inter-chip multicore communication doesn't react: 10ns synchronous communication? How nice, we can have 40 asynchronous messages across 2.5" chip in those 10ns.
sorry, but My Hoare is buried in academic research. No idea how it works out there now.
@@veramentegina I disagree. He's saying, "Look at this great stuff! Why are you ignoring it? If you want something generic, make it a layer on top of this fast stuff." While Joe and Carl are saying, "We want a single scalable mechanism." But really, in my view, the difference between CSP and Actors is CSP requires a channel and Actors don't, they wrap arbitration; and that's what's driving those opinions.
These were my takeaways.
1. What Carl Hewitt is talking about is extreme parallelism. This puts a huge burden on the programmer to design a system that conforms to this. But the underlying physics assumptions that Joe talks about to support this is absolutely right.
2. But, when Tony talks about is, if you can indeed ignore that assumption (which you can do in a single chip multi-core systems), it has a huge benefit of thinking problems in terms of synchronised IO and also a lot more additional properties you can prove. This reduces the programmers need to worry about everything being async. Joe himself talks about using extreme parallelism and it's downsides later in the vid.
Now which is more right, as Tony said both are necessary. You just have to chose the right one based on the usecase. Infact networks are expected to be so good that the advantages that synchronized IO (CSP) maybe worth ignoring the assumption that everything is asynchronous for most cases and use CSP, but as of now we're not there yet.
For Sir Tony Hoare, Joe Armstrong, and Carl Hewitt please take a look at IPFS.