Re:Introducing Tent - the decentralized social web
by mmahemoff on Wednesday August 22, 2012 @03:50PM
(#4419208)
I upvoted this as it's good to question the assumption and it's easy to imagine this working better without HTTP. In practice though, I feel the project is orders of magnitude more likely if it's built with HTTP and REST. A social product needs to reach critical mass and any hurdles you're putting in front of users and developers need to be seriously justified. In this case, workarounds like cookies are Good Enough.
[ Parent
| on HN ]
Re:Introducing Tent - the decentralized social web
by flatline3 on Wednesday August 22, 2012 @03:52PM
(#4419227)
I'd argue that -- for native client authors -- HTTP and REST are actually larger barriers than a simple non-HTTP protocol.
I also don't understand the concern over "inventing" a new binary protocol. It's not like it's any more complicated or difficult than "inventing" a non-binary protocol. The framing rules just don't use ASCII/UTF-8 field delimiters. There are plenty of existing encodings to use, no more difficult than JSON.
Re:Introducing Tent - the decentralized social web
by TazeTSchnitzel on Wednesday August 22, 2012 @04:23PM
(#4419373)
Only if you're using C/C++ and only native OS libraries (or on windows, afraid to use WinHttp for fear of lack of SP1 support)
Re:Introducing Tent - the decentralized social web
by flatline3 on Wednesday August 22, 2012 @06:08PM
(#4419886)
Marshalling back and forth between HTTP requests and in-app models via the like of NSURLConnection is far more ugly to do cleanly and efficiently than a simple state machine modeled on message passing over a socket.
Re:Introducing Tent - the decentralized social web
by TazeTSchnitzel on Wednesday August 22, 2012 @06:12PM
(#4419911)
That's only for Objective-C, it's nicer in Go.