- From: Mark Nottingham <mnot@mnot.net>
- Date: Mon, 16 Jul 2012 11:20:05 +1000
- To: HTTP Working Group <ietf-http-wg@w3.org>
FYI. Begin forwarded message: > From: "James A. Morrison" <jim.morrison@gmail.com> > Subject: HTTP/2.0 proposal feedback > Date: 15 July 2012 11:30:14 AM AEST > To: mnot@mnot.net > Cc: Mike Belshe <mike@belshe.com> > > Hi, > > This is a response to the call for expressions of interest: > http://trac.tools.ietf.org/wg/httpbis/trac/wiki/Http2CfI > > I’m the developer of the spdy for iPhone library. I’m making this > initial response on behalf of myself, not any previous or current > employers. I work for Twist which is currently developing an iPhone > application. > > I chose to implement a SPDY client library rather than other HTTP/2.0 > proposals because of existing server support. Mainly existing support > through Google for applications running on Google Appengine. I was > also familiar with SPDY from previous work at Google. Along with > existing server implementations, I built my library as an Objective-C > wrapper around spdylay. Spdylay is an existing C spdy library that > has both a client and server example built already. The spdylay > library is mature enough that it transparently supports SPDY/2 and > SPDY/3. Neither the Twist code, nor the SPDY for iPhone library > required changes when Google switched their servers from SPDY/2 to > SPDY/3. > > One of the promising features of SPDY is the gzip requirement. This > will allow us to save significant upload bandwidth at Twist by > uploading compressed json blobs instead of the uncompressed json text. > > The other improvement is multiplexing multiple streams over a single > connection. This allows us to keep our api simple by having small > specific requests for different types instead of creating large > complicated requests and responses. The multiplexed streams keeps the > cost of all the small requests done by limiting the time the requests > are waiting for a connection to be established. > > With the spdy for iphone library I have not experimented or used any > of the flow control features. > > -- > Thanks, > Jim > http://phython.blogspot.com -- Mark Nottingham http://www.mnot.net/
Received on Monday, 16 July 2012 01:20:24 UTC