- From: Mark Nottingham <mnot@mnot.net>
- Date: Tue, 17 Jul 2012 09:55:40 +1000
- To: Phillip Hallam-Baker <hallam@gmail.com>
- Cc: "ietf-http-wg@w3.org Group" <ietf-http-wg@w3.org>
Phillip, I gave my feedback on that way back when we started all of this, in <http://www.w3.org/mid/0C615921-7EE0-4E53-93F9-8B406D1561A1@mnot.net> (I see we're rehashing that thread in more ways than once now). Regards, On 16/07/2012, at 11:43 PM, Phillip Hallam-Baker wrote: > How easy is it to implement SPDY, does the 'implement in a day' > criteria make any sense? > > I have written a LOT of HTTP clients and more than a few servers. I > know that it is possible to hack 'something' together in a day and > that implementing a full HTTP/1.1 will take considerably longer. So > (1) what level of implementation are we talking about and (2) what > support libraries are we including? > > Implementing a text based protocol like HTTP is easy in C because > stdlib gives you 70% of the code ready made. Implementing HTTP in Perl > or the like is even easier as it provides the parsing ready made. > > Take away those support libraries and implementation takes longer, a > lot longer. Give people a similar library for SPDY and maybe the > difficulty of implementation becomes equal. It might become easier. > > > What I really care about is not how long it takes to code HTTP/2.0. > There are going to be so many libraries floating about that it does > not matter much. What I care about is not how long it takes to > implement but if I can implement on restricted chips like embedded > control systems. So code footprint is more important to me than > time-to-implement. And I think it is a more objectively fair test. > > > > -- > Website: http://hallambaker.com/ > -- Mark Nottingham http://www.mnot.net/
Received on Monday, 16 July 2012 23:56:07 UTC