This is the mail archive of the
kawa@sourceware.org
mailing list for the Kawa project.
Re: GSoC 2011
On 04/07/2011 06:48 PM, Fattom Riddle wrote:
Dear Per
Thanks for your specific comments!!!
It clears lots of questions.
But in my original thought, I don't imagine (2) a communication protocol.
I am not sure what a communication protocol does.
Whenever you use AJAX, you need to define a protocal: The browser makes
a request, and the server returns a result. you need to specify some
conventions for how the server interprets the request, and what the
browser does with the result.
With WebSockets, it's the same, except it's bi-directionally asynchronous.
(I.e, it provides a Comet solution.)
I think there are several steps to process the task.
1. A browser send my input lines(scheme syntax) through POST/GET to server.
2. Server eval the inputs and return result (XML or JSON format ).
3. A browser get the pure text results and show.
You don't show "pure text results". The result from the server may be
XML/HTML that gets inserted. The whole point of the exercise (well, one
of the goals) is to support "rich text", pictures, etc in the output.
Of course, the basic case is simple: server (i.e. Kawa) sends XHTML,
and the browser just inserts at the current insertion position (which
is basically a "current node").
So does a communication protocol mean we should have a request format in step1?
Basically yes, though a single input line is enough to start with.
Or you mean the web console can show rich results but not pure text results?
Both, but we don't need to explicitly deal with pure text - the server can
be responsible for emitting the necessary escape characters.
Otherwise, should we have a configuration to define which IP address
can access the web console?
Yes, there are definitely some security issues. For now I'm most interested
in local use - just using the browser as the GUI on the same system.
Non-local
use should should require logging in and HTTPS.
--
--Per Bothner
per@bothner.com http://per.bothner.com/