- From: <ronan@roasp.com>
- Date: Wed, 24 Nov 2004 10:00:26 -0500 (EST)
- To: www-svg@w3.org
- Cc: "Jan-Klaas Kollhof" <jan@kollhof.net>
> > Hello, > > here the proposal for the Socket Connection section. > Thanks for putting that together, Jan. Wouldn't it be nice to have the recommendation also simply echo the existing socket implementations and provide a recommendation for full-blown socket support? Here are Perl's particular low-level implementation in Perl [1] and in C [2], built on top of the C socket support for linux. This approach is well proven and highly flexible. [1] http://seth.positivism.org/man.cgi/3/Socket in Perl [2] http://seth.positivism.org/man.cgi/socket in C I appreciate that this may be too far reaching for the WG to take on at this time and that full-blown socket support may be more appropriate for SVG-1.3. But just think how wonderful it would make the SVG UI layer to finally get away from the yoke of client-pull http connections and to get access to server-push and p2p capabilities without a server... There are some that suggest that sockets are inappropriate for SVG and should be relegated to YetAnotherNamespace... This is of course a valid point that has plenty of weight behind it. But nonetheless, SVG's main appeal to application developers is the ability to implement transparent communications without refreshing the page. Without the getURL/postURL that the Adobe SVG team doped into ASV3, SVG would be much less attractive to all of us as an application layer for web applications. Because of the great potential for useful work with SVG as an application GUI layer on the web, I very much would like to continue to see communications in 1.2, and would even push for (optional?) full-blown socket support matching either the above linux socket behaviour or a more Object-oriented implementation like the Perl one. Ronan -- Founding Partner Ronan Oger Vectoreal.com
Received on Wednesday, 24 November 2004 15:00:28 UTC