W3C home > Mailing lists > Public > www-lib@w3.org > October to December 1998

Re: "HTTransport_find" in "HTNet_newClient"

From: Henrik Frystyk Nielsen <frystyk@w3.org>
Date: Sun, 13 Dec 1998 19:59:57 -0500
Message-Id: <3.0.5.32.19981213195957.009f7100@localhost>
To: cchan2@notes.cc.bellcore.com, www-lib@w3.org
At 02:01 13/12/1998 -0500, cchan2@notes.cc.bellcore.com wrote:

>     In HTNet_newClient, it requires that there has to have a transport
>object associate with it; otherwise it return "NO".  But is it true that,
>there may be a future access scheme that does not require to specify any
>transport layer object? And there will be specified when it is registered
>with the core by a application.

Libwww has a stackable transport layer mechanism which looks like this: A
URI scheme is linked to an application layer protocol module which in turn
is linked to a transport module. An active instance of a transport object
with in- and output streams is called a channel.

That is, it is in fact possible to write a UDP module for HTTP, if that it
what you want. The only transports currently available are TCP and local
file access. There are parts of a WebMux transport but it is not working.

>      Well, that's what it takes me a while to trace the problem when I'm
>trying to experiment with a new scheme like that for my school project...

If you want to register a new URI scheme then you first link it to a
protocol module using the HTProtocol_* methods. Then you register which
transport this protocol should use using the HTTransport_* methods.

You can see how this is done for the existing transports and protocols in

	http://www.w3.org/Library/src/HTInit.c

Henrik
--
Henrik Frystyk Nielsen,
World Wide Web Consortium
http://www.w3.org/People/Frystyk
Received on Sunday, 13 December 1998 21:41:16 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 17:59:24 EDT