Re: problem of amaya


> I tested some servers.
> amaya accessed them a little quickly except one server.
> but amaya did not accessed my math bbs properly well.
> amaya accessed bbs and stopped and accessed and stopped.
> that is, amaya  accessed bbs very slowly.
> some case is that amaya accessed bbs for a little time and
> stopped with a message "can't access server"
> but IE4.0 accessed my bbs very quickly.
> whY? Why?
> Why amaya access my bbs very slowly?
> how can i  check my using server's http version?
> i don't  know.

To check your server's HTTP version, just do a telnet to port 80, using
the server's name. Then type in what should be send using the HTTP
request (look at the HTTP standard or a book for more details :) :

>telnet 80
Connected to
Escape character is '^]'.
GET /~s952011/wwwboard/index.cgi HTTP/1.1
Connection: Keep-Alive

HTTP/1.0 200 OK
Server: Netscape-Communications/1.12
Date: Monday, 04-May-98 12:30:08 GMT
Content-type: text/html


user: 0.1   system: 0.0166666666666667

Your server is an HTTP/1.0 one. In addition, it seems to be a rather old one.
as it doesn't understand the keep-alive connection (maybe it's not well

About Amaya being slower than Netscape and IE... This may come from a 
combination of the way HTTP/1.1 works, the kind of document your server
returns, and the way Amaya parses documents.

As we pointed out before, Amaya uses libwww for its HTTP/1.1 implementation.
This is a resource-friendly implementation in that only one connection is
opened to a server at a time. Older versions of the above browsers (don't
know if it's still the case with the latest ones), open multiple connections
to the same browser. Under these conditions, Amaya is slower than other
browsers when accessing an HTTP/1.0 server (and not using Keep-Alive).

The document you're serving seems to be generated on the fly. I see this
in that the server's response doesn't have a Content-Length header. The only
way that the server may tell the client that it finished sending the document
is by breaking down the connection. Thus, a new connection has to be set up.

Finally, Amaya first downloads the complete document before starting
to parse it. The other browsers do support an on-the-fly parsing of the
document, so they can send requests for the images, before the document
finished loading.

In conclusion, because of the above points, Amaya is indeed slower than
other browsers while accessing HTTP/1.0 servers. However, it's a testbed and
our goal is to show the benefits you can gain from HTTP/1.1. It works much
faster then. We could have implemented multiple connections on Amaya to have
a better performance on HTTP/1.0. However, it consumes time to have two
different behaviors on libwww and we prefered to invest our time on the
HTTP/1.1 implementation.

Hope this clears up things up,


Received on Monday, 4 May 1998 09:21:17 UTC