httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Randy Terbush <ra...@zyzzyva.com>
Subject Re: HTTP/1.1
Date Fri, 26 Jul 1996 15:47:31 GMT
One thing to interject into this debate...

By applying these patches to the CVS tree, they are _less_ public
than if we leave patches lying about.

Except of course that Jim's spy will probably then leak the code.... :-)


> > No, this is untrue. "Keep-Alive" support is not required by the HTTP/1.1
> > specification. It's in an appendix under "HTTP/1.1 Compatibility". So if I
> > have a browser that does not support Keep-Alive, doesn't understand it,
> > and I get back "HTTP/1.0 200 OK", I'm going to assume (assuming that I at
> > least know what HTTP/1.0 is) that it is a non-persistent connection,
> > because every HTTP/1.0 response since 1993 has been. So if Apache tries to
> > open a persistent connection, the browser will hang.
> > 
> > Hopefully, that's clear.
> 
> Absolutely. What the "normal" HTTP/1.1 client will do in that case
> is immediately close the connection after receiving the content-length
> amount of bytes, which will be properly interpreted by the server
> as an end to the persistent connection.  It is therefore safe for an
> HTTP/1.0 server to act like an HTTP/1.1 server in all things except
> the HTTP-version. That is not an optimal way to behave, but it is
> sufficient for testing.  If you want to make it more optimal, key the
> persistence off of an unusual "apache-test" user-agent field.
> 
> ....Roy




Mime
View raw message