httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Alexei Kosut <>
Subject Re: 1.1b3 and things???
Date Sun, 09 Jun 1996 01:25:19 GMT
On Thu, 6 Jun 1996, Robert S. Thau wrote:

> A few more lines, but I'm pretty sure I know how it will behave no
> matter *what* the incoming data looks like.  (Or I thought I was...
> so Alexei, how did I blow it this time?)

Not sure. While that code may work, I'm quite sure (is that more sure
than pretty sure?) that my sscanf() code will behave correctly no
matter what the incoming data looks like. And it's only three lines of
code, as opposed to your eleven. *shrug* At any rate, it should be in
read_request_line() and stored in the request_rec, not in
set_keepalive() - this I'm sure of, since other functions may need
access to the information.

On a related note, I am concerned about one thing... I think this is a
question for Roy; we are a HTTP/1.0 server that wants to allow
HTTP/1.1 clients to connect to us and have persistent connections. Can
we do that?

The current apache-XX sends Connection: Keep-Alive back to the client
in all cases, which I'm quite sure is wrong - it should only do this
for HTTP/1.0 requests that had a Connection: Keep-Alive. And we can't
do nothing, because then the client will assume that we're not
persistent-capable, and will wait forever for the connection to close.

I think the answer is going to have to be something like this,
assuming the protocol numbering part of spath.patch is installed:

int http_11_or_better = 0;
int http_11_or_better = (r->proto_num >= 1001) ? 1 : 0;

i.e. the code can only be activated once we switch to HTTP/1.1 as the
protocol version.

Unless it's my turn to have mis-read the spec.

Alexei Kosut <>      The Apache HTTP Server
      "War does not determine who is right, only who is left."

View raw message