Mailing List Archive

HTTP/0.9 - still needed?
I'm going to use this to track the user-agents that don't claim to be
"HTTP/1.0" compliant in the first line of the request, yet still give a
"User-Agent" field, on hotwired and hyperreal - hopefully we'll get a
nice list we can post on www-talk to embarrass people :) The crux is, if
we can show that no one's using HTTP/0.9 anymore, I'd make the argument
for throwing out the assbackwards shit.

Brian


*** http_mime.c.old Sat Apr 29 16:57:18 1995
--- http_mime.c Sat Apr 29 17:33:29 1995
***************
*** 487,492 ****
--- 487,493 ----
}

void get_mime_headers(int fd, FILE *out) {
+ char errstr[MAX_STRING_LEN];
char w[MAX_STRING_LEN];
char l[MAX_STRING_LEN];
int num_inh, num_processed;
***************
*** 527,532 ****
--- 528,537 ----
note_client_accept_enc(l, out);
if(!strcasecmp(w,"Accept-language"))
note_client_accept_lang(l, out);
+ if(!strcasecmp(w,"User-Agent") && assbackwards) {
+ sprintf(errstr, "HTTP/0.9 User Agent: %s",l);
+ log_error(errstr);
+ }

http2cgi(w);
if(in_headers_env) {
*** http_request.c.old Sat Apr 15 11:05:40 1995
--- http_request.c Sat Apr 29 17:31:30 1995
***************
*** 216,227 ****

getword(w,l,'\0');
init_header_vars();
! if(w[0] != '\0') {
assbackwards = 0;
- get_mime_headers(in,out);
- }
else
assbackwards = 1;

if(!strcmp(m,"HEAD")) {
header_only=1;
--- 216,227 ----

getword(w,l,'\0');
init_header_vars();
! if(w[0] != '\0')
assbackwards = 0;
else
assbackwards = 1;
+
+ get_mime_headers(in,out);

if(!strcmp(m,"HEAD")) {
header_only=1;
Re: HTTP/0.9 - still needed? [ In reply to ]
Date: Sat, 29 Apr 1995 17:55:52 -0700 (PDT)
From: Brian Behlendorf <brian@organic.com>
Precedence: bulk
Reply-To: new-httpd@hyperreal.com


I'm going to use this to track the user-agents that don't claim to be
"HTTP/1.0" compliant in the first line of the request, yet still give a
"User-Agent" field, on hotwired and hyperreal - hopefully we'll get a
nice list we can post on www-talk to embarrass people :) The crux is, if
we can show that no one's using HTTP/0.9 anymore, I'd make the argument
for throwing out the assbackwards shit.

Brian

Well, if assbackwards is set, then the server doesn't even try to read
MIME headers (they aren't supposed to be present for "simple", i.e.
HTTP/0.9, requests, so it would hang forever if it tried), and so
there is no User-Agent field to check.

rst
Re: HTTP/0.9 - still needed? [ In reply to ]
On Sat, 29 Apr 1995, Brian Behlendorf wrote:
>
> I'm going to use this to track the user-agents that don't claim to be
> "HTTP/1.0" compliant in the first line of the request, yet still give a
> "User-Agent" field, on hotwired and hyperreal - hopefully we'll get a

You can't look at the headers in an HTTP 0.9 request - there aren't any!

Simon
Re: HTTP/0.9 - still needed? [ In reply to ]
On Sun, 30 Apr 1995, Simon Spero wrote:
> On Sat, 29 Apr 1995, Brian Behlendorf wrote:
> >
> > I'm going to use this to track the user-agents that don't claim to be
> > "HTTP/1.0" compliant in the first line of the request, yet still give a
> > "User-Agent" field, on hotwired and hyperreal - hopefully we'll get a
>
> You can't look at the headers in an HTTP 0.9 request - there aren't any!

True, if this is a valid HTTP/0.9 request. But my thesis is that there
are some browsers out there who are really HTTP/1.0 browsers and not
stating it. And I was validated - all the following browsers do not send
"HTTP/1.0" in at least some of their requests:

AIR_Mosaic(16bit)/v3.10.06.07
AIR_Mosaic(16bit)/v1.00.198.07
IBM WebExplorer /v950331 Beta
Lynx/2.3 BETA libwww/2.14
Lynx/2.3.6 BETA libwww/2.14
Lynx/2.3.7 BETA libwww/2.14
Lynx/USiS Lynx 2.3 libwww/2.14
Lynx/ALynx_2.3 AmigaPort 1.13 libwww/unknown
MacWeb/1.00ALPHA3 libwww/2.17
Mozilla/0.96 Beta (Windows)
Mozilla/1.0N (Windows)
Mozilla/1.1b3 (Macintosh; I; 68K)
NetManage Chameleon Mosaic+ 4.03

Uh.....

Brian

--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--
brian@organic.com brian@hyperreal.com http://www.[hyperreal,organic].com/