Re: bug in libwww-perl-5.05

• Messages sorted by: [ date ][ thread ][ subject ][ author ]
• Next message: Clinton Wong: "LWP::UserAgent bug with relative url redirection with URI::URL::strict(1)"
• Previous message: Johannes Kabisch: "MOMSpider"
• In reply to: Martijn Koster: "Re: bug in libwww-perl-5.05"
• Next in thread: Martijn Koster: "Re: bug in libwww-perl-5.05"

I wrote:
 
> > 'www.informatik.uni-kiel.de' sends an empty last line. 
> > 'www.netuse.de' does not. I checked the RFC and I think this 
> > behavior is legal.
 
Martijn Koster and Guy Decoux told me it wasn't. From a formal point 
of view this is correct. Dropping the Entity-Body does not mean dropping 
CRLF too. My fault, sorry.
 
Keith Jones wrote:
 
> It would be good if this was fixed though, since browsers are 'supposed' 
> to be robust enough to cope with servers not sending responses =exactly= 
> right.
 
Appendix B of RFC 1945 tells us to be tolerant against applications.
So maybe it's time for tolerance :-)
 
Martijn Koster wrote:
 
> I should look more closely at the .pm's, but from a glance at your patch
> it might break servers who send legal responses that contain headers
> and body in a single write, and thus have more than just the header
> read in the first read_until?
 
from RFC1945:
 
[...]
 
8.2  HEAD
 
   The HEAD method is identical to GET except that the server must not
   return any Entity-Body in the response. 
 
[...]
 
Sending a _legal_ response to a HEAD request contaning header _and_ body 
is not possible.
 


Willi

• Next message: Clinton Wong: "LWP::UserAgent bug with relative url redirection with URI::URL::strict(1)"
• Previous message: Johannes Kabisch: "MOMSpider"
• In reply to: Martijn Koster: "Re: bug in libwww-perl-5.05"
• Next in thread: Martijn Koster: "Re: bug in libwww-perl-5.05"