httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dean Gaudet <>
Subject Re: [PATCH]: "Undying keepalives in 1.2b7" on Solaris 2.x (fwd)
Date Thu, 13 Mar 1997 03:37:25 GMT
Only the existence of a Transfer-Encoding header is checked for, not its
content... this problem exists without the patch too.  So it's assuming
that all Transfer-Encodings properly delimit the length of the response in
some way. 


On Wed, 12 Mar 1997, Roy T. Fielding wrote:

> >On reading rfc2068 it isn't clear to me that a Transfer-Coding is
> >*required* to provide "safe transport" such as that required by
> >keepalive...  It's strongly hinted at but I didn't see a MUST or a SHOULD. 
> I don't understand.  The patch only sets chunking if there is no
> message-length mechanism already defined for the message.  A defined
> message length is always desired in HTTP/1.1, even without persistence,
> since it is the only way the recipient can tell if they received the
> entire message or just a truncated portion.  This can be of significant
> concern in situations like when HTTP is used to transfer hospital patient
> records from a central database to a Nurse's station.
> .....Roy

View raw message