httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Roy T. Fielding" <field...@kiwi.ics.uci.edu>
Subject Re: chunking too often?
Date Sun, 03 Aug 1997 04:23:22 GMT
>No it's not a waste to be able to reliably tell the client that they
>received the entire message and not part of the message.  Consider a proxy
>which is caching a response from a server when the network hiccups and
>it's only 50% of the way through the response.  Without chunking,
>Content-Length, or multipart/byteranges there is no way for the proxy to
>know that it received the entire response.  (In practice Content-MD5 can
>also do the same thing.) 

Exactly.

>I personally like us chunking on all dynamic responses.  It sets a very
>good example. 
>
>I'm suprised that rfc2068 doesn't say this is a MUST ... Roy?  I thought
>it was a MUST but I can't find it.

Requiring it would add to the minimum complexity of an HTTP server.
Since it is not necessary in order to achieve communication, it isn't
required.  Besides, it took almost a year just to convince people that
it was necessary for a client to be able to read it.

....Roy

p.s.  I am not fond of Content-MD5.  Error detection is handled at the
TCP/IP layers, and Content-MD5 does not add any security since the
field-value could be replaced just as easily as the response content.

Mime
View raw message