- From: Scott Lawrence <lawrence@agranat.com>
- Date: Wed, 2 Feb 2000 13:38:18 -0500
- To: jordi ros <jros@ece.UCI.EDU>, http-wg@cuckoo.hpl.hp.com
> From: jordi ros > Suppose the following case: > > A- Our message includes a message-body, which means that > condition 1 in > section 4.4 is false, and > B- There is no Transfer-Encoding, which means that > condition 2 in 4.4 is > false, and > C- There is no Content-Length field, which means that > condition 3 in 4.4 is > false (note that this is possible since from section > 14.13 we have that > applications SHOULD (only SHOULD) use this field to indicate the > transfer-length when it is not prohibited by rules in > section 4.4) and > D- The content-type is not "multipart/byteranges", which > means that condition 4 in 4.4 is false, and > E- The server is not closing the connection, which means > that condition 5 is false. Your point A is where the problem lies. If you don't send a Content-Length, and you don't send a 'Transfer-Encoding: chunked', and your Content-Type is not some kind of 'multipart', then your message does not include a message-body; it ends at the CRLF following the last header. If you are a client, then anything you send following that will be interpreted as the beginning of a new request; if you are a server then I have no idea what clients might do. SHOULD does not mean 'you can ignore this rule' - it means that you don't have to do it to be compliant, but you aught to unless there is an excellent reason.
Received on Wednesday, 2 February 2000 10:42:31 UTC