Re: status code 100 (continue)

rlgray@raleigh.ibm.com:
>
>
>The way I read section 8.2 (message transmission requirements) is: 
> 
>IF ((client is 1.1) AND (NOT connection:close)) then I MUST send 
>status 100 or an error. 

The connection:close conjunct does not come into it.  It is 

IF (client is 1.1) then ....

but only for responses to certain methods.  In the text of 14.10:

|HTTP/1.1 defines the "close" connection option for the sender to signal
|that the connection will be closed after completion of the response.

`completion of the response' does not mean completion of the 100
response, but completion of the entire HTTP/1.1 response, possibly
consisting of a 100 followed by something else.  A 100 response is an
`interim response' only.

>Will clients get confused receiving 100 even if connection:close? 

Not if they are correctly implemented 1.1 clients.

>Do I *really* have to send a Date header? 

Yes, but not in the preliminary 100 response part.  The spec wants you
to send:

 -begin-
 100 Continue

 200 OK
 Date: Tue, 15 Nov 1994 08:12:31 GMT
 ....

 ....
 -end--

and the whole thing should be seen as a single HTTP response.  I
notice that the spec is not very clear on this, there should probably
be more explanatory text in 10.1.

[...]
>And, (once again) what is the rationale for the 100 response? 
>It seems unnecessary and wasteful; what am I missing here? 

I never heard a rationale that convinced me we needed it.

>Thank You, 
>Richard L. Gray

Koen.