RE: multipart/@boundary

>   In the current CR, the section "7.1.10.2 Request Entity body
> conversion" defines c:multipart/@boundary as "The boundary attribute
> is required and is used to provide a multipart boundary marker."  So
> it is required.
> 
>   But the section does define the element for both request and
> response, and I guess the attribute is required only for requests.
> Actually, I think it should be required in requests, and forbidden in
> responses.  At least, I don't see what possibly could be its purpose
> in a response.

Hi,

Hmm, interesting point... One possible advantage of having the boundary
information in the response could be that it makes it possible to
"re-post" the response data to the server as-is: you just replace the
c:response wrapper with c:request and pass it to p:http-request, without
being forced to set the boundary information in c:multipart.

Also, what if somebody was interested in what the boundary string sent
by the server looked like? (I know, probably an obscure use case, but
still a use case... :)

But I guess that the real question is: Does/can having the boundary
information in the response cause any issues? 

Regards,
Vojtech

Received on Wednesday, 14 January 2009 11:14:06 UTC