RE: interesting note

> -----Original Message-----
> From: Alex Rousskov [mailto:rousskov@measurement-factory.com]
> Sent: Friday, 9 February 2001 10:13 AM
> To: Robert Collins
> Cc: Squid-Dev (E-mail)
> Subject: Re: interesting note
>
> W3C document misquotes RFC 2616. The paragraph they point to refers
> only to the situation when a cache is trying to serve a hit to the
> client using a previously cached partial response. W3C text seems to
> imply that RFC requirement is "generic" (i.e., applied to all
> conditions). It is not.
>
> Note that in cases where Content-Length is available, the browser can
> detect aborted transfer.
>
> Alex.
>

I should have caught that... Thanks.

My understanding is that the browser should be able to detect aborted
transfers in every case with http/1.1 - if the length is known
(content-length header), a dropped connection at the wrong point will
signal that, other wise chunked encoding should be used, and that also
allows detection of early termination.

The only 'hole' is the allowance for dynamic content to be non-chunked
and the connection dropped as the terminator - as per http/1.0

So in a nutshell we can ignore that article :] - should we let them
know?

Rob
Received on Thu Feb 08 2001 - 16:58:34 MST