Re: A modest proposal

> For the transition phase I still think that we should add Lou's SIZE
> parameter to I-M-S.  Most of the cache corruption is truncation due to
> the fact that HTTP uses closing of connection as EOF and far too many
> implementations leave truncation unnoticed (and it's not even possible
> to notice it if there is no C-L header).

Why is getting someone to pay attention to 'size' easier than getting
them not to cache truncated data or data without C-L?

If the data didn't have a C-L in the first place, how would the server
be able to check the length against a I-M-S parameter?