RE: Legal tokens

> -----Original Message-----
> From: Koen Holtman [mailto:koen@win.tue.nl]
> Sent: dinsdag 12 september 2000 7:11
> To: Joris Dobbelsteen
> Cc: http-wg@cuckoo.hpl.hp.com
> Subject: Re: Legal tokens
>
>
> >If
> >
> >Connection: Close, keep-alive
> >
> >is legal, how about
> >
> >Cache-Control: Public, Private, max-age=30, s-maxage=30
> >
> >They are both buggy, and from my opinion not to be considered legal.
>
> Well, when I say legal I mean legal according to what is written in
> the specifications.  Your examples are clearly strange, but the
> specifications do not disallow these strange examples, so they are
> legal.
>
> Not that I would like to see these things on the wire, mind you.
>

Agree,

But although it's not explicitly forbidden by the RFC, logic sense (you know
what I mean) would forbid such constructions. The cache-control example was
from another discussion, and Jeffrey Mogul (& John Strake) agreed such
responses should not be cached by a public cache, since this points to a
buggy server.
But not how do you explain illegal and legal responses?
If you have software that receives in a response the connection header(s)
above, you would suppose the header is buggy and cannot be used. And since
it's not used by the software, can't we just say it's illegal?
And since it's not in a request...

>
> >
> >
> >- Joris
>
> Koen.
>
>

- Joris

Received on Tuesday, 12 September 2000 07:40:17 UTC