Re: Negotiating Window Limits for Content Encodings

Hi Martin,

On Thu, Aug 05, 2021 at 02:18:10PM +1000, Martin Thomson wrote:
> On Thu, Aug 5, 2021, at 12:38, Willy Tarreau wrote:
> > Given what you said about Brotli, it seems that it should be generic,
> > though it's not certain that such an argument easily applies to all
> > algorithms.
> 
> I would avoid making this truly generic, but there is value in establishing a
> convention.  If brotli wants to use the same concept, but set w=18 to mean
> something else, that's probably OK.  What the convention will mean is that
> you can build software that knows that there is a w parameter and that
> Accept-Encoding: foo;w=n is an acceptable match for Content-Encoding: foo;w=m
> for any m <= n.  That is generally applicable logic, even if it has to be
> applied on a per-format basis.
> 
> (It can't be truly generic as identity;w=3 means nothing, there are other
> content-codings for which w would be meaningless, and we haven't defined
> gzip;w=4.)

I agree with your points, which is also why I suggested that a level could
be more generic. In addition, encoding doesn't necessarily mean compression,
we could for example imagine encrypting the document with a known key, and
then probably that the key ID could be passed the same way. I know it's a
bit far-fetched, and we ought to be careful not to start to reinvent the
wheel, but that seems to indicate that the argument is probably something
specific to the token itself, though the format ought to be generic.

Willy

Received on Thursday, 5 August 2021 14:13:18 UTC