Re: #282: Recommend minimum sizes for protocol elements

Just to be clear -- we're talking about the total size of the *entire* header block here, not a single header limit.

Were the folks arguing for 4k assuming the former or the latter?

On 25/06/2011, at 4:13 AM, Willy Tarreau wrote:

> On Fri, Jun 24, 2011 at 11:05:18AM -0700, David Morris wrote:
>> Sounds like a good idea to me. I still professionally live in a dial-up
>> world so less is better.
>> Since I've not followed this topic from the beginning, I appologize if
>> the following is a repeat ... but it seems to me that the early/original
>> cookie specifcations allowed for a 4k cookie value? Folding multiple
>> values into a single header might be an issue. Might want to note that
>> if folding would cause the field length to be violated, then the
>> mulitple headers shouldn't be folded?
> Note that we're not speaking about a strict rule which could be violated,
> just a hint about what size should have a high chance of passing without
> complex verifications.
> Many products will support more, but at least senders will know that if
> they want to rely on more than 4kB they're among the rare people doing
> so and that a validation of the whole chain would be a good idea before
> thinking that everything will work out of the box. In short, this
> recommendation should progressively motivate implementations to try to
> send smaller headers, or at least not to waste too much space for no
> reason.
> Regards,
> Willy

Mark Nottingham

Received on Saturday, 25 June 2011 02:38:50 UTC