Re: Comments on draft-ietf-httpbis-http2-04

[Julian Reschke <[e-mail]>]

On 2013-07-10 16:42, Jeff Pinner wrote:
> Julian, if you recall (or have easily accessible) I'd love to hear the
> rational for the "ought to be handled as an error" line as opposed to a
> "SHOULD return a 400" or "MUST return a 400."
>
> Perhaps it is due to some subtlety that should lead us to reconsider the
> 2.0 requirement? Or perhaps it is due to legacy implementations and it
> could guide us in wording to add to a HTTP/1.1 <--> HTTP/2.0 section?
>
> - Jeff

Generally, when we say "ought to" it's because it's a really good idea, 
but for some reason we really can't require it. In this particular case 
I suspect that it's a problem common enough so that existing UAs simply 
can't enforce it (the error handling).

Best regards, Julian