W3C home > Mailing lists > Public > ietf-http-wg@w3.org > July to September 2014

Re: handling bad priority parameters

From: Matthew Kerwin <matthew@kerwin.net.au>
Date: Sat, 23 Aug 2014 11:53:04 +1000
Message-ID: <CACweHNCrDH6Ehapp5ev7tGOiMzskw9BSzbHaLgk80h8LDtadbw@mail.gmail.com>
To: Martin Thomson <martin.thomson@gmail.com>
Cc: "ietf-http-wg@w3.org" <ietf-http-wg@w3.org>
On 23 August 2014 03:52, Martin Thomson <martin.thomson@gmail.com> wrote:

> On 21 August 2014 20:30, Matthew Kerwin <matthew@kerwin.net.au> wrote:
> > 1. What's the appropriate reaction if a HEADERS or PRIORITY frame
> includes
> > an invalid stream dependency? i.e. too large, odd when it should be even,
> > etc? I can see a case for allowing future IDs into the tree, completely
> > ignoring bad priority data, or for throwing a stream error. Which should
> it
> > be?
>
> My intent, which I failed to capture, was to say that if you don't
> have priority state for the parent stream, the dependent stream
> instead is given default priority (that means stream 0, weight 16).
>
>
Understood. I guess it doesn't hurt anyone too much if the tree gets
mangled because of a desync.



> > 2. What do we do if we receive a trailing HEADERS frame that contains
> > priority info?
>
> I think that we should fix that too.
>
>
> https://github.com/http2/http2-spec/commit/620348d708aba641e4e09b0b7abd6960c199a764
>

Much better, thanks.‚Äč


-- 
  Matthew Kerwin
  http://matthew.kerwin.net.au/
Received on Saturday, 23 August 2014 01:53:32 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 30 March 2016 09:57:10 UTC