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

Re: h2 padding

From: Amos Jeffries <squid3@treenet.co.nz>
Date: Wed, 03 Sep 2014 23:33:26 +1200
Message-ID: <5406FC86.7010707@treenet.co.nz>
To: ietf-http-wg@w3.org
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 3/09/2014 6:15 p.m., Brian Smith wrote:
> On Tue, Sep 2, 2014 at 11:07 PM, Brian Smith wrote:
>> I actually think it is worth evaluating whether the padding
>> mechanism is practically useful as a security mechanism as
>> specified and with the above issues addressed. Has anybody
>> actually used frame padding to solve a real-world problem yet?
>> Has anybody tried to write a terrible-but-conforming
>> implementation that effectively undoes all the protection that
>> padding is supposed to offer? It seems likely that the answer to
>> both questions is "no."
> 
> Also, what is a HTTP1.1 <-> HTTP/2 proxy supposed to do with
> padding?

Gateway itself and 1.1 representation is a non-issue. Chunked encoding
trailers are available to relay padding after each DATA chunk.

What happens *inside* the 1.1-only hops is the big issue. The gateway
may need to use a CONNECT tunnel to communicate directly to the server
over 1.1 hop(s).

Amos

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.22 (MingW32)

iQEcBAEBAgAGBQJUBvyGAAoJELJo5wb/XPRjZGYH/iVs84CI7Ogv5IB9gjwqjCIF
c5WdvTCqUyE9hE8rreGPS7JfBRCU5kbmJF3W8tZwA56NpLlplB7JprO0rpR2Vpm7
/OcXaYf1HpNRF/D5CV4iSYOMBMDto/R6zt0bzJ1t5zQvSvgjcDENqA/1qex+BIBX
ueQRPr8fgpdJSUxUb0TqyuBr+91L77R+9UvdME0SC7Ws9xSsqcZKaYkGJVHs/MA7
9PcveysDM9OKdznD4Yp2X7Ugsdm7CMQoInHwhDLEa6nUPMlAM4eDtVC+CT3NpDhz
Xsldc4OY4ZKNmkbUnL4o84PqK1xLpg2OXBkVWSW3UhegdTZngtwncRET2BC4/J8=
=Ogmf
-----END PGP SIGNATURE-----
Received on Wednesday, 3 September 2014 11:34:01 UTC

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