RE: http/2 & hpack protocol review

On Monday,05 May 2014 18:51, martin.thomson@gmail.com<mailto:martin.thomson@gmail.com> wrote:
> On 5 May 2014 08:50, Erik Nygren <erik@nygren.org<mailto:erik@nygren.org>> wrote:

>> I do think that the idea of a light-weight HTTP/2 profile may make sense.

>

> This has been proposed, but it wasn't considered interesting enough to specify.  See [1].

> ...

> [1] tools.ietf.org/html/draft-montenegro-httpbis-http2-server-profiles



I could only find one place this draft was even mentioned on the mailing list [1]. Apparently nobody responded to Gabriel and Roberto?  Is that the definition of not interesting?  For the record, I'm now expressing an interest :)



I think it sounds like a great idea, though it doesn't solve the problem for non-TLS case.



As Peter pointed out in his review, an endpoint can "simplify" its implementation by setting the SETTINGS_* to appropriate values.  But some of it still can't be simplified e.g. in HPACK we can set the table size to 0, but we can't disable Huffman. Why not?



I think the default SETTINGS_* should be at the levels similar to what Gabriel and Roberto proposed to call the "constrained server profile" [3].  This would only affect the connection for one RTT until the server is able to upgrade its settings to the "normal server profile".  Changing the SETTINGS_* in the other direction would "overrun any of the server's preferences" (as Gabriel and Roberto called it).



(For the TLS case I think the Gabriel and Roberto's idea is the better way as you can immediately start at the server's settings preferences.)



[1] http://lists.w3.org/Archives/Public/ietf-http-wg/2013JulSep/1141.html


[2] http://lists.w3.org/Archives/Public/ietf-http-wg/2014AprJun/0602.html


[3] http://tools.ietf.org/html/draft-montenegro-httpbis-http2-server-profiles-00#section-2






This email message is intended only for the use of the named recipient. Information contained in this email message and its attachments may be privileged, confidential and protected from disclosure. If you are not the intended recipient, please do not read, copy, use or disclose this communication to others. Also please notify the sender by replying to this message and then delete it from your system.

Received on Tuesday, 6 May 2014 16:49:48 UTC