Re: ENABLE ⇒ PROVIDE | Re: SETTINGS_PRIORITY_SCHEME | Re: Setting to disable HTTP/2 Priorities

[Matthew Kerwin <[e-mail]>]

On Thu, 1 Aug 2019 at 04:10, Kari Hurtta <[e-mail]>
wrote:

> > On Wed., 31 Jul. 2019, 02:56 Lucas Pardue, <[e-mail]>
> > wrote:
> >
> > > Hi Kari,
> > >
> > > On Tue, Jul 30, 2019 at 4:52 PM Kari Hurtta <
> [e-mail]>
> > > wrote:
> > >
> > >> Why boolean ("ENABLE") ?
> > >
> > >
>
> >
> > At the risk of bike-shedding, I think calling it "enable" is a bit of an
> > issue. The setting, as an advertisement of the sender's capability,
> should
> > say something like "will ignore" (for disabling 7540 priorities) or "can
> > understand" (for enabling some other scheme).
> >
> > Unless we also feel the need to advertise "will not send"?
> >
> > Cheers
> > --
> > Matthew Kerwin
>
> Yes,
>
>
> SETTINGS_ENABLE_* typically changes how HTTP/2 connection is
> prosessed or allows some usages which otherwise produce
> protocol error (example: SETTINGS_ENABLE_CONNECT_PROTOCOL).
>
> Therefore these these typically allows only 0 ⇒ 1 transitions.
> Feature which is enabled, is not allowed to be turned off because
> it may effect handling of frames which are already sent by
> other peer.
>
>
"Typically"?  In the old testament there is only one 'ENABLE' type setting,
and its initial value is 1.  The extended canon introduces a single new
setting that works the other way.

Also remember: the initial handshake includes settings, so these can be
disabled from the beginning; and: settings have an ACK, so you have a
definite epoch where "already sent" no longer applies.

But yeah. what's significant about them is, 1 or 0, what they say is: "the
recipient of this setting must/must not send a frame of this type".  I.e.
the sender of the setting says, "I will/won't understand these frames".
It's not "I will/won't *send* them myself."  I think that means we're
agreeing that 'ENABLE' isn't the right word.



> Therefore I think that
>
>     SETTINGS_PROVIDE_HTTP2_PRIORITIES
>
> is better name than
>
>     SETTINGS_ENABLE_HTTP2_PRIORITIES
>
>
> In other words this does not chahneg parsing of HTTP/2
> frames. Even when this is set to 0, sending priorities
> does not cause protocol error.
>
> In SETTINGS_PROVIDE_HTTP2_PRIORITIES server asks
> client to provide (or not provide) http/2 tree
> priorities.
>
>
> ( Another possible name is
>
>   SETTINGS_PROVIDE_TREE_PRIORITIES
>
> )
>
> / Kari Hurtta
>

Cheers
-- 
  Matthew Kerwin
  https://matthew.kerwin.net.au/