Re: Clarification on Alternative Service Connection Failures

On Fri, Sep 29, 2017 at 3:48 PM, Ryan Hamilton <rch@google.com> wrote:

> On Fri, Sep 29, 2017 at 6:27 AM, Patrick McManus <mcmanus@ducksong.com>
> wrote:
>
>> The text is referring to any failure - the client MAY want to try a
>> different alternative in that case. Its just saying don't feel pinned down
>> by an Alt-Svc announcement just because you started to use it - you can
>> always use a different one (or the default) if you think it would work
>> better. A 500 seems a reasonable input to that choice to me.
>>
>
> ​Wow, really? Hm. While I understand your reasoning here, I don't love it.
>

I think I should underscore my point - these hosts form a set of equivalent
alternates. If one of them isn't working well for you - using another one
is a sensible thing. Errors that seem to be about the host, rather than the
resource, seem like valid input into that decision. There is never a
directive that you have to use any particular alternate.. the ordering
indicates the server's preference but the client is not bound by that.


> For example, what about 4xx errors?
>

other than the special case of 421, that's not related to the host or
connection.. it should be a function of the origin and the request. so I
wouldn't think that's a reasonable input to the algorithm. But it wouldn't
be out of spec to panic and flush the alt-svc cache on seeing it (though
not very helpful either).


>   Is it "safe" to advertise an alternative which is not current reachable
> (I think so).
>

I think so too. That's definitely a connection error :)


> What about one which 404s for all requests, or one which 500s every
> request? Should this be considered "safe" for a server to advertise?
>

500 is an unexpected error on the server, that's pretty different than a
404.


> I'm not sure I have a terribly well thought out idea here, but this makes
> me a bit nervous.
>
>