- From: Roland Zink <roland@zinks.de>
- Date: Fri, 12 Jun 2015 22:51:09 +0200
- To: Martin Thomson <martin.thomson@gmail.com>
- CC: HTTP Working Group <ietf-http-wg@w3.org>
On 12.06.2015 22:38, Martin Thomson wrote: > On 12 June 2015 at 13:12, Roland Zink <roland@zinks.de> wrote: >> So the additional input is missing so far in the text and there are no plans >> to add them. I find it confusing if sometimes proxies should be used and >> sometimes not depending on how the proxy to use is configured. > Currently: "A client configured to use a proxy for a given request > SHOULD NOT directly connect to an alternative service for it, but > instead route it through that proxy." > > I take that to mean that you would query proxy.pac with the request > URI. That's limited, but no worse than what we have today. If you > are motivated, you could fix the shortcoming, but the pragmatist in me > notices the efficacy of Mark's text in avoiding the need to specify > something that depends on a non-existent specification for proxy > discovery. My concern is about a setup routing plain text http through a different proxy than TLS traffic. In this case this approach may fail. Don't know how often such an approach is used in practice. Roland
Received on Friday, 12 June 2015 20:51:44 UTC