Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:72357
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: pass (pb1.pair.com: domain cypressintegrated.com designates 173.1.104.101 as permitted sender)
Date: Thu, 6 Feb 2014 19:43:22 -0500
Reply-To: Sanford Whiteman <swhitemanlistens-software@cypressintegrated.com>
Message-ID: <765772707.20140206194322@cypressintegrated.com>
To: Daniel Lowrey <internals@lists.php.net>
In-Reply-To: <CAH8Mpn=LOWn1EkaoeZBd+Rp9WM7fLMMLm-LJAnSNohH19Hmf_g@mail.gmail.com>
References: <CAH8Mpn=LOWn1EkaoeZBd+Rp9WM7fLMMLm-LJAnSNohH19Hmf_g@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Subject: Re: [PHP-DEV] Re: Re: Windows Peer Verification
From: swhitemanlistens-software@cypressintegrated.com (Sanford Whiteman)

>     If you don't trust the data or want to be more certain: run the
>     script yourself. Offering the data over HTTPS would give you a
>     chicken-and-egg problem as which CAs would you trust when
>     you download the bundle?

Thanks for posting this. I find their excuse entirely fallacious.
There is no chicken-and-egg "problem" as all OSes and browsers ship
with trusted CAs, and every single one of us works under the
assumption that it is possible to securely download and/or purchase an
OS. You don't assume that the first site you visit in IE or Safari
(probably to get Firefox or Chrome!), or the first connection to the
Microsoft or Apple Update server, is not a peer-verified TLS
connection to the best of the browser's current ability. To suggest
that the list of trusted CAs might as well be initialized over
plain-text is an embarrassment.

If we didn't have any starting point of trust, whether that be
sneakernet from Red Hat HQ or whatever, there could be no such thing
as "trusted CAs". 

They also imply the main vulnerability is that _their copy_ will be
altered in transit from their site... ? That's not the point. The
problem is using plain text *broadens* the attack surface so that it
is even easier to make someone download your substitute copy rather
than hitting their server at all (DNS cache poisoning, IP spoofing to
direct you to attacker's DNS, straight-up compromising a DNS server
and adding a fake zone, or one-second mod of HOSTS when someone is at
the water cooler, for some examples). All that and *also* being able
to rewrite the plain-text response if you intercept it.

-- S.