- From: Close, Tyler J. <tyler.close@hp.com>
- Date: Tue, 8 Jan 2008 19:44:02 +0000
- To: Mary Ellen Zurko <Mary_Ellen_Zurko@notesdev.ibm.com>
- CC: "public-wsc-wg@w3.org" <public-wsc-wg@w3.org>
- Message-ID: <C7B67062D31B9E459128006BAAD0DC3D145A3CBB@G6W0269.americas.hpqcorp.net>
I think the browser hook is the major missing infrastructure bit. For example, assume the browser can be configured with a URL identifying the notification service. If the user clicks the "report problem" button, the browser posts all the relevant certificate chains to the notification service. In an enterprise, browsers would be configured to notify the IT department of misconfigured servers. For home users, the browser manufacturer may offer a default notification service that makes a best effort attempt to notify the sysadmin for the server, and otherwise just keeps the information in a publicly accessible database that sysadmins can check. I'm sure people will want to criticize this particular design and perhaps offer alternatives. The important point is to put the hook in the browser to enable these kinds of services. It is a good thing to quickly identify and resolve spurious MITM-like attacks. They provide noise in which the attacker can hide. --Tyler ________________________________ From: public-wsc-wg-request@w3.org [mailto:public-wsc-wg-request@w3.org] On Behalf Of Mary Ellen Zurko Sent: Tuesday, January 08, 2008 9:23 AM To: Close, Tyler J. Cc: public-wsc-wg@w3.org Subject: RE: ISSUE-127: Safe Form Bar: Separate MITM handling? [Techniques] It's working through viable deployable patterns for the infrastructure behind it that has me doubtful (not the coding in the user agent, which, as you rightly point out, is quite modest). Is there enough infrastructure bits lying about for this? Is there place to put a mailto: url that could be used, for example? Just what would it be associated with? Mez From: "Close, Tyler J." <tyler.close@hp.com> To: Mary Ellen Zurko <Mary_Ellen_Zurko@notesdev.ibm.com> Cc: "public-wsc-wg@w3.org" <public-wsc-wg@w3.org> Date: 01/07/2008 06:36 PM Subject: RE: ISSUE-127: Safe Form Bar: Separate MITM handling? [Techniques] ________________________________ There's a chicken and egg problem here. If the browser provides the hook, then maybe the service will be developed. If the browser doesn't provide the hook, then we're stuck with the pitiful options we have now. It's not like it's such an incredible implementation burden, that we need to ensure a browser can be "conditionally compliant" without it. It's one configuration option and another button in a rarely seen dialog if the option is set. --Tyler ________________________________ From: Mary Ellen Zurko [mailto:Mary_Ellen_Zurko@notesdev.ibm.com] Sent: Monday, January 07, 2008 3:17 PM To: Close, Tyler J. Cc: public-wsc-wg@w3.org Subject: RE: ISSUE-127: Safe Form Bar: Separate MITM handling? [Techniques] >From my point of view, because we don't have an alert service that's useful. That's why I was OK with MAY. I get that it would be a nice thing to have. But the infrastructure doesn't exist to make it work often enough for a SHOULD. Mez From: "Close, Tyler J." <tyler.close@hp.com> To: Mary Ellen Zurko <Mary_Ellen_Zurko@notesdev.ibm.com>, "public-wsc-wg@w3.org" <public-wsc-wg@w3.org> Date: 01/07/2008 06:11 PM Subject: RE: ISSUE-127: Safe Form Bar: Separate MITM handling? [Techniques] ________________________________ The text of ISSUE-160 includes the statement: "I'm still not buying the notification stuff. MAY at best." I understand there are other points bundled up in ISSUE-160, but I'ld like to get some more details on this particular point. Why exactly is offering notification a problem? I actually had a whole series of relevant experiences with the internal intranet at work this morning. Here's a story for ISSUE-160. I clicked a hyperlink to an intranet web service I use once in a while. It's certificate chain is rooted at one of the custom CAs used here. Normally, these custom CA certificates are auto-magically distributed to our desktops by the same software that does security updates. For some reason, this web service has changed certificate chains and is now using a CA cert that I don't yet have. I don't want to click through the cert warning to the service because that will reveal my username/password, which are kept in a cookie. So I can't find out who to complain to by looking at the hosted web pages. Wouldn't it be nice if the software which updates my browser's CA list could also configure a URL to be pinged when I encounter such a potential MITM attack. That way the dialog shown by the browser could offer me a nice button to click: "get someone else to deal with this problem". Instead, the button it offers me is "click here to ignore this MITM attack and turn over your password to some random computer on the intranet". --Tyler -- [1] "Web Security Context: Experience, Indicators, and Trust" <http://www.w3.org/2006/WSC/drafts/rec/#safebar-mitm> ________________________________ From: public-wsc-wg-request@w3.org [mailto:public-wsc-wg-request@w3.org] On Behalf Of Mary Ellen Zurko Sent: Friday, January 04, 2008 6:59 AM To: public-wsc-wg@w3.org Subject: Re: ISSUE-127: Safe Form Bar: Separate MITM handling? [Techniques] ISSUE-160 makes the same basic proposal, perhaps for the same basic reasons, but I'm leaving both open and cross referenced, in case the resolutions of the underlying issues turn out to be different. I agree that there should be only one place this is discussed. And from the logic of the document, it is in other places. If there is something in section 7 that should inform those other places, proposals for changes to those other places should be made. I'll give other folks a little more time on this issue to discuss, then do a straw poll of any concrete proposals on the table (so far there is one, to remove 7.9, but I'm certain there could be others that respond to the issues raised). http://www.w3.org/2006/WSC/track/issues/127<http://www.w3.org/2006/WSC/track/issues/127> http://www.w3.org/2006/WSC/track/issues/160<http://www.w3.org/2006/WSC/track/issues/160> Mez
Received on Tuesday, 8 January 2008 19:45:03 UTC