- From: Yngve N. Pettersen (Developer Opera Software ASA) <yngve@opera.com>
- Date: Sun, 09 Dec 2007 21:59:58 +0100
- To: "public-wsc-wg@w3.org" <public-wsc-wg@w3.org>
Hello all, Below is my comments/questions based on my own review of the WSC-xit document. Other comments from Opera will be sent Monday. ------- 5.2 Relaxed certificate path validation This is a bit on the vague side. A couple of examples would help understanding the definition, as would using "revocation" instead of "status". Also, the use elsewhere in the document should be clearer. 5.3 Certificates Should this be before 5.2? 5.3.4: should this be removed? No specification, at least yet. 5.3.7 What should selfsigned certificates be indicated as after the probation (and for that matter, how should they be presented during the probation). (potential ISSUE?) 5.4: "HTTP tansactions" -> "HTTP transactions" "This definition implies that inline images, stylesheets, script content, and frame content for a secure page need to be retrieved through strongly TLS protected HTTP transactions in order for the overall page to be considered TLS-secured." Suggest rewrite to "This definition implies that inline images, stylesheets, script content, frame content, and all other content retrieved by the user agent for a secure page need to be retrieved through strongly TLS protected HTTP transactions in order for the overall page to be considered strongly TLS-secured." This will include applets and content requested by applets. I would also prefer that similar language be added about AA sites. (potential ISSUE?) Reference: http://my.opera.com/yngve/blog/2007/06/19/it-aint-ev-til-its-ev-all-ev Reference: "What is a secure page", recommendation #6/#16 5.5.1 Need more precise language. It is too short now. 5.5.2 (from other reviewer) conditions for changed security level not quite clear enough. Particularly the "terminates on a Web Site different" point 5.5.3 ", or a no-interactoin certificate;" -> "no-interaction" (although see 5.3.4) 5.5.4 https always means TLS protection is expected. even if the user also expects it to be done using a selfsigned cert. 6.1.2 What does " the identity signal must be indistinguishable from one that would be shown for an unprotected HTTP transaction, unless a change of security level has occured" mean? In which direction did the level change? "Extended validation"-> "AA cert"? 6.2 RFC 2965 is the current cookie spec, but the Netscape "spec" is the de-facto one. 6.4.2 I think this could be embellished a little bit more, in particular about the constructed URL. 7.1 "For each certificate chain received from SiteB". Does this mean that the client will have to store the history of certificates it has seen for a given site? (1) "currently valid": does that mean the client should perform a full certificate check, including revocation checks? (2) Need clarification about what "same name" means (4) What about ("OU") department match/non-match? Potential problem: Some root CAs are using a root they bought from another (often defunct) company, but are in the process of moving to a new one. This will cause match failures. Can't do anything about that. Should there be a way to detect Domain Validated certs? Usuaully "CN" == "O". What about hostname/domain matching? I am not sure I like the implied ability to share data between sites with different names. Also, we may need special handling of self-signed certificates And: How do we defend against spoof roots? Either installed by the user (intentionally, or not), or by malware? Filtering proxies can be configured to present a generated certificate that permit the proxy to filter secure content. 7.3 (from internal comment) The actions listed and the text does not match. Actions listed as numbers, text assume "(a)", "(b)". Those are listed above as (a) and (b), but the mixing is confusing. 7.8 What if the user want to refurbish the UI? 9.2 Add "key" as alternative to padlock? Should it be clarified as being written for deployment of TLS-protected services? 9.3 "and servers MUST NOT set credential cookies from secure servers that can be sent unencrypted" Should the be reordered? "and secure servers MUST NOT set credential cookies that can be sent unencrypted". 9.4 "redirect to an unsecure page and then redirect again to a secure page" Might want to make this clearer. 10 Primary problem of this is how to encourage the user to switch to the mode. -- Sincerely, Yngve N. Pettersen ******************************************************************** Senior Developer Email: yngve@opera.com Opera Software ASA http://www.opera.com/ Phone: +47 24 16 42 60 Fax: +47 24 16 40 01 ********************************************************************
Received on Sunday, 9 December 2007 21:00:38 UTC