MisterSSL, I'm rather appalled that you are ignoring the realities of
US government user requirements.  I can state with enough knowledge of
regulation and policy behind me that I believe that it is primarily
due to your lack of acknowledgement of the requirements in-place that
Firefox has not enjoyed greater US government agency penetration.
These requirements are set not merely in stone but in the Rock of
Gibraltar.

You, in your office of NSS component manager, continue to insist that
there is no reason to allow any central Administrator-defined policy
to make any change whatsoever to:

a) the choice of root certificates used
b) the choice of PKCS#11 modules used
c) the prevention of unapproved PKCS#11 modules from being used
d) any cryptographic requirement whatsoever

To the US government, users have no privacy, because the computers are
never under any circumstances to be used for non-US-government
business or for personal use.  This goes for civilian agencies as well
as for military agencies.  Users, when they are on the clock, are
agents of the US government, subject to monitoring and auditing of
resource usage.  For all intents and purposes, what the user "prefers"
is irrelevant.  Computers are tools, and are provided as such.

This is an extreme version of fairly-common requirements of corporate
and educational policies.  Educational policies are often as difficult
to change as governmental policies (since many schools are public
institutions and are thus bound by requirements common to all public
institutions in the given jurisdiction); corporate policies often
appear just as difficult to change to people "in the trenches" who
have to make things work in the face of mandates from on high.  It
must be noted that these mandates often have Governmental origin; for
example, Sarbanes-Oxley.  In addition, private educational and
health-care entities are subject to additional requirements (the
Buckley amendment in the former case, and the Health Insurance
Portability and Accountability Act in the latter) -- as well as the
standard regulations affecting all fiduciary entities including banks,
insurance companies, accountancies, lawyers, etc.  In all
circumstances, there are two conflicting needs: the need to ensure
that an approved cryptographic algorithm is used to protect the
information to a minimum standard, and the Judiciary's mandate to
ensure that any and all electronic communication can be readily and
reliably retrieved.

Microsoft makes available tools to ensure, for example, that only
FIPS-validated cryptography can be used by the CryptoAPI.  This is one
reason why they have sought FIPS validation for their cryptographic
modules so stringently -- among other things, the CSRC lists
certificate numbers 1010, 1009, 1008, 1007, 1006, 1005 and 1004 for
Windows Server 2008, and certificate numbers 1003, 1002, 1001, and
1000 for Windows Vista.  11 certifications of validity spanning two
products.  With the administrator-mandated policy available "require
FIPS-compliant cryptography", there is little wonder why Windows (for
the OS platform), Internet Explorer (for web browsing), and Outlook
(for messaging, calendaring, and all other forms of inter-user
interaction) are so widely adopted.

One of the reasons is that they are so easy to update, at a Domain
level.  The Domain contains the Policy.  The Policy may include a list
of certificates to be installed into either the machine's root
authority store or the user's root authority store.  The Policy may
also include a machine-level "require FIPS-compliant cryptography"
setting.  Upon logon to the Domain, the machine obtains the Policy and
applies it -- as well as applying it repeatedly over the course of the
logon session.  (Among other things, this provides for a means of
updating the list of root authorities dynamically.)

Such a Policy application cannot be said to exist for Firefox.  There
is no way, upon logon, to enforce the addition of a FIPS-validated
PKCS#11 module.  There is no way, upon logon, to enforce the addition
of organization-policy mandated certificates.  There is no way, upon
logon, to remove unapproved PKCS#11 modules.  There is no way, upon
logon, to prevent the installation of new PKCS#11 modules.  There is
no configuration file that NSS reads, and there is no enforcement in
NSS itself.

The commonly-proposed alternative is to use the Client Customization
Kit.  This is a kit that creates an XUL file to customize the
installation of Firefox, and which creates an extension which applies
the XUL-formatted policy file.  At first glance, this would appear to
meet the requirements.  However, there are some MAJOR problems which
prevent it from being useful.

1) It is only available as source code.
2) The source code is only available from CVS.
3) It apparently requires building a full, customized version of
Firefox (which, among other things, requires a full compiler toolchain
installation, which is itself by no means a trivial task)
4) There is no version which claims to support Firefox 3.
5) There is no documentation which suggests that it affects
Thunderbird or any other Mozilla product.
6) There is no documentation which suggests that it precludes the
addition or usage of unapproved PKCS#11 modules.

Simply put, the CCK is not an option for people who prefer to use
Firefox 3, or for anyone who wishes to use or deploy any other Mozilla
product.  For most people it has been recommended to, it never truly
was an option.

I respect your tenacity, MisterSSL, but I sincerely hope that you
realize that it is solely YOUR office and YOUR office's mandates --
mandates which you have repeatedly been requested to change, each and
every request completely ignored and the requestor directed to
something which cannot meet their needs -- which is preventing wider
adoption of Firefox, Thunderbird, and all other Mozilla Foundation
products.

Sincerely,

Kyle A Hamilton

(For reference:

http://www.mozilla.org/projects/cck/
http://www.mozilla.org/projects/cck/firefox/

These are the only pages I could find which came up on a "Client
Customization Kit Mozilla" search on google.)
_______________________________________________
dev-tech-crypto mailing list
dev-tech-crypto@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-tech-crypto

Reply via email to