From: Charles Lindsey (chl@clw.cs.man.ac.uk)
Date: Tue Jun 01 1999 - 04:42:12 CDT
In <19990531161617.08803@main.templetons.com> Brad Templeton <brad@templetons.com> writes:
>On Mon, May 31, 1999 at 08:16:50PM +0200, Brad Knowles wrote:
>> At 10:37 AM -0700 5/31/99, Brad Templeton wrote:
>>
>> Given that multipart/signed is already a MIME standard, if
>> anybody wants to talk about signing articles, you'd have a very hard
>> time convincing me that we should even bother considering any other
>> alternative.
That is not suitable for header-signing.
We have three urgent applications for signing headers:
1. Newgroup messages (currently uses pgpverify)
2. Moderated articles (currently uses pgpmoose, incompatible with pgpverify)
3. Third party cancels (needed so we can release the cancel lock proposal)
Other applications will follow, but those are the urgent ones, and what
they have in common is that the headers in question (Control, Approved,
etc) MUST remain in the main headers of the article (for compatibility
with present usage) and NOT be moved into some multipart.
I am quite happy for people who just want to sign ordinary articles to
prove their authorship to use multipart/signed (though even that is not
proof against changes of Content-Transfer-Encoding en route). I know Brad
T does not share this view.
>The general feeling is that since all existing newsreaders would display it
>in an ugly fashion, it is not a suitable format for the transition to
>signed articles. It is also wasteful (double all the headers) and
>complex (ignore "real" headers, use doubled headers extracted from body),
>but the main reason is it might be rejected by users of existing newsreaders.
Multipart/signed is not so ugly as all that on old newsreaders (no worse
than plain PGP body signing, and at least it avoids that "-- -" stuff).
And with Mime-compliant readers (even poor ones that do not properly
understand it) it is presented quite reasonably. I shall sign this message
that way, so people can see how it appears.
>>
>> In fact, since I thought we were supposed to be going through
>> the draft chapter-by-chapter, and since cryptographic signing is not
>> yet something that is in the draft (to my knowledge), I think that
>> this is a discussion that we should table until such time as we've
>> gotten the base level work out of the way.
>The question at hand was should the standard require that the static
>headers be invariant and at the bottom of the header, and the dynamic
>headers be at the top. To address the questions of why and why not do this,
>one issue brought up is that doing it allows better options for
>compression, signature and efficiency.
Quite so. I agree that header invariance is a desirable aim, though I do
not believe we are in a position to rely on it for any signature scheme.
We are goping to have to consider this issue fairly soon anyway (as soon
as the present pass through the draft is complete), so I see little harm in
some preliminary look at the possibilities available.
-- Charles H. Lindsey ---------At Home, doing my own thing------------------------ Email: chl@clw.cs.man.ac.uk Web: http://www.cs.man.ac.uk/~chl Voice/Fax: +44 161 437 4506 Snail: 5 Clerewood Ave, CHEADLE, SK8 3JU, U.K. PGP: 2C15F1A9 Fingerprint: 73 6D C2 51 93 A0 01 E7 65 E8 64 7E 14 A4 AB A5