Re: [dss] Comments to draft 6

["Simeon Falk Sheye" <Simeon.Sheye@cryptomathic.com>]

A technical rationale for delayed verification responses is that the signature service
may be used by a backend system for validating bulks of incoming signed data (forms, mail etc).

An entirely different reason to make delayed response a general option is simplicity in the solution.
I think the final service becomes simpler to use if behaviour is uniform across all protocols,
without too many special cases or exceptions to the rule.

- Any opinions on this?

Simeon
Trevor Perrin <trevp@trevp.net>
06/02/2003 10:36 MST

To: simeon.sheye@cryptomathic.com, dss@lists.oasis-open.org
cc:
bcc:
Subject: Re: [dss] Comments to draft 6

At 09:01 AM 6/2/2003 +0000, simeon.sheye@cryptomathic.com wrote:

>Section 3.4.3 states that "The server may return the signature
>immediately, or may return a transaction identifier, which instructs the
>client to call back periodically and see if the signature is ready yet".
>
>I propose to move this into section 3.3 (generic request reqs). This would
>allow the response to any request to be delivered either immediately or
>delayed. The bullet in line 330 (section 3.5.2) should go with it.

As I recall there were 2 rationales for delayed response on the "signing"
request/response exchange (I think we're in agreement this exchange is also
going to be used for time-stamping):

- where a human or other process has to approve the signature
- linked timestamping may require it, in certain cases

I'll add those rationales to the document itself.  Is there a good
rationale for allowing delayed response to the "verifying" request/response
exchange?


>In section 3.7.4 ("time-stamp/time mark"), the text says "If the
>verification was carried out for some time in the past (see 3.6.2, bullet
>2), a time-stamp or time mark for that time should be returned as well."
>I think we should be cautious here because this makes it possible to
>produce timestamps that dates back before the signature was actually created.
>I propose that back-dated time-stamps are ruled out completely.

Sounds good.  I'll take that sentence out unless someone disagrees.

Trevor