Some remarks below.
I'm sorry for the late response.
From: Denis Pinkas
Sent: Monday, September 21, 2009
To: Ezer Farhi
Cc: Juan Carlos Cruellas; dss-x
Subject: RE: Late comments on
OASIS DSS-X Profile on visible signatures CommitteeDraft
I am fine after the
surgery,working from home.
See my responses
embedded in the text.
RE: Late comments on OASIS DSS-X Profile on visible signatures CommitteeDraft
If by any chance you did not receive this email, it is right ahead.
Appreciate any remark.
From: Ezer Farhi
Sent: Sunday, August 02, 2009
To: 'Denis Pinkas'
Cc: Juan Carlos Cruellas; stefan;
Subject: FW: Late comments on
OASIS DSS-X Profile on visible signatures Committee Draft
I'm very sorry for the late response.
Please see my comments below.
Thank you very much for the detailed review and the detailed
From: Denis Pinkas
Sent: Friday, July 24, 2009 10:38
To: Juan Carlos Cruellas; stefan;
Subject: Late comments on OASIS
DSS-X Profile on visible signatures Committee Draft
I sent the comments
below to Juan-Carlos, but I got no reply from him.
So I send them directly
to the people that have their e-mail address on the front page of the document.
BTW, the e-mail address
from Juan-Carlos on teh front page is wrong: it should end with "edu"
rather than "ede".
Juan Carlos and all,
Please find below late
Juan-Carlos would you be
able to post them to the DSS-X committee
since the link indicated in the mail does not work ?
Late comments on version
1.0 of the "Visible Signature Profile".
Source: Denis Pinkas. Bull SAS.
After two readings of the document, I still have difficulties to understand
how the protocol may be usable in practice with signature formats like CAdES
The Visible Signature profile performs is aimed to embed visible information
upon a given content such as PDF document, MS Office documents, etc in addition
to the actual digital signature operation that is handled through the DSS core
In the case
that the client wishes to use an advanced signature (CAdES, XAdES), the client
will need to use two profiles in addition to the core:
signature profile, which is specified at http://docs.oasis-open.org/dss/v1.0/oasis-dss-profiles-AdES-spec-v1.0-os.html
and the visible signature profile.
As an example,
you can incorporate a visible signature to a PDF document in addition to an
advanced digital signature embedded in the PDF document (at the level of
conformance that is applicable today by PDF).
ETSI standardization process that is aimed for enabling advanced signatures in
PDF files will probably incorporated into a new version of the advanced
signature profile (which also in combination with the visible signature
profile, enable supporting visible advanced signature in PDF files).
Whatever the protocol may be, the end result should be incorporated either into
the signed document itself
or in the electronic signature.
the protocol tries to be aligned with existing document implementations such as
PDF or Office 2007. And indded it is incorporated into the signed document
are in disagreement here.
As I said: "... should be incorporated either
into the signed document itself or
in the electronic signature itself.
[Denis] I see
difficulties to incorporate the visible signature inside the document. It means that the visible signature API must be applied
before the document is signed. It modifies the document to be signed
and the "What you see is what you sign" property is no more
applicable. Or should the visible signature become part of "What you see
is what you sign" property ? Does it mean that the end user should be able
to "see" the document with the visible signature field before
the document is signed ?
I see it differently and it is also implemented that way in existing product
that implement visible signature such as
Adobe-PDF and MS Office.
As part of the digital signature ceremony both visible and non visible content
will be incorporated into the document (BTW, many of the existing documents
implementation contained non visible data this is also signed as part of the
digital signature operation - the "What you see is what you
The visible signature profile goes along with this direction (which I believe
is a correct one) and performs the both adding of visible content that is
related to the digital signature act and performs a digital signature operation
within a single transaction.
The abstract is only addressing the
embedding of "visible signature characteristics into documents".
Should the abstract be left unchanged, then this would mean that support of
would not be supported for detached signatures and embedding signatures.
- The protocol mainly addresses documents types and not native XML, that
inherently addresses visibility of information.
[Denis] You do not answer to my question. Do you believe that the document
does not support visible signature
for detached signatures and embedding signatures ?
EZER - Denis, Can
you refer me to a standard, a document format or a data format that enables
digital signature along with a matching visible content or pointing to a
Currently the only existing implementations/standards that I'm aware
of are the PDF standard and the Microsoft OOXML standard (along with some
additional existing implementation).
The aim of the profile is not to create a standard for encoding visible
Secondly, the internationalization of such
visible signatures is not addressed. Since electronic signatures
are intended to be made visible, an electronic signature generated in one
country using a given language
should be made visible in another country using a different language. Each
field should thus be made visible
using a local language. This document is silent about this aspect.
It is possible to include textual information in a variety of languages by
indicating fonts for every textual element. Text can be either supplied by the
client or incorporated by the server (for example, a common name can be
extracted by the server from the user's certificate).
is not what I was thinking.The idea is to include only OIDs into the
visible signature that can be locally translated into any language.
As an example an OID for the "claimed signature time", that will be
translated into "date présumée de la signature" if the French
language is being used by the viewer of the visible signature field.
EZER - Your remark is mainly relevant to
"static" information such as the above example, and the protocol
this issue by enabling the client specify the ID of the item to be incorporated
to the visible content (in this case the identity
Of the label
is <xs:enumeration value="SignatureTime" />).
It is up to the server implementation and the type of relevant document to use
to either incorporate a real text or an OID.
However, in the case of a dynamic type of data (such as the signature time itself
or a reason for signing), it is a bigger problem to implement a multi lingual
Maybe you actually mean that you expect the protocol that in any case that a
string is passed from the client, the client will have the possibility to pass
the string in several languages (for example, a reason in English and French),
so that the server
and document implementation will be able to incorporate several strings into
visible content is signed by existing implementation that forbids any
modifications to the visible content, therefore it will not be possible to
translate visible content from one language to another.
I think that supporting different languages based on universal strings is
[Denis] See my comment above. I got the impression that you want to be
backwards compatible with existing implementations.
However, it does not mean that existing implementations are "good".
Thirdly, all the characteristics of a signature should not necessarily be made
visible at once. There should be
a concept of a first level of visibility, and for each first level, a second
level of details. At the very end,
all the visible characteristics (as requested by the signer) should be made
This "visibility level" is not embedded in the protocol since
currently no existing implementation (PDF, MS Office) enables that. The
protocol does enable you to control (in the time of the digital signature act)
which information should be incorporated to the visible signature.
It is possible
to incorporate such a parameter to the protocol, which may be used by future
[Denis]. Once again, I got the feeling that you want only to support
existing implementations whether teh approach they have taken is
"good" or "bad".
EZER - Right.
However, the protocol is not limited only for these implementations, so it is
possible to incorporate additional
functionality. However, it is important to separate between the protocol and
the implementation of the server along with the
Certain document standard.
These observations let me to conclude that the signer should be able to
incorporate a template for the
which will describe the first level, secondary level, etc ... of the
visible signature in a language independent manner.
Applications able to handle these visible signatures fields should be able to
interpret the template in a local language.
Please see previous remarks about languages. Since most existing and future
implementations sign the visible content, I think that the protocol should
handle actual textual content and not templates.
It is possible to bind a "display level" for every
displayable field (as described above).
[Denis] Since all people do not understand English, I do not see how the
approach you are taking may work.
all languages would not be acceptable either.
EZER - I do not see any problem of using
any language. The issue that you raised is how a certain visible signature can
several languages in parallel. As said, there is no existing implementation
that enables you to visualize different
signature depending on the current language the user is using.
You may require
that the protocol will enable future implementations to support such
functionality by supplying different
text for the
same item as part of the ItemValueStringType
As already stated above, whatever the protocol may be, the end result should be
incorporated either into the signed document itself or in the electronic
Do you really agree: "into the signed document itself or into the electronic signature"
I think it will be more accurate to say that the visible signature is
incorporated into the document. This is dependant on the document
If incorporated into the
electronic signature, the basic question is where to place that information ?
The protocol address that by either supplying a position or an existing
signature field, which is a placeholder for the visible signature.
Should it be as signed attributes (or signed properties) or as unsigned
attributes (or signed properties) ?
If you referring to whether the visible content is signed. It is very much
depended on the document implementation. The protocol addresses both
implementation types (that sign the visible content and do not sign the visible
content). One of the implementations that do not sign the visible content
actually incorporate to the visible content the actual digital signature in
bar-code representation or other type of representation. In this case the
visible signature content cannot be signed.
[Denis] I am confused here.You mean that there are two flavors. One that is
signed, the other one that is unsigned.
The unsigned one can be
chahged by anybody and so is not trustable. Do you plan to add a visible
saying whether it is signed or
unsigned ? Anyway, I do not see the value of an unsigned visible
signature: it can fool a verifier.
EZER - There are
existing implementations that include a certain particle of the visible
information that is a direct output
of the digital signature
itself (for example, a bar-code image that is a representation of the digital
element cannot be signed.
change the profile description to be more precise in this case, so that only
this type of element can be not signed.
Whatever the response to the question is, this means that new signed attributes
(or signed properties)
or new unsigned attributes (or signed properties) should be defined
respectively for CAdES and XAdES.
This issue will be covered as well by the implementation and not the protocol.
In most existing applications today (PDF, Office), the visible signature is
incorporated into the document's content and not to the digital signature
content (for example, signed attributes).
[Denis] If so, then the visible signature is signed
Yes, in most implementation. Please read my previous remark.
If incorporated into the document, the incorporation depends upon the type of
document. However, this approach
would be restricted to some types of documents and would not be as general as
the other approach.
Even if the electronic signature is embedded into the document (as it is the
case for PDF signatures), it appears better
to add the information related to the visible signature into the electronic
signature itself (placed within the document),
rather than placing it somewhere else in the document.
This issue should be addressed by the documents standards. The specification
(which is also true for other aspects of the DSS specification) addresses the
protocol and not the actual implementation. As I know there is an ETSI
committee that will address aspects of incorporating visible content into the
PDF specification that may address the issue that you raise.
[Denis] I do no beleive that is possible to define a protocol without thinking
at the same time what will/may be done
by implemtations. The actions should be done in parallel.
As stated above, this means that new signed attributes (or signed properties)
or new unsigned attributes
(or signed properties) should be defined respectively for CAdES and XAdES.
As said above, this direction should be handled by the document standards
themselves. The protocol is not aimed only for documents types that use
advanced signatures, but if the documents types do support CAdES/XAdES, the DSS
server implementation may use signed properties to include visible content.
In the draft document, there is intent to add information like a "scanned
image of the user's hand-written signature".
While this is an interesting idea, this information, if added should be
incorporated as a signed attribute and a signed property.
As said above, it is up to the document implementation. Today, both PDF and
Office sign this content as well as any other content in the visible signature.
The visible signature fields should not contain any value. Values should always
be computed by a verification service
using the signed attributes (or signed properties) or/and the unsigned
attributes (or signed properties), so that
there is no conflict or ambiguity between values placed in the electronic signature
and asserted values that would be placed
in the visible signature fields.
In most of the cases the claim is correct and most existing implementation
assure that by signing the visible content and make sure it is aligned with
values that are not part of the visible content. As I said, above, the protocol
also addresses implementation that produces a digital signature image. In this
case this image is not signed and content data.
[Denis] This does not work. The verification MUST always be done by the
signature verification software.
creation software MAY also do it, but the verification software cannot know
whether that software was "good" or "bad"
I agree with you. I will be more specific about the digital signature image,
which is the only possible visible item that may be unsigned since it is a
direct representation of the digital signature itself.
To summarize the suggested approach:
when used by a signer in a signing protocol, the service should be able to
incorporate a "visible signature template"
into the electronic signature, preferably as a signed attribute or a signed
property. This template would support
different levels of visibility.
EZER - Please see the comments
above. We can highly consider incorporation a visible display level attribute
for every item in the visible content. The application can use this attribute
to define the level of visibility.
[Denis] Adding the notion of a visible display level would only solve one of my
when used by a verifier in a verifying protocol, the service should be able to
interpret in a local language
the "visible signature template" placed into the electronic
signature, to the desired level of visibility and say
whether the information that has been retrieved has been verified or has simply
been copied and pasted
from the unverified signature. In the later case, if the signature is not
verified as being valid, this would allow
to know more about the electronic signature. This interpretation would also be
faster in practice.
Application displaying the visible signature fields should take care to make
the difference between "verified values"
and "unverified values".
EZER - The
visible signature profile mainly addresses the digital signature operation and
has limited functionality in the validation processing.
have difficulties to understand what you mean here. I believe that what is
important is what a verifier will see. Any field that is
displayed should be marked verified or claimed. Let us take the example of
a scanned manual signature. If can simply be part of the document (or of the
signature) and is whatever manual signature has been added by the signer (so it
is unverified). Another possibility is to include the jpg of the scanned
signature and verify that it matches a hash included into the signer's
certificate. In such a case, if the digital signature has been verified,
then the scanned manual signature has been verified.
EZER - I refer to
a scanning image of the digital signature itself (not a scanning of the
signer's handwritten signature). For example, this visible string
12389413824388123468ABABABAB728342732379. This item can be signed by the same
only processing that is included is to mark the verification status of the
digital signature, but this act by itself is mainly aimed for documents types
that will not violate the digital signature.
don't understand what you mean here.
The DSS core performs that actual digital signature verification act. The
profile is currently doing some limited functionality, which is to add some
visible marking to the document that specifies whether the digital signature is
valid. We have some debates on this issue, since adding this marks will
invalidate the digital signature in existing implementations.
you raise is interesting, but I need to consult with the DSS committee to
whether this approach should be defined and described in the visible signature
familiar with such an implementation? What types of documents are used?
don't understand your questions.
You described functionality in the above paragraph ("when used by a verifier in a
verifying protocol, the service should...."). I asked whether you are familiar with any
implementation or standard that describes or implement it.
[ESI] Deadline for commenting OASIS DSS-X Profile on visible
Just these few lines to remark that the window for raising comments on
the the DSS-X Profile on Visible Signatures **ENDS THE 30TH OF JUNE 2009**.
The OASIS DSS-X Technical Committee strongly encourages feedback
from you for the sake of improving the interoperability and quality of
OASIS work. Please feel free to distribute this announcement within
your organization and to other appropriate mail lists.
This profile is aimed to incorporate visible information into documents
as part of the digital signature operation.
Incorporating a visible signature as part of a digital signature
operation is mandatory for the acceptance of digital signatures in many
business scenarios and applications.
Today, there is an existing support in several document types such as
PDF and MS Office 2007, as well as non standard support in other
documents types. Eventually, many document types will support the both
visible and non-visible digital signature embedded in the document.
The target of the Visible Signature profile is to provide a general
interface to a digital signature service for incorporating a visible
signature to any type of document.
To ease the collaboration of a visible signature into applications, the
profile defines several types of scenarios/policies. Two of the most
outstanding usages of implementing a digital signature into documents
are application workflow scenario and a document submission scenario.
The committee will appreciate any submitted component as part of
reviewing the profile specifications.
For more information of how to access the content of the profile use the
The specification document and related files are available here:
Non-normative information about the specification and the technical
committee may be found at the public home page of the TC at:
Comments may be submitted to the TC by any person through the use of the
OASIS TC Comment Facility which can be located via the button marked
"Send A Comment" at the top of that page, or directly at:
Submitted comments (for this work as well as other works of that TC) are
publicly archived and can be viewed at:
submitted to OASIS are subject to the OASIS Feedback License, which
ensures that the feedback you provide carries the same obligations at
least as the obligations of the TC members.
OASIS and the DSS-X TC welcome your comments.
Mail archive for ESI can be browsed at the following url: