[openpgp] Re: [EXTERNAL] Re: draft-ietf-openpgp-nist-b p-comp: add Category-5/P-384 combinations

NGG <[email protected]> Tue, 7 Jul 2026 16:29:38 +0200
Newsgroups gmane.ietf.openpgp
Message-ID <CA+z5aXcbKtLEXimP2a+VNoQtB2=6U1X+WwPrfAw84tcCAucPfg@mail.gmail.com>
--===============0478771040097885020==
Content-Type: multipart/alternative; boundary="0000000000002b80160656063925"

--0000000000002b80160656063925
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

Hi Quynh, Falko,

Thanks, that clarification helps.

My intention in raising this was mainly to see whether people in the WG, or
implementers, see deployment value in these combinations. I do not
personally have an immediate hard requirement for them.

The case I had in mind is a product or library using OpenPGP that wants to
choose a default PQ/T OpenPGP key type for broad deployment.

For such a default, category-5 ML-KEM / ML-DSA seems like the conservative
choice for CNSA 2.0 and similar future high-assurance PQ requirements. At
the same time, P-384 seems to be the classical curve that is already
profiled in a number of existing high-assurance environments, while P-521
is often not profiled there.

So the reason this combination looks useful to me is that it could be a
good fit for both sides: category-5 ML-* for the PQ part, and P-384 for
compatibility with existing environments that already profile P-384 but not
P-521.

The references below are not OpenPGP profiles, and most are
PKI/X.509/TLS/credential-related, so I am not claiming that they directly
impose an OpenPGP requirement. They are just examples of the kind of
surrounding policy environment where P-384 is familiar and accepted, but
P-521 is not part of the profile.

References where P-384 is profiled and P-521 is not:

   - CNSA 1.0 TLS profile / RFC 9151:
   https://www.rfc-editor.org/rfc/rfc9151.html
   <https://www.rfc-editor.org/rfc/rfc9151.html?utm_source=3Dchatgpt.com>
   - NIST SP 800-78-5 / PIV algorithms and key sizes:
   https://csrc.nist.gov/pubs/sp/800/78/5/final
   <https://csrc.nist.gov/pubs/sp/800/78/5/final?utm_source=3Dchatgpt.com>
   - U.S. Federal PKI Common Certificate Profiles:
   https://www.idmanagement.gov/docs/fpki-x509-cert-profile-common.pdf
   <https://www.idmanagement.gov/docs/fpki-x509-cert-profile-common.pdf?utm=
_source=3Dchatgpt.com>
   - U.S. Federal Public Trust TLS PKI Certificate Policy:
   https://www.idmanagement.gov/docs/us-federal-public-trust-tls-cp.pdf
   <https://www.idmanagement.gov/docs/us-federal-public-trust-tls-cp.pdf?ut=
m_source=3Dchatgpt.com>
   - DoD NIPRNet Certificate and CRL Profiles:

   https://dl.dod.cyber.mil/wp-content/uploads/pki-pke/pdf/unclass-dod_pki_=
nipr_cert_profiles.pdf
   <https://dl.dod.cyber.mil/wp-content/uploads/pki-pke/pdf/unclass-dod_pki=
_nipr_cert_profiles.pdf?utm_source=3Dchatgpt.com>
   - DoD External Certification Authority Certificate Policy:

   https://dl.dod.cyber.mil/wp-content/uploads/eca/pdf/unclass-eca_cp_v4-7_=
final_signed.pdf
   <https://dl.dod.cyber.mil/wp-content/uploads/eca/pdf/unclass-eca_cp_v4-7=
_final_signed.pdf?utm_source=3Dchatgpt.com>
   - NATO NISP Baseline 16, including FMN-related material:

   https://nhqc3s.hq.nato.int/apps/nisp/NISP_Baseline_16_Catalogue_5SEP2024=
_enclosure_only.pdf
   <https://nhqc3s.hq.nato.int/apps/nisp/NISP_Baseline_16_Catalogue_5SEP202=
4_enclosure_only.pdf?utm_source=3Dchatgpt.com>

Best regards,
Gergely

Falko Strenzke <[email protected]> ezt =C3=ADrta (id=C5=91pont: 2026. j=
=C3=BAl. 7., K,
14:58):

> Indeed, it's possible to write a draft and then have code points assigned
> via the "specification required" rule for assigning code points. An RFC i=
s
> not necessary for assigning code points. WG approval is also not required=
.
>
> Falko
>
> On Tue, 2026-07-07 at 10:40 +0000, Dang, Quynh H. (Fed) wrote:
>
> Hi Gergely,   One thing I know you can do is to write a short draft
> specifying the options that you need/want, then ask the working group to
> adopt it.  If the working group does not adopt it, you could ask for only
> code points (I don=E2=80=99t know if the working group would approve that=
).
> Regards, Quynh.   *From:* NGG [email protected]
> *Sent:* Monday, July 6, 2026 11:05 AM
> *To:* Falko Strenzke [email protected]
> *Cc:* [email protected]
> *Subject:* [EXTERNAL] [openpgp] Re: draft-ietf-openpgp-nist-bp-comp: add
> Category-5/P-384 combinations
>
>   You don't often get email from [email protected]. Learn why
> this is important <https://aka.ms/LearnAboutSenderIdentification>
>
>     Hello Falko,
>
> Thanks for the question. I did not mean that any single one of the
> existing profiles requires both hybrid use and category-5 ML-* alone. My
> point is interoperability across profiles: some environments allow P-384
> but not P-521, while other requirements may call for category-5 ML-*.
> Implementations may need one standardized hybrid option that works across
> as many such environments as possible. So the goal is not to satisfy a
> single profile with one algorithm choice, but to define an optional
> P-384/category-5 combination that is broadly usable across multiple
> profiles and avoids private, non-standard combinations. Best regards,
> Gergely
>
>   Falko Strenzke <[email protected]> ezt =C3=ADrta (id=C5=91pont: 202=
6. j=C3=BAl.
> 6., H, 16:35):
>
> Hi Gergely,
>
>
>
> On Sat, 2026-06-20 at 22:56 +0200, NGG wrote:
>
> Hello OpenPGP WG,
>
> I would like to suggest adding the following optional combinations
> to draft-ietf-openpgp-nist-bp-comp:
>
>    - ML-KEM-1024+ECDH-NIST-P-384
>    - ML-DSA-87+ECDSA-NIST-P-384
>
> The current draft pairs the category-5 ML-* parameter sets only
> with P-521, while P-384 is paired with category-3.
>
> The motivation is compliance interoperability. Some deployed
> profiles, including CNSA 1.0 and NATO FMN certificate profiles, admit P-3=
84
> but not P-521. Separately, other policies or deployment requirements may
> call for the category-5 ML-* parameter sets.
>
> What requirements of profiles are you exactly referring to? CNSA does not
> require multi-algorithm at all, so for this purpose the security level of
> the traditional component should not matter. In fact it could simply be
> ignored for the signature at least.
>
>
>
> Falko
>
> These are not necessarily requirements imposed by the same profile,
> but software may need to operate across several such environments.
>
> Standardized P-384/category-5 combinations would provide a
> useful intersection.
>
> There is already relevant IETF precedent:
>
>    - The TLS ECDHE-MLKEM specification defines SecP384r1MLKEM1024.
>    - The LAMPS composite-signature specification defines
>    id-MLDSA87-ECDSA-P384-SHA512.
>
> These combinations could be added alongside the existing P-521
> variants, without changing or replacing them.
>
> Would the WG consider adding them?
>
> Best regards, Gergely Nagy
>
> ------------------------------
>
> openpgp mailing list -- [email protected]
>
> To unsubscribe send an email to [email protected]
>
>
> >
> >
> >   ```
> -- ```
>   MTG AG
> >
> >   Dr. Falko Strenzke
> >
> >
> >
> >   Phone: +49 6151 8000 24
> >
> >   E-Mail: [
> > [email protected]](mailto:[email protected])
> >
> >   Web: [mtg.de](http://mtg.de/)
> >
> >
> >
> >   MTG AG - Dolivostr. 11 - 64293 Darmstadt, Germany
> >
> >   Commercial register: HRB 8901
> >
> >   Register Court: Amtsgericht Darmstadt
> >
> >   Management Board: J=C3=BCrgen Ruf (CEO), Tamer Kemer=C3=B6z
> >
> >   Chairman of the Supervisory Board: Dr. Thomas Milde
> >
> >
> >
> >   This email may contain confidential and/or privileged information. If=
 you are not the correct recipient or have received this email in error,
> >
> >   please inform the sender immediately and delete this email.Unauthoris=
ed copying or distribution of this email is not permitted.
> >
> >
> >
> >   Data protection information: Privacy policy
> >
> >
> >
> >
> >
> >
> >
>
>
>
>
>
> --
> MTG AG
> Dr. Falko Strenzke
>
> Phone: +49 6151 8000 24
> E-Mail: [email protected]
> Web: mtg.de
>
> MTG AG - Dolivostr. 11 - 64293 Darmstadt, Germany
> Commercial register: HRB 8901
> Register Court: Amtsgericht Darmstadt
> Management Board: J=C3=BCrgen Ruf (CEO), Tamer Kemer=C3=B6z
> Chairman of the Supervisory Board: Dr. Thomas Milde
>
> This email may contain confidential and/or privileged information. If you
> are not the correct recipient or have received this email in error,
> please inform the sender immediately and delete this email.Unauthorised
> copying or distribution of this email is not permitted.
>
> Data protection information: Privacy policy
>
>

--0000000000002b80160656063925
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><p class=3D"gmail-PDq2pG_selectionAnchorContainer">Hi Quyn=
h, Falko,<span aria-hidden=3D"true" class=3D"gmail-PDq2pG_selectionAnchor">=
</span></p>
<p>Thanks, that clarification helps.</p>
<p>My intention in raising this was mainly to see whether people in the WG,=
 or implementers, see deployment value in these combinations. I do not pers=
onally have an immediate hard requirement for them.</p>
<p>The case I had in mind is a product or library using OpenPGP that wants =
to choose a default PQ/T OpenPGP key type for broad deployment.</p>
<p>For such a default, category-5 ML-KEM / ML-DSA seems like the conservati=
ve choice for CNSA 2.0 and similar future high-assurance PQ requirements. A=
t the same time, P-384 seems to be the classical curve that is already prof=
iled in a number of existing high-assurance environments, while P-521 is of=
ten not profiled there.</p>
<p>So the reason this combination looks useful to me is that it could be a =
good fit for both sides: category-5 ML-* for the PQ part, and P-384 for com=
patibility with existing environments that already profile P-384 but not P-=
521.</p>
<p>The references below are not OpenPGP profiles, and most are PKI/X.509/TL=
S/credential-related, so I am not claiming that they directly impose an Ope=
nPGP requirement. They are just examples of the kind of surrounding policy =
environment where P-384 is familiar and accepted, but P-521 is not part of =
the profile.</p>
<p>References where P-384 is profiled and P-521 is not:</p>
<ul>
<li>
CNSA 1.0 TLS profile / RFC 9151:<br>
<a rel=3D"noopener" class=3D"gmail-decorated-link" href=3D"https://www.rfc-=
editor.org/rfc/rfc9151.html?utm_source=3Dchatgpt.com">https://www.rfc-edito=
r.org/rfc/rfc9151.html</a>
</li>
<li>
NIST SP 800-78-5 / PIV algorithms and key sizes:<br>
<a rel=3D"noopener" class=3D"gmail-decorated-link" href=3D"https://csrc.nis=
t.gov/pubs/sp/800/78/5/final?utm_source=3Dchatgpt.com">https://csrc.nist.go=
v/pubs/sp/800/78/5/final</a>
</li>
<li>
U.S. Federal PKI Common Certificate Profiles:<br>
<a rel=3D"noopener" class=3D"gmail-decorated-link" href=3D"https://www.idma=
nagement.gov/docs/fpki-x509-cert-profile-common.pdf?utm_source=3Dchatgpt.co=
m">https://www.idmanagement.gov/docs/fpki-x509-cert-profile-common.pdf</a>
</li>
<li>
U.S. Federal Public Trust TLS PKI Certificate Policy:<br>
<a rel=3D"noopener" class=3D"gmail-decorated-link" href=3D"https://www.idma=
nagement.gov/docs/us-federal-public-trust-tls-cp.pdf?utm_source=3Dchatgpt.c=
om">https://www.idmanagement.gov/docs/us-federal-public-trust-tls-cp.pdf</a=
>
</li>
<li>
DoD NIPRNet Certificate and CRL Profiles:<br>
<a rel=3D"noopener" class=3D"gmail-decorated-link" href=3D"https://dl.dod.c=
yber.mil/wp-content/uploads/pki-pke/pdf/unclass-dod_pki_nipr_cert_profiles.=
pdf?utm_source=3Dchatgpt.com">https://dl.dod.cyber.mil/wp-content/uploads/p=
ki-pke/pdf/unclass-dod_pki_nipr_cert_profiles.pdf</a>
</li>
<li>
DoD External Certification Authority Certificate Policy:<br>
<a rel=3D"noopener" class=3D"gmail-decorated-link" href=3D"https://dl.dod.c=
yber.mil/wp-content/uploads/eca/pdf/unclass-eca_cp_v4-7_final_signed.pdf?ut=
m_source=3Dchatgpt.com">https://dl.dod.cyber.mil/wp-content/uploads/eca/pdf=
/unclass-eca_cp_v4-7_final_signed.pdf</a>
</li>
<li>
NATO NISP Baseline 16, including FMN-related material:<br>
<a rel=3D"noopener" class=3D"gmail-decorated-link" href=3D"https://nhqc3s.h=
q.nato.int/apps/nisp/NISP_Baseline_16_Catalogue_5SEP2024_enclosure_only.pdf=
?utm_source=3Dchatgpt.com">https://nhqc3s.hq.nato.int/apps/nisp/NISP_Baseli=
ne_16_Catalogue_5SEP2024_enclosure_only.pdf</a>
</li>
</ul>
<p>Best regards,<br>
Gergely</p></div><br><div class=3D"gmail_quote gmail_quote_container"><div =
dir=3D"ltr" class=3D"gmail_attr">Falko Strenzke &lt;<a href=3D"mailto:falko=
[email protected]">[email protected]</a>&gt; ezt =C3=ADrta (id=C5=91pont=
: 2026. j=C3=BAl. 7., K, 14:58):<br></div><blockquote class=3D"gmail_quote"=
 style=3D"margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);p=
adding-left:1ex"><p>Indeed, it&#39;s possible to write a draft and then hav=
e code points assigned via the &quot;specification required&quot; rule for =
assigning code points. An RFC is not necessary for assigning code points. W=
G approval is also not required.</p>
<p>Falko</p>
<p>On Tue, 2026-07-07 at 10:40 +0000, Dang, Quynh H. (Fed) wrote:</p>
<blockquote type=3D"cite">
<p>Hi Gergely,
=C2=A0
One thing I know you can do is to write a short draft specifying the option=
s that you need/want, then ask the working group to adopt it.=C2=A0 If the =
working group does not adopt it, you could ask for only code  points (I don=
=E2=80=99t know if the working group would approve that).
=C2=A0
Regards,
Quynh.
=C2=A0
<strong>From:</strong> NGG <a href=3D"mailto:[email protected]"=
 target=3D"_blank">[email protected]</a><br>
<strong>Sent:</strong> Monday, July 6, 2026 11:05 AM<br>
<strong>To:</strong> Falko Strenzke <a href=3D"mailto:[email protected]=
" target=3D"_blank">[email protected]</a><br>
<strong>Cc:</strong> <a href=3D"mailto:[email protected]" target=3D"_blank">=
[email protected]</a><br>
<strong>Subject:</strong> [EXTERNAL] [openpgp] Re: draft-ietf-openpgp-nist-=
bp-comp: add Category-5/P-384 combinations</p>
<p>=C2=A0
You don&#39;t often get email from <a href=3D"mailto:[email protected]=
etf.org" target=3D"_blank">[email protected]</a>. <a href=3D"ht=
tps://aka.ms/LearnAboutSenderIdentification" target=3D"_blank">
Learn why this is important</a></p>
<pre><code>    Hello Falko,
</code></pre>
<p>Thanks for the question.
I did not mean that any single one of the existing profiles requires both h=
ybrid use and category-5 ML-* alone.
My point is interoperability across profiles: some environments allow P-384=
 but not P-521, while other requirements may call for category-5 ML-*. Impl=
ementations may need one standardized hybrid option that works across as ma=
ny such environments as possible.
So the goal is not to satisfy a single profile with one algorithm choice, b=
ut to define an optional P-384/category-5 combination that is broadly usabl=
e across multiple profiles and avoids private, non-standard combinations.
Best regards,<br>
Gergely</p>
<p>=C2=A0
Falko Strenzke &lt;<a href=3D"mailto:[email protected]" target=3D"_blan=
k">[email protected]</a>&gt; ezt =C3=ADrta (id=C5=91pont: 2026. j=C3=BA=
l. 6., H, 16:35):</p>
<blockquote type=3D"cite">
<pre><code>Hi Gergely,
</code></pre>
<p>=C2=A0</p>
<p>On Sat, 2026-06-20 at 22:56 +0200, NGG wrote:</p>
<blockquote type=3D"cite">
<p>Hello OpenPGP WG,</p>
<p>I would like to suggest adding the following optional combinations to=C2=
=A0draft-ietf-openpgp-nist-bp-comp:</p>
<ul>
<li>ML-KEM-1024+ECDH-NIST-P-384</li>
<li>ML-DSA-87+ECDSA-NIST-P-384</li>
</ul>
<p>The current draft pairs the category-5 ML-* parameter sets only with=C2=
=A0P-521, while P-384 is paired with category-3.</p>
<p>The motivation is compliance interoperability.
Some deployed profiles,=C2=A0including CNSA 1.0 and NATO FMN certificate pr=
ofiles,=C2=A0admit P-384 but not P-521.
Separately, other policies or deployment requirements=C2=A0may call for the=
 category-5 ML-* parameter sets.</p>
</blockquote>
<p>What requirements of profiles are you exactly referring to? CNSA does no=
t require multi-algorithm at all, so for this purpose the security level of=
 the traditional component should not matter. In fact it could simply be ig=
nored for the  signature at least.</p>
<p>=C2=A0</p>
<p>Falko</p>
<blockquote type=3D"cite">
<p>These are not necessarily requirements imposed by the same profile, but=
=C2=A0software may need to operate across several such environments.</p>
<p>Standardized P-384/category-5 combinations would provide a useful=C2=A0i=
ntersection.</p>
<p>There is already relevant IETF precedent:</p>
<ul>
<li>The TLS ECDHE-MLKEM specification defines SecP384r1MLKEM1024.</li>
<li>The LAMPS composite-signature specification defines id-MLDSA87-ECDSA-P3=
84-SHA512.</li>
</ul>
<p>These combinations could be added alongside the existing P-521 variants,=
=C2=A0without changing or replacing them.</p>
<p>Would the WG consider adding them?</p>
<p>Best regards,
Gergely Nagy</p>
<pre><code></code></pre>
</blockquote>
</blockquote>
<hr>
<pre><code></code></pre>
<p>openpgp mailing list -- <a href=3D"mailto:[email protected]" target=3D"_b=
lank">[email protected]</a></p>
<pre><code></code></pre>
<p>To unsubscribe send an email to <a href=3D"mailto:[email protected]=
" target=3D"_blank">[email protected]</a></p>
<pre><code>=20
&gt;   =C2=A0
&gt; =20
&gt;   ```
-- ```
  MTG AG
&gt; =20
&gt;   Dr. Falko Strenzke
&gt; =20
&gt;   =C2=A0
&gt; =20
&gt;   Phone: +49 6151 8000 24
&gt; =20
&gt;   E-Mail: [
&gt; <a href=3D"mailto:[email protected]" target=3D"_blank">falko.stren=
[email protected]</a>](mailto:<a href=3D"mailto:[email protected]" target=3D"_=
blank">[email protected]</a>)
&gt; =20
&gt;   Web: [<a href=3D"http://mtg.de" target=3D"_blank">mtg.de</a>](<a hre=
f=3D"http://mtg.de/" target=3D"_blank">http://mtg.de/</a>)
&gt; =20
&gt;   =C2=A0
&gt; =20
&gt;   MTG AG - Dolivostr. 11 - 64293 Darmstadt, Germany
&gt; =20
&gt;   Commercial register: HRB 8901
&gt; =20
&gt;   Register Court: Amtsgericht Darmstadt
&gt; =20
&gt;   Management Board: J=C3=BCrgen Ruf (CEO), Tamer Kemer=C3=B6z
&gt; =20
&gt;   Chairman of the Supervisory Board: Dr. Thomas Milde
&gt; =20
&gt;   =C2=A0
&gt; =20
&gt;   This email may contain confidential and/or privileged information. I=
f you are not the correct recipient or have received this email in error,
&gt; =20
&gt;   please inform the sender immediately and delete this email.Unauthori=
sed copying or distribution of this email is not permitted.
&gt; =20
&gt;   =C2=A0
&gt; =20
&gt;   Data protection information: Privacy policy
&gt; =20
&gt;   =C2=A0
&gt; =20
&gt; =20
&gt; =20
&gt; =20
&gt;




</code></pre>
</blockquote>
<p>-- <br></p>
<div>MTG AG</div><div>Dr. Falko Strenzke</div><div><br></div><div>Phone: +4=
9 6151 8000 24</div><div>E-Mail: <a href=3D"mailto:[email protected]" t=
arget=3D"_blank">[email protected]</a></div><div>Web: <a href=3D"http:/=
/mtg.de" target=3D"_blank">mtg.de</a></div><div><br></div><div>MTG AG - Dol=
ivostr. 11 - 64293 Darmstadt, Germany</div><div>Commercial register: HRB 89=
01</div><div>Register Court: Amtsgericht Darmstadt</div><div>Management Boa=
rd: J=C3=BCrgen Ruf (CEO), Tamer Kemer=C3=B6z</div><div>Chairman of the Sup=
ervisory Board: Dr. Thomas Milde</div><div><br></div><div>This email may co=
ntain confidential and/or privileged information. If you are not the correc=
t recipient or have received this email in error,</div><div>please inform t=
he sender immediately and delete this email.Unauthorised copying or distrib=
ution of this email is not permitted.</div><div><br></div><div>Data protect=
ion information: Privacy policy</div><div><br></div>
</blockquote></div>

--0000000000002b80160656063925--


--===============0478771040097885020==
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: base64
Content-Disposition: inline

X19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX19fX18Kb3BlbnBncCBt
YWlsaW5nIGxpc3QgLS0gb3BlbnBncEBpZXRmLm9yZwpUbyB1bnN1YnNjcmliZSBzZW5kIGFuIGVt
YWlsIHRvIG9wZW5wZ3AtbGVhdmVAaWV0Zi5vcmcK

--===============0478771040097885020==--