[Ietf-krb-wg] New text for section 3

[Leif Johansson-2]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Here is my suggestion for new text for section 3 which I hope makes
the interpretation of terms more clean. Obviously I need to go through
the attributes and clean up the 2119 language there too but first lets
agree on this:

- ----------

This document describes an information model for kerberos 5 but does not
directly describe any mapping onto a particular data model (described by
a schema- or data description language).

For this reason the precise interpretation of terms from [RFC2119]
require some extra explanation.

The terms MUST, MUST NOT, REQUIRED, SHALL, SHALL NOT mean that a
Kerberos implementation MUST, MUST NOT etc implement a feature as
described in this specification but does not imply that this feature
must be REQUIRED by all data models implementing this information model.

For instance an attribute marked as MUST in the information model MUST
be implemented by all Kerberos implementations and MUST be included in
an LDAP schema but MAY be marked as OPTIONAL in the LDAP schema. Any
deviation from this general principle will be clearly described in the
specification text for instance by saying ”Feature X SHOULD be REQUIRED
by all data models”.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkr3nGwACgkQ8Jx8FtbMZnc7+QCfb1/8VCLfyAsK2v28fSkC9/Sg
EoAAn0h9hkEXQbeAGO2oIhk6XFbDAYOk
=6RRI
-----END PGP SIGNATURE-----
_______________________________________________
ietf-krb-wg mailing list
ietf-krb-wg@...
https://lists.anl.gov/mailman/listinfo/ietf-krb-wg

[Sam Hartman-5]

>>>>> "Leif" == Leif Johansson <leifj@...> writes:


    Leif> For this reason the precise interpretation of terms from
    Leif> [RFC2119] require some extra explanation.

    Leif> For instance an attribute marked as MUST in the information
    Leif> model MUST be implemented by all Kerberos implementations
    Leif> and MUST be included in an LDAP schema but MAY be marked as
    Leif> OPTIONAL in the LDAP schema. Any deviation from this general
    Leif> principle will be clearly described in the specification
    Leif> text

The above seems like a reasonable thing.  I'm not sure yet whether I
agree with it.

    Leif> The terms MUST, MUST NOT, REQUIRED, SHALL, SHALL NOT mean
    Leif> that a Kerberos implementation MUST, MUST NOT etc implement
    Leif> a feature as described in this specification but does not
    Leif> imply that this feature must be REQUIRED by all data models
    Leif> implementing this information model.

However this paragraph does not say something similar to the paragraph
that comes next in your text (and that I transposed above).  When you
say that an attribute is required by a data model, in my mind that
means that a data model must provide a way of representing the
attribute, *not* that the objects of that data model must include the
attribute.

--Sam
_______________________________________________
ietf-krb-wg mailing list
ietf-krb-wg@...
https://lists.anl.gov/mailman/listinfo/ietf-krb-wg

[Leif Johansson-2]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Sam Hartman wrote:

>>>>>> "Leif" == Leif Johansson <leifj@...> writes:
>
>
>     Leif> For this reason the precise interpretation of terms from
>     Leif> [RFC2119] require some extra explanation.
>
>     Leif> For instance an attribute marked as MUST in the information
>     Leif> model MUST be implemented by all Kerberos implementations
>     Leif> and MUST be included in an LDAP schema but MAY be marked as
>     Leif> OPTIONAL in the LDAP schema. Any deviation from this general
>     Leif> principle will be clearly described in the specification
>     Leif> text
>
> The above seems like a reasonable thing.  I'm not sure yet whether I
> agree with it.

Those two paragraphs don't appear in that order in the text I just
sent to the email - this has me potentially confused wrt what you
mean by "this paragraph" and "the paragraph that comes next in your
text" below.

>
>     Leif> The terms MUST, MUST NOT, REQUIRED, SHALL, SHALL NOT mean
>     Leif> that a Kerberos implementation MUST, MUST NOT etc implement
>     Leif> a feature as described in this specification but does not
>     Leif> imply that this feature must be REQUIRED by all data models
>     Leif> implementing this information model.
>
> However this paragraph does not say something similar to the paragraph
> that comes next in your text (and that I transposed above).  When you
> say that an attribute is required by a data model, in my mind that
> means that a data model must provide a way of representing the
> attribute, *not* that the objects of that data model must include the
> attribute.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkr3qv0ACgkQ8Jx8FtbMZneeWwCfT0xkCAe332LfpLrcxbiYYRMx
NaYAoJ++ThQWrDlcvEBBUj74fnPlLrSW
=vdjq
-----END PGP SIGNATURE-----
_______________________________________________
ietf-krb-wg mailing list
ietf-krb-wg@...
https://lists.anl.gov/mailman/listinfo/ietf-krb-wg

[Leif Johansson-2]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Leif Johansson wrote:

> Here is my suggestion for new text for section 3 which I hope makes
> the interpretation of terms more clean. Obviously I need to go through
> the attributes and clean up the 2119 language there too but first lets
> agree on this:
>
> ----------
>
> This document describes an information model for kerberos 5 but does not
> directly describe any mapping onto a particular data model (described by
> a schema- or data description language).
>
> For this reason the precise interpretation of terms from [RFC2119]
> require some extra explanation.
>

After talking to Sam AFK this is what we came up with:

> The terms MUST, MUST NOT, REQUIRED, SHALL, SHALL NOT mean that a
> Kerberos implementation MUST, MUST NOT etc implement a feature as
> described in this specification but does not imply that this feature
> must be REQUIRED by all data models implementing this information model.

The terms MUST, MUST NOT, REQUIRED, SHALL, SHALL NOT mean that a
Kerberos implementation MUST, MUST NOT etc implement a feature and
that all data models MUST etc represent said feature as described in
this specification but does not imply that this feature must be REQUIRED
in all objects represented by data models implementing this information
model.

>
> For instance an attribute marked as MUST in the information model MUST
> be implemented by all Kerberos implementations and MUST be included in
> an LDAP schema but MAY be marked as OPTIONAL in the LDAP schema. Any
> deviation from this general principle will be clearly described in the
> specification text for instance by saying ”Feature X SHOULD be REQUIRED
> by all data models”.
>

For instance an attribute marked as MUST in the information model MUST
be implemented by all Kerberos implementations and MUST be included in
an LDAP schema but MAY be marked as OPTIONAL in the LDAP schema. Any
deviation from this general principle will be clearly described in the
specification text for instance by saying ”Feature X SHOULD be REQUIRED
by all objects”.

        Cheers Leif
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkr3vQIACgkQ8Jx8FtbMZndA/QCgm8lOTbmbsj8+6G51z8sCqYMm
7ZsAniwdFIT0I7mvG3rpXXEPF9wCYhw1
=m/wX
-----END PGP SIGNATURE-----
_______________________________________________
ietf-krb-wg mailing list
ietf-krb-wg@...
https://lists.anl.gov/mailman/listinfo/ietf-krb-wg