[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Is UML an option for the IETF?

Title: RE: Is UML an option for the IETF?
NIM List:
I received an enquiry about where to get info on XMI. The specification is at:
Slides are at:
Hope this is helpful.
Kirk Shrewsbury
-----Original Message-----
From: owner-nim@ops.ietf.org [mailto:owner-nim@ops.ietf.org] On Behalf Of Kirk Shrewsbury
Sent: Tuesday, September 05, 2000 2:06 PM
To: 'Weiss, Walter'; 'Harald Alvestrand'; 'Tom Scott'; nim@ops.ietf.org; 'Avri Doria'
Subject: RE: Is UML an option for the IETF?

I think XMI and UML are intended to meet the two needs for machine  and human readability in cooperation with one another. It is the OMG standard for XML interchange of UML models between tools. So one can make a model in Rational Rose, export to XMI, then import it into Paradigm Plus, or vice versa. But the human interface is at the UI of the modeling tool, not at the exported XMI file.
However, XMI would not be a bad standard to use to publicly specify UML models, because it eliminates commitment to any particular UML tool vendor. TeleManagement Forum led the way on introducing UML modeling in the telecom industry. (Two guys from there received an award from the Smithsonian for it.) However, because the effort was pre-XMI, they had to choose collectively a single UML tool vendor and use that vendor's file format as their medium of model exchange. XMI now makes it possible use UML across tool vendors, while staying (technically) within the IETF policy for ASCII documents.
Other than the lack of human readability, I don't know if XMI has other shortcomings. Does anyone have any experience?
Kirk S.
-----Original Message-----
From: owner-nim@ops.ietf.org [mailto:owner-nim@ops.ietf.org]On Behalf Of Weiss, Walter
Sent: Tuesday, September 05, 2000 12:36 PM
To: 'kirk.shrewsbury@wcom.com'; 'Harald Alvestrand'; 'Tom Scott'; 'nim@ops.ietf.org'; 'Avri Doria'
Subject: RE: Is UML an option for the IETF?

There are two main reasons for an ASCII text representation:
1. Machine readable (Tools should be able to read/compile the model).
2. Human readable (All IETF documents are [currently] written in a fixed format ASCII file).

Both needs must be satisfied, preferably in a single document. In other words, it is conceivable that a standard comprises two documents (one for machines and one for humans). However, the overhead in ensuring consistency between the two makes this alternative less attractive.



> -----Original Message-----
> From: Kirk Shrewsbury [mailto:shrews@post.harvard.edu]
> Sent: Tuesday, September 05, 2000 12:27 PM
> To: 'Harald Alvestrand'; 'Tom Scott'; 'NIM List'; 'Avri Doria'
> Cc: 'Walter Weiss'
> Subject: RE: Is UML an option for the IETF?
> How about using XMI as the standard? That is ASCII, and the
> various UML
> tools can export and import it. It's not really for human
> beings, but it is
> -----Original Message-----
> From: owner-nim@ops.ietf.org
> [mailto:owner-nim@ops.ietf.org]On Behalf Of
> Harald Alvestrand
> Sent: Tuesday, September 05, 2000 3:20 AM
> To: Tom Scott; NIM List; Avri Doria
> Cc: Walter Weiss
> Subject: Re: Is UML an option for the IETF?
> At 11:02 04/09/2000 -0400, Tom Scott wrote:
> >1. A combination of text and UML graphics would be a practical
> >solution, keeping in mind that the "official" version would be the
> >text and possibly a few simplified "ascii art" representations.
> there is an ASCII representation format for UML documents.
> The VB Visual Modeller saves its .mdl files in ASCII, doing
> things like
>              (object Class "CustomerManager"
>                  quid            "389020F8019E"
>                  superclasses    (list inheritance_relationship_list
>                      (object Inheritance_Relationship
>                          quid            "38902107016D"
>                          supplier        "Logical View::Customer"
>                          quidu           "389002BC0253")))
> When I tried to find the basis for this in the UML
> documentation, I did not
> get very far, but there were references suggesting that the format was
> based on an SGML DTD, and WAS part of the standard.
> Publishing the ASCII that allows reproduction of the diagrams is an
> acceptable method.
>                   Harald
> --
> Harald Tveit Alvestrand, alvestrand@cisco.com
> +47 41 44 29 94
> Personal email: Harald@Alvestrand.no