Clarify unit string specification

[casella]

The beginning of Chapter 19 of the MLS states:

Unless otherwise stated, the syntax and semantics of unit expressions in Modelica conform with the international standards International System of Units (SI) by BIPM superseding parts of ISO 31/0-1992 General principles concerning quantities, units and symbols and ISO 1000-1992 SI units and recommendations for the use of their multiples and of certain other units.

Later on, I read:

The units required to be recognized are the basic and derived units of the SI system, as well as some units compatible with the SI system listed below, but tools are allowed to additionally support user-defined unit symbols.

I guess the "Unless otherwise stated" only refers to the last clause of this sentence. Do I understand that correctly?

If so, what is the semantics and the syntax of the unit in that case? User-defined is by definition user-specific, not only tool-specific. Are we fine with that?

Later on, in the same chapter I read:

It is recommended that non-SI units are only used for the displayUnit-attribute in order to reduce impact of unrecognized unit symbols when using another Modelica tool.

so I understand that the semantical and syntactical fuzzyness of non-SI units is restricted to displayUnits attributes, which I understand will be ignored and replaced by the unit attribute if they are not "recognized". But yet, the semantics and syntax of display units will be user-specific. Is that our intention?

I would really feel more comfortable if there were a (not necessarily compulsory) way to specify the semantics of the non-SI units used in a Modelica model inside the model itself, or in its dependencies, e.g. the MSL, by referring them to SI units, for which the semantics is well-defined by the BIPM.

Could we go in that direction?

[henrikt-ma]

If so, what is the semantics and the syntax of the unit in that case? User-defined is by definition user-specific, not only tool-specific. Are we fine with that?

I don't think anybody finds the current situation satisfying, we are just used to the fact that unit checking is a part of the language that remains to be defined.

In case you are interested, we have moved away from user-defined unit symbols, replacing the per-user configuration with per-library configuration, meaning that what we have is better described as library-author-defined unit symbols. On the other hand, when reading the Modelica specification I don't think there is a clear distinction between "user" and "library author", so I'd say that what we do is a possible interpretation of "user-defined".

Until there is a standardized way of configuring unit symbols, I would suggest that we leave the specification the way it is, and that it should be up to tools to reach the same conclusion as us, namely that unit configuration which is not tied to a library is a bad idea in terms of unit checking.

[henrikt-ma]

It is recommended that non-SI units are only used for the displayUnit-attribute in order to reduce impact of unrecognized unit symbols when using another Modelica tool.

so I understand that the semantical and syntactical fuzzyness of non-SI units is restricted to displayUnits attributes, which I understand will be ignored and replaced by the unit attribute if they are not "recognized". But yet, the semantics and syntax of display units will be user-specific. Is that our intention?

No. It is not restricted since the specification is only giving a recommendation. Both unit and displayUnit can make use of the user-defined unit symbols, but it is a worse idea to do so for the unit than for the displayUnit. However, if you have no intention of using your library with other Modelica tools you can take full advantage of tool-specific unit customization.

[henrikt-ma]

I would really feel more comfortable if there were a (not necessarily compulsory) way to specify the semantics of the non-SI units used in a Modelica model inside the model itself, or in its dependencies, e.g. the MSL, by referring them to SI units, for which the semantics is well-defined by the BIPM.

Could we go in that direction?

Yes, we have already agreed to have a separate meeting to get started, where at least I have promised to demonstrate how we do it in System Modeler. We just haven't planned that meeting yet, but I would say that now (after the conference) is a good time to plan it.

[HansOlsson]

To me "Unless otherwise stated" refers to:

• Using "." instead of center-dot "·" (or space).
• Writing exponent as plain numbers instead of as super-scripts
• "deg" and "Ohm" instead of degree-symbol and Ω

They are all different from the SI-standard, and explicitly stated. (I hope I didn't miss anything.) So, the syntax for non-SI units isn't clearly defined, and not related to that sentence.

[HansOlsson]

To me "Unless otherwise stated" refers to:

• Using "." instead of center-dot "·" (or space).
• Writing exponent as plain numbers instead of as super-scripts
• "deg" and "Ohm" instead of degree-symbol and Ω

They are all different from the SI-standard, and explicitly stated. (I hope I didn't miss anything.) So, the syntax for non-SI units isn't clearly defined, and not related to that sentence.

Does that make sense, @casella ? Can you see another formulation that makes it clearer? (Perhaps "Except for the minor changes stated below, "?).