1071   Length of DO name

This tissue has following status: green

Created: 25 Mar 2013

Links:

Page: 182

Clause: 22.2

Paragraph:

Category: Issue for edition 2 of this part

Issue: Here data object (DO) related definitions for the lengths from 61850-7-2:

DataObjectClassName = 12 characters (as, for example, used in IEC 61850-7-4);
start with upper case alpha character;
one numeric last character indicates multiple instantiation capability
DataObjectName = 12 characters:
DataObjectClassName without last numeric character,
followed by [Data-Instance-ID], if the last character was numeric
Data-Instance-ID = n numeric characters, optional;
n shall be equal for all instances of the same data class in the same logical node

With the UML effort, we have streamlined the presence conditions and removed the need to use the naming convention (with "1" suffix) in the DO names. We also now produce machine-processable arguments to new presence conditions that specify multiple instances of DOs.

However, it still seems not clear how to interpret the number 12 above.

1) One possible interpretation (preferred, because it is bound to the type system and not instances):
- length 12 applies to the DO name without any instance ID (Data-Instance-ID)
- this means the designers of LNs (and their DOs), are free to combine abbreviations up to 12 characters
- presence conditions Mmulti/Omulti mean instance number can be any between 1 and 99
- presence conditions MmultiRange(min,max)/OmultiRange(min,max) can be more specific about both min and max; min>=1.
In essence, this means we CAN safely validate the model (type system as in 61850-7-4), without bothering with instance numbers.

2) Another possible interpretation heard of:
- length 12 applies to the full instantiated DO name, potentially including any instance ID (Data-Instance-ID)
- this means the designers of LNs (and their DOs), are free to combine abbreviations up to 'xx' characters, but what is 'xx'?:
* if my DO name has 12 characters, I automatically preclude it from usage in multiple instances
* if my DO name has 11 characters, I allow it potentially to be used for instances 1 through 9 only
* if my DO name has 10 characters, I allow it potentially to be used for instances 1 through 99 only,
* etc.
This means we can NOT validate the model (type system as in 61850-7-4) if this rule is applied.

Note that for some domains, already 12 letter DO names (without any instance number) may be pretty tough constraint, because we're running out of meaningful 2-4 letter abbreviations. This has been proven as problematic in joint modelling calls with other WG model managers.

Another side note: I am not sure why the third constraint above (on the same length of the Data-Instance-ID in an LN). If there is a good reason for it, it should be stated more explicitly when modifying the rest.

Proposal: In every case, the text in 7-2 needs to be updated (to remove the reference to the naming convention with "1" suffix).

It is proposed to follow the rule given above as (1), which would result into the updated text similar to:
DataObjectClassName = 12 characters (as, for example, used in IEC 61850-7-4);
start with upper case alpha character;
Data-Instance-ID = n numeric characters, optional;
as given by presence conditions Mmulti/Omulti/MmultiRange(min,max)/OmultiRange(min,max)
DataObjectName = (12 + n)

Discussion Created Status
?
Ballot until Editor
24 May 13 green
(2013-04-19, 7-2/3 editors call)
Change the description of the three mentioned items to read so:
...............................
DataObjectClassName = x characters, such that x+i <= 12 characters (as, for example, used in IEC 61850-7-4);
start with upper case alpha character
...............................
DataObjectName <= 12 characters: DataObjectClassName,
followed by [Data-Instance-ID] if the presence condition allows multi-instantiation (one of "multi").
...............................
Data-Instance-ID = i numeric characters, optional (note: instance IDs "1" and "01", and similar, are not allowed at the same time because they correspond to the same ID)
19 Apr 13 final proposal
Proposal for an 7-2 Ed2.1:
22.2
DataObjectClassName
= 12 characters (as, for example, used in IEC 61850-7-4);
start with upper case alpha character; a presence condition "multi" indicates multiple instantiation capability

DataObjectName - no change

12 Apr 13 final proposal 2013-May-17
The important requirement is that DO instance names are restricted to 12 characters, which is then also checkable with the SCL schema.
From this follows that DataObjectClassNames must have less than 12 characters, else a number behind is not possible. If the DO name is a DO class name is easily checkable by the M/O/C condition.
It is up to the designer of the LN class to decide how many instances of the DO class he wants to allow.
08 Apr 13 red
The proposed resolution text:
"For 7-2 I would propose:
22.2
DataObjectClassName
= 12 characters"
is what the initial Issue statement says (as preferred option 1 :-).

However, given the rest of the proposed resolution, AND if DataObjectName length <=12, then I am afraid we have to use in 7-2 the definition:
"- data object classes that can be instantiated within a range from 1 to 9 must have a maximum DataObjectClassName length of 11
- data object classes that can be instantiated within a range from 1 to 99 must have a maximum DataObjectClassName length of 10
- data object classes that can be instantiated within a range from 1 to 999 must have a maximum DataObjectClassName length of 9
... "

This is exactly the implication of saying "DataObjectName length <=12", i.e., binding the validation to be dependent on instance (as opposed to the type system). And it is not good to have this limitation when defining a type system - we have to limit then DataObjecClassName to the length of 9, to allow its usage in the multi-context.

Also the statement "It shall include an index only if the compatible logical node class allows multiple instantiation of the data object class." is not really clear to me. How can the LN "decide" to allow multiple instantiations of a DO (class)? "Multi" is not a characteristic of a DO (class, as in 61850-7-4), I think.

I am again lost :-)
27 Mar 13 red
The definition is:
The length of a DataObjectName is always limited to 12 characters.
Those 12 characters may or may not include an instance number. It shall include an index only if the compatible logical node class allows multiple instantiation of the data object class.
For 7-4, the restriction is then:
- data object classes that can be instantiated within a range from 1 to 9 must have a maximum DataObjectClassName length of 11
- data object classes that can be instantiated within a range from 1 to 99 must have a maximum DataObjectClassName length of 10
- data object classes that can be instantiated within a range from 1 to 999 must have a maximum DataObjectClassName length of 9
...

For the modeling, there is only one check possible:
- data object classes with an O/M-multi presence condition must have a maximal DataObjectClassName length of 11
- data object classes without an O/M-multi presence condition must have a maximal DataObjectClassName length of 12

Within an implemented model, the check to perform is that ANY DataObjectName has a maximal length of 12.
(as for instance in SCL with tDataName)
The DataObjectName includes the instance number of the data Object class only if the compatible logical node class allows multiple instantiation of the data object class.

7-1 and 7-2 need to be updated since the UML model changed the specification of the multiple instantion of data object class with mean of the presence condition and nit anymore with the presence of a numerical.

For 7-2 I would propose:
22.2
DataObjectClassName
= 12 characters (as, for example, used in IEC 61850-7-4);
start with upper case alpha character; a presence condition "multi" indicates multiple instantiation capability

DataObjectName - no change

27 Mar 13 red

 

Privacy | Contact | Disclaimer