Correct usage of unknown values in dREL code

[vaitkus]

See issue COMCIFS/dREL#20 for more details.

[jamesrhester]

I suppose explicitly setting 'missing' (the default value anyway) is OK here to simplify transformation of the dREL - but the value will be 'missing' if dREL fails to set it to something else.

[vaitkus]

On second though, it might be better to change the ? to unspecified, that is, to "figure it out yourself"?

Also, this is the first dREL snippet which I have seen that modifies the definition (units) for each entry in the loop, i.e. different values of the same data item in the same CIF file will get different units. I guess this is allowed?

[jamesrhester]

First time I've seen it as well. But that is the point of these Definition methods, they are for situations where the values are row-dependent. The default conversion matrix to Cartesian coordinates is also cell-dependent, i.e. depends on _structure.id, so is row-dependent if you were to tabulate cells against _structure.id.

[vaitkus]

I changed the unknown values to "unspecified" since it better describes the situation.

First time I've seen it as well. But that is the point of these Definition methods, they are for situations where the values are row-dependent. The default conversion matrix to Cartesian coordinates is also cell-dependent, i.e. depends on _structure.id, so is row-dependent if you were to tabulate cells against _structure.id.

I have encountered 1) dREL that calculates row dependent values and 2) dREL that modifies the definition of an item from a Set category based on the values of other items (most often measurement units are modified), but not the combination of 1) and 2) as in this case. I was simply surprised, that's all.