p103 and 104 of IT Vol G 1st Edition state that occupational disorder can either be described as above, or else a dummy atom type can be created and the atom_type loop used to assign weighted values for the relevant parameters. While not stated, I think the former approach retains more information and is to be preferred.
Since the 1st edition was published a "restraints and constraints" dictionary has been developed, which allows the restraints on atom occupancies and positions to be made explicit. Use of those data names would allow checking that the positions in data files match when they correspond to occupational disorder.

Thank you for the useful reference!

Slight deviation regarding the terminology. It seems that what I called substitutional disorder (based on [1]) is referred to as compositional disorder in IT Vol. G 1st Edition so I will be using this term from now on. Also, occupational disorder seems to be a synonym of positional disorder at least from the way these terms are used in p105 and p212 (definition of _atom_site_disorder_group) so you probably meant "compositional disorder" instead of "occupation disorder" in your previous message. It might be useful to write these definitions up (including the synonyms) in a resource that is slightly easier to search through than IT Vol. G. (e.g. the Online Dictionary of Crystallography [2]), though, I am slightly hesitant to do this myself due to the lack of expertise.

Regardless, now it is completely clear that there are two formalised ways of describing compositional disorder (same site approach and dummy site approach). However, one question still remains -- can the _atom_site.disorder_* items be used in combination with the same site approach to describe compositional disorder or is their usage strictly restricted to the situations of occupational (positional) disorder?

[1] https://ocw.mit.edu/courses/chemistry/5-067-crystal-structure-refinement-fall-2009/lecture-notes/MIT5_067F09_lec4.pdf
[2] https://dictionary.iucr.org/Main_Page

Interesting. I think that the authors of the Volume and definitions had in mind groups of bonded atoms when defining the _atom_site_disorder_* tags. However, as noted in the original issue, these tags would work perfectly well for describing compositional disorder and are easier for machines to work with. Unfortunately both of the other methods must also be supported anyway. I'll check to see if the core DMG have any objections.

The core DMG have no objections to compositional disorder being described using _atom_site.disorder_* tags. We should update the relevant definitions and I will take care of the relevant chapter of the next edition of Volume G. It would be useful if an example could be prepared of the use of these tags to describe compositional disorder.

Relevant core DMG discussion can be found at https://www.iucr.org/__data/iucr/lists/coredmg/msg00414.html.

The definitions were updated and a few example CIF files were added by merging PR #343.

I will leave to @jamesrhester to close this issue in case he still wants to keep it open as a reminder to also update the text in Volume G.

Also, I will reiterate my question from the last comment of PR #343. I noticed that the disorder example that was provided in the DDL1 version of the dictionary as part of the ATOM_SITE category description (see https://github.com/COMCIFS/DDL1-legacy-dictionaries/blob/28e20dc928790dceb889716d2bed435fc10c4c79/dictionaries/cif_core.dic#L850) is no longer available in the DDLm dictionary. Maybe this example should be restored or a similar one reintroduced?

Yes, the missing example should be returned, it is very helpful.