Issue 427: Reuse identifiers of obsolete entities never published
Posted by Vincent Alamercery on 13/6/2019
Dear all,
during the SIG meeting in Paris, we added the new property "P177 assigned property type" (see http://www.cidoc-crm.org/sites/default/files/CIDOC%20CRM_v6.2.6_Definiti...).
This property reuses the already given identifier of the property "P177 ends within" which has been deprecated without ever belonging to a published version (see http://www.cidoc-crm.org/Property/p177-ends-within/version-6.2.2 and http://www.cidoc-crm.org/sites/default/files/CIDOC%20-%20CRM_v6.2.6_%20A...)
We had a little discussion on whether or not to reuse this identifier already given. Maybe I'm picky but I'm not really comfortable with this practice. I suggest never to reuse an identifier for the following non-exhaustive reasons:
Even it's highly not recommended to use a draft version of CIDOC CRM, an entity exists from the moment it appears on a public document. It could then be potentially used by anyone. In a given namespace, an identifier must have to be unique.
For documentation reason, it's easier to have unique identifiers too to avoid speaking of "the old P177" or "the new P177". For instance, in the issue #345, how to know of which P177 property we are talking about? Think "the new P177" could be deprecated too one day...: http://www.cidoc-crm.org/Issue/ID-345-properties-having-domain-or-range-...
Numbers are infinite, we don't need to save them. ;-)
Posted by Richard Light on 13/6/2019
Vincent,
I strongly support your view that we should not re-use identifiers. The only argument I could give for this practice is the desire for a nice neat sequence of identifiers: and we have already scuppered that aspiration by deprecating previously-published classes and properties (thereby causing gaps to appear). So, please, don't do it!
Posted by Robert Sanderson on 14/6/2019
I also agree with Vincent and Richard. Given the very slow rate of change between “official” versions, and the prominence of the intermediate versions, I agree that the condition should be “in a public document” not “in an official version”.
http://www.cidoc-crm.org/get-last-official-release lists 5.0.4, dated 2011, as the last official release.
The “Current Version” link in the website sidebar lists version 6.2.3.
And the top most link in the home page under What’s New, refers to the upload of 6.2.6.
And http://www.cidoc-crm.org/versions-of-the-cidoc-crm lists 6.2.1 as the most recent published version, and the most recent published RDFS file.
So I believe that it is entirely reasonable for people to be confused as to which identifiers are stable and which are not, and thus we should treat the assignment of a number to a class or property as final. While in draft, it can be xxx as per our typical practice.
Posted by George Bruseker on 18/6/2019
I also support there not being reuse of numbers. There is no end of numbers to choose from.
In the 45th joint meeting of the CIDOC CRM SIG and SO/TC46/SC4/WG9; 38th FRBR – CIDOC CRM Harmonization meeting, given the distress and concerns caused by reusing identifiers in the crm, the sig decided against this practice –even when reuse concerns numbers that were previously assigned to deprecated CRM classes and properties, which, in their turn, have never made it to an official version.
From this point on, identifiers will never get reused. Of course, classes/properties already making use of a “reused” identifier are not going to change numbers now.
The issue closed
Heraklion, October 2019