Wikidata:Property proposal/software version of

software version of edit

Originally proposed at Wikidata:Property proposal/Creative work

Not done

Description	the subject item is a version of the object item
Represents	software (Q7397)
Data type	Item
Domain	item
Allowed values	software (Q7397)
Example	Firefox 4 (Q1950119) → Mozilla Firefox (Q698)
See also	software version identifier (P348)

Motivation

Inverse of software version identifier (P348). There are items that are software versions of other items. Currently, there isn't a property that links from the former to the latterMalore (talk) 14:57, 18 May 2018 (UTC)[reply]

Discussion

Support David (talk) 16:38, 18 May 2018 (UTC)[reply]
Support actually it's not an inverse of software version identifier (P348) since that is a string-valued property and can't have an inverse in wikidata. I think this is the better direction to have this anyway, so definitely I support this. ArthurPSmith (talk) 17:29, 18 May 2018 (UTC)[reply]

Comment@ArthurPSmith: For this reason too, I proposed in project chat to create a "software version" property with items as datatype and to rename software version identifier (P348) in "software version name string".--Malore (talk) 19:07, 18 May 2018 (UTC)[reply]

WikiProject Informatics has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead.ChristianKl ❪✉❫ 15:09, 24 May 2018 (UTC)[reply]

~~On hold~~ Can't we use a more general one? -- MichaelSchoenitzer (talk) 12:08, 26 May 2018 (UTC)[reply]

What does it mean a more general one? subclass of (P279) or instance of (P31)?--Malore (talk) 15:47, 27 May 2018 (UTC)[reply]

We should be creating and using generic properties wherever possible, so as to avoid creating many 1000s of very specific properties. The concept of version is not limited to software, and we do already have based on (P144) and subclass of (P279) which can be used in many situations. Danrok (talk) 01:28, 29 May 2018 (UTC)[reply]

@Danrok: You could say the same about edition or translation of (P629), or a bunch of other properties that could be substituted by instance of (P31) and subclass of (P279).--Malore (talk) 16:57, 30 May 2018 (UTC)[reply]

After thinking about it for long time, I Oppose, since I see two problems here:

It's to specific. "software version of" makes it usable to only software. If at all it should be "version of" since there are other things that are versioned too (Licenses, 3D-Models, technical Specifications, etc.) and there is a fluid transition. But I would like to have even more abstract properties and share them with other areas where different versions/editions/iterations of things exist, including editions of books. The reason is that the more specialized properties your have for each area the harder it's getting to write queries that (also) work in the border-areas and/or in the bigger picture. An example in this case would be technical specifications and standards which are in between of software and books, but are none of both. Should one then use "software version of" or has edition or translation (P747) or yet another to be created property? When writing queries and would need to know and think of all these special cases, which doesn't work…
It's semantically not fully correct in most cases – including the reference example: Firefox 4 (Q1950119) → Mozilla Firefox (Q698). Firefox 4 is not one version of Firefox. It's a version series. Firefox 4.0.0, 4.0.1, etc. are versions of Firefox. This would be the issue basically always since we so far basically only have items about version-series not single versions. One could bypass this by renaming it, but that would make it uglier and lead to problems if/when we have items about singe versions. And in the end this increases the need for having a more general property.

I see the need to improve the ontology of software versions, but I'm against this property. Sorry! I'm thinking about writing down all the open questions & issues about software versioning (there are a lot!) and make a working group or a big RFC, but I sadly don't have the time right now. -- MichaelSchoenitzer (talk) 11:49, 8 June 2018 (UTC)[reply]

@MichaelSchoenitzer: There's already an RFC about software versioning problems. As regards your points:

In order to simplify the queries, we could make "software version of" and edition or translation of (P629) (and eventually other properties such as "specification version of") subproperties of "edition or version of".

"software version of" and edition or translation of (P629) should be separate properties because they have different qualifiers and the model that describes books - the FRBR model - is different from the software version one (in fact, there isn't a universal versioning model, but every software can adopt its own).

Firefox 4 is a major version of Firefox, while as far as I know Firefox 4.0.0 and Firefox 4.0.1 are called releases. They are different type of versions, but they are both versions.

However, even if "software version of" wasn't the best solution, it's better than the current management of software version and I think it doesn't create problems if we want to change the model in future.--Malore (talk) 16:08, 8 June 2018 (UTC)[reply]

Oppose Use generic properties whenever possible. Many types of things are versions of something, this concept is not limited to software. Danrok (talk) 01:32, 29 May 2018 (UTC)[reply]
~~Oppose~~ I think we would be better off expanding the scope of edition or translation of (P629) to include "or version". Is there a reason this wouldn't work? --99of9 (talk) 01:24, 1 June 2018 (UTC)[reply]

@99of9: edition or translation of (P629) would no longer be the inverse of has edition or translation (P747) - unless we change it into "edition or version" - and it would no longer be an instance of version, edition or translation (Q3331189) - unless we change it into "version, edition, translation or version". Probably other things should be modified because these properties are thought for books, not for software. I think we can use similar structures in order to describe software and books, but not the same one--Malore (talk) 04:02, 1 June 2018 (UTC)[reply]

Ok, I've struck my opposition. --99of9 (talk) 12:26, 14 July 2018 (UTC)[reply]

~~Support~~ Constraints for this property would be different than those for edition or translation of (P629), for example, requiring target items to have properties such as version type (P548). Dhx1 (talk) 13:41, 6 June 2018 (UTC)[reply]
- Oppose I agree with the proposal to expand edition or translation of (P629) instead of creating this new property. Dhx1 (talk) 10:53, 14 November 2018 (UTC)[reply]
Oppose edition or translation of (P629) + has edition or translation (P747) is good enought for this kind of relationship. It's not hard to change labels. FRBR is not only about books. -- Vlsergey (talk) 17:20, 29 July 2018 (UTC)[reply]
Oppose use edition or translation of (P629) and extend scope if needed. The difference between "version" and "edition" is too fuzzy to be of use. -- JakobVoss (talk) 17:59, 29 October 2018 (UTC)[reply]
{{
Oppose no more inverse properties. --Midleading (talk) 04:09, 11 December 2018 (UTC)[reply]

Not done No consensus for creation.--Micru (talk) 15:07, 19 December 2018 (UTC)[reply]