Talk:Q727439

From Wikidata
Jump to navigation Jump to search

Ranking of legacy branch 1.2[edit]

Hello @AMDmi3:,

I've seen that you have re-introduced the latest version of SDL "legacy" branch (1.2.15) as "preferred" (here, and a few weeks back there) with the following message "This change is incorrect: 1.2.15 is an actual release of SDL 1.x branch".

I was under the impression that the rank "preferred" should be used for the most recent stable release for any branch; there should not be one "preferred" version per release branch (see software version identifier (P348) talk page). For example, for PostgreSQL (Q192490), even though they maintain 5 branches (from 9.4 to 11), only the latest release for branch 11 is set as "preferred".

Maybe I missed something but I think for SDL there should only be the latest release of branch 2.0 set as "preferred" (currently 2.0.10). What do you think?

Naked8Snake (talk) 13:05, 3 October 2019 (UTC)[reply]

When upstream supports and develops multiple branches in parallel, there's a reason to it (API stability, compatibility etc.), there are users for each branch, and for these only releases in specific branch are important, so they all should be marked as preferred to convey an actual list of all latest releases and to inform data consumers on update in every supported branch. From practical standpoint, it, for example, allows to list all latest versions on wikipedia pages (most software infoboxes already support multiple branches in form of stable/development ones for which you need multiple preferred versions - technically, multiple stable branches is no different from this). Marking only one branch with preferred rank would just lead to incomplete data. Lost ranks should be fixed for PostgreSQL as well, but for SDL it's even more important as 1.x and 2.x are completely different incompatible libraries, 1.x still being used by a lot of software. --AMDmi3 (talk) 14:02, 3 October 2019 (UTC)[reply]