בתאריך יום ה׳, 12 בדצמ׳ 2019 ב-23:37 מאת Pine W <wiki.pine@gmail.com >:
I'm thinking out loud here. Are there any estimates of would be required in terms of time (both staff time and community time) and money to make templates and other tools be much easier to globalize across wikis and across skins? I'm looking for an answer that is more specific than "a lot", but isn't a promise or a detailed estimate.
Difficult to say.
I won't make an actual time estimation, because I'm very bad at doing it, and because I have too many conflicts of interest ;)
However, I do hope to give you something more specific than "a lot". I envision the following feasible plan for "global modules and templates, phase 1": * Make a localization framework for modules. ( https://phabricator.wikimedia.org/T238417 ; probably, but not necessarily, mostly by staff) * Develop a documentation page and a framework for making robust modules ( https://phabricator.wikimedia.org/T238532 ; probably, but not necessarily, mostly by staff). * Make modules storable and loadable from a global repository, and *actually enable it on all Wikimedia projects* ( https://phabricator.wikimedia.org/T41610 ; probably, but not necessarily, mostly by staff). * Migrate most local modules from all the wikis to using global modules, and deleting all the migrated local modules. This will have to be done by the editors communities in many wikis, and it will only be feasible if all the points above are planned and executed well. The challenges I expect at this step are: ** Making sure that just the right amount of things are global and everything that communities want to override locally can be conveniently overridden. ** Making tough choices about which modules to use when several communities developed modules with similar functionality. For example: English, French, Russian, Spanish, and Hebrew Wikipedias have modules for loading Wikidata values. They aren't the same, but they probably should be. Merging them into a global module will require a lot of good-faith collaboration.
Note that I only mentioned modules. Templates have some extra challenges. But once modules are done well, a "phase 2" of this project, that would tackle templates, will become possible. Also, global gadgets will have to be a separate project. Maybe the same localization framework can be used for both modules and gadgets, but I cannot think of anything else that they really have in common.
All of the above is my interpretation of discussions in the recent Tech Conf in Atlanta (other people may have a significantly different interpretation). See these Phab tasks, and the web of other tasks linked to them: https://phabricator.wikimedia.org/T234661 https://phabricator.wikimedia.org/T52329