Topic on Talk:Wikimedia Developer Summit/2017/Topic ideas

WikiDev17 topic: Distribution and Analysis

9
RobLa-WMF (talkcontribs)
This is the text of the "Distribution and Analysis" section of WikiDev17/Topic ideas as of this writing

How can we better distribute the information on our websites? What data should we make available? How should we offer it? What APIs should we offer to manipulate our content? Topics to discuss: Kiwix, ORES, Bots, RESTbase.

Fora: xmldumps, mediawiki-api, analytics-l, and research-l readers and participants

Qgil-WMF (talkcontribs)

"Distribution and Analysis" seems a too wide umbrella for a main topic? One common denominator there are the Wikimedia APIs, which has been a relevant topic during all this year and will continue to be a hot topic. What do you think about defining a main topic related to the Wikimedia APIs? Ping @BDavis (WMF) since this also affects the Labs users.

BDavis (WMF) (talkcontribs)

I'm not sure that "APIs" is much more narrow. I guess it could mean focusing more on how to improve moving data into and out of the wikis (distribution) rather than the business cases for doing so (analysis).

One topic that could be discussed both broadly and in depth would be bulk data dumps. This is an area that @ArielGlenn works in heavily and was discussed in the 2015 conference. I know that discussion lead to creating a phabricator project. Maybe there are topics in that area that are ready for additional discussion or evangelization?

Another general topic that crosses over a bit with the topics of code quality and future user interface enhancements is ensuring that all new business logic is available via the Action API or other externally automatable means. If we could make a collective decision that a SpecialPage must just be a user interface over an API that can be manipulated programmatically both inside and outside of MediaWiki+extensions then we would be turning a corner towards the possibility of radical new user interfaces (single page app, service workers, rich desktop/mobile apps).

Cscott (talkcontribs)

@BDavis (WMF) I think one of the structural issues w/ bulk data dumps has been who owns the service, and who decides what formats we should dump (or archive). I think @GWicke has taken the lead with the services teams on aspects of this, but I'm not sure that services wants to own all our dumping.

Qgil-WMF (talkcontribs)

I am still proposing "a main topic related to the Wikimedia APIs". Not "APIs" but a main idea related to our APIs that could serve as driver for many related discussions. I'll throw an idea:

Useful, consistent, and well documented APIs

Tgr (WMF) (talkcontribs)

That seems to me like a sub-topic of DX. The requirements for useful, consistent and well documented web APIs is largely the same as for useful, consistent and well documented PHP service interfaces or useful, consistent and well documented puppet roles for Labs (although the way these requirements will be implemented will of course be completely separate).

Qgil-WMF (talkcontribs)

All the main topics proposed could be a sub-topic of something wider.  :) If you are interested in this topic, you are encouraged to think of a proposal and/or promote it in your context so we get the right people and the right topics to discuss.

Tgr (WMF) (talkcontribs)

I'd be interested in gathering feedback about the usability of the action API. For third-party users, the summit is probably not the best place for that (the hackathon or Wikimania are more interesting events for third-party data reusers). For Wikimedia/MediaWiki developers, I think we should gather feedback more generally about what pain points are there. See my comment about DX in the Quality thread.

Tgr (WMF) (talkcontribs)

Offline distribution would be an interesting topic (although maybe too narrow?)

Reply to "WikiDev17 topic: Distribution and Analysis"