I've scheduled the session for Sunday morning at 10 in Powidl.

Any needed follow-ups from the session?

Notes from the session.

About 10-15 people total, as far as I could tell about half or them from WMF.

jr: an general overview of The Document

• the fact that mediawiki has to run in various environments makes it more robust
• how do we engage with the community?
• for example, should we make sure that mediawiki works in shared hosting?
• if we only focus on mediawiki running on wikimedia production, it would make the job easier
• is there a way that the community can contribute?
• some of the tests are running on travis, not only on wikimedia ci
• is shared hosting support a priority and what concerns the community has?
• how much do we know on how shared hosting has changed in the recent years?
• can we get some metrics about shared hosting (php version....)?
• wikiapiary or some other solution (ori has been working on something)
• it would be easier for the foundation if we have data and know if a smaller of bigger portion of the community is affected by the change

jr: I hate to bring up containers, but a Docker based solution could help the community (pulling the images the foundation has created)

• how much the diversity is intentional, and how much is accidental, because it is the easy way
• it would be good if a small installation would be a similar to wmf

jr: what is the structure of the audience: mediawiki developers, people installing mediawiki for organizations, people using the api

• rewriting stuff (xtools is rewritten for the third time)
• it would be great if the tools directory displayed metrics about the repository visible (test coverage...)
• what would be people looking for when looking for a tool (number of downloads, ratings...)

greg: cloud team is working on best practices for tools (code in a git repository, it is public, it has a license...), the next step would be surfacing their testing practices
jr: what are some other things that would make your life easier developing/deploying?

• the topic of test data would be interesting - I am testing an extension that is doing something unusual, and I need test data

jr: a first step could be collecting as much data as possible, then later figure out how to use it (dashboards), analytics could help

• one thing that comes to mind are wordpress plugin beta testers, volunteers applying for pre-release versions of plugins

greg: we have beta features
jr: wordpress over time has moved to hosting wordpress.com but for a long time their goal was for wordpress to be easily installable

• code quality of wordpress is horrible as well (like mediawiki), but the users (even developers developing plugins) do not see that

jr: how to provide pointers to design patterns?

• tools like striker (?) helps with creating new tools (creates a repo with license, readme file, language specific files), similar to what github does when creating a repo