Jump to content

Topic on Manual talk:Adding support for new filetypes

Point to example of new-type-workflow on Commons?

8
Sj (talkcontribs)

I know this is for Mediawiki generally, and only parts of the man page are about integration with WM wikis. But it would be illustrative in that section to point to where these sorts of requests are handled on Commons (which should likewise point back here).

The various para-technical reasons that adding new filetype support is hard on Commons may also be relevant to other implementers.

Jdforrester (WMF) (talkcontribs)

There isn't a process. Each time is bespoke, and needs to run a different gamut of concerns depending on who objects that week, it seems. There's also no team that's the owner of all the different aspects of media support for MediaWiki, let alone Wikimedia production, so I'm not sure who would be best-placed to write up such a workflow, sorry.

Sj (talkcontribs)

Ok. Sounds like there should be a checklist for proposing and speccing our a new request (RfFT?) on Commons; for evaluating a new file type to be added to MW options; and for evaluating a new file type for WM production.

And there should be at least a single owner of new multimedia issues in Core, to respond to the last...

Sj (talkcontribs)

I filed an umbrella request to add data-format and document-format file support : https://phabricator.wikimedia.org/T297514

This seems like the most impactful subset of the broader range of requests, hopefully sharing enough context and requirements to be addressable as a body.

There should absolutely be a page describing how to initiate, structure, and see to completion one of these requests; adding relevant filetype support is Step 0 of bringing on a new community of practice, helping them integrate free-content-license workflows into their work, &c. I have no sense of where that should live, but it should link to the elder tracker https://phabricator.wikimedia.org/T44725

Jdforrester (WMF) (talkcontribs)

Linking ignored Phabricator tasks to other ignored Phabricator tasks isn't going to get a ~US$2m annual budget line created, unfortunately.

Sj (talkcontribs)

You think the obstacle here is budget? It sounded above like prioritization and ownership.


$2M a year is not a lot for multimedia. But this is just about adding a dozen new file formats. Is that a $2M ongoing cost? Elsewhere someone suggested $20k + $2k/yr per type... and even there the recurring costs w/in a category should have returns to scale.

Jdforrester (WMF) (talkcontribs)

No, US$2m is just my rough guess at an annual budget just to keep the lights on and do basic maintenance (1 tech lead, 2 engineers, 2 SREs, a community liaison, an engineering manager, hardware capex, etc.).

To do anything actually feature-ful (which is what everyone wants, and so where all the demand to executives will come in) would require possibly double that (mostly around treating the security concerns, plus also hiring a product manager, a designer, etc.).

Adding a novel file type into production is trivial (it's maybe a few dozen lines of code). Adding one in a way that random people can't take down the production cluster is intensely hard.

Sj (talkcontribs)

That is a ... truly intriguing and provocative answer which it will take me some time to digest.

Reply to "Point to example of new-type-workflow on Commons?"