Change 592695 had a related patch set uploaded (by Tchanders; owner: Tchanders):
[mediawiki/extensions/JsonConfig@master] Introduce basic edit dialog for editing tabular data

https://gerrit.wikimedia.org/r/592695

This looks amazing, @Tchanders . Thank you :) Just waiting on review?

@1Veertje 's spreadsheet-to-json tool is still needed for the many datasources that begin life as a spreadsheet -- we still want some way to paste that into a sheet, and then generate a Tab file.
Is there a way to connect that to this workflow somehow?

Change 592695 merged by jenkins-bot:
[mediawiki/extensions/JsonConfig@master] Introduce basic edit dialog for editing tabular data

https://gerrit.wikimedia.org/r/592695

I just tried it at https://commons.wikimedia.beta.wmflabs.org/wiki/Data:ISO15924/trans.tab -- looks great, and offers an amazing base for further features! Did you remove the "delete row" button?

@Yurik Thanks for testing this! Yes, the ability to add and remove rows was removed for this first iteration, in favour of getting something basic up sooner. There's a follow-up task next up though: T252442: Allow adding and removing rows in dialog for editing JSON tabular data

@Sj @1Veertje The spreadsheet tool looks really useful - thanks for sharing.

I'd be open to suggestions about integrating the workflows, but for now I think they complement each other pretty well. The spreadsheet app looks very useful for generating the JSON and making larger structural changes. Then once the page has been created, I think it's helpful to have the dialog in the wiki for any quick fixes.

I think the right way to integrate spreadsheets and .tab editor is the proper copy/paste handling. It should be possible to paste a spreadsheet directly into the editor, and handle when the paste data is a table from spreadsheet, or has considerable size, or has multiple lines, or has tabs (or commas?), and treat it as a table of text, and ask the user how to deal with it: replace data or append to it. Other cases to handle would be shape mismatch (different number of columns), and type mismatch (e.g. paste data has text in the numeric column). I do not think the special cases like "replace just one column while keeping others intact" should be handled -- for that people can just copy/paste the whole table.

Multilingual column may need additional thinking -- one solution is to flatten them. For example a multilingual column namecould be copied as multiple columns -- name:en , name:fr, ... - for all languages that exist in the data.