Thinking about using Playground for editing WordPress Documentation, one of the possible data flows is importing markdown documentation files from a GitHub repository into Playground, making edits, and then creating a Pull Request directly from Playground.

I wonder how we could make the two workflows compatible, so enable translating documentation pages brought over from markdown files from GitHub. We have a two-way markdown<->blocks conversion pipeline at https://github.com/dmsnell/blocky-formats so we can reason about blocks instead of text paragraphs. Would that suffice? Or would we have store some additional structured data in the repo to cross-reference paragraphs/blocks from different markdown files?

@amieiro and I discussed that the current implementation on writing <data> tags to block HTML currently breaks editing the translated blocks as they are all flagged for invalidity.

Possible solution: add the metadata to the block attribute json (for example as fake css class names so that they remain after editing the page gp-original-123) and convert it to a element upon output by getting the data from the database. A positive aspect of this is that we can do this transformation only for people with permission (= logged in users).

Would the meta attribute work? Or storing it as a block binding?