AtJSON
AtJSON is a Conde Nast-sponsored open source project that is intended to be a 100-year document format.
They already have support for ingesting documents through their sources and converting to other formats through renderers. AtJSON truly intends to be an intermediate format:
we need a format that can be rich, extensible, and portable
They currently support the following sources: commonmark (markdown), google docs paste, html, mobiledoc, prism (xml?), and url embeds (unfurling social previews).The currently support the following renderers: commonmark (markdown), plain text, react, graphviz (visualizes the hierarchy of a doc).They don't seem to do any "cleaning up" of the doc at all:
It is important that the content source provide little ambiguity about how a story should be displayed.
We would need to do our own parsing to encode readingOrder (difficult?!) and write our own renderer (not difficult!) to re-order to text into the appropriate reading order.
AtJSON uses vendor prefixes for all the annotations on the doc to prevent collisions:
AtJSON stores annotations with vendor prefixes. These are meant to prevent collisions between different types of documents.
Open Questions
- What do we gain from using AtJSON if we can't use their source or the renderer?
- What if we put postlight's parser in front of AtJSON? Don't we need to write our own parser in order to encode reading order?