Open ntjess opened 9 months ago
I think there are many generalizations that are worth considering from the start:
--input
s or output formatsI think the last one may be attainable, but may also have the fewest legitimate use cases and/or it might be an anti-pattern to structure one's workflow in a way where this saves a lot of work.
There has been some discussion on Discord in this direction: https://discord.com/channels/1054443721975922748/1245821833308012614
One of the suggestions was the option to start Typst as a agent and access it using an ID:
id=$(typst agent diagrams.typ)
typst query --id "$id" --field value metadata
typst c --id "$id"
typst c --id "$id" 'diagram{n}.svg'
typst agent --kill "$id"
this would allow the first three scenarios in a natural way.
Description
I have a large document that I would like to compile and see the results of some labeled content. Currently, I query and compile separately for 2 total calls to typst. It would be ideal to allow querying during compilation and print the query result to stdout while exporting files as usual.
Use Case
Any documents where both metadata and outputs are equally desirable. A common use case is slideshows. In one compilation, the svg slides can be generated and their metadata for transitions, speaker notes, etc. will also be available.