issue_comments: 618070541

This data as json

html_url issue_url id node_id user created_at updated_at author_association body reactions issue performed_via_github_app
https://github.com/simonw/datasette/issues/731#issuecomment-618070541 https://api.github.com/repos/simonw/datasette/issues/731 618070541 MDEyOklzc3VlQ29tbWVudDYxODA3MDU0MQ== 9599 2020-04-22T22:20:38Z 2020-04-22T22:20:38Z OWNER

I started thinking about this while building https://github.com/simonw/datasette-publish-now

The Vercel hosting platform is built on ASGI - so I need a way to provide an app ASGI object for it to use rather than Datasette's normal pattern of starting up the server using the datasette serve command.

The plugin currently does that by generating an index.py file that looks like this:

python from datasette.app import Datasette import json metadata = dict() try: metadata = json.load(open("metadata.json")) except Exception: pass app = Datasette([], ["fixtures.db"], metadata=metadata).app() But that means I have to replicate the different CLI options - --plugins-dir etc - and figure out how to pass them to the Datasette() constructor.

This would be a whole lot easier if I could use app = Datasette(configure_directory=".") and provide a magically configured instance based on the presence of files in the specified directory.

 
{
    "total_count": 0,
    "+1": 0,
    "-1": 0,
    "laugh": 0,
    "hooray": 0,
    "confused": 0,
    "heart": 0,
    "rocket": 0,
    "eyes": 0
}
 605110015