how are you meant to turn something that runs from another script (e.g.

gunicorn

) into a

.pex

? setting

script="gunicorn"

works... but doesn't allow specifying the

myproject.wsgi

argument.

Maybe have a stub script for each service that calls into common setup code that invokes gunicorn? (That is what we do at my company for our Django servers.)

You can also just use the entry point exported by gunicorn. This works for things like gunicorn that use an defined entry point to create scripts (https://github.com/benoitc/gunicorn/blob/master/setup.py#L116). I doubt this would work for django or something that doesn't use that mechanism. so you'd have something like...

pex_binary(
    name="app",
    dependencies=[":lib" ,"//:gunicorn"],
    entry_point="gunicorn.app.wsgiapp:run",
    ...
)

And then you could run this like

./app.pex $GUNICORN_OPTIONS <http://your.gunicorn.app:entry_point|your.gunicorn.app:entry_point>

@flat-zoo-31952 Yeah, this is where I'm at. But I feel like that's "broken". as it only ever works if you provide

<http://your.gunicorn.app:entry_point|your.gunicorn.app:entry_point>

as the command line argument.

No, a stub Python script.

To clarify, are you wanting to bake arguments into the pex?

yes. I think

./django-app.pex

should require 0 additional augments to run, even if it's a basic runtime configuration.

That's https://github.com/pantsbuild/pex/issues/987

ah yes. that's the one.