Open lwjohnst86 opened 2 weeks ago
I think it could work and I can see the reasons justifying it, 1 and 2 very concretely and 3 in a vaguer way.
I feel like once we decide what exactly will be included in the checks and what the API will be, it will be easier for me to imagine how exactly this package would look like.
Based on some initial impressions, I think we could get away with very few dependencies/assumptions if metadata was input to the functions as a dict
or JSON (as opposed to a *Properties
class). And then we would just need to decide how we want to input the actual data (data frame, from file, etc.) to the functions.
Yea, judging from many of the issues open at frictionless, I can see this being something people might use a lot.
We decided that we will implement within sprout first and if need be, it would be fairly easy to split out into own package.
What would you like to discuss?
I was struggling to fall asleep last night and one of the many thoughts swirling around was about moving the check functions into another Python package.
The main reasons to do that are that:
Thankfully, this package would be fairly small and very focused, so building it and then publishing to PyPI could be done much sooner than for sprout. It would be nice to be able to show an more tangible output of our work so far, sooner than later.
Some ideas for the name could be:
datapackage-lint
ordatapackage-check
.