Skip to content

Getting your Daisi "Verified"

What are "Verified" Daisies?

"Verified" Daisies have been through a verification process which guarantees their usefulness and operational readiness. These Daisies are presented at the forefront of each section of the catalog and a badge is added on their card.

Verification is a manual process undertaken by the Daisi team. To get a Daisi "Verified", the Daisi creator has to make a request to the Daisi team.

"Verified" Daisi have unit tests, which are run automatically on a regular basis. If at some point a Verified Daisi fails to pass the unit tests, it will be removed from the "Verified" section until the problem is fixed.

Criterias to get a Daisi "Verified"

A "Verified" Daisi has to be fully useful to the community and functional at all time. It needs to fill the following requirements:

  • Usefulness: Characterizing the usefulness of the Daisi to the community is at the sole discretion of the Daisi team. Trivial and obvious Daisies won't get a Verfied status (For instance a "Hello World" or "Add Two Numbers" Daisi can be great as an example but is intrinsically useless).
  • Uniqueness: Two identical Daisies can't be both "Verified". If a candidate Daisi for verification overlaps an existing "Verified" Daisi, the Daisi creator will have to explain how this Daisi is a valuable alternative to the already verified Daisi.
  • Documentation: "Verified" Daisies must be complete with an image on their card, a short description, docstrings for their key endpoints and a Readme. A Streamlit app is not mandatory as long as the automated UI is functional
  • Operational readiness: A "Verified" Daisi must have at least one documented endpoint. A dataset for testing needs to be provided to the Daisi team. This dataset (endpoints inputs and expected outputs) will be used to buil unit tests for the Daisi. These unit tests will be run periodically.

Submit your Daisi to add it to the "Verified" collection

  • Step 1: Make sure that your Daisi is complete, execute the endpoints with the pydaisi client, make it Public (the GitHub repository of the code doesn't have to be Public)
  • Step 2: Prepare a dataset for unit tests, providing for each endpoint one example of input data and expected result. If your Daisi takes in input and/or returns arbitrary Python object, save them in the Pickle format
  • Step 3: prepare a one page document exposing the usefulness and the uniqueness of your Daisi
  • Step 4: prepare and save a Jupyter notebook showing how to load the input and output of your Daisi and how to invoke the endpoints.
  • Step 5: prepare a Zip file with all the above (Input/Output data, in Pickle format if needed, Jupyter Notebook, Document) and post it to the channel "Daisies verification" on Include a link to your Daisi in your post.

Daisi update

You can update your Daisi a long as the unit tests of your endpoints pass sucessfully. If an update breaks the unit tests, your Daisi will be automatically removed from the "Verified" Daisies collection and you will need to contact us again to obtain a new "Verified" status.

Note : we reserve the right to update this process and policies in the future, at our sole discretion, either with or without notice