TypeError: '<' not supported between instances of 'NoneType' and 'NoneType' in `_sort_summary_list`

[mansenfranzen]

TypeError: '<' not supported between instances of 'NoneType' and 'NoneType' in `_sort_summary_list`

[mansenfranzen]

Hi @nchaly,

thanks a lot for reporting this bug at such detailed level. It made it very is to reproduce.

It is definitely a bug due to an unhandled edge case. I will get back to you soon.

[nchaly]

@mansenfranzen I agree. In fact current behaviour creates a duplicate block of data, which confused me the first time I saw it - it looked at a first glance as if DataTransformer itself has now got its own schema.

So yes, a reference to the original pydantic model should be fine.

[mansenfranzen]

I merged the changes in the v1.6.0 release. There is a FAQ section in the updated documentation describing your issue. Feel free to reopen or comment. Thanks for reporting this once again.

[mansenfranzen]

It turns out that the bug is not difficult to fix. However, which is more important, your example reveals a more general question. How should autodoc_pydantic document pydantic models/settings which are used as class attributes just like in your example.

While autodocumenting DataTransformer it is not relevant to show every detail of PARAMS_CLS (like json schema, field and validator summary lists etc...). Instead, only a short description is enough. The main focus is on DataTransformer in its entirety and not on the most detailed documentation of all its individual potentially complex attributes.

Moreover, when DataTransformerConfig is used as an attribute, sphinx automatically adds a link to its main auto-documentation section which allows deep dive inspection. That's yet another reason why a brief description is sufficient.

Hence, I suggest to remove all additional sections such as json schema, field & validator summary lists and config summary from pydantic models/settings if they are documented as an attribute. What do you think?

This will also automatically resolve the bug you're experiencing.

[mansenfranzen]

#79 addresses the issue.

Before merging the related PR, it would be great if you could test the bug fix on your site to confirm that my test case did indeed capture the incorrect behavior which you've encountered. To do so, please install the current dev release in your doc-building-environment via pip install git+git://github.com/mansenfranzen/[email protected] and rebuild your docs.

[mansenfranzen]

@all-contributors please add @nchaly for bug and tests

[allcontributors]

@mansenfranzen

I've put up a pull request to add @nchaly! 🎉