Maxime Liquet

*Recording some issues.* This snippet used to run fine and `allow_None`, despite being set to `False` in the Parameter constructor, is actually dynamically re-set to `True` internally (`P().param.s.allow_None` is `True`)....

Replying to my own comment above https://github.com/holoviz/param/pull/605#issuecomment-1438862047. This is the same behavior as *main*, which is correct, with `allow_None` being automatically set to `True` if the provided `default` is `None`....

I think I am going to stop there for now :) Early on I felt that the test suite was not good enough to capture all the changes that would...

Thanks for the review @jbednar, I applied your suggestions and renamed another function so that they follow a consistent naming scheme `_compute_...`. I also needed to make a small change...

Just noting that this will need some tests and documentation before being merged.

Indeed, because there's a lot of code in `__init__.py` the auto-completion suggests many irrelevant objects like `param.sys` or `param.OrderedDict`. Moving that code to another module would help. Or importing them...

My plan was to do this refactoring once there are a sufficient number of PRs merged, as that will cause a lot of conflicts.

@philippjfr do you need this in 2.0?

Does it fix https://github.com/holoviz/param/issues/398 too?

I no longer see much value in the changes I made in this PR. As stated above`@param.depends()` is valid. `@param.depends` could either raise an error or be accepted and made...