st--

@Joshuaalbert @hughsalimbeni I'm assuming this issue would apply the same to gpflow 2.0?

I'm not quite sure I follow. Isn't the point of the White kernel to decorrelate different measurements at the same input location? Also, either way, why would the mean change...

Should the docs version actually include the minor version? The [Sphinx documentation on `version` and `release`](https://www.sphinx-doc.org/en/master/usage/configuration.html#confval-version) makes me think these should both be `2.0` (instead of tracking minor version)...

@willtebbutt do you mean this one: [Spline bases, regularization, and generalized cross validation for solving approximation problems with large quantities of noisy data](https://pages.stat.wisc.edu/~wahba/TRarchive/TRsEARLY2/tr597_1.pdf) ? Or the 1990 [Spline models for...

At what stage does the reduction from `TypeVar("T", int, float)` to `int | float` happen? Would it be possible to postpone that until after looking at the concrete types of...

@mschauer could you please make a new release (0.5.2?) for compatibility with Distributions=0.25 ?

@chhoumann this seems related so instead of opening yet another issue I assumed it might be better to build on this one (let me know if you'd prefer me to...

> Should we define `approx_log_evidence(::FiniteGP, ...) = logpdf(...)`? This doesn't quite make sense to me anymore, as it would require a different signature from the approximate GPs. Instead, we could...

Hi, could you turn this into a [minimal reproducible example](https://stackoverflow.com/help/minimal-reproducible-example) (include some definitions of the variables you're using, e.g. defined using rand())?

Hi, thanks for the quick response! To follow the [terms of the license](https://github.com/GPflow/GPflow/blob/develop/LICENSE#L89), for the copied/adapted files (conditionals.py, kernels.py, likelihoods.py as far as I can see) you should keep in...