st--

@joelberkeley-secondmind > this ticket is introduced as: here's a problem, implement this solution. Can I suggest: here's a problem, solve the problem, here's one possible way of doing it. So...

@joelberkeley-secondmind we're caching the quantities depending on the training data. There's no caching of test points.

Probably not anytime soon by the core maintainers, I'm afraid!

Thanks for looking into this and sharing your concerns! It's a general challenge with these stochastic tests - there's a trade-off between comparison tolerance, test runtime, and flakiness - we...

Hi @crawlingcub, thanks for looking into this (and bearing with slow replies!).:) I'm a bit hesitant to merge changes that reduce the tolerances. Would it not be better to have...

This is related to #1569, which will be closed by #1582; this issue is about actually implementing the full_(output_)cov cases.

@mohitrajpal1 - would you still be up for looking at how to implement this ?

Great! Looking forward to it.

@SamTov hi, thanks for the prod & apologies for not getting back to this sooner! I think the issue is in the shapes: ```python print(configurations.shape) # == (5, 108, 50)...

Hi @BL7X4 ! thanks for catching this, I suspect you are right. You need to run `make format` (or manually call the `black` formatter) to get the format-checker to pass...