Jonah Gabry

If changing `repair_variable_names()` and `unrepair_variable_names()` is all that's needed to get this working and the change is simple, I'm OK with adding it. But if other things also need to...

Thanks. If you can get that test to pass then you're probably on the right track, but we should test this with parameters that have more than one element. I...

@yizhang-yiz Thanks for reporting this. Does #1120 solve this for you?

The issue is a bit more complicated than I thought. We can get `fread` working using #1120, but unfortunately there are some other problems after that that I don't have...

Setting `simplify=FALSE` when calling the print method also errors due to a weird interaction with the `p_worse`. But I think we can get rid of the `simplify` argument. I've never...

I fixed the tests so this is passing now. I will run reverse dependency checks over the weekend hopefully. I just put one question for you @avehtari in the review...

Starting a comment to keep track of reverse dependency issues. The reverse dependency checks are still running (they will take a while). I will update here as I discover new...

> 1. About the diagnostic messages > Should we use > `N `|elpd_diff| `khat_diff > 0.5` > or > `N `|elpd_diff| `khat_diff > 0.5, possible outliers` > ... > Maybe...

Another question: `loo_subsample` has its own `loo_compare` method and print method. I guess there were enough differences that it made sense for @MansMeg to implement a separate method. @avehtari Do...

> Considering above points, I feel like `diag_p_worse` focuses too much on the probability and not all `diff` columns. As the returned data frame has also `elpd_loo`, I would now...