Matthew Kay

Yeah good point. The two options to my mind are: - throw an error in `check_existing_variables()` if any elements of `variable` are `NA`. - something consistent with how variable selection...

Yeah this is hard --- would probably need something like #217 so that thickness values could be translated into `y` values for that geom.

I can't reproduce the bug on the latest versions of brms and tidybayes. From the output it looks like the error is coming from functions in brms, so it is...

Hmm, at some point I should probably support more arbitrary indexing schemes in those functions. Just added an issue for that here: #322 Unfortunately I don't have cycles for it...

Okay, I think this is ready for review, pending two things: 1. Do we want to add an attribute distinguishing MC vs MCMC to this PR? I would suggest we...

Thanks for this! Is there a fix that doesn't introduce a direct dependency on abind? Might be able to use NROW instead of nrow for the first bit.

This suggests to me that the output is being bound into a matrix (which requires a single type) and then converted to a data frame later, which it should probably...

I think since those args don't apply to all the functions anyway I'd be inclined to wrap the relevant functions in a lambda, like define `p = c(.01,. 99)` then...

Personally I think anonymous functions make the code clearer as they move the args closer to the function they apply to: ``` r summarise_draws( example_draws(), \(x) quantile2(x, probs = 0.1),...

Because `summarise_draws()` can take a named list of functions, you can construct partially-applied versions of your desired functions and pass them as lists. This allows you to apply the arguments...