Matthew Kay

Hmm yeah, to be consistent with other cases where this comes up, conversion functions from `draws_df` (which I believe is the only format that supports unequal chain lengths) to other...

Hmmm... What if summarise_draws passed a single-variable rvar down to convergence functions instead of a draws_array? `draws_of(x, with_chains = TRUE)` could be used in convergence functions to basically get the...

Ah yeah. I think this kind of relates to other questions about modifying summarise_draws to support things like weighted draws. See this comment: https://github.com/stan-dev/posterior/issues/273#issuecomment-1440350034 That is, if we're going to...

Another solution might be to check for an rvar implementation on the summary function generic and pass that format if present, and otherwise use an array. That way rvar support...

Thanks! if it helps, we had to do something similar in `as_draws`, so there's already a utility function for checking for S3 method definitions: https://github.com/stan-dev/posterior/blob/e23467b87aa5d4b5df2f7f5bc545c201186ea2c5/R/as_draws.R#L51

Yeah, good point... That could lead to counterintuitive behavior for users (and hard to track down bugs) if they start with a summary function then add arguments to it by...

Yeah good point. This lends some priority to #184 then. I can take a stab at that when I next have time...

I would do `draws_of(x, with_chains = TRUE)` then use `apply()` It's a good question if we should export something - my thinking so far has been to keep that stuff...

If I understand what you're asking, that sounds about right for how to do what you want --- compute the relevant columns then map them to the sub-geometry colors. Let...

Is this another example of the general problem of how to incorporate information from special variables (like weights - see #184) into summary calculations? Maybe the solution to this should...