I like bayesplot and try to use it for most of my plotting tasks. Here are a few cases which could be considered:
ppc_* are great functions for model checking, but sometimes I want to show predictions of the model in absence of data. Could we make the y argument optional? Right now I just pass in the median of yrep so that it does not bother me, but I think this would be generally a good thing to be able to plot model predictions.
ppc_intervals (+ribbon) … how about a prob_outer argument just like mcmc_intervals?
I really like the “x” argument in ppc_intervals/ribbon; more if these in other functions if sensible would be great.
If this sounds reasonable I am happy to transfer these as issues to the bayesplot repo.