@n-kall I think the changes in this PR were introduced in #14, and we are now using the simpler form udraws <- x$unconstrain_draws(format = "draws_matrix"). Can this be closed, or is there something I missed?

This PR also has updates using the new are_log_weights =TRUE for pareto smoothing, which could be added. I can look again next week, or you can check if this PR can be selectively merged in the meantime

Those were also added already. In hindsight it would have been much clearer to merge those changes in separate PRs. I'll try to have smaller and better separated PRs going forward.

I actually tried to pull the changes in this branch into the current main, and there does not seem to be any meaningful changes left (see here), unless I missed something in solving merge conflicts.

Great! Yeah I realised that I shouldn't have put these in the same PR. I didn't realise it would be hanging for so long due to delayed posterior release, but we eventually manages to get that on CRAN

Yeah the delay complicated things. But good that it's now clear, I will close this one.