Maximilian Scholz

Results 33 comments of Maximilian Scholz

Uhm I think Stefan (the external who'll join ploen) said that.

It's hard to decide if a change in a cEP is "API breaking" and needs a major version or if it Is enough to increase the minor version.

commits, reviewed PRs and issues. The idea was to instead of showing the overall stats, we show the monthly stats. I don't really have time atm, you can ask the...

Just curious, is there a rough timeline for the 1.0 release? Im currently writing a package that I'd want to use polars in and pondering if I should wait (if...

Wondering if it'd be cleaner to have an explicit `split_chains` function or option for the `mcmc_trace_data` function, rather than implementing this for each of the plotting functions individually. Lemme know...

Oh nice :D good to know. Agree with adding it as a default False option to keep the interface stable.

I'll try to come up with a new version when I find the time this week :) What do you think about renaming the chains/legend in the split chain case?...

Small thing I noticed, `split_chains` requires a `draws` object. But `mcmc_rank_xx` worked with raw arrays so far. So the input would have to be cast via `as_draws` first.

Hmm I feel like IF rank plots SHOULD be used with split chains by default, it should be supported by bayesplot. If using split chains is a niche thing that...