Carson Sievert

I don't think this is a problem I can solve from plotly. You're using supplying different data for the x/y positions (discrete for the heatmap and numeric for the dendrogram)....

> So the issue seems to be sharing axes Sure, but why should `subplot()` do the thing you're expecting it to do when `h$p` has *discrete* x/y scales and `h$py`/`h$px`...

Sorry, I meant to say the data that you're supplying to x/y are different: ```r unique(h$p$data$column) #> [1] Sepal.Length Petal.Length Sepal.Width Petal.Width #> Levels: Sepal.Length Petal.Length Sepal.Width Petal.Width unique(h$py$layers[[1]]$data$x) #>...

It would have to be implemented at the plotly.js level

> My only hesitation is with applying the bslib theme by default, since reactable isn't a Bootstrap themed table. I understand the hesitation, but keep in mind that this implementation...

> I like the idea of an opt-out, but am still on the fence about applying the {bslib} theme by default. I'm just worried there'll be enough cases to opt...

Note to self: if new `getCurrentOutputInfo()` styles are available, those should take precedence over sass

I hadn't realized that compilation of datatables Sass isn't even relevant for Bootstrap styling, only for the `default` styling approach. 🤦 As a result, I might drop this idea entirely...

> For the slider I retrieve the dependency with htmltools::findDependencies Maybe you could use `htmltools::htmlDependencies()` instead? This way the functional dependency won't get resolved early (we could also look into...

cc @meztez