Tom Vo

> Just weighing in that I usually lean toward convenience, but in this case I'm not so sure. It could lead users to systematically ignore the bounds that the dataset...

@pochedls @lee1043 do you guys typically use bounds from the dataset as the default option in most cases (if they exist) or mainly use xCDAT's add bounds/add missing bounds APIs....

> Oh,,,, @pochedls thanks for clarifying, I must have confused it with the `add_missing_bounds`! Sorry for the confusion. What I suggested [here](https://github.com/xCDAT/xcdat/issues/435#issuecomment-1992240392) is about the `add_missing_bounds`, not `add_bounds`. I think...

I was considering the idea of a `ds.bounds.drop_bounds()` API for convenience and special cases where bounds are incorrect. It could be nice to quickly drop bounds without having to use...

@pochedls Sure, we can re-open this issue to gather ideas. If we don't have any substantial reason to have this feature, then we can re-close it.

Just following up after some time of inactivity in this ticket. > If I loop over 50 models, I might reproduce the correct bounds for the vast majority of models...

I think the simplest, most convenient path is to implement a `drop_bounds()` API that automatically identifies the bounds (via CF `bounds` key) to drop based on the specified axis/axes. That...

If both of you think this type API would be generally useful to the broader user community, I'm happy to add it to the backlog as a lower priority enhancement.

> This seems like a pretty risk-free feature to me. We could add a `regenerate` flag if we wanted to regenerate the bounds using xcdat. This also seems like a...

> Basic driver and plotting scripts are working. Through only with `MULTIPROCESSING=FALSE`, if switching it on, I hit errors as follows. It appears from `ds = xc.open_mfdataset(**args)` which is newly...