Feature request: set continuous limits for discrete scales

[Yunuuuu]

is it possible to set limits in coord_radial()? Just like what we do in coord_cartesian(xlim = *, ylim = *)

[teunbrand]

This has been suggested before for coord_polar() in the ancient #43. I think setting the limits in the coord_*() hails from a time when there was no scales::oob_keep() to indicate that out of bounds data shouldn't be discarded. I'd gladly remove the xlim and ylim arguments in other coords if ggplot2 didn't advertise it as 'zoom limits' and it weren't a backwards compatibility issue.

EDIT:
To clarify, I think the xlim/ylim mechanism is outdated and hails from a time before we could use oob argument in scales to preserve out-of-bounds values and outdated mechanisms shouldn't be encouraged.

[Yunuuuu]

Coord limits provide an effective way to precisely align axis values, especially since ggplot2 currently does not support setting limits by numerical values alone. This feature enables the alignment of plots with different axis types (continuous and discrete). Removing it would break ggalign, as the package relies heavily on it to align both discrete and continuous axes. With this capability, ggalign can generate highly complex figures.

[teunbrand]

Removing it

It probably won't be removed, precisely because it would cause a lot of breakage all over the ecosystem.

ggplot2 currently does not support setting limits by numerical values alone.

Perhaps it should be easier to set continuous limits for discrete scales.

[Yunuuuu]

It probably won't be removed, precisely because it would cause a lot of breakage all over the ecosystem.

Thanks for the clarification.

Perhaps it should be easier to set continuous limits for discrete scales.

It would be great if this could be possible. While I initially tried aligning plots using scale limits, I found it challenging to align various plots consistently and eventually, I switched to using coord limits.

[teunbrand]

It would be great if this could be possible

Alright I'll repurpose this issue for setting the continuous range of discrete scales then.

[Yunuuuu]

Thanks! This is awesome. I really appreciate it!