Custom-margin-resetting causing issues when adding points to some plots

[thomased]

The margin-reset-on-exit changes cran wanted as part of 2.6.1 have created a little issue when adding points to plots that require custom margins. E.g. the following should add a point at the origin, but it's offset owing to margin differences (I think?).

data(flowers)
vis.flowers <- vismodel(flowers,
                        visual = "apis", qcatch = "Ei", relative = FALSE,
                        vonkries = TRUE, achromatic = "l", bkg = "green"
)
hex.flowers <- colspace(vis.flowers, space = "hexagon")
plot(hex.flowers)
points(x = 0, y = 0, col = 'red', pch = 20)

Not sure what the simplest solution would be yet.

(As an aside, some reliable way to test images would be lovely...)

[Bisaloo]

The best solution IMO would be to stop messing with margins. Users can set par() themselves if they wish. But I'm not even convinced it's necessary in most cases.

[thomased]

Hah, well that's a fair point. I'll scroll through them all & check but you might well be right.

I'm not sure if we needed to tweak the 3d plot(s) for good reason, but perhaps not...

[Bisaloo]

I think the current code calls for margin tweaking in some cases but my sentiment is that it's a suboptimal behaviour and we should transform this code to avoid it rather than treating the symptom and tweaking the margins

[Bisaloo]

On a related note, par(pty = "s") should be removed to use plot(asp = 1) (which seems closer to the desired effect anyways).

[thomased]

Ah, cheers. Will do.

So after a quick look we can remove all the margin-fiddling for the 2d-plots without issue. I guess we need to do the proper deprecation procedure? Or can we just ditch it since I doubt many people really use it anyway...

[thomased]

Fixed in #220 — thanks.