Upcoming scales 1.4.0 release

[teunbrand]

The upcoming release of scales (r-lib/scales#474) prompts a few tasks on ggplot2's side.

• Prepare for future incorporation of breaks_exp() into transform_exp() (Use snapshot expectations for transformation breaks #6171)
• Remove shims like the following

  ggplot2/R/utilities.R

  Lines 792 to 808 in b29b831

  	# Shim for scales/#424
  	col_mix <- function(a, b, amount = 0.5) {
  	input <- vec_recycle_common(a = a, b = b, amount = amount)
  	a <- grDevices::col2rgb(input$a, TRUE)
  	b <- grDevices::col2rgb(input$b, TRUE)
  	new <- (a * (1 - input$amount) + b * input$amount)
  	grDevices::rgb(
  	new["red", ], new["green", ], new["blue", ],
  	alpha = new["alpha", ], maxColorValue = 255
  	)
  	}
  	on_load({
  	if ("col_mix" %in% getNamespaceExports("scales")) {
  	col_mix <- scales::col_mix
  	}
  	})

• Tweak col_mix() calls to continue to produce the same hex codes, because the shim has a grDevices::rgb() backend while scales::col_mix() is based on {farver}, producing tiny differences.
• Rework default colour/fill scales to accept new palette specifications (More palette flexibility #6112)
• Ensure that default palettes can be retrieved from the theme (Scale palettes from theme #5946)

[yutannihilation]

I found the dev version of scales breaks most of the visual tests by a subtle difference of fill (w/ CRAN version: #4D4D4D, w/ dev version: #4E4E4E). Is this expected? Do you observe this on your local? (I just tested on my Windows, so this might be platform-specific problem).

diff --git a/tests/testthat/_snaps/aes/alpha-set-in-alpha.svg b/tests/testthat/_snaps/aes/alpha-set-in-alpha.svg
index 218cedbd2..1e08edd47 100644
--- a/tests/testthat/_snaps/aes/alpha-set-in-alpha.svg
+++ b/tests/testthat/_snaps/aes/alpha-set-in-alpha.svg
@@ -31,11 +31,11 @@
 <rect x='45.03' y='22.78' width='669.49' height='522.33' style='stroke-width: 1.07; stroke: #333333;' />
 </g>
 <g clip-path='url(#cpMC4wMHw3MjAuMDB8MC4wMHw1NzYuMDA=)'>
-<text x='40.10' y='548.14' text-anchor='end' style='font-size: 8.80px; fill: #4D4D4D; font-family: sans;' textLength='22.02px' lengthAdjust='spacingAndGlyphs'>0.950</text>
-<text x='40.10' y='417.56' text-anchor='end' style='font-size: 8.80px; fill: #4D4D4D; font-family: sans;' textLength='22.02px' lengthAdjust='spacingAndGlyphs'>0.975</text>
-<text x='40.10' y='286.98' text-anchor='end' style='font-size: 8.80px; fill: #4D4D4D; font-family: sans;' textLength='22.02px' lengthAdjust='spacingAndGlyphs'>1.000</text>
-<text x='40.10' y='156.39' text-anchor='end' style='font-size: 8.80px; fill: #4D4D4D; font-family: sans;' textLength='22.02px' lengthAdjust='spacingAndGlyphs'>1.025</text>
-<text x='40.10' y='25.81' text-anchor='end' style='font-size: 8.80px; fill: #4D4D4D; font-family: sans;' textLength='22.02px' lengthAdjust='spacingAndGlyphs'>1.050</text>
+<text x='40.10' y='548.14' text-anchor='end' style='font-size: 8.80px; fill: #4E4E4E; font-family: sans;' textLength='22.02px' lengthAdjust='spacingAndGlyphs'>0.950</text>
+<text x='40.10' y='417.56' text-anchor='end' style='font-size: 8.80px; fill: #4E4E4E; font-family: sans;' textLength='22.02px' lengthAdjust='spacingAndGlyphs'>0.975</text>
+<text x='40.10' y='286.98' text-anchor='end' style='font-size: 8.80px; fill: #4E4E4E; font-family: sans;' textLength='22.02px' lengthAdjust='spacingAndGlyphs'>1.000</text>
+<text x='40.10' y='156.39' text-anchor='end' style='font-size: 8.80px; fill: #4E4E4E; font-family: sans;' textLength='22.02px' lengthAdjust='spacingAndGlyphs'>1.025</text>
+<text x='40.10' y='25.81' text-anchor='end' style='font-size: 8.80px; fill: #4E4E4E; font-family: sans;' textLength='22.02px' lengthAdjust='spacingAndGlyphs'>1.050</text>
 <polyline points='42.29,545.11 45.03,545.11 ' style='stroke-width: 1.07; stroke: #333333; stroke-linecap: butt;' />
 <polyline points='42.29,414.53 45.03,414.53 ' style='stroke-width: 1.07; stroke: #333333; stroke-linecap: butt;' />
 <polyline points='42.29,283.95 45.03,283.95 ' style='stroke-width: 1.07; stroke: #333333; stroke-linecap: butt;' />

[teunbrand]

Yeah this is anticipated in the 'Tweak col_mix() calls ...' bullet. The reason is {scales} uses farver, which is slightly different than the current base shim. I have a branch ready to open a PR once {scales} is released. CRAN ggplot2 shouldn't be affected, as it doesn't use the new theme mechanism that relies on this.

[yutannihilation]

Ah, thanks for the details! I didn't notice that bullet is about this and there's a plan. Great.