Should not strip calc(var(--cp)) expressions

[lunelson]

Currently cssnano assumes that calc(var(--myvar)) will always be equal to var(--myvar) but this is not a safe assumption. var() expressions can contain arbitrary numeric expressions, and may need a calc() wrapper in order to be valid. Example:

p {
  --fz: (4em / 2);
  font-size: calc(var(--fz));
}

cssnano will strip this down to the following, which is not valid:

p{--fz:(4em/2);font-size:var(--fz)}

[alexander-akait]

@lunelson can you provide example when this optimization doesn't work as expected (codepen or other services)

[lunelson]

If you check the following pen you can see that font-size: calc(var(--fz)); works, but font-size:var(--fz); (which is what you get with cssnano's minification) does not:

https://codepen.io/lunelson/pen/OqGEgG

[alexander-akait]

Yep, bug in postcss-calc, thanks for issue

[alexander-akait]

Fixed https://github.com/postcss/postcss-calc/releases/tag/7.0.2