Section 8.1.2 **calc() Type Checking does not explicitly indicate how a unitless 0 inside a calc() expression is treated. For example:

left: calc(0 - 5px);

It's not clear whether this expression is invalid. Inside calc((), the types of arguments to the - operator must be compatible, and the rhs is of type dimension<length>. But what is the type of the lhs: number or dimension<length>?

Since section 5 says:

for zero lengths the unit identifier is optional (i.e. can be syntactically represented as the <number> '0'

I would argue that interpreting 0 as being of type dimension<length> would be valid, and would be more in line with developer expectations. However, this was discussed in 2010 with @tabatkins agreeing the 0 should cause a parse error. That's fine, but it would be helpful if the fact that 0 is never treated as a dimension inside calc() were explicitly noted in the spec (which was also agreed upon in that discussion).