Hi Dmitry, thanks for your review.

On Sun, Mar 6, 2016 at 8:10 PM, Dmitry Gutov <dgutov@yandex.ru> wrote:
> Doesn't the latter allow for several indirections? It seems to me the 
> two functions are equivalent.

It does, sorry for being unclear. I'll try to reformulate it:

There are two lists that must be traversed to find completion
candidates: the property alist and the value class alist. company-css.el
supports indirections from the property alist into itself, and into the
value class alist. With this patch, indirection can happen from the
property alist into itself and into the value class alist, but also from
the value class alist into itself and back into the property alist.

> One question: is there a reason to do the (symbol-name value) 
> conversion before doing the css--property-values lookup?
> 
> Maybe the css-value-class-alist should have symbol keys, not strings.

With the change mentioned above, the property alist and the value class
alist are now treated like the same kind of data structure by
`css--property-values'. The only difference is that the CARs of the
entries in the property alist are also valid property names. So I
thought it simpler if the keys were of the same data type. I don't have
an opinion on whether that should be string or symbol, just that they
are the same.

Did this answer your question?

-- Simen