Hi all, there is still potential for optimisation: To be able to create an index which gives all rules for a tag like surface=gravel, we order the expressions in the rules so that they start with a tagkey=* or tagkey=value. We use the existing tags of an element to find the rules that have to be checked. Now, if we find two rule like surface=gravel | surface=cobblestone {...} surface=* & surface ~'(abc|xyz)' {...} after searching for rules with tagkey surface=gravel we should save the info that the 1st term in the expression is true. However, there is more to it: Other rules may delete or chang the tag surface=gravel, so this may require too much logic. Maybe it can be done with the new cacheId. Gerd