[mkgmap-dev] Possible optimization of StyledConverter

Fri Jan 6 12:20:07 GMT 2012

Hi
> thanks for the explanation, I'll verify my assumptions. Regarding your
> patch:
> I see a change in the number of evaluated rules, but it is is slightly
> higher with the patch. I think this proves that my understanding is wrong,
> but the patch is probably too special, means, it has to be modified for each
> set of styles.

I'd expect it to vary based on the actual input data, the tag/values with
the lowest probability of being matched should be moved towards the
beginning so that as few rules as possible are selected to begin with.

My example of amenity was probably a bad one to include, since I was
only looking at amenity=restaurant, and it may not be benificial with
other values.  I'm would be suprised that the highway one did not make
an improvement. But in any case, I didn't see a difference in
performance and I suspect that there are better places to look.

For example there are 56 hardwired calls to getTag() in
StyledConverter.

..Steve