Attribute Parser

  • + 1 comment

    I suppose in practical terms a tree could only be useful if memory was concern. You'd have to calculate how much memory can a tree save you depending on various characteristics of data. It would seem that if tagnames are relatively long and tags are rather branchy, you could end up with an arbitrarily high memory overuse with the flat map, because a lot of your keys would be very long. Then again, you could try using hashed keys to get rid of that problem.

    On than memory, tree could possibly have a very bad impact if this was some intense code, as I imagine it would lead to a lot of CPU cache misses.