[mkgmap-dev] Do we have a memory leak in mkgmap recently introduced?

Sat Jan 2 11:03:19 GMT 2010

I noticed that mkgmap since a few revistions memory requiresments at 
least doubled. Two weeks ago everything was still okay, rev >=1350 is 
however consuming a lot more memory. The thing that makes me think that 
there is a big leak however is that the memory builds up tile after 
tile. When rendering Germany with max-jobs=2 I needed 5GB of java heap. 
Some month ago I did fine with 3GB on max-jobs=3 (without any big 
changes to either my style, nor an increase in the max-nodes for the 
splitter).

I'll try to find some time to find out if there was a specific revision 
from which onwards the memory requirements increased.