[mkgmap-dev] [PATCH v1] make DP filter avoid removing pointslocated at nodes

Mon Nov 23 21:02:46 GMT 2009

Hi Johann,

> Mark Burton schrieb:
> >> As the object sizes are rounded to multiples of 8, it
> >> looks like a case of buy one, and get 7 for free!
> >>     
> >
> > Yes, just tried that and the same amount of memory is used for 1
> > boolean and 8 booleans.
> >   
> >
> Sorry, I cannot confirm this.

But your figures below do confirm that, don't they?

> Out of personal interest I have run your test programm with OpenJDK6 and 
> get following results:
> 
> johann at ubuntu910:~/workspace>java Test
> Size for LotsOfBooleans: 87978536
> Average size: 87.978536
> Size for LotsOfInts: 336000000
> Average size: 336.0
> 
> 
> This means 1Byte/Boolean and 4Bytes/Int plus 8 Byte per Instance of 
> object, as I expected it.
> Maybe I had to run the vm with a java option?

Yes, but if you only have 1 boolean in the class with the ints it still
uses 336 bytes. So you get 7 for free! If you add no booleans to the
class, the size is 328. So I guess that's why they say that some
people think the class is half empty while the others say that the
class is half full!

Cheers,

Mark