[Liblas-devel] Point limit in the software

[Howard Butler]

On Feb 23, 2009, at 5:09 AM, Neil Woodhouse wrote:

> Hi,
>
> 	I am using the libLAS software extensively and have came up with
> an oddity.


> If the file is created with over 450 million points the file
> cannot be read by the PointVue viewer

I have check through the code and it looks like we are using uint32_t  
(unsigned long) throughout the code to describe the number of points  
in the file (this is in lasheader.hpp/cpp).  This is what the spec  
said to use, so I don't think we have a problem there.

> so I
> used the LAS tools to create the file from the XYZ file. It works.

the LAS tools that come with libLAS, or Martin Isenberg's tools?

> I do
> not know the number of points that I am writing until the very end of
> the processing and I write the header at this point.

It would be helpful to see some code.  The LASWriter keeps track of  
how many points it is writing into the file as it writes them, and  
there is a function called UpdateHeader that is supposed to take care  
of writing the number of points in the file for you when the  
LASWriter's destructor is called.  You shouldn't have to manually keep  
track of the point count, and if you were doing that plus  
inadvertently clipping off precision by not using unsigned long, this  
might be the problem.  We would need to see some example code though.

http://liblas.org/browser/trunk/src/detail/writer10.cpp#L218

I wouldn't rule out that this isn't libLAS' problem though, as I've  
never attempted to write that many points with libLAS before (my quick  
python example to test writing 425 million points is still grinding as  
we speak ;).

Thanks for using libLAS, we hope it is working well for you.

Howard