[Liblas-devel] libLAS' Boost infection

[Howard Butler]

On Jul 14, 2010, at 11:39 AM, Mateusz Loskot wrote:
>> and boost::array will be working its way into a number of areas where we
>> are using simple static arrays (system identifier and user id in the
>> header, for example).
> 
> boost::array is portable equivalent to std::array defined in upcoming
> C++0x (GCC 4.3.x or 4.4.x as well as Visual Studio 2010).
> It may be a good idea to use a liblas::array alias defined as
> either by boost::array or std::array depending on compiler used.

Will do.

> 
> > Boost also provides many other interesting
>> possibilities, but for now, I will try to limit the requirements to
>> header-only libraries.
> 
> What's the state of the large files and streams issue?

Mostly untouched at this point <http://trac.liblas.org/ticket/147>.  This is only a problem when *seeking* to the end of the file to determine how many points are in it, and even then only in the case where we're updating an existing file's header, not writing into a new file.  This all only happens in the case where the file is greater than 4gb on Windows/OSX/Linux.

I think the most reasonable option is to just believe what the header says in that case, and if the user updating a header on an existing file is not keeping track of the point count, it's their own fault.  LAS 1.3 files can have waveform data after the point data, so this strategy of seeking to the end of the file and subtracting the header to determine the number of points isn't going to be valid anymore.  We're going to need to believe the header's point count and size information to get the count.

Howard