The memoryview reader is a special stage that allows the reading of point data arranged in rows directly from memory – each point needs to have dimension data arranged at a fixed offset from a base address of the point. Before each point is read, the memoryview reader calls a function that should return the point’s base address, or a null pointer if there are no points to be read.

Note that the memoryview reader does not currently work with columnar data (data where individual dimensions are packed into arrays).


The memoryview reader cannot be used from the command-line. It is for use by software using the PDAL API.

After creating an instance of the memoryview reader, the user should call pushField() for every dimension that should be read from memory. pushField() takes a single argument, a MemoryViewReader::Field, that consists of a dimension name, a type and an offset from the point base address:

struct Field
    std::string m_name;
    Dimension::Type m_type;
    size_t m_offset;

void pushField(const Field&);

The user should also call setIncrementer(), a function that takes a single argument, a std::function that receives the ID of the point to be added and should return the base address of the point data, or a null pointer if there are no more points to be read.

using PointIncrementer = std::function<char *(PointId)>;

void setIncrementer(PointIncrementer inc);