Adding Eigen::VectorXd

[PepMS]

Hello team,

First of all, thank you for this great tool and for making it open source!

I have a simple question: have you considered introducing support for logging Eigen::VectorXd?

In my own fork, I implemented this in a very straightforward way:

inline RegistrationID LogChannel::registerValue(const std::string& prefix,
                                                const Eigen::VectorXd* value)
{
  return registerValueImpl(prefix, ValuePtr(value), {});
}

and

inline ValuePtr::ValuePtr(const Eigen::VectorXd* vect)
  : v_ptr_(vect)
  , type_(GetBasicType<double>())         // element type
  , type_index_(typeid(Eigen::VectorXd))  // identifies the "container type"
  , memory_size_(sizeof(double))          // element size (consistent with containers)
  , is_vector_(true)
{
  serialize_impl_ = [vect](SerializeMe::SpanBytes& buffer) -> void {
    const uint32_t n = static_cast<uint32_t>(vect->size());
    SerializeMe::SerializeIntoBuffer(buffer, n);

    // contiguous block of doubles
    const size_t bytes = static_cast<size_t>(n) * sizeof(double);
    if(bytes > 0)
    {
      std::memcpy(buffer.data(), vect->data(), bytes);
      buffer.trimFront(bytes);
    }
  };

  get_size_impl_ = [vect]() -> size_t {
    return sizeof(uint32_t) + static_cast<size_t>(vect->size()) * sizeof(double);
  };
}

There may be some caveats that I’m not aware of. In any case, I’m open to discussion and happy to create a PR if you think this would be useful.

Regarding Eigen compatibility in general, we could also add a CMake option to enable or disable building with Eigen support.

Let me know what you think.

Cheers

[henrygerardmoore]

Hello @PepMS thank you for making this! I think that looks good to me. I think adding it behind a CMake option would be best to keep the size down, so please feel free to make that PR! I will review it.