Hi Barend, Thanks for the information. Nothing related but, have you done any comparison with GeographicLib ( http://geographiclib.sourceforge.net/)? In term of performance and accuracy for geodesic computation? Thank you, Romain On 23 March 2010 01:22, Barend Gehrels <barend@geodan.nl> wrote:

Hi Romain,

Thanks for your interest.

Does anyone know if the Boost Geometry Library is going to be shipped in

...

the next release of Boost?

We're doing our best but probably the next release is still too early.

If you want to "early adopt" it, it is possible.

Regards, Barend

_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

Hi Romain,

Nothing related but, have you done any comparison with GeographicLib ( http://geographiclib.sourceforge.net/)? In term of performance and accuracy for geodesic computation?

I know this library but didn't test these aspects. All our projections are converted from proj4 to C++, and by this conversion alone they are slightly faster. It is indeed interesting to see if there are differences with this library. About accuracy: we will support both double and high-precision arithmetic types and it is also interesting if there are differences here. We don't have yet official performance tests for projections in the suite Mateusz pointed to. Actually the projections are an "extension" and were not part of the review, even if Boost.Geometry is in the next release or in 1.44, it might be that the projections are still not there yet, we have to harmonize some things as earth models. Regards, Barend

Hi Mateusz, Barend No worries about the previous post :-) GeographicLib is a set of C++ classes, there is no such dependencies with PROJ4 (from my understanding). I actually asked Barend if you guys had a look to this library because the library can do geodesic computation: http://geographiclib.sourceforge.net/html/geodesic.html <http://geographiclib.sourceforge.net/html/geodesic.html> http://geographiclib.sourceforge.net/html/classGeographicLib_1_1Geodesic.htm... <http://geographiclib.sourceforge.net/html/classGeographicLib_1_1Geodesic.html>However, I am not sure that Boost Geometry is able to perform "geodesic direct / inverse" computation (cf. links before). Any advice on this part? Thank you. Romain On 25 March 2010 03:31, Mateusz Loskot <mateusz@loskot.net> wrote:

Barend Gehrels wrote:

...

Hi Romain,

Nothing related but, have you done any comparison with GeographicLib (

...

http://geographiclib.sourceforge.net/)? In term of performance and accuracy for geodesic computation?

I know this library but didn't test these aspects. All our projections are converted from proj4 to C++, and by this conversion alone they are slightly faster. It is indeed interesting to see if there are differences with this library.

AFAIS, Geographiclib is not a standalone library but it uses PROJ.4

We don't have yet official performance tests for projections in the suite

...

Mateusz pointed to.

I have to confess I replied to Romain's post assuming, incorrectly, Geographiclib is a yet another geometry library.

Sorry for confusing.

Best regards, -- Mateusz Loskot, http://mateusz.loskot.net _______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

Hi Romain,

GeographicLib is a set of C++ classes, there is no such dependencies with PROJ4 (from my understanding).

I agree, I didn't see any reference.

I actually asked Barend if you guys had a look to this library because the library can do geodesic computation:

http://geographiclib.sourceforge.net/html/geodesic.html <http://geographiclib.sourceforge.net/html/geodesic.html> http://geographiclib.sourceforge.net/html/classGeographicLib_1_1Geodesic.htm...

<http://geographiclib.sourceforge.net/html/classGeographicLib_1_1Geodesic.html>However, I am not sure that Boost Geometry is able to perform "geodesic direct / inverse" computation (cf. links before). Any advice on this part?

Boost Geometry can do distance calculations on the ellipsoid, which are modelled as "strategies". But not this specific geodesic, neither direct nor inverse. It can currently do Vincenty and Andoyer. Because Boost.Geometry is designed with strategies, you might even mix the two libraries, using Boost.Geometry for normal calculations, creating a strategy using geographiclib. Didn't do this in practice yet. Regards, Barend

Hi Barend, Is there any plan to include this kind of computation in Boost Geometry? Another solution that I was thinking about - using Boost Geometry as it is right now - is to convert Geographic coordinates in Cartesian coordinates (it is possible to do that right?), and do the same kind of computation but based on Cartesian reference system (computation of distance, azimuth, and line should be easy). Would it be possible? Thank you, Romain On 25 March 2010 18:54, Barend Gehrels <barend@geodan.nl> wrote:

Hi Romain,

GeographicLib is a set of C++ classes, there is no such dependencies with

...

PROJ4 (from my understanding).

I agree, I didn't see any reference.

I actually asked Barend if you guys had a look to this library because the

...

library can do geodesic computation:

http://geographiclib.sourceforge.net/html/geodesic.html <http://geographiclib.sourceforge.net/html/geodesic.html>

http://geographiclib.sourceforge.net/html/classGeographicLib_1_1Geodesic.htm...

< http://geographiclib.sourceforge.net/html/classGeographicLib_1_1Geodesic.htm...

...

However, I am not sure that Boost Geometry is able to perform "geodesic direct / inverse" computation (cf. links before). Any advice on this part?

Boost Geometry can do distance calculations on the ellipsoid, which are modelled as "strategies". But not this specific geodesic, neither direct nor inverse.

It can currently do Vincenty and Andoyer.

Because Boost.Geometry is designed with strategies, you might even mix the two libraries, using Boost.Geometry for normal calculations, creating a strategy using geographiclib.

Didn't do this in practice yet.

Regards, Barend

_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost

Hi Romain,

Is there any plan to include this kind of computation in Boost Geometry?

At this moment geodesics not planned. However, as Mateusz said, we have the extension model so if it is somehow needed, it can be added extended functionality. As in my previous post, it can be implemented as an additional strategy

Another solution that I was thinking about - using Boost Geometry as it is right now - is to convert Geographic coordinates in Cartesian coordinates (it is possible to do that right?), Yes, it is possible.

and do the same kind of computation but based on Cartesian reference system (computation of distance, azimuth, and line should be easy). Would it be possible?

It is probably possible but I don't think it is the right way to do it. The geodesic calculation is a high accuracy calculation. As soon as you project to a Cartesian projection, you will loose that accuracy in the calculated distance. The same will be true for other properties as azimuth. Regards, Barend