KDL wishlist


For some time now the user base of KDL has been expanding significantly,
and the use cases are becoming increasingly more complex. Practical
experience is showing that parts of the KDL API are difficult to extend in
a modular fashion. A particular example that has been quite active recently
is that of Inverse Kinematics solvers [1, 2, 3], where extensions
frequently end up resorting to code duplication and little reuse. Apart
from the obvious problem of maintaining fragmented extensions that do make
it upstream, many just don't make it back to KDL, and people start to
duplicate efforts over and over again.

Possible solutions to these issues have been proposed on the list [4], but
I believe the more general subject of how to improve the KDL developer
experience deserves some attention, so I'd like to ask fellow KDL
developers and users for their opinion on what they would like to see in a
future release of the library. I just created a wiki page [5] and seeded it
with a couple of ideas. Feel free to contribute your own. If there is
enough community interest, maybe we could organize a KDL developers meeting
in the future.

[2] http://www.orocos.org/forum/rtt/rtt-dev/new-solvers-welcome
[5] http://www.orocos.org/orocos/kdl-wiki/kdl-wishlist