Sounds like an excellent idea to me.
On Fri, 2010-04-02 at 10:23 -0600, Jack O'Quin wrote:
> I just committed some changes to the experimental camera_info_manger
> implementation currently packaged with camera1394 in the
> camera_drivers_experimental stack.
>
> With this change, camera1394 starts cleanly even if the
> ~camera_info_manager/url parameter points to a non-existent file. That
> allows starting an uncalibrated camera, so the file will be written on
> the call to set_camera_info when calibration completes.
>
> This fix is basically a hack to catch exceptions thrown by the
> yaml_cpp package. If the file is in .ini format, it will not work.
>
> Patrick Beeson's sr4k driver uses a separate copy of the
> camera_info_manager. I have not updated that version yet.
>
> We discussed this issue and decided it would be cleaner to temporarily
> pull camera_info_manager into a separate camera_drivers_experimental
> package, used by both, until it is ready for release. Blaise suggested
> adding it to camera_calibration_parsers, which makes sense to me. But,
> we still need a readily available common implementation to support
> testing of the experimental drivers with the boxturtle release.
>
> Unless there are objections, I plan to move the code to that new
> experimental package shortly.
(text/plain)