rosbag: auto-generated docs don't include rosbag/bag.h

[trainman419]

Most of the links here are broken: http://docs.ros.org/indigo/api/rosbag/html/player_8h.html

It sort of looks like the generated python and C++ API docs are generated on top of one another because they have the same class names.

This makes it rather difficult to get started working with the rosbag C++ API.

[trainman419]

Apparently these docs have moved to rosbag_storage: http://docs.ros.org/indigo/api/rosbag_storage/html/c++/index.html

And there seems to be some mismatch between the doc generation that is preventing the cross-linking from working properly.

[trainman419]

It looks like rosbag is missing the rosdoc.yaml that should be splitting the python and C++ doc generation into separate paths.

[dirk-thomas]

The rosdoc.yaml and mainpage.dox was moved from the rosbag package to rosbag_storage during the refactoring.

I have readded a rosdoc.yaml and mainpage.dox file for the rosbag package in ebd9e49.

Please recheck the updated docs.
If you think further changes are necessary please comment with specific proposals.

Regarding the links in the dependency graph: I expect them to fix get fixed after multiple doc generation runs. I will keep an eye on it.

[trainman419]

That looks much better! Thanks for the quick fix.

It looks like the cross-links to rosbag/bag.h here: http://docs.ros.org/indigo/api/rosbag/html/c++/player_8h.html point to http://docs.ros.org/indigo/api/rosbag/rosbag_storage/html/classrosbag_1_1Bag.html instead of http://docs.ros.org/indigo/api/rosbag_storage/html/c++/bag_8h.html . There are lots more examples of this throughout the API docs, but this is an easy example to describe and reproduce.

That said, the broken cross-linking is less of an issue now that the top-level docs indicate that most of the rosbag classes have now been moved to a new package ( #299 ).

[dirk-thomas]

I ran the doc job multiple times but the cross-referencing is still there. But this is not limited to rosbag. It seems to be a generic problem. I created a rosdoc_lite ticket for it: ros-infrastructure/rosdoc_lite#51

Are you ok with closing this ticket with the current state of the docs?

[trainman419]

Yeah, I think the remaining issue here is a bug in rosdoc_lite.