Hi,
I would like to write this mail in a constructive, positive way, so please,
understand my complaints as "warnings" to improve ROS.
I have spent a lot of hours trying to package ROS fuerte for a Debian distro
(Squeeze), and I have to admit that I desist. First because the difficulties
to package a version of ROS that probably will expire soon (November: Groovy
release?). Second, to package a version of ROS that has a lot of issues (pcl,
perception ...) means a lot of work patching at least to have something that
compile!!!
I understand the gentoo guy about the question of download source tarballs. My
feeling is that ROS is overflowed of repos, versions, packages, etc. I admit
that I don't understand the policy of the releases of ROS. I have found
several versions of ros-fuerte. For example, I could compile pcl_percepcion
from a version of source of July (with a pcl1.5) but I couldn't repeat the
same in September. Ros install download a different version.
Also, not obey fhs structure is a problem. I hope that catkin could help to
solve that issue. As a Debian user, and amateur packager I try to follow the
Debian guidelines. There are very good. I have packaged ros-fuerte-base and
ros-fuerte-rx, but I have stuck with the stacks, specially perception_pcl.
I don't know if in general, the ROS users just use Ubuntu and make apt-get and
doesn't take any more question and just to develop their robotics application.
Maybe it's just because I do a half robotic /system job but maintain a group
of boxes (worst in a multiuser environment) with ROS it's not easy.
So, it does worthwhile to spend more time to try to have a Fuerte version for
Squeeze and up? Or it's better to wait and concentrate effortss in Groovy?
Regards,
Leopold
--
--
Linux User 152692
Catalonia
(text/plain)