I've noticed similar confusion from less experienced folks about when
to clean. Is there a reason that rosmake doesn't run make clean by
default? We're going to be setting up one of our computer labs with
ROS for the fall semester, and I'm wondering if I should alias rosmake
to rosmake --pre-clean to avoid the inevitable slew of emails from
frantic students.
-Dan
On Jul 14, 2010, at 12:54 PM, Jack O'Quin wrote:
> On Mon, Jul 12, 2010 at 7:40 PM, Ken Conley <kwc@willowgarage.com>
> wrote:
>> More information here:
>>
>> http://www.ros.org/news/2010/07/ros-c-turtle-alpha-5-released.html
>>
>> -- your friendly neighborhood ROS C Turtle team
>
> We have been upgrading many of our students from boxturtle to cturtle.
> The binary installs seem to work fine. But several students stumbled
> over problems of binary incompatibility for external packages built
> with the two releases.
>
> After installing the cturtle packages, they update $ROS_PACKAGE_PATH
> and try to run their code. It generally segfaults, and they don't know
> what to do (these are smart, but relatively inexperienced
> undergraduates).
>
> Even running rosmake again does not work. Apparently, the build
> subtree is full of incompatible files. So, they need to at least use
> the rosmake --pre-clean option. I've been recommending that they check
> out a whole new tree and build that.
>
> So, this is not a big deal, but I was surprised how much confusion it
> created. It should probably be documented somehow in the release
> notes.
>
> If it were possible for rosmake and rosrun to detect packages built
> with the old ABI and issue a clear warning, that would help a lot.
> Even better if rosmake could automatically do the right thing (e.g.
> "make clean && make").
>
> I consider this more of a documentation issue than a defect, but
> probably many more users will stumble over it soon.
> --
> joq
> _______________________________________________
> ros-users mailing list
> ros-users@code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users
(text/plain)