[ros-users] urdfpy - URDF parser in Python

Thomas Moulard thomas.moulard at gmail.com
Mon Mar 19 21:58:00 UTC 2012


On Mon, Mar 19, 2012 at 8:06 PM, Kelsey Hawkins <kphawkins at gatech.edu> wrote:
> I agree that yours is probably the best bet for us to push forward with.  If
> you could formally release a version, I'd be happy to help add to it.

Me too!

>> 1) Should we split the interface and parser/writer?
>
> I like the way you have kept them together, it's easier to read.

Ok, it seemed to me there was an effort to split interfaces from parser in ROS
in general so I avoided mixing both. For such a lightweight parser it
is probably
not that important.

>> 2) Which name should we unite under?
>
>  The C++ equivalent to this package is named urdf_parser.  Realistically, it
> would make sense if eventually we could integrate it into this official
> package.  Until then, I think we should keep with a similar name.
> urdf_parser_py might make sense but it doesn't really matter to me.  What
> would be nice for me is if we could have a stack (like my
> robot_model_python) in a repository we could all maintain where my KDL
> parser and other similar packages could reside.

Ok, the only thing that I would prefer is having a stack that we can
release as a Debian package
for at least Electric and further as this is what I use on the robot.

>> 3) I think that we might want to merge into my package, but would be open
>> to other suggestions. Does that sound good?
>
> Sounds good to me.  It might also be nice to incorporate Thomas' string
> pretty printing.  I'd also personally make a few changes to the URDF class
> you have if you think they're reasonable.  First, make the parse and load
> methods static so you can load the URDF object in one line.  Secondly, make
> those two more intuitive by renaming them parse_xml_string and load_xml_file
> or something similar.  Not making these changes isn't a game-killer though.

Agree.

-- 
Thomas Moulard



More information about the ros-users mailing list