Scraping error when building drake_ros package via Bazel
#281
Comments
adeeb10abbas
changed the title
drake_ros package via Bazel drake_ros package via Bazel
|
Is this still an issue after #268? If so, it looks like there's an issue parsing plugin description xml files where possibly it's not valid XML. Maybe |
|
@sloretz Yes I am using the latest version of everything so I do believe this is still an issue. I agree 100% that's exactly what I suggested a while back on drake-slack I believe Suggested snippet. I can wrap it up in a PR if you'd like. Thanks! |
|
do ya'll want me to PR it up as well? I can do that no worries! cc: @EricCousineau-TRI |
|
I hit this as you can see I modified it to print the file name so I could isolate it, and it turns out in fact that is invalid XML I'll just uninstall it to move on but given that its a distribution provided file ... not many great options. |
|
I suppose ros-perception/laser_filters#183 hasn't percolated into the repositories yet? |
|
No |
|
Would it be desirable to somehow make the scraping more robust especially when the packages are not directly required for drake-ros to function? Otherwise any one package even if it's a non dependency doesn't allow the user to 1) use drake-ros via bazel 2) help them get around it without adding some logic (similar to the snippet I shared above) to find the root cause. IMO this affects the bazel workflow UX a lot. Just my 2 cents. I personally moved all my ros/drake-ros code away from bazel due to how often I ran into this problem. |
|
Yeah, ideally, we just squelch the scraping error if the package is not in the transitive set of deps specified - sorry you had to run into that! |
|
running into this again at bdai now hehe. Is there a possible near term solution? cc: @jwnimmer-tri @EricCousineau-TRI |
|
I think we would welcome any pull requests with improvements. As I understand it, this metadata is often unused anyway. So, it seems like reifying the error and proceeding would be the way to go. Something like this: --- a/bazel_ros2_rules/ros2/resources/ros2bzl/scraping/metadata.py
+++ b/bazel_ros2_rules/ros2/resources/ros2bzl/scraping/metadata.py
@@ -131,8 +131,14 @@ def collect_ros_package_metadata(name, prefix):
if not os.path.isabs(path_to_desc):
path_to_desc = os.path.join(prefix, path_to_desc)
if os.path.exists(path_to_desc):
- plugin_libraries.extend(parse_plugins_description_xml(
- path_to_desc)['plugin_libraries'])
+ try:
+ new_plugin_libraries = parse_plugins_description_xml(
+ path_to_desc)['plugin_libraries']
+ except Exception:
+ new_plugin_libraries = [
+ f"{path_to_desc}-had-a-parse-error-so-is.missing",
+ ]
+ plugin_libraries.extend(new_plugin_libraries)
if plugin_libraries:
metadata['plugin_libraries'] = plugin_libraries
A more nuanced approach would be to add a field like |
|
Or maybe even the calls to |
Hey everyone, I am running into to scraping errors when I try building with Bazel via
bazel build ...I had been away for a bit from drake stuff, so I am not sure if I am missing anything. Thanks for all the help!
Setup -
apt-get; path to source is/opt/ros/humble)Error log -
The text was updated successfully, but these errors were encountered: