Tomoya Fujita

> ('Invalid parameter value ([]) for parameter. Reason: Passed value is of type 5, but ParameterDescriptor declared type 9' i do not see this error with mainline https://github.com/ros2/ros2/commit/9597c36591c643a1ff6d87964ddd3d999f191baf. but actual...

@travipross > Is there truly no way to declare a STRING_ARRAY parameter with rclpy whose default value is an empty list? Not all empty arrays are BYTE_ARRAYS, and by ignoring...

@travipross really appreciate your detailed explanation. > But when I change my default parameter value in the node, or the parameter value in my YAML file, to instead be an...

@travipross thanks for iterating with me, lets keep this open for more feedback from other developers 👍 it would be better to get more information or ideas before implementation. (saying...

> [] defaults to BYTE_ARRAY this is weird i think all problems and complains come back to this point. `[]` empty list tuple and array will be deduced to `BYTE_ARRAY`...

@clalancette @wjwwood friendly ping.

CI: * Linux [](http://ci.ros2.org/job/ci_linux/20753/) * Linux-aarch64 [](http://ci.ros2.org/job/ci_linux-aarch64/15052/) * Windows [](http://ci.ros2.org/job/ci_windows/21432/)

@InvincibleRMC this seems including unrelated changes to this PR addressing? i think we need to rebase?

CI: * Linux [](http://ci.ros2.org/job/ci_linux/21324/) * Linux-aarch64 [](http://ci.ros2.org/job/ci_linux-aarch64/15637/) * Linux-rhel [](http://ci.ros2.org/job/ci_linux-rhel/830/) * Windows [](http://ci.ros2.org/job/ci_windows/22054/)

@wjwwood requesting another review when you have time.