Describe wildcard usage in parameter files

[jacobperron]

Wildcards were introduced in ros2/rclcpp#762
And further usage proposed in ros2/rclcpp#1265.

[ivanpauno]

Both additions seem like reasonable options (matching all nodes in a namespace, matching all nodes with a name).

I'm not sure if this really has many use cases or not.
If not, I wouldn't add it.

[hidmic]

Overall LGTM, but I will say that along with ros2/rclcpp#762, ros2/rclpy#370 was introduced. Whatever changes are proposed have to be applied to both client libraries IMHO.

[jacobperron]

Whatever changes are proposed have to be applied to both client libraries

I'll follow-up to make sure that this happens.

[jacobperron]

IMO, it would be nicer if the implementation for wildcard matching was done in rcl, then the client libraries would get it for free.

[ivanpauno]

IMO, it would be nicer if the implementation for wildcard matching was done in rcl, then the client libraries would get it for free.

I think that's possible, it only requires a refactor of the current implementation.

[hidmic]

I think that's possible, it only requires a refactor of the current implementation.

Hmm, I think I fiddled with it some time ago. Or @Lobotuerk did. We require extra API, as the wildcard matching cannot happen until we have the fully qualified node name.

[ivanpauno]

LGTM!

[ivanpauno]

Checking ros2/rcl#809, I have some doubts if some things are valid or not:

1. /*
2. /asd/*/bsd
3. /asd/**/bsd
4. /*asd
5. /asd*
6. /a*sd
7. /**asd
8. /asd**
9. /a**sd

4 to 9 should be all invalid IMO, * and ** should always be between slashes.
1 should work, because /asd/* works (and I don't think namespace / should be special).
2, 3 also sound like reasonable things to support, but I'm not sure if they are really useful or not.

[jacobperron]

@ivanpauno I completely agree with all your examples. IIUC, the logic added to rcl in ros2/rcl#809 does consider partial matching invalid (e.g. examples 4 to 9).

[jacobperron]

It might be the case that example 1 isn't considered valid in ros2/rcl#809, I'd have to check.

[hidmic]

It might be the case that example 1 isn't considered valid in ros2/rcl#809, I'd have to check.

We should follow-up.

[iuhilnehc-ynos]

@ivanpauno @jacobperron

I didn't notice your comment before. #303 (comment)

ros2/rcl#809 expects items from 4 to 9 to be invalid (EXPECT_FALSE(res) with the rcl_parse_yaml_file).

Am I missing something?

[fujitatomoya]

@iuhilnehc-ynos i think everything is covered as expected on #303 (comment).

@jacobperron @ivanpauno i will go ahead to merge this. if anything is missing, let us know!