use regex for wildcard matching

[iuhilnehc-ynos]

to fix ros2/rcl#954

My intent is to use regex for supporting some complicated wildcard, such as /**/a/b/*/c/d/*/node.

I think that the same param name in a node of param file parsed by order seems more reasonable than the order(/**, specific_node). Because if there are more wildcards items, such /**/node, /ns/**/node, etc, I don't think users would like to memory these special rules.

Note: I can also use a 'std::set<..,std::less>' to store the node keys(* < / < alpha/num), and then move all the relative items iterator into {specific_node, {}}

Signed-off-by: Chen Lihui lihui.chen@sony.com

[iuhilnehc-ynos]

I use your test cases from https://github.com/ros2/rcl/pull/809/files#1280, could you help review this PR? @rpaaron

[fujitatomoya]

@iuhilnehc-ynos is this PR for #1265 and ros2/rcl#954? could you add a description about what we are trying fix and related issue?

[fujitatomoya]

@iuhilnehc-ynos so you kinda have done rebase for #1280 with some changes, right? (new code for support wildcards and borrowing test cases from #1280)

[rpaaron]

I use your test cases from ~https://github.com/ros2/rcl/pull/809/files~#1280, could you help review this PR? @rpaaron

Hi! Its been a little while for me, so need to re-familiarize myself with it..
Should we add a couple of basic tests for parameter_map_from(const rcl_params_t * const c_params, const char * node_fqn = nullptr); with a non null node_fqn ?

[iuhilnehc-ynos]

@rpaaron

Should we add a couple of basic tests for parameter_map_from(const rcl_params_t * const c_params, const char * node_fqn = nullptr); with a non null node_fqn ?

Thank you, I'll add some for it.

[fujitatomoya]

CI:

• Linux
• Linux-aarch64
• Windows

[fujitatomoya]

@alsora @SteveMacenski could you help us review on this?

[tonynajjar]

This would be really nice to have, any reason it's blocked?

[fujitatomoya]

No, we are just waiting for another maintainer's review.

[alsora]

Hi, looks good to me.
Sorry for the long wait.
Can we run again CI before merging?

[fujitatomoya]

CI:

• Linux
• Linux-aarch64
• Windows

[iuhilnehc-ynos]

Rebased branch and re-run CI:

• Linux
• Linux-aarch64
• Windows

[fujitatomoya]

@iuhilnehc-ynos thanks for checking on this! all green, i will go ahead to merge this.

[eranroll]

Hi,

Will this be pulled into Galactic or Humble?

[ivanpauno]

@iuhilnehc-ynos @fujitatomoya is there an equivalent PR for rclpy?

[fujitatomoya]

No i dont think so, I am not sure if the same problem (ros2/rcl#954) can be observed in rclpy.

I think this can be backported to Galactic and Humble since it extends the argument with default value.

CC: @iuhilnehc-ynos

[iuhilnehc-ynos]

@ivanpauno

is there an equivalent PR for rclpy?

Not yet. I will open a new PR for it.
Note for me. https://github.com/ros2/rclpy/blob/1d0cc63579cdcd4c923dd7e04c3c60372fb445f3/rclpy/src/rclpy/node.cpp#L348

@fujitatomoya

I think this can be backported to Galactic and Humble

I think so.

[fujitatomoya]

@Mergifyio backport humble galactic

[mergify]

backport humble galactic

✅ Backports have been created

• #1986 use regex for wildcard matching (backport #1839) has been created for branch humble
• #1987 use regex for wildcard matching (backport #1839) has been created for branch galactic

[ivanpauno]

@fujitatomoya @iuhilnehc-ynos have the new wildcard rules been documented somewhere?

[fujitatomoya]

No i do not think so. can you tell us where it is supposed to be, if you have idea?

[ivanpauno]

Maybe here https://github.com/ros2/design/blob/e160af4bf8955b8cfd735fc65e3d97ff4c18abb2/articles/160_ros_command_line_arguments.md#multiple-parameter-assignments.

Here could be https://github.com/ros2/ros2_documentation/blob/73f10d917bb0a2bdda553970f9988e119ef5548a/source/How-To-Guides/Node-arguments.rst#setting-parameters-from-yaml-files as well.