Implement rcl_clock_time_started#1021
Conversation
b3c2184 to
d3a766b
Compare
fujitatomoya
left a comment
There was a problem hiding this comment.
although this change itself looks good to me, i would suggest that the following procedure.
- revert #1018 only, and merge this revert commit in rolling.
- then apply
rcl_clock_time_started.
it is up to package maintainers, but highly recommended. each PR should have complete fix as less dependencies as possible.
Fair point! #1022 I've reconfigured this PR @fujitatomoya , the revert change is split off. I will merge on green CI |
3aaab01 to
78f1934
Compare
fujitatomoya
left a comment
There was a problem hiding this comment.
lgtm with one comment
Signed-off-by: methylDragon <methylDragon@gmail.com>
7e31df0 to
51a0841
Compare
|
@Mergifyio backport humble |
✅ Backports have been createdDetails
|
Co-authored-by: Tomoya Fujita <Tomoya.Fujita@sony.com> (cherry picked from commit 09d86e2)
This PR reverts #1018 and implements
rcl_clock_time_startedinstead, which checks if anrcl_clock_thas started (i.e., outputs a non-zero time when queried for current time.)This was from an internal discussion that concluded that checking for time validity on a time point doesn't make sense (std::chrono's time points have no concept of validity, a time point reflecting (0, 0) is valid.) Instead, what matters is if a clock has a time point or not.