Fix date_time_with_ms millisecond formatting

[ahuazuipiaoliang]

Summary

Fix {date_time_with_ms} formatting so the subsecond field is computed as milliseconds instead of taking the first three digits from an unpadded nanosecond string.

Details

The previous implementation converted the nanosecond remainder to a decimal string without zero-padding and then used the first three characters. For values with leading zeros, such as 085860000 ns, this produced .858 instead of .085.

This PR computes milliseconds directly with nanoseconds / 1000000 and formats the value with %03.

Testing

Added coverage for a timestamp whose nanosecond remainder has leading zeros.

Closes #569

[fujitatomoya]

good eye! thanks for the PR.
this is real bug. any timestamp whose subsecond portion is under 100 ms produced wrong output, which is a substantial fraction of timestamps in practice.

[fujitatomoya]

Pulls: #570
Gist: https://gist.githubusercontent.com/fujitatomoya/e35582465a12e4a61963c3387a965caf/raw/bea084ce649b28a197429b2f6d526ff131e4b133/ros2.repos
BUILD args: --packages-above-and-dependencies rcutils
TEST args: --packages-above rcutils
ROS Distro: rolling
Job: ci_launcher
ci_launcher ran: https://ci.ros2.org/job/ci_launcher/19389

• Linux
• Linux-aarch64
• Linux-rhel
• Windows

[mergify]

Tick the box to add this pull request to the merge queue (same as @mergifyio queue).

• Queue this pull request

[fujitatomoya]

@ahuazuipiaoliang can you rebase this? and then i will start the CI before merge.

[ahuazuipiaoliang]

@ahuazuipiaoliang can you rebase this? and then i will start the CI before merge.

@fujitatomoya I've already rebased.

[fujitatomoya]

Pulls: #570
Gist: https://gist.githubusercontent.com/fujitatomoya/7493b8086e3cfbd1f657174a2eb9f275/raw/2a72bda5f06266dd105395f2ee57d80a55606411/ros2.repos
BUILD args: --packages-above-and-dependencies rcutils
TEST args: --packages-above rcutils
ROS Distro: rolling
Job: ci_launcher
ci_launcher ran: https://ci.ros2.org/job/ci_launcher/20002

• Linux
• Linux-aarch64
• Linux-rhel
• Windows

[fujitatomoya]

failure tests are unrelated.