maint: Replace usage of folly in clock.hpp

[jjerphan]

Reference Issues/PRs

Part of #1412.

What does this implement or fix?

folly::chrono::clock_gettime_ns's definition uses the vDSO on Linux (calling the vDSO is not considered a good practice (see the "Description" section of vdso(7)'s man page). Alternatively, it calls clock_gettime.

std::chrono::system_clock::now has nanosecond precision. and also calls clock_gettime with CLOCK_REALTIME. libstdc++ does not use CLOCK_REALTIME_COARSE at all..

Hence, the only drawback is that coarse_nanos_since_epoch who used to return coarse precision timestamp (see CLOCK_REALTIME_COARSE) now returns an accurate timestamp. Must we keep coarse precision or not?

Any other comments?

Checklist

Checklist for code changes...

• Have you updated the relevant docstrings, documentation and copyright notice?
• Is this contribution tested against all ArcticDB's features?
• Do all exceptions introduced raise appropriate error messages?
• Are API changes highlighted in the PR description?
• Is the PR labelled as enhancement or bug so it appears in autogenerated release notes?

[vasil-pashov]

Is is possible to check if the resolution of the clock is the same?

[jjerphan]

Both have nanoseconds resolution seemingly calling clock_gettime.

[jjerphan]

folly::chrono::clock_gettime_ns's definition uses the vDSO on Linux (calling the vDSO is not considered a good practice (see the "Description" section of vdso(7)'s man page). Alternatively, it calls clock_gettime.

std::chrono::system_clock::now has nanosecond precision. and also calls clock_gettime with CLOCK_REALTIME.

The only drawback is that coarse_nanos_since_epoch who used to return coarse precision timestamp (see CLOCK_REALTIME_COARSE) now returns an accurate timestamp. Must we refine it or not?

[jjerphan]

No regression by the ASV benchmarks are observed.