Use float instead of Date for protocol types for higher precision.

[maciejwalkowiak]

📜 Description

Use float instead of Date for protocol types for higher precision.

To retrieve more precise timestamps, we are using System#nanoTime as it's the only method available both on Android and JVM.

When Span is created with given start timestamp (for example with Android app start time), we are not able to measure high precision timestamp, we fall back then to regular timestamp with milliseconds precision.

How it looks in UI:

💡 Motivation and Context

Fixes #1689

💚 How did you test it?

Unit & integration tests.

📝 Checklist

• I reviewed the submitted code
• I added tests to verify the changes
• I updated the docs if needed
• No breaking changes

[codecov-commenter]

Codecov Report

Merging #1737 (035274a) into 6.x.x (23913ee) will increase coverage by 0.05%.
The diff coverage is 89.39%.

@@             Coverage Diff              @@
##              6.x.x    #1737      +/-   ##
============================================
+ Coverage     80.93%   80.98%   +0.05%     
- Complexity     2822     2835      +13     
============================================
  Files           205      205              
  Lines         10360    10405      +45     
  Branches       1370     1376       +6     
============================================
+ Hits           8385     8427      +42     
  Misses         1493     1493              
- Partials        482      485       +3     

Impacted Files	Coverage Δ
sentry/src/main/java/io/sentry/SentryTracer.java	88.88% <66.66%> (+0.07%)	⬆️
...ain/java/io/sentry/protocol/SentryTransaction.java	89.60% <80.00%> (-0.50%)	⬇️
...y/src/main/java/io/sentry/protocol/SentrySpan.java	83.70% <88.23%> (-0.43%)	⬇️
sentry/src/main/java/io/sentry/DateUtils.java	100.00% <100.00%> (ø)
sentry/src/main/java/io/sentry/Span.java	96.42% <100.00%> (+2.48%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 23913ee...035274a. Read the comment docs.

[bruno-garcia]

Looks so much better in the UI 💯

[marandaneto]

btw: System#nanoTime

This clock is guaranteed to be monotonic, and is suitable for interval timing when the interval does not span device sleep

that means when the device is in sleep mode, the system does not increase the timer, is it possible that it'd cause off-by-1 sort of problems?

https://developer.android.com/reference/android/os/SystemClock search for nanoTime

[maciejwalkowiak]

that means when the device is in sleep mode, the system does not increase the timer, is it possible that it'd cause off-by-1 sort of problems?

Are our transactions meant to span over the system sleep? If so, we should at least check if the endTime - startTime isn't a negative number.

[marandaneto]

that means when the device is in sleep mode, the system does not increase the timer, is it possible that it'd cause off-by-1 sort of problems?

Are our transactions meant to span over the system sleep? If so, we should at least check if the endTime - startTime isn't a negative number.

good question and I don't know the answer right now, have to check if it makes sense or not

[marandaneto]

should we target the v6 branch then?

[maciejwalkowiak]

Yes but this PR is anyway on hold until we have JSON serialization merged.

[bruno-garcia]

JSON PR is getting to its final form so might be time to point this to that branch and merge?

[maciejwalkowiak]

Just to avoid double work - once JSON PR is merged to 6.x.x, this branch will be rebased and pointed against 6.x.x.

[bruno-garcia]

Gson PR merged into branch 6.x.x. Lets get this one merged in there too? We'll release an alpha soon.

[maciejwalkowiak]

6.x.x is now up to date with main, this PR is now pointing to 6.x.x. Ready for the final review.

[marandaneto]

@romtsn would you like to have a look?

[romtsn]

@marandaneto checking 👍

[romtsn]

LGTM