Commit Graph

5 Commits

Author SHA1 Message Date
Valentin Shergin 26d3bcae72 Fabric: Increasing threshold for MountingTelemetryTest
Summary:
We neede that to make the test less flaky.

Changelog: [Internal] Fabric-specific internal change.

Reviewed By: mdvacca

Differential Revision: D21261487

fbshipit-source-id: afec45e03be10ba071a5ab808ce13f95b5f475e3
2020-04-27 12:57:09 -07:00
Valentin Shergin 5dc6eded1b Fabric: Using proper clock in MountingTelemetryTest
Summary:
Apparently, `std::this_thread::sleep_for` uses a different clock to measure time which causes ofter misalignment with the clock which Telemery uses which makes the test flaky. Using the same clock should fix it.

Changelog: [Internal] Fabric-specific internal change.

Reviewed By: JoshuaGross

Differential Revision: D21116058

fbshipit-source-id: 52dde2e325776d365431a2a957dcc12dfe53f890
2020-04-19 22:55:09 -07:00
Joshua Gross 24b79c1182 Core telemetry tests: update so they pass on my machine
Summary:
`std::this_thread::sleep_for` is not really precise and will attempt to sleep for "at least" that much time, but may sleep much longer depending on what CPUs are doing and scheduling policies.

To get this to pass on my machine, I had to substantially increase the thresholds.

Changelog: [Internal]

Reviewed By: shergin

Differential Revision: D20689571

fbshipit-source-id: f159420d24a95da2b5d95d49ed7882e783291e98
2020-03-27 19:07:53 -07:00
Valentin Shergin eadc2ac493 Fabric: Improving precision and type-safety of Telemetry
Summary:
Collecting Telemetry is a crucial part of building a performant UI framework; we do that but we need to improve it to make the data more reliable, actionable and trustful.

Now we collect time points as the number of milliseconds from the start of the CLOCK_MONOTONIC epoch. That's fine but it also has problems:
Sometimes a millisecond is an eternity. We have only 16 (or fewer) of them on each frame. What if some operation takes 1ms (according to telemetry) but we have to run it a dozen times? Does it mean that it's 12 ms in total? So, we lack precision.
This is not type-safe. Do you know how many milliseconds in a microsecond? I don't. We multiply that on magical constants hoping that we copied that from some other place right.
The current implementation is not cross-platform. We have ifdefs for iOS and Android and Unix and Windows (which is now implemented).

So, this diff replaces that with using `std::chrono` which is part of the standard library that designed to fix all those concerns. We also define our type-aliases on top of that to express our concrete constrains:
We use `std::chrono::steady_clock` as the base clock which is according to the standard using `clock_gettime(CLOCK_MONOTONIC, ... )` if available. So, it's fast and compatible (the same under the hood) with Android infra.
We use nanoseconds when we store time durations (TelemetryDuration type).
 Changelog: [Internal] Fabric-specific internal change.

Reviewed By: JoshuaGross, mdvacca

Differential Revision: D19184569

fbshipit-source-id: 7a44688f4bb3bfc6e3009874f0075c531c8569a1
2020-01-27 13:26:49 -08:00
Valentin Shergin b3a8706c2f Fabric: Basic tests for MountingTelemetry
Summary:
Some tests which will be useful for the next diff.

Changelog: [Internal] Fabric-specific internal change.

Reviewed By: mdvacca

Differential Revision: D19482255

fbshipit-source-id: 3384662477f750620a37acc4a277a55c2dbd8d0e
2020-01-27 13:26:49 -08:00