epics/epics-base
0
0
Fork 0
Code Issues Pull Requests Projects Wiki Activity

Change the epicsTimerTest delayVerify failure condition

Browse Source 
This test verifies timer delays between 1.0 and 3.4 seconds.
A test failure now means the measured delay was more than
0.25 seconds different than the request, instead of being
5% of the request. This should now pass on GHA macOS.

On the 7.0 branch it should call testImpreciseTiming() and
use a smaller absolute delay threshold for better targets.
This commit is contained in:
Andrew Johnson
2021-09-17 11:16:00 -05:00
parent 9b69e63a69
commit 5cddcea829
1 changed files with 8 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
src/libCom/test/epicsTimerTest.cpp
View File

@@ -108,13 +108,14 @@ inline double delayVerify::delay () const
double delayVerify::checkError () const
{
const double messageThresh = 5.0; // percent
double actualDelay = this->expireStamp - this->beginStamp;
double measuredError = actualDelay - this->expectedDelay;
double percentError = 100.0 * fabs ( measuredError ) / this->expectedDelay;
testOk ( percentError < messageThresh, "%f < %f, delay = %f s, error = %f s (%.1f %%)",
percentError, messageThresh,
this->expectedDelay, measuredError, percentError );
const double minError = 0.25;
double measuredDelay = this->expireStamp - this->beginStamp;
double measuredError = measuredDelay - this->expectedDelay;
double absoluteError = fabs(measuredError);
double percentError = 100.0 * measuredError / this->expectedDelay;
testOk(absoluteError < minError,
"Delay %.3f s, error = %+.6f ms (%+.3f %%)",
this->expectedDelay, measuredError * 1000, percentError);
return measuredError;
}