#360 plot cmd blocking problem demoed by test001
Occasionally Test001 will fail because of a bug with blocking in the writeToPng. It looks like at the end of the script, there are two plot commands before the write command, and my guess is that when the first plot is finished it clears the block and then a race occurs between the writeToPng and the second plot command. The failure rate is maybe once every 20 tests.
My guess is this bug does not affect the interactive version of the software, but it could mess up the server and certainly running scripts.
This smaller script probably demos the probllem:
reset();
plot( "file:/home/jbf/ct/hudson/data.backup/xls/2008-lion and tiger summary.xls?sheet=Samantha+tiger+lp+lofreq&firstRow=53&column=Complex_Modulus&depend0=Frequency" );
plot( "file:/home/jbf/ct/hudson/data.backup/qds/hist.qds" );
writeToPng( "test001_003.png" );
Discussion
I write a program to run Test001 100 times and it would fail about once per invocation. In the failure I saw, the plot showed the first plot command instead of the second.
Putting a loop in waitUntilIdle definitely improves things, but there is still an issue. We consistently have with the short demo cases (1 in 150) where the first dataset is plotted on the second dataset's ranges, as if the autoranging were done but there is no busy flag set to block until the new data is used.