Workaround for bug (?) in ProcessorScheduler jv
authorJan Vrany <jan.vrany@fit.cvut.cz>
Sat, 11 Mar 2017 21:41:58 +0000
branchjv
changeset 327113031dfa759
parent 326 7a039308efa5
child 335 a7f6c7bc62a8
Workaround for bug (?) in ProcessorScheduler

"Due to some bug (?) in ProcessorScheduler, we need to set
exitWhenNoMoreUserProcesses to prevent process dispatcher
to exit prematurely.

Sometimes, ProcessorScheduler>>checkForEndOfSispatch thought
there's no runnable user proless even though test runner did not
finish (was merely waiting for external command to finish).
Happened sometimes. not always.
reports/report-runner.st
     1.1 --- a/reports/report-runner.st	Thu Feb 23 13:36:24 2017 +0000
     1.2 +++ b/reports/report-runner.st	Sat Mar 11 21:41:58 2017 +0000
     1.3 @@ -38,4 +38,9 @@
     1.4      ].
     1.5  ].
     1.6  
     1.7 -(Smalltalk at:#'Builder::ReportRunner') start.
     1.8 +"/ Due to some bug in ProcessorScheduler, we need to set 
     1.9 +"/ exitWhenNoMoreUserProcesses to prevent process dispatcher
    1.10 +"/ to exit prematurely. 
    1.11 +Processor exitWhenNoMoreUserProcesses: false.
    1.12 +
    1.13 +(Smalltalk at:#'Builder::ReportRunner') start.
    1.14 \ No newline at end of file