Just for the record and to inform for those who do not follow the ooRexx trackers: after creating a proper bug report (*[bugs:#1554] <https://sourceforge.net/p/oorexx/bugs/1554/>*) in the past days patches got developed and tested that address and solve this problem.
---rony On 18.09.2018 17:04, Rony G. Flatscher wrote: > > OK, I have spent quite some time on the Linux box. Created debug versions for > ooRexx and > BSF4ooRexx and tried to use gdb to get at the data when the hanging or > exceptions occur. > > Unfortunately, gdb reports SIGSEGV (segmentation faults) in the JVM (using > Oracle's Java 1.8) in > both scenarious: loading a Rexx script via rexx and then loading Java to run > the Rexx script, and > loading a Java program that will load Rexx to run the Rexx script. :( > > Running the (BSF4)ooRexx programs outside gdb works, no matter whether rexx > loads Java or Java > loads rexx to execute the scripts. > > --- > > Some observations: > > * the concurrent updating of awt/swing and JavaFX GUIs hangs as on Wndows. > Sometimes the "Stop" > button (causing an asynchroneous call back on the awt/JavaFX GUI thread) > works, i.e. on the > GUI thread the ooRexx control variable gets reset that controls the Rexx > thread that > dispatches runLater messages to the Java GUI thread on a separate Rexx > thread, which is > probably interesting, here a sketch of the execution (RT1=primodal Rexx > thread, JGT2=Java GUI > thread, RTn=Rexx thread dispatching messages to Java GUI using > invokeLater): > RT1: | JGT2: > | RTn: dispatcher thread > | | > sets up the GUI, waits until | Java GUI thread: sends Rexx objects > | "Start" button press causes new > GUI closes using a Rexx control | messages if "Start" or "Stop" buttons > | Rexx thread (using reply) which > variable that gets set when | get pressed, or the window gets > | dispatches GUI messages until > user closes window | closed; sets the Rexx control vars > | "Stop" gets pressed which will > | > | cause the Rexx control var to be reset > * the BSF4ooRexx testunit excercising multithreaded executions that hangs > and causes an > exception on Windows, causes hangs on Linux as well. So far I have not > been able to cause an > exception. > > Also, a few times the test script ran successfully! > > (This test script creates 23 Java threads that each create an ooRexx > interpreter instance and > run a Rexx script that each will cause 29 Rexx threads to spin off which > will interact with a > Rexx array object directly and indirectly via a Java object proxying a > Rexx object. This test > as currently configured will create a total of 712 threads in its run.) > > Short of success in creating the necessary information on Linux, I will have > to resort to Windows, > but do not know whether I will be able to do that today. > > ---rony >
_______________________________________________ Oorexx-devel mailing list Oorexx-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/oorexx-devel
