On Apr 10, 2020, at 2:27 PM, Mark Rivers <
rivers at cars.uchicago.edu> wrote:
Sorry I missed what you were suggesting at the end of your message.
No problem, thank you for pointing out that the IOC’s carriage-return and line-feed can fully explain my “missing” characters. I could have spent a good few hours trying to find the cause of them!
I will now find a suitable place to set the main thread’s okToBlock flag, and this fix will go into the next releases from the 3.15 and 7.0 branches of Base.
Users of previous Base releases can fix the problem (which I believe should only ever occur on IOCs running on VxWorks or RTEMS with GeSys, never on any of the targets such as Linux or Windows that use the iocsh) by adding the call like Mark did, just
after the IOC’s munch file or core binary has been loaded:
I added the following at the top of my startup script:
epicsThreadSetOkToBlock(1)
value = 0 = 0x0
Indeed that fixes the problem!