Experimental Physics and
Industrial Control System

Michael Davidsaver <[email protected]> · Wed, 9 Mar 2016 14:43:15 -0500

I think the issue is that the test is not waiting for the initial
monitor update to be delivered.  It later does wait for an update to be
delivered, which is assumed to be the result of a dbPutField().

I'm not quite sure how to handle this.  With the present API, by the
time I can hook into caLink::monitor I don't know if the initial update
has been delivered or not.  Maybe add another counter to caLink (eg.
nUpdate)?

I'd like to sync. with the "CAC-event" thread's queue, but the structure
of dbContext.cpp doesn't make it easy to get the dbEventCtx pointer.  I
suppose the way around this would be to keep a global list of all
dbEventCtx and sync. with them all.  Seems a bit heavy handed, but
acceptable for unittest code.

On 03/04/2016 06:52 PM, Michael Davidsaver wrote:
> FYI I have been able to reproduce this a couple of times.
>
> On 03/04/2016 12:13 PM, Michael Davidsaver wrote:
>> On 03/04/2016 11:56 AM, APS Jenkins wrote:
>>> src/ioc/db/test/O.linux-x86/dbCaLinkTest.tap ............. 
>>> not ok 14 -  == hello
>>> Failed 1/99 subtests 
>> To put this in context.  This is effectively failing in the same place
>> as previously.  The same test is repeated with native and string CA links.
>>
>>> src/ioc/db/test/O.linux-x86/dbCaLinkTest.tap ............. 
>>> not ok  7 - temp (0) == 42 (42)
>>> Failed 1/99 subtests 
>> Zero and empty string are the previous values of the link, so my attempt
>> to wait for for an update is probably racy.
>>

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System