Re: Problems with dbLinkIsConstant

[Torsten Bögershausen via Core-talk <core-talk at aps.anl.gov>]

On 30/06/20 17:55, Johnson, Andrew N. wrote:
> Hi Torsten,
>
> On Jun 30, 2020, at 3:09 AM, Torsten Bögershausen via Core-talk 
> <core-talk at aps.anl.gov <mailto:core-talk at aps.anl.gov>> wrote:
> > I have a problem with dbLinkIsConstant(), it seems.
> >
> > If I run the motor module with this code:
> >
> > #if LT_EPICSBASE(7,0,0,0)
> > #define dbLinkIsConstant(lnk) ((lnk)->type == CONSTANT)
> > #else
> > #include "dbLink.h"
> > #endif
> >
> > And later
> >
> >    if (dbLinkIsConstant(&pmr->dol))
> >    {
> >        pmr->udf = FALSE;
> >        Debug(pmr,3, "init_record %s\n", "dbLinkIsConstant=true set 
> > UDF=FALSE");
> >
> >        recGblInitConstantLink(&pmr->dol, DBF_DOUBLE, &pmr->val);
> >    }
> >
> > I get this log message:
> > [motorRecord.cc:906 <http://motorrecord.cc:906>  IOC:m1] init_record 
> > dbLinkIsConstant=true set UDF=FALSE
> >
> > However, the .DOL field is empty.
> > Is this the expected behavior ?
>
> Yes, an empty or unpopulated link field has always been represented with 
> type==CONSTANT which is a macro with value 0 so this is also the default 
> if the link field is not set at all. In 3.16.1 I introduced extensible 
> links and the dbLinkIsConstant() routine to Base because extensible 
> links can also provide constant values, and they don’t have type==CONSTANT.
>
> Note that recGblInitConstantLink() has always returned TRUE if the link 
> is a constant and provided a value (else FALSE), so there is no actual 
> need for the conditional around it – that was merely an optimization for 
> earlier versions of Base but is no longer helpful. My suggested 
> replacement for the code above is to not call dbLinkIsConstant() at all, 
> just do this instead:
>
> >    if (recGblInitConstantLink(&pmr->dol, DBF_DOUBLE, &pmr->val))
> >    {
> >        pmr->udf = FALSE;
> >        Debug(pmr,3, "init_record %s\n", "recGblInitConstantLink set 
> > UDF=FALSE");
> >    }
>
> That should be backwards compatible with older Base versions, and I 
> believe all of the record types  have now had that modification done 
> already. This was documented in the Release Notes for version 3.16.1 
> <https://epics.anl.gov/base/R3-16/1-docs/RELEASE_NOTES.html>, look for 
> the section titled “Support Routine Modifications for Extensible Link 
> Types”.
>
> HTH,
>
> - Andrew
>
> --
> Complexity comes for free, simplicity you have to work for.

Thanks Andrew for the good explanation
I (now) realized that I had mis-used the .UDF field to keep track
if there was ever a communication to the controller, or if it is still
"offline". The .UDF seemed good for this purpose, but it isn't.