RE: const-constraints on struct rset method arguments

[Ron Sluiter <sluiter@aps.anl.gov>]

Hello Ben,

I won't comment on the wider topic of your proposed changes, but with regard to the motor record specifically, I agree with the changes I see in your motor module fork; https://github.com/bfrk/motor.

I suggest that these changes be added to motorRecord.cc

537c537
< static long init_record(dbCommon* arg, int pass)
---
> static long init_record(motorRecord *pmr, int pass)
539d538
<     motorRecord *pmr = (motorRecord *) arg;
1188c1187
< static long process(dbCommon *arg)
---
> static long process(motorRecord *pmr)
1190d1188
<     motorRecord *pmr = (motorRecord *) arg;

since this workaround is no longer needed, and without the above changes, init_record() and process() become unresolved external references.

I have successfully built your motor module fork against your EPICSbase 3.16 branch (after making the above changes to motorRecord.cc). I plan on running an IOC and actually driving a motor. I'll let you know my results when available.

The most controversial issue is that the use of the conditional compilation macro EPICS_VERSION_INT will make this version of the motor record incompatible with EPICSbase versions < 3.15. It should not be too difficult to change conditional compilation macro usage so that EPICSbase 3.14, 3.15 and 3.16 are supported.

Ron