Experimental Physics and
Industrial Control System

Chip Watson <[email protected]> · Thu, 17 Jul 1997 08:45:01 -0400

Jeff said:
>I propose that we add dbr_ulong_t and dbr_ushort_t to the CA client API.
>Requests for these types would of course be refused when they are
>issued to old servers.

I agree this is the best long term solution, however, I wish to also patch my
3.12 system (large systems have a lot of inertia against upgrades). 

>Converting unsigned32 to signed32 is of course not a good
>idea and would result in a sloppy API - which would almost certainly 
influence
>clients in a way that could cause bugs.

Is anyone aware of a use of ULONG to hold a number larger than 2^31 which would
fail if the value were contained in a 32 bit int?

I would think this would only fail in the case of true positive numbers (not 
masks)
greater than 2^31, and these sorts of numbers do not occur in control systems,
IMHO (the exception being the Q value of our RF cavities, and we've had plenty 
of
problems dealing with those :). Control and status registers can be int's with 
no ill
effects, since masking and shifting operations still would work. 

The only possible ill effect I can see (as Jeff alluded to) is applications 
taking
advantage (unwisely) of the fact that a double may be a ULONG and using that to
infer record type.

Chip

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System