Experimental Physics and
Industrial Control System

Jeff Hill <[email protected]> · Thu, 21 Nov 2013 01:37:18 -0000

> I take it that the use of reinterpret_cast<>() is meant to bypass the
> restriction that offsetof() won't work on complex types?

The reinterpret cast is used to convert the pointer to an Octet pointer so that we can get an Octet offset. Since we are crossing fields of different data types we need to have that. if you look under the hood in the offsetof macro you will see the same type of implementation.

The offsetof() is implemented as a macro, and therefore less suitable for use in templates. 

This implementation should work just fine with user defined types, and I believe that the regression test should be testing that situation.

> What, if any, trouble could I expect when using "enclosureOf" on a class w/ inheritance?

I use enclosureOf frequently with embedded members that have simple single base class inheritance, and I doubt that there could be any issues with the enclosing class using simple inheritance. I don't currently use multiple inheritance very often but doubt that there would be any issues. If you used virtual base classes, especially double diamond virtual base classes, then proceed with caution, but that goes with the territory DKN.

-- 
https://code.launchpad.net/~johill-lanl/epics-base/epics-base-enclosure-of/+merge/196010
Your team EPICS Core Developers is requested to review the proposed merge of lp:~johill-lanl/epics-base/epics-base-enclosure-of into lp:epics-base.

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System