Experimental Physics and
Industrial Control System

[email protected] (Tim Mooney) · Mon, 7 Nov 94 16:48:12 CST

Deb Kerstein writes:
> 
> Ned writes:
> >Such changes to the CALC record routines would impact things other than the
> >CALC records !  The handy routines "postfix" and "calcPerform" have been used
> >by other record types (at least two that I know of). So to say that such changes
> >could be implemented and the known effects minimized by a script that changes
> >CALC expressions in CALC records is quite dangerous.
> 
> Although I agree with Steve's and Tim's comments and requests for change in
> the CALC record, I have to support Ned in this one.  Having been a veteran
> of the GTA project which was done during a time of rapid EPICS development and 
> change, I can vouch for the painful and unexpectedly far-reaching effects
> of a few "simple" changes.  If you really want these changes right now, the
> suggestion of implementing a new "CCALC" record seems like the best bet--maybe
> Tim and Steve could collaborate on this one.  And, of course, be sure and let 
> everyone know when it's available!

There are two kinds of change under discussion; let's not lump them together.

1) Steve and I suggest changes (I'd call them bug fixes, actually) that can,
if properly implemented, have essentially no effect on existing databases.

2) John Winans suggested doing CALC fields the way they probably would end up
being done if the CALC record were first proposed yesterday and we all got to
vote on it.

Change (2) clearly is not going to fly for the CALC record, and is a
good reason for producing a CCALC record, in my opinion.  (I won't be
working on this one :)

Change (1) is quite similar in implementation to changes already made
in the CALC record's calculation engine to support the TRANSFORM
record.  I doubt anyone even noticed those changes.  (For about a year,
I had to have my own private copy of postfix.c and calcPerform.c to
implement the TRANSFORM record.  I didn't stray too far from the
official version to preserve the possibility of a merge.  If a merge
had been out of the question, I would have made LOTS of changes, and
we'd now have at least two incompatible calc engines running around.
I think such a proliferation would be worse for EPICS than any
conceivable effect of the changes Steve and I suggest.)

Tim

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System