Experimental Physics and
Industrial Control System

Dirk Zimoch <[email protected]> · Wed, 16 Feb 2005 16:50:14 +0100

Hi all,
what's the calc record supposed to do if it calculates infinity?
I have tested the following with R3.13.2/vxWorks5.5/ppc and
R3.14.6/Linux/x86
1)
field (CALC, "A/B")
If B is 0.0, the record becomes invalid (as expected) but the value does
not change to 'inf' but stays the previous value. In my opinion this is
wrong behaviour. (On the other hand, the value IS invalid, thus it might
be any number.)
2)
field (C, "inf")
field (CALC, "B?A/B:C")
This works. VAL is 'inf' and the record is invalid.

If one argues in that the record is invalid in 1) because VAL couldn't
be calculated, why is it invalid in 2) even though it is accurately inf?

If one argues that the value IS inf in 2), why is the record invalid and
why is the value wrong in 1)?
3)
field (CALC, "B?A/B:inf")
Does not work. The string inf is not accepted as a constant in the CALC
field. This is inconsistent with the behaviour of DBF_DOUBLE fields.

I know that there has been some discussion about infinity and what it
means in the 'real world', in EPICS and for a driver. But there are
cases, where infinity had a real meaning. E.g. the resistance if a cable
is broken. I miss the possibility to have a valid infinity.
Dirk
--
Dr. Dirk Zimoch
Swiss Light Source
Paul Scherrer Institut
Computing and Controls
phone +41 56 310 5182
fax   +41 56 310 4413

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System