fieldScaling : missing for sampling in ensight format

Functionality to add/problem to solve

With commit f8ffee81 field scaling was extended, however it is still missing for ensight format.

added feature label

assigned to @mark

In retrospect it would have been better to have the scaling at the base surfaceWriter level. Makes the base class even fatter still, but avoid having similar copy/paste (and possibly forget) code everywhere.

Makes sense to add in a fieldLevel dictionary at the same time. Eg,

formatOptions
{
    ensight
    {
        fieldLevel
        {
            p  1e5;  // Absolute to gauge pressure [Pa]
            T  273.15;
        }
    }
}

In contrast to the fieldScale dictionary, I don't think that we should support wildcards here. The fieldLevel has a type associated with it (scalar, vector, tensor), whereas fieldScale is simply a scalar.

I like the idea. Perhaps just for clarification - level value specified is still a scaling factor and not 'offset' or any other operation...

And also, does it apply for fetched data e.g. normal of face (in case of 'raw' format?)

The level value is an offset for a particular field, which will be applied after sampling and before scaling.

Eg,

• Temperature: fieldLevel 273.15 -> Kelvin to Celsius conversion
• Pressure: fieldLevel 1e5 -> absolute to gauge pressure
• Pressure: fieldLevel 1e5, fieldScale 0.01 -> absolute to gauge pressure [Pa], followed by [Pa] -> [mbar]

The area normal of the faces is a geometric quantity, not sampled. So unaffected by fieldLevel or fieldScale, but is still adjusted according to scale (geometric scale) if the writer supports that.

We can keep things general and allow any fieldLevel - eg, U (1 0 0) or we be slightly more restrictive and only permit uniform levels - eg U 1.0 instead of (1 1 1) - maybe could then use a single scalar and allow regex.

mentioned in commit 1a55829e

closed

@Prashant noticed a bug during testing: "access of non-const tmp"

• should be fixed now

reopened

closed

mentioned in commit eb676f08