Evaluate the use of fixed number of decimal places in numeric representations

[Piero Campalani]

While the current policy on numerical representations (both coordinates and data values) is to reach the highest precision possible, it is common practice to deliver data with a fixed number of either decimal places or significant figures.

As far as I understand, a fixed representation is useful when interpreting the data uncertainty, and it can help a software implementation choosing the best data type to use (float, double, etc) to store values associated to a given data component.

Regarding data values, this feature should be embedded in the encoding phase of rasdaman directly (via a parameter which shall be taken from petascopedb::ps_quantity.significant_figures).

Regarding coordinates (both absolute and relative), a deeper inspection should be done. Maybe using fixed 15 decimal places like GDAL does.

[Peter Baumann]

sounds like an interesting idea to deliberate in the future. Tricky numeric effects are luring, so serious investigations will be required for sure.

[Bang Pham Huu]

It has been used BigDecimal in Petascope for a long time, so can be closed here.