#11 'improveprecision' over-used

[Paul Hodgkinson]

'improveprecision' is used to tweak every floating point number read from the CIF. This leads to subtle problems when a value happens accidentally to trigger adjustment e.g. 0.3333 to exactly 1/3 etc.

Keith Refson comments: "I would suggest that there is a case for confining the rounding used specifically for fractional co-ordinates. I can't think of any case where an irrational fraction is needed to describe a crystallographic special position."

Ideally 'improveprecision' would only be triggered when checking specifically for atoms at special positions - detected when they "land on top of each other"

A quick work round would be to have 'coordinate-tolerance' as an input parameter or, as you have suggsted, a "--strict” flag which would disable heuristics in the cell generation.