There are numerous potential was to observe the handling of numeric values read from and written to time series data files. During file type conversion they are subject to type casts and such can suffer from round-off errors, numeric overflow or numeric underflow. For users it might by helpful to produce diagnostic output indicating the danger of loss of information. Below I sketch several possibly helpful approaches. They will only be implemented upon a users request.

Remarks: Some type casts take place outside this library and cannot be trapped by the convert function template. This is the case for SFF read functions like datrw::isffstream::dseries(), which use the external InputWaveform module from libsffxx.

Todo:: Think of using file magic to identify data file formats

Todo:: <typeinfo> provides extended information on intrinsic numeric types

Todo:: <limits> provides numeric_limits template structs for numeric types. They can be used to study the potential effect of type casts.

Todo:: Distinguish between internal represenation of samples in intrinsic data types and representation of data in the data file format. For some formats the exact representation can only be deduced for the specific file (like PDAS format or the planned binary format or seife format).

See also: Format, File, data, and stream properties

Date: 20.12.2010