Methods for compiling formulas stored in an electronic spreadsheet system

In an electronic spreadsheet system having an IEEE-compliant floating-point unit, said floating-point unit having processor registers comprising sign, exponent, and mantissa bit fields, each field being arranged from a most-significant bit to a least-significant bit, a method for storing in the floating-point unit spreadsheet values representing Not-a-Number (NAN) values, the method comprising:

• (a) reserving a plurality of bits in the mantissa field for encoding a cell identifier for indicating a particular cell of the spreadsheet where the spreadsheet value originated, so that said cell identifier propagates to dependent cells;
• (b) for a spreadsheet cell giving rise to an NaN value, representing the NaN value in a register of the floating-point unit by:
• (i) setting all bits in the exponent field of the register to high,
• (ii) setting at least one bit of the mantissa field in the register to high, and
• (iii) setting said plurality of bits in the mantissa field of the register to encode a cell identifier for the particular cell of the spreadsheet where the NaN spreadsheet value originated;
• whereby, when the spreadsheet cell representing the NaN is combined with a non-NaN value, the result is the NaN value with the cell identifier, so encoded, thereby propagating the cell identifier giving rise to the NaN value to a dependent cell.