| 摘要 |
<p>It is made possible during the processing of an IEEE type binary floating point number irrespective of precision to judge whether it is a non-numeric character. An IEEE type binary floating point number having a sign portion, an exponent portion and a mantissa portion, is defined as a non-numeric character under the conditions that the bits of the exponent portion are all '1' and that not all the bits of the mantissa portion are '0'. When the character has the maximum precision, the data thereon are used as they are as interior expressed type data. When the character does not have the maximum precision, on the other hand, the sign portion is used as it is; the exponent portion is expanded to the maximum precision by filling the vacant bit(s) with '1'; and the mantissa portion is expanded to the maximum precision by filling the vacant bit(s) with '0'. The results of this conversion are used as interior expressed type data. The interior expressed type data, all as data of maximum precision, are subjected collectively to non-numeric character discrimination.</p> |
