[Cfp-interest 1380] normalized?

[Jim Thomas]

Action item:
>     Jim: Add in the normalized discussion from Fred (CFP 1341, CFP 1342) to the agenda for next meeting.
> 
For discussion at the August 21 CFP meeting …

The term normalized suggests a representation. However, C does not specify representations for floating types, but only their values. The representations in the model are rightly used to describe the values in a type, but must not be mistaken for the representations in the type. The following is a possible change in 5.2.4.2.2#4 to replace “normalized” with “normal”. The term normal here has the same meaning as in IEC 60559.

5.2.4.2.2

….

[4] A normal floating-point number is a number defined by the model with f1 > 0. Floating types shall be able to represent zero (all fk == 0) and all normal floating-point numbers. In addition, floating types may contain other kinds of floating-point numbers,22) such as negative zero and subnormal floating-point numbers (model numbers x!= 0 with e = emin, f1 = 0) and values that are not floating-point numbers, such as infinities and NaNs.

... 

NOTE	Each non-zero numbers defined by the model is equal to a normal or a subnormal model number, obtained by shifting the digits d1d2 … dn to the left to remove any leading zero and decrementing the exponent e accordingly, until either d1 is nonzero or e is emin.

 

Replace all instances of “normalized”, related to floating point, with “normal”, except in the logb spec, which needs a different change.


- Jim Thomas

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.oakapple.net/pipermail/cfp-interest/attachments/20190820/f7d9f204/attachment-0001.html