Claris CAD stores object sizes and locations in a special data format that is usually equal to or better than 32 Bit floating point format. This compares favorably with other Macintosh drawing and CAD applications, which generally use integer (16 bit - roughly half the accuracy of Claris CAD) or long integer (32 bit - roughly the same precision).
VersaCAD and AutoCAD have double precision floating point accuracy (roughly twice as good as Claris CAD) - but because of this require:
- more memory for a given drawing
- floating point coprocessor for acceptable performance (redraw, rotate, etc.)
- double precision is necessary in only a few, very demanding applications
- if the drawing is to be used to drive CAM process
- if the drawing requires an exceptional level of detail
A great majority of users do not need this level of accuracy.
Claris CAD uses SANE (Standard Apple Numeric Environment) to access a math co-processor (if one is available).
Claris CAD uses a fixed format to store numbers (as opposed to a floating point format). Claris CAD uses one integer for the left side of the decimal and one integer for the right side of the decimal. An integer on the Macintosh is a number between -32767 and 32768.
So, the largest number that Claris CAD can store in its database of objects is 32,768. The smallest number it can store to the right side of the decimal place is 1/32767 = 0.0000305185094.
A one-inch line on the screen is 72 pixels. So, how close can Claris CAD come to an inch without actually being an inch?
Answer: 0.0000004238681872 or 1/(32767 * 72)
Using this same type of logic, we can calculate how accurately Claris CAD can approximate a 1 inch line when represented at various scales.
A one inch line on the screen represented as:
1 foot ..... is accurate to ..... 40 millionths = 0.00000508641 inches
100 feet ..... is accurate to ..... 5 ten-thousandths = 0.000508641 inches
1 mile ..... is accurate to ..... 1/32 = 0.026856288 inches