A resolution attribute's value consists of two items, the cross feed direction resolution and the feed direction resolution. A resolution attribute may be constructed by supplying the two values and indicating the units in which the values are measured. Methods are provided to return a resolution attribute's values, indicating the units in which the values are to be returned. The two most common resolution units are dots per inch (dpi) and dots per centimeter (dpcm), and exported constants DPI and DPCM are provided for indicating those units.

Once constructed, a resolution attribute's value is immutable.

Design

A resolution attribute's cross feed direction resolution and feed direction resolution values are stored internally using units of dots per 100 inches (dphi). Storing the values in dphi rather than, say, metric units allows precise integer arithmetic conversions between dpi and dphi and between dpcm and dphi: 1 dpi = 100 dphi, 1 dpcm = 254 dphi. Thus, the values can be stored into and retrieved back from a resolution attribute in either units with no loss of precision. This would not be guaranteed if a floating point representation were used. However, roundoff error will in general occur if a resolution attribute's values are created in one units and retrieved in different units; for example, 600 dpi will be rounded to 236 dpcm, whereas the true value (to five figures) is 236.22 dpcm.

Storing the values internally in common units of dphi lets two resolution attributes be compared without regard to the units in which they were created; for example, 300 dpcm will compare equal to 762 dpi, as they both are stored as 76200 dphi. In particular, a lookup service can match resolution attributes based on equality of their serialized representations regardless of the units in which they were created. Again, using integers for internal storage allows precise equality comparisons to be done, which would not be guaranteed if a floating point representation were used.

The exported constant DPI is actually the conversion factor by which to multiply a value in dpi to get the value in dphi. Likewise, the exported constant DPCM is the conversion factor by which to multiply a value in dpcm to get the value in dphi. A client can specify a resolution value in units other than dpi or dpcm by supplying its own conversion factor. However, since the internal units of dphi was chosen with supporting only the external units of dpi and dpcm in mind, there is no guarantee that the conversion factor for the client's units will be an exact integer. If the conversion factor isn't an exact integer, resolution values in the client's units won't be stored precisely.