1.4 Data Type Labels

VICAR uses system label items to keep track of the machine type the file (both image and binary label) was written on. These label items are summarized below.

• HOST, string : The type of computer used to generate the image. It is used only for documentation; the RTL uses the INTFMT and REALFMT label items to determine the presentation of the pixels in the file. Nevertheless, HOST should be kept consistent with INTFMT and REALFMT.
  See Table 1: Valid VICAR HOST Labels and Machine Types for a list of the currently valid host labels and the machine types they represent. New values for HOST will appear every time the RTL is ported to a new machine, so the table is not necessarily a complete list. Programs should not be surprised by values other than those in Table 1: Valid VICAR HOST Labels and Machine Types appearing in the label.

* Host machine is not officially supported
† No longer officially supported

Table 1: Valid VICAR HOST Labels and Machine Types

• INTFMT, string : The format used to represent integers in the file. INTFMT, REALFMT, and HOST should all match. If you change one change all three.
  The valid values of INTFMT may change as the RTL is ported to new machines. However, the currently valid values are listed. See Table 2: Valid VICAR Integer Formats.

Table 2: Valid VICAR Integer Formats

• REALFMT, string : The format used to represent floating point numbers in the file. INTFMT, REALFMT, and HOST should all match, so if you change one please change all three. Note: Compaq’s Alpha supports three floating point formats: D, G, and IEEE. D is used for compatibility with VAX, though G is the default. IEEE is used for Alphas running Unix, but not for VMS.
  The valid values of REALFMT may change as the RTL is ported to new machines. However, the currently valid values are listed. See Table 3: Valid VICAR Real Number Formats.

Table 3: Valid VICAR Real Number Formats

• BHOST, string :The type of computer used to generate the binary label. It is used only for documentation; the RTL uses the BINTFMT and INTFMT label items to determine the representation of the binary labels in the file. Nevertheless, it should be kept consistent with BINTFMT and BREALFMT.
  The valid values of INTFMT are exactly the same as for the HOST label item.
• BINTFMT, string : The format used to represent integers in the binary label. BINTFMT, INTFMT, and INTFMT should all match, so if you change one please change all three.
  The valid values of BINTFMT are exactly the same as for the INTFMT item above.
• BREALFMT, string : The format used to represent floating point numbers in the binary label. BINTFMT, INTFMT, and INTFMT should all match, so if you change one please change all three.
  The valid values of BREALFMT are exactly the same as for the REALFMT label item above.
• BLTYPE, string: The type of the binary label. This is not type in the sense of data type, but a string identifying the kind of binary label in the file. The RTL does not do any interpretation or checking of BLTYPE. It is intended for documentation, so people looking at the image will know what kind of binary label is present. It may also be used by application programs to make sure they can process the given type of binary label, or to make sure it is processed correctly.
  Valid BLTYPE values are maintained in a name registry, so that all possible kinds of binary labels can be documented in one place. Only names that are registered should be used in BLTYPE. See 1.7.2 Programming and Binary Labels for more details.
• FORMAT, string : The data type of the pixels in the file:
  • BYTE: Single byte, unsigned integer pixel type.
  • HALF: Signed short integer pixel type (often 2 bytes).
  • FULL: Standard size signed integer pixel type (often 4 bytes).
  • REAL: Single precision floating-point pixel type.
  • DOUB: Double precision floating-point pixel type.
  • COMP: Two single precision floating-point numbers representing a complex pixel type, in the order (real, imaginary).

The label item FORMAT is used to specify the data type. It is unfortunate that the name FORMAT is used for the data type, rather than for the host representation (which would better be called format), but the names cannot be changed for historical reasons.