Allen Farris
ST ScI
The purpose of this set of C++ classes is to give programmers the necessary tools to manipulate FITS files for a broad range of applications. Its primary purpose is to import and export data and to adhere strictly to the FITS standards. On input, all errors are flagged but corrections are attempted. On output, detailed error checking is used to prevent writing an invalid FITS file. These classes support two processing models, sequential and random, integrated into one coherent framework. The particular processing model is selected whenever the file is opened. The sequential processing model is useful for reading or writing from tapes, pipes, or standard I/O. Within the sequential model no assumptions are made that require random access media. The random processing model is restricted to devices that support lseek. With the random model, one may access individual header-data-units or any portion of the header or data. The random model allows the user considerable flexibility in manipulating random access FITS files, including the ability to append data to the end of a file and update the size information in the FITS header at a later time. All aspects of the FITS standard are supported, including image extensions, random groups structure, general extensions, ASCII and binary tables. The variable length array facility and the multidimensional array convention, used in conjunction with binary tables, are also supported.