As has already been mentioned, it cannot be assumed that all programs have been prepared for accessing large objects, i.e. that they can address 4-byte block numbers and block counters, although it must be borne in mind that the interfaces available to user applications only relate to accessing and processing files and their metadata. The following description is therefore confined to large files. Large disks will not be handled here, since there are no user interfaces available, as has been mentioned. Program behavior can thus be classified as follows:
Class A: | A program is able to process large files without restrictions. This behavior is defined as capable of LARGE_FILES (LARGE_FILES-capable). |
Class B: | A program has not been explicitly prepared for processing large files and/or their metadata. It is, however, able to reject corresponding access that it regards as illegal, or alternatively, it does not access files or their metadata. This behavior is defined as compatible with LARGE_FILES (LARGE_FILES-compatible). |
Class C: | A program has not been prepared for processing large files and cannot perform a defined rejection of corresponding access. This behavior is defined as incompatible with LARGE_FILES (LARGE_FILES-incompatible). |
For configurations that contain large files, programs that are compatible with or capable of LARGE_FILES must be presupposed. LARGE_FILES compatibility is to be regarded as the norm. Growth over 32 GB will initially be limited to a relatively small number of files. Only programs that access these must be capable of LARGE_FILES.