The term access method is normally used to refer to a set of interfaces which permit access to data from files and thus offer a particular logical view of this data. Generally this logical view will differ to a greater or lesser extent from the physical storage in the file (the data repository). Here access methods which enable record-oriented processing of the data in a file are of interest; the logical view is thus restricted. The access methods considered here thus implement the following:
Definition of a record and mapping of this logical view of the data onto a physical storage form (file type).
Definition of the order of the records from the viewpoint of the user or program; this order need not necessarily have anything to do with the physical order of the data in the file.
Interfaces to read and write records in their entirety.
Elementary access methods are a special type of access method. These are distinguished by the fact that the file system (in which they are effective) knows of them and, for example, a file and access method can already be assigned to each other via file types. Such access methods are provided in DMS, albeit not reversibly unique. Other file systems (for example UFS) know only a single elementary access method which generally offers the raw physical view of the data and logically also has no content-oriented file types.
However, there can also be any further access method desired, these generally being implemented using one of the elementary access methods and offering further logical views of the data. These access methods have one problem in common. As the file system has no knowledge of them, it is not possible to tell from a file whether and with which of these access methods it can be successfully processed. Interpretation errors will thus only be recognized during processing, if at all.
In UFS there is no elementary access method which offers record-oriented processing. However, the following access methods, for example, are conceivable:
TEXT - Access method which regards text files as record-structured files with records of variable and unlimited length. The physical record separator would be the new line character, which would be masked out in the logical view.
CISAM - ISAM-type access method for Unix file systems.
In DMS there are several elementary access methods, of which the following are supported directly in JRIO:
SAM - sequential access method
ISAM - indexed-sequential access method
UPAM - block-oriented access method
In DMS, too, there are access methods which are based on one of the elementary access methods and supply a different logical view. A prominent example of these is an access method based on ISAM which is used by various tools (editors, compilers, ...) to render ISAM files usable for normal texts. For this purpose, ISAM files with standard keys are used. In the logical view these keys (which in ISAM are a part of the record) are masked out, and are generated by the access method when records are written.
In the JRIO interfaces you always will encounter the access methods when you must take a decision as to how access to a file is to be implemented.
Currently JRIO only supports the DMS file system, and in this only the access methods SAM, ISAM and UPAM. However, the JRIO architecture will in future permit extension by the addition of further file systems without the user interfaces needing to be modified.