Persistence of Data
Independent Documents, page 5
One of the most common methods used by applications to persist data is to store the required information in document files on disk. There are a number of standard formats for disk files, for example: .BMP files (windows bitmap format) is a standard file format for bitmap images. RTF (rich text format) is a standard formatted text document format. In many cases, applications define their own private file formats - in fact, many standard formats originated in that manner. Whether a private or public file format is appropriate for a given application is one of the design choices that must be made by the developer.
With document files, the contents of the file are the responsibility of the application, as are all of the file operations. File operations are implemented using either API commands or a wide variety of Visual Basic commands that allow you to work with lines, records or binary data.
Are document files appropriate for saving application configuration information? Yes - after all, a private initialization file is, essentially, a document on disk. But document files are rarely used in this way. If the configuration information is simple, it is easier to use an initialization file. If the configuration information is complex or requires the persistence of data that cannot be easily represented as text, it is often easier to use the system registry.
Document files are easy to manage on disk. They can be copied, moved, sent via Email, and managed using standard file management software.
The task of reading and writing document files can range from very simple to extremely complex, depending on the type of document. A simple text document can be created by simply printing lines of text into the file, each one separated by a carriage-return line-feed combination. If, however, you wish to insert a line into the middle of the file, or remove a line from the file, you will generally need to write the entire file from scratch - there is no easy way to shift information in a disk file.
What if you wish to save many different types of information in the document? For example: a word processing file that contains both text and pictures? In that case you need to define a more complex file format that keeps track of where individual objects are located within the disk file. You might need to actually define an internal directory within the disk file that lists the various parts of the file and where they are located. In fact, if you think about it, you might reach a level of complexity in file organization that is used by database systems (which routinely track large numbers of arbitrarily sized objects within a disk file).
Which brings us to the subject of records. Is a document file appropriate for storing multiple records of data that are similar to each other? The answer here is yes and no. Let's say your application uses a user defined type (data structure) internally to store data, and needs to persist an array of these structures. In this case, a document file can be an extremely efficient way to store this information. Visual Basic includes record I/O file operations that are fast and have low overhead.
The document approach towards storing records does, however, have a number of potential disadvantages. Insertion and deletion of individual records can be complex. If there is a chance that the structure format might change, your application will need to keep track of document versions and implement a scheme for updating documents from one version to the next. Use of variable length records requires a more complex storage scheme that includes some mechanism for finding the start and end location of each record. Also, any searching or filtering schemes need to be implemented by the application.
