declare no-default-io

The no-default-io declaration is provided to make it easier for you to develop server and similar transaction-based applications in OmniMark. This declaration keeps the programs in which it is used from using environment features such as "standard input" and "standard output", because these features differ in their behavior between batch processing and server environments. The no-default-io declaration allows you to maintain consistency in how these different types of programs run. Since no-default-io programs don't have any established input sources or output destinations, these sources and destinations must be explicitly established within the program.

Note that the no-default-io declaration cannot be used in any of the aided translation type programs.

Since no-default-io programs do not have any established input sources or output destinations, the following streams cannot be used:

• #process-output
• #main-output
• #error
• #console

Additionally, neither the #process-input nor #main-input source can be used. The #command-line-names shelf is still available in no-default-io programs, but #main-input is not bound to the files named in that shelf as it is in other programs.

Since the current output set (#current-output) is not initially established, it is an error to use an output action or to make any reference to #current-output until a current output set is established. Note that the #current-output is attached test is now available, allowing you to determine the current state of #current-output prior to writing to it.

Error and status reporting in a no-default-io program cannot be done by writing to #error because any reference to this stream is an error. OmniMark provides the log-message function (part of the #builtin function library) as a general mechanism for error and status reporting in a no-default-io program.