REDUCE is designed as an interactive system, but naturally it can also operate in a batch processing or background mode by taking its input command by command from the relevant input stream. There is a basic difference, however, between interactive and batch use of the system. In the former case, whenever the system discovers an ambiguity at some point in a calculation, such as a forgotten type assignment for instance, it asks the user for the correct interpretation. In batch operation, it is not practical to terminate the calculation at such points and require resubmission of the job, so the system makes the most obvious guess of the user’s intentions and continues the calculation.
There is also a difference in the handling of errors. In the former case, the computation can continue since the user has the opportunity to correct the mistake. In batch mode, the error may lead to consequent erroneous (and possibly time consuming) computations. So in the default case, no further evaluation occurs, although the remainder of the input is checked for syntax errors. A message ~Continuing with parsing only~ informs the user that this is happening. On the other hand, the switch ERRCONT, if on, will cause the system to continue evaluating expressions after such errors occur.
When a syntactical error occurs, the place where the system detected the error is marked with three dollar signs ($$$). In interactive mode, the user can then use ED to correct the error, or retype the command. When a non-syntactical error occurs in interactive mode, the command being evaluated at the time the last error occurred is saved, and may later be reevaluated by the command RETRY.