NOTE: This library is deprecated; use racket/sandbox, instead. The racket/sandbox library provides a more general way to simulate running a new Racket process.

The vector of strings init-argv is read first with the standard Racket command-line parsing. Flags that load files or evaluate expressions (e.g., -f and -e) are ignored, but flags that set Racket’s modes (e.g., -c or -j) effectively set the default mode before argv is parsed.

Before argv is parsed, the procedure adjust-flag-table is called with a command-line flag table as accepted by parse-command-line. The return value must also be a table of command-line flags, and this table is used to parse argv. The intent is to allow adjust-flag-table to add or remove flags from the standard set.

After argv is parsed, a new thread and a namespace are created for the “restarted” Racket. (The new namespace is installed as the current namespace in the new thread.) In the new thread, restarting performs the following actions:

• The init-namespace procedure is called with no arguments. The return value is ignored.

• Expressions and files specified by argv are evaluated and loaded. If an error occurs, the remaining expressions and files are ignored, and the return value for restart-mzscheme is set to #f.

• The read-eval-print-loop procedure is called, unless a flag in init-argv or argv disables it. When read-eval-print-loop returns, the return value for restart-mzscheme is set to #t.

Before evaluating command-line arguments, an exit handler is installed that immediately returns from restart-mzscheme with the value supplied to the handler. This exit handler remains in effect when read-eval-print-loop is called (unless a command-line argument changes it). If restart-mzscheme returns normally, the return value is determined as described above.

Note that an error in a command-line expression followed by read-eval-print-loop produces a #t result. This is consistent with Racket’s stand-alone behavior.