This is a copy of Ken Thompson's original documentation for the Unix shell. It is taken with permission from http://man.cat-v.org/unix-1st/1/sh, home of the whole set of 1st ed manual pages at http://man.cat-v.org/unix-1st/.
11/3/71 SH (I)
NAMEsh - shell (command interpreter)
SYNOPSISsh [ name [ arg1 ... [ arg9 ] ] ]
sh is the standard command interpreter. It is the program which reads and arranges the execution of the command lines typed by most users. It may itself be called as a command to interpret files of command lines. Before discussing the arguments to the shell used as a command, the structure of command lines themselves will be given.
Command lines are sequences of commands separated bycommand delimiters. Each command is a sequence of non-blank command arguments separated by blanks. The first argument specifies the name of a command to be executed. Except for certain types of special arguments discussed below, the arguments other than the command name are simply passed to the invoked command. If the first argument represents the path name of an executable file, it is invoked; otherwise the string "/bin/" is prepended to the argument. (In this way the standard commands, which reside in "/bin," are found.) If this search too fails a diagnostic is printed. The remaining non-special arguments are simply passed to the command without further interpretation by the shell.
There are three command delimiters: the new line, ";" , and "&". The semicolon ";" specifies sequential execution of the commands so separated; that is, coma; comb causes the execution first of command coma, then of comb.The ampersand "&" causes simultaneous execution: coma & comb causes coma to be called, followed immediately by comb without waiting for coma to finish. Thus coma and comb execute simultaneously. As a special case, coma & causes coma to be executed and the shell immediately torequest another command without waiting for coma.
Two characters cause the immediately following string to be interpreted as a special argument to the shell itself, not passed to the command. An argument of the form "arg" causes file "arg" to be used as the standard output file for the given command. If any argument contains either of the characters "?" or "*", it is treated specially as follows. The current directory is searched for files which match the given argument. The character "*" in an argument matches any string of characters in a file name (including the nullstring); ? matches any single character in a file name. Other argument characters match only the same character in the file name. For example, "*" matches all file names; "?" matches all one-character file names; "ab*.s" matches all file names beginning with "ab" and ending with ".s".
If the argument with "*" or "?" also contains a "/", a slightly different procedure is used: instead of the current directory, the directory used is the one obtained by taking the argument up to the last "/" before a "*" or "?". The matching process matches the remainder of the argument after this "/" against the files in the derived directory. For example: "/usr/dmr/a*.s" matches all filesin directory /usr/dmr which begin with "a" and end with ".s" In any event, a list of names is obtained which match the argument. This list is sorted into alphabetical order, and the resulting sequence of arguments replaces the single argument containing the "*" or "?". The same process is carried out for each argument with a "*" or "?" (the resulting lists are not merged) and finally the command is called with the resulting list of arguments.
For example: directory /usr/dmr contains the files al * s,a2.s, ..., a9.s. From any directory, the command as /usr/dmr/a?.s calls as with arguments /usr/dmr/al.s, /usr/dmr/a2.s, .../usr/dmr/a9.s in that order.
The character "\" causes the immediately following character to lose any special meaning it may have to the shell; in this way <, and other characters meaningful to the shell may be passed as part of arguments. A special case of this feature allows the continuation of commands onto more than one line: a new-line preceded by "\" is translated into a blank.
Sequences of characters enclosed in double (") or single(') quotes are also taken literally.
When the shell is invoked as a command, it has additional string processing capabilities. Recall that the form in which the shell is invoked is
sh [ name [ arg1 ... [ arg9 ] ] ]
The name is the name of a file which will be read and interpreted. If not given, this subinstance of the shell will continue to read the standard input file. In the file, character sequences of the form "$n", where n is a digit 0, ..., 9, are replaced by the nth argument to the invocation of the shell (arg ). "$0" is replaced by name.
An end-of-file in the shell's input causes it to exit. A side effect of this fact means that the way to log out from UNIX is to type an end of file.
[reference], which gives the theory of operation of the shell.
"?", in case of any difficulty. The most common problem is inability to find the given command. Others: input file ("<") cannot be found; no more processes can be created (this will alleviate itself with the passage of time). Note that no diagnostic is given for inability to create an output (">") file; the standard output file has already been closed when the condition is discovered and there is no place to write the diagnostic.
If a "*" or "?" is used, the "glob" routine is invoked; it types "No command" if it cannot find the given command, and "No match" if there were no files which matched an argument with "?" or "*".
Better diagnostics should be provided. If a "*" or "?" isused, the command must be in /bin (Not, for example, in the user's directory.) This is actually a glob bug.