Network Computingwww.networkcomputing.com

By Augie Hansen

A Survey of students in my UNMIX system classes shows that somewhere between one-third and one-half had prior experience with DOS before attempting to learn about the UNIX system. This range is significant, though not surprising, given the large number of DOS-based computer systems, which some estimates place at more than 20 million.

It is heartening that so many of you are converting to UNIX or at least learning about UNIX as an alternative to DOS for some computing tasks. If you're among those who are converting, this month's column is for you. It describes the similarities and differences between the two operating environments and summarizes the major commands and application programs that you need to know to obtain a reasonable level of proficiency.

To simplify the task, I assume that you have a basic knowledge of DOS commands and files, knowledge that goes beyond having your system come up running Lotus 1-2-3 or dBase. If you've used your DOS system in a way that has prevented you from ever seeing the DOS prompt, this month's column will be of little or no value to you. Windowing and menu-style user interfaces, a topic for a future column, will undoubtedly serve you better than a command-oriented shell.

Commands Common to DOS and UNIX

Table 1 lists DOS and UNIX system commands that have similar purposes in both operating environments-not identical behavior, to be sure, but close enough for the commands to be considered almost equivalent. Fundamental differences between the design philosophies and capabilities of the two operating systems preclude precise functional equivalence.

The DOS and UNIX system commands that have similar purposes in both operating environments-not identical behavior, to be usure, but close enough for the commands to be considered almost equivalent. Fundamental differences between the design philosophies and capabilities of the two operating systems preclude precise functional equivalence.

The DOS and UNIX directory-related commands are very similar. The primary difference is that the DOS MKDIR and RMDIR commands can deal with only a single directory name for each invocation. The UNIX mkdir and rmdir commands accept directory name lists,so a single command invocation can create or delete multiple directories.

Under DOS, all file and directory names are converted to uppercase on input. Such names are conventionally shown in all capital letters, but you can type them with either lowercase o///// uppercase letters. Most UNIX program names, on the other hand, contain only lowercase letters, which you must type exactly as shown because the UINIX system is sensitive to letter case.

The DOS CD command serves the purposes of two UNIX commands. When invoked with a directory path name argument, it acts lid cd, which changes to the specified directory if the directory is correctly named. However, CD without a directory argument displays the name of the current directory, which is the function of pwd (print working directory) under UNIX. The UNIX cd command, when invoked without a directory argument, changes to the user's home directory - a source of considerable frustration for many DOS-to-UNIX converts. Of course, DOS has no notion of a home directory since it's not a multiuser system like UNIX.

Several other DOS commands have identical names and purposes as their UNIX namesakes, but not quite the same operation. The following examples describe the differences you should be aware of.

The DOS DATE command allows you to set and read the system date value. A separate TIME command deals with system time values. The UNIX date command combines both the system date and time functions into one program. Issuing a date command without any arguments produces a fixed-length string that contains date and time information in an abbreviated form, like so: Tue Jul 26 11:45:14 MST 1988.

You can use an optional format-control string argument with the System V UNIX date command to alter the presentation of the date and time information. A variety of field descriptors determine the appearance of date and time strings and permit you to extract the components of dates and times individually. Each field descriptor is introduced by a percent sign (5).

Listing 1 shows how you can use the and %Tfield descriptors intermixed with free text to customize the date/time output. The single quotes around the format-control argument are required to present the entire string, including embedded spaces, as a single argument to the date command. If you need to print a percent sign in the test, use two (%%) to turn off the characters special meaning.

More detailed information about the date>date command and all other UNIX commands is available in the UNIX User Reference Manual. See the sidebar for information about the manual's organization and contents, and some tips on how to use the manual effectively.

The DOS SORT command is a simplistic version of the UNIX sort command. Both help you order text in one or more files, but the DOS version is limited in its ability to isolate text strings in portions of a line in a file and in its ability to handle patterns rather than fixed strings.

Listing 2 shows how easy it is to create and use a database under UNIX. A simple text file containing the data is created with a text editor (or a word processing program). The file contains the data organized for easy access. Each line is a record. Each record is a collection of fields that are separated from each other by a blank or some other other character that is not likely to be part of the text of any field. I have used the colon (:) as a field separator in the sample phone list database file shown in Listing 2A. The default field separator is a blank (space or tab).

The sort command without any sort-ordering arguments prints the records out ordered by first name, which is the content of the first field. The entire record, starting with the first character, is used as a sort key unless you specify a different key. The results of a simple sort on the the entire record produces the output shown in Listing 2B.

To sort using a different position in the records, use an offset to specify the starting field. For example, typing sort -t: +1 phonelist sorts the records starting with the last name field. The +1 option syas to use the field offset by one from the left-most end of the record (which has zero offset). The -T: option tells sort to use the colon (:) as a field separator instead of the default blank,

Note that sorting a file this way does not change it. It simply displays the contents in the specified order. If you want to reorder the original file, use command-line redirection of the output (sort phonelist >temp) to create a temporary file; then overwrite the original file with the temporary file (mv temp phonelist). The UNIX sort command also recognizes a -o option followed by an output file name. In this case the output name can be the same as the input name, as in sort -t: +1 phonelist -o phonelist to sort phonelist ``in place.''

A Rose by another Name...

Other commands have different names in the two operating environments but do approximately the same job. An example is the DOS FIND command, which is similar to the UNIX grep command. These commands help you find text in a file or in a group of files. For example, suppose you need to find all references to the word administration in a set of files related to a project you are working on. You could, of course, use an editor (or word processor) to open each file in turn and use its built-in search feature to look for the word. But there is an easier and faster way.

Use grep. This cryptic name is derived from the ed editor terminology g/re/p, which means do a global search (g) for a regular expression (re-a text pattern) and print (p) any lines that contain a string of characters matching the pattern specified by re. Listing 2C shows a command that does the job. The search for ``johnson'' finds one matching line, which is printed in its entirety.

Caution: The UNIX system has a command named find, too, but it has a completely different purpose that has no standard DOS equivalent.

Both the UNIX find and grep commands will be covered in detail in a future ``New to UNIX'' column about commands that help you find text in a set of files, find files somewhere in the UNIX file system, and so on.

Reading the UNIX User Reference Manuals

The UNIX system is a large and complex product, and thousands of pages of documentation are but one piece of evidence in support of this claim. Beginning UNIX users are often put off by thesheer bulk of this terse documentation.

Don't let the manuals overwhelm you. Once you know where to look, most of what you need to know is there for the reading. Learning where to look involves learning about the manual sections, the permuted index, and the format of the manual pages.

The cornerstone of the UNIX documentation library is the various reference manuals. UNIX system vendors are free to package them any way they wish, but most adhere to the original spirit-if not the exact form-of the AT&T Bell Laboratories reference manuals.

The reference manuals are divided into sections. Each section has a primary intended audience. Nearly all sections are of interest to a system administrator, whereas beginning users will, at first, care only about Section 1 (Commands and Applications), kand possibly Section 6 (Games) of the User Reference Manual.

A few commands have few or no options, do very simple, clearly defined jobs, and therefore, have brief manual pages. The cd command is a good example. It takes a simgle argument to specify the target directory, or none at all to indicate that the target is the user's home directory