L^ATEX on UNIX

The only special problems in using L^ATEX caused by the UNIX operating system involve the way UNIX handles files. The first problem arises because, when a program starts to write a file, UNIX destroys the previous version of that file. Thus, if an error forces you to stop L^ATEX prematurely (by typing Control-C or Control-Z), then the files that L^ATEX was writing are incomplete, and the previous complete versions have been destroyed. You probably don't care about the output on the dvi file, but, if you are making a table of contents or using cross-referencing commands, then L^ATEX also writes one or more auxiliary files that it reads the next time it processes the same input file. If the auxiliary files are incomplete because L^ATEX was stopped before reaching the end of its input file, then the table of contents and cross-references will be incorrect the next time L^ATEX is run on the same input file. You will have to run L^ATEX a second time to get them right. If you want to avoid having to run L^ATEX twice after making an error--for example, if your input is very long--then you should save copies of these auxiliary files before running L^ATEX. An input file named myfile.tex and all the auxiliary files produced by L^ATEX from it are included in the UNIX file specifier myfile.*. Use the UNIX cp command to save copies of these files.

The second problem in using L^ATEX on UNIX involves the files that L^ATEX reads. The file whose name you type with UNIX's latex command is called the root file. In addition to reading the root file, L^ATEX also reads the files specified by \input and \include commands. With the UNIX directory system, L^ATEX must know not only the names of these files but also in what directories they are. It will have no problem finding the correct files if you follow two simple rules:

1.

Run L^ATEX from the directory containing the root file.

2.

Keep all files specified by \input and \include commands in the same directory as the root file.

If you follow these rules, you never have to type an UNIX path specifier when using L^ATEX.

You should never break the first rule, otherwise L^ATEX will have trouble finding auxiliary files. (To run L^ATEX on someone else's file, copy the file to your directory.) If you break the second rule, specifying a file from another directory in an \input or \include command, you must use a complete path name. For example, to include the file hisfile.tex from Jones' directory /foo/bar, you can type

     \include{/home-theory/jones/foo/bar/hisfile}

A ~ character may not appear in the argument of an \input or \include command, so you can't use a file name such as ~jones/foo/bar/hisfile.

For people who don't like to obey rules, here is exactly how L^ATEX finds its files. The root file is found by UNIX according to its usual rules. L^ATEX's auxiliary files are read and written in the directory from which it is run. All file names specified in the L^ATEX input, including the names of document-style (sty) files specified by the \documentstyle command, are interpreted relative to the directory from which L^ATEX is run. If L^ATEX does not find a file starting in this directory, it looks in the system directory /usr/physics/lib/tex. You can change the directories in which L^ATEX looks for its input files by setting the environment variable TEXINPUTS. Putting the command

     setenv TEXINPUTS :.:/home-theory/jones/myown:/usr/physics/lib/tex:

in your .login file causes L^ATEX to look for files first in the current directory, then in Jones' myown directory, and then in the system directory. You might want to do this if your name is Jones and you have your own personal document-style files in your myown directory.