File hierarchy

makes it easy to structure a document as a set of separate files. There are good arguments for making use of this. Among others these include:

• such files are easily maintained if the file structure reflects the document structure;

• sections can be moved within the document, or replaced with new versions, simply by changing links;

• diagrams etc can be developed and tested separately;

Experience with the development of large reports suggests that the following structure is useful during development as well as being easily maintained:

1. A root file which contains
   1. option-setting commands that will have global effect, such as \pagestyle and any locally defined commands such as the \dspaceon and \dspaceoff commands shown in Chapter . Also title-setting commands.

   2. Commands to create the tableofcontents and to set the page numbering forms (see later).

   3. Commands to `pull in' the text of the chapters.

   4. Comments to remind the author of tasks to be performed in the future.
   Essentially, this file reflects the top-level design of a document.

2. A file for each chapter, that in turn pulls in the files containing the text of each section. Use of this added level of indirection makes it easy to re-organise parts of a document if the original design proves incorrect.

3. A file of text for each section.

4. A file for each diagram/table. This allows these to be developed separately, using a temporary root file, so reducing processing time considerably, as these items often need a disproportionate number of re-tries in order to get them right.