🚀 go-pugleaf

In a recent misc.writing article, Len Olszewski wrote:
>
>In article Thomas Beagle writes:
>
>|> ObWriting : When writing a manual, do you index as you go (and risk
>|> all sorts of incompatibilities) or do you index after the bulk of the
>|> manual is written?
>
>After its finished. The way we lay it out here, setting the final page 
>breaks is an important pre-index requirement. Also, you never know 
>what's going to be in or out in some books, really, until the last
>minute. You've already mentioned consistency as good reason to wait.
>Also, the editors get the book last, and they can introduce all kinds of 
>changes.

As we go. We use SGML and our document structure includes index elements.
You can put them in as you go. ("Oh, I'm writing about error messages. That
should go into the index. BANG! There it is.")

When the document is done and printed, the printing software doesn't print
these elements, it just notes them and the page number on which they appear.
So the index is created as the document is printed. ("Look, Sadie, there's
a discussion of error messages on page 67!") Usually, we massge the index
entries when we look at the first page proofs, which are intended to be
drafts anyway.

The index element (in case anyone cares) allows you to specify the index
entry, including a parent entry if you want. As a result, we get index
entries which look like:


    Equation delimiters  . . 36
    Errors
       conditions  . . . . . 72-75
       messages  . . . . . . 67, B.1-9
       testing in macros . . B10

I've worked in shops where the indexing was done by hand at the end of the
process. This is much, much better. Much better. 

Marcy

-- 

Marcy Thompson		
  SoftQuad (West)			
    marcy@sqwest.wimsey.bc.ca (preferred) or marcy@sq.com