Name: rlT66838 Date: 07/23/97


In question 7 of the javadoc tips at
http://java.sun.com/products/jdk/javadoc/javadocfaq.html
one can read that a Perl script called

javadocindexsplit.pl

has been used to break up the unusably long
index file that javadoc creates for the jdk.

Why is the feature that this script supplies
not included in javadoc, possibly as an option
one can switch on and off?

The contra argument that this were useful only
for huge libraries like the jdk and cumbersome
for smaller packages (since it would obstruct
navigation within the index) can be easily
confronted with suggesting the following:

javadoc gets a new option, maybe called
-maxindex xxx
where xxx denotes the maximum size that each
of the index pages may have.

Thus, the index is broken up only when in needs
to, not 26 times, once for each letter.

The resulting, broken up pages should be
automatically interconnected, as it is done now
with this Perl script. And it should be taken
care to break up the index only at "alphabet
boundaries", where an alphabet boundary is
the transition between one letter and the next
one, as the first letter of a word.

, email - ###@###.###
======================================================================