What is the difference between \PassOptionsToPackage and inline options passing?

Question

I have noticed that, for example, ClassicThesis config file uses this style:

\PassOptionsToPackage{utf8}{inputenc}
\usepackage{inputenc}

instead of

\usepackage[utf8]{inputenc}.

What is the difference between these two options of passing the options to a package? In other words, what is the goal of supporting both types of syntax for arguments in TeX (or LaTeX?)?

Answer 1

\documentclass{beamer}
\usepackage[colorlinks]{hyperref}
\begin{document}

foo

\end{document}

gives an option clash, because a package can be loaded only one time! And that is done for hyperref by beamer itself internally, but without an optional argument. LaTeX marks hyperref as loaded after the first usepackage or RequirePackage and then throughs an error if a package should be loaded a second time with different optional arguments. In such a case a \PassOptionsToPackage helps, or only for beamer one can use hyperref=colorlinks:

 \documentclass[xcolor=table,hyperref=colorlinks,...]{beamer}
 \begin{document}

 foo

 \end{document}

A user often didn't know what packages are already loaded by the documentclass or other packages or by LyX in its defined preamble. Then he/she realizes only by the option-clash error that a package should be loaded more than once.