The chm output in FPDoc is actually inherited from the HTML backend so everything that applies to the HTML backend applies to the chm backend, except that all generated HTML files are written directly to a stream. After all the HTML files are generated the compression is begun. Once all the auto generated files are compressed, if the –other-files option is used these files are collected and compressed. Now if the –auto-index or –auto-toc are used the Index and Table of Contents are created and the compression is finished.