I have input files in ISO-8859-1 and have set INPUT_ENCODING accordingly. Doxygen output is in UTF-8, which is, in principle, OK. However, the generated LaTeX files seem to wrap at column 306, chopping in two whatever word happens to fall on that boundary. In the worst case, the first half of a UTF-8 two-byte sequence will be still on the "old" line, while the second half will go to the next. This is no longer valid UTF-8, and consequently, LaTeX will choke when trying to handle this.

To reproduce, use the attached files. The problem will show up in doc/latex/test_8h_source.tex, lines 11/12.

As a side note, there seem to be remains of a config parameter "LATEX_OUTPUT_ENCODING" in the sources for LaTeX output, but the corresponding lines n the LaTeX output handler are commented out via "#if 0". I have found no way to ask doxygen for non-UTF-8 output (which might circumvent the problem), but I may have overlooked an option to achieve this.