Hmm ... no! I can't think of a single use case where i would need to treat the BOM as part of the content. Actually, i can only come to the conclusion that a BOM within the content would be a serious bug. After all, your appication should _never_ deal with the binary representation, only with code points. What _code point_ do you get for BOM?

i just download HTML pages using Java functions into Java strings. then i use CL-PPCRE to extract some information from it. certainly, i don't care about BOM, but CL-PPCRE crashes on it trying to aref array beyong char-code-limit. i can pre-filter data removing BOM, but i'm not guaranteed that i won't get some other wild character.

well, there are better ways to tokenize HTML, but i've made quick and dirty solution via CL-PPCRE :)