If you save the .htaccess file using ISO-8859-1 or Windows 1252 encodings, it works fine. But if you use utf-8 recommended and universal encoding, Apache triggers an "Internal Server Error". I have tried several hours creating some rules, and when I found out nothing worked, all of the sudden I saw Apache did not accept utf-8. Most operating systems and applications are migrating to utf- 8 to support international languages, so it would be a good thing to work properly with utf-8.
That'll be a BOM, inserted by your editor. I'm not convinced we gain anything by accepting a BOM in .htaccess files, but I'll leave it open as an enhancement request. You might want to submit a bug report to the developers of your editor, if it inserts binary stuff in a text file without telling you.
I thought I had hacked ap_popen/read eons ago to address this, wonder if it fell off the map, or I'm misremembering. I'll take a peek.
http://mail-archives.apache.org/mod_mbox/httpd-dev/200303.mbox/%3cn2m-g.ncsueipv7xln$ndparker@news.perlig.de%3e This was done for Win32 script shebang parsing, since httpd has a rule to ignore leading whitespace, I don't see why avoiding these first three bytes would be a bad thing for conf file parsing (minimal overhead). That PR was 16687, see http://svn.apache.org/viewvc?view=rev&revision=98775
Whoops, wrong patch, the patch in the above comment relates to CGI (and could be applied to perl python and possibly other environments when on unix.) The patch you wanted was; http://svn.apache.org/viewvc?view=rev&revision=94446 and is wrapped in an #ifdef WIN32 that is easily stripped off to apply to any platform at all. By rights this is a dup of 10125, but you are asking across platforms.