This has been reported before. The problem is that "if there is UTF8 in the text" isn't something you can reliably check. What you define as a UTF8 bitstream can easily be a bunch of ANSI chars to me. No "option" will work, because you end up with the exact same case in reverse-- a bunch of unicode chars in a script would mess with the ANSI chars in a line. I'm also not quite sure if your suggestion is to have an option to selectively ignore a specific set of ANSI chars- if so, that's a solution that only seems to help your specific scenario, not something that helps mIRC as a whole.
There was a compromise suggestion to have mIRC treat each word (space delimited) as its own encoding. That way, for instance, a unicode timestamp could sit next to ansi text without changing the encoding. Unfortunately this is likely to make the display a lot slower, and we're still waiting to see what Khaled does about it.
