said by Johnny 
:said by cacroll :said by Johnny :Any parser should be able to tell that a left angle bracket by itself is not HTML and ignore it.
Define "by itself". Do you mean unpaired, or surrounded by spaces?
I mean without being part of a valid HTML expression.
If the former, the parser would have to scan the entire message that you're writing, to check for pairs of < and > ? If the latter, you could write "if a < b" but couldn't write "if a<b"?
It has to scan it anyway, to look for the expression that it must assume the left angle is a part of. As you said, if it reaches the end without identifying a valid expression, it can either throw an error, throw away the text scanned up to that point (which is what it is doing now), or treat the text scanned as plain text.
The feature request is to treat scanned text that does not constitute a valid HTML expression as plain text and just output it.
The same is true for square brackets - the metalanguage or BNF for them is fairly limited, so it isn't hard to determine if the characters following a left bracket are making a valid expression. As soon as the first invalid character is encountered, it can bail and treat it as plain text.
