How using a large number will cause an error exactly? At best he will get a loss of precision, perhaps that is what you meant to say.
I should also point out that the regex version won't cause any loss in precision, which should also be the case for $bytes in this case, imo.
I'm guessing $bytes has been implemented in a generic way that create a loss in precision even when it could be avoided, but it sounds like a bug in $bytes.
