I know there is of are a good amount of financial systems written in COBOL, but that's because no one has had the guts to rewrite them, not because COBOL is some amazing perfect language.
And there are a plethora of other important, reliable systems that are still in perl because no one has had the guts to rewrite them. But I would guess that there are just as many that have been rewritten to python and ruby. And many more that have been written for the first time in those languages. And in the next few years we're going to be in the same place with those sacred cow scripts written in python and ruby.
I'm not trying to say that code is wrong. But it isn't productive to claim that we can slap a new version number on a library and claim it's going to satisfy everyone's hopes and dreams. At some point, we need to let our cows die, even if they're sacred.