That's how I've come to view Borders these days.
I'm not here to wax nostalgic about my book-consumption record of late, I'm here to throw wonky speculation together.
It has occurred to me recently that Perl has a mutability that other languages, say C++, lack. This is because Perl (and your Pythons and Rubys and whatnot) are based on either linguistics (like Perl) or Perl (like Ruby). Language has been mutable, over the millenia; math less so. It makes sense that something based on calculus cannot truly be given an overhaul; calculus cannot be overhauled!
The argument goes back and forth; C++/Java/whatever this this this, Perl/Python/whatever that that that. It's irrelevant. The resounding cry of "Use The Best Tool For The Job!" is lost on people who do not realize there's more than one type of job. They're arguing that C is algebraically complete, or they're arguing that Perl is more compact.
Would I ever have a conversation in mathematical symbols? Not unless it was for the sake of a joke. Would I ever attempt to communicate deep mathematics without the lovely shorthand provided by mathematical symbols? Not unless it was for the sake of a joke. (I'd do almost anything for the sake of a joke.) The point is that some languages are more comfortable for different things, and it has nothing at all to do with good or bad or ugly. It has to do with useful. Useful trumps everything else.
Regardless of these things, mathematically-based languages will probably always play the card of Running Fast, and linguistically-based languages will probably always play the cards of Easier, Faster, More Portable. The only problem for math-languages is that Running Fast doesn't matter if you're not in a race. It isn't going to be such a big deal as the years go by; speed becomes irrelevant after a certain point. Eventually, Running Fast just won't be enough.
The other problem is that math-languages seem to push people to closed viewpoints. Note the Inline::* modules. Note that no such thing exists for, say, Java. (Well, sure, there are a few efforts -- but Perl's got the monopoly right now, AFAIK.) If speed becomes a true non-issue, and Perl can parse C, who needs C? Any old things can be wrapped into "inlined C" Perl scripts.
I didn't have a point, sorry. I've just been having this thought since earlier in the day, when I was scanning _Code Complete_ at Borders, and I wanted to get it recorded somewhere.