And every ALGOL installation or project can (and will) define its own set of procedures, that (sic) will make their programs readable for everyone else.

If people don’t include those procedures when they show their source code to other people, how will anyone read and understand ALGOL programs?

It’s impossible. We should stick with FORTRAN.

;-)

It’s pretty clear to me that Lisp macros failed for purely social reasons though, since C++ templates, which are basically a macro facility from hell, have “succeeded”, and for any possible technical problem with a Lisp macro one can come up with, templates have 10 worse problems. I think people like M-expressions, or think that they like M-expressions, and that pretty much sums it up.

I think the “lingua franca” business has some truth to it, and that’s why CLOS became “built-in”, eventually, but I’ll never believe that this issue influences anyone’s language decisions. For example, C++ doesn’t have a built-in list, and uses an awful form of macro instead; Lisp has a built-in list. Few people seem to mind.

P.S. I get PHP errors instead of entry/comment contents. I only knew what I was replying to because RSS seems to work.

@Yossi: sorry about PHP troubles, WP is crap, but I only have PHP on this host :(. As to your comment, it’s related to the surgeons things — I just don’t understand how suggesting that we cater for the worst doctors (perhaps without proper education) around would be rightly taken as crazy talk, yet in programming it’s somehow important that you write tools for the dumbest. Perhaps the failure of Lisp to gain popularity (measured in C++, Java and other mainstream languages’ terms) is no failure, but actually a sign of success. You don’t see cranial surgery becoming a fad either.

As a sidenote, there seems to be discussion on reddit about this post, at http://programming.reddit.com/info/65gco/comments

As to “problems with abstraction”: C /does/ have a problem of making it hard to debug in interlaced assembly mode, and making it hard to implement a compiler that makes such debugging tolerable (forget about “easy”) for optimized code. And don’t get me started about C++ and its templates. I’ve had lots of crap for lunch because of that on this very day…

But there are tons of such locally valid remarks on the subject. I still don’t see what they converge to, globally, if you consider the “pros and cons of abstraction” issue. I know wish I had good macros…

However, to paraphrase a great writer, “Hell is other people’s macros.” True, bad coders will always write bad code, but the ability to overload operators and arbitrarily extend language syntax at will… it’s more than most programmers will handle gracefully. Add to that the fact that two individuals may choose the same naming convention and syntax to perform the same task (say, a chain of command), but may have subtly different

While I take solace in the fact that most code is code I will never have to read, if I do have to read crappy code, I would prefer those coders had just a bit less rope…

There are other issues with macros, though. They’re hard to support effectively in debugging tools, for one.

I feel lazy evaluation is a superior substitute for many common uses of macros.

If I had to sum up the post in one sentence, it’d be “Dismissing macros as too difficult is dumb, because you’re confusing the complexity inherent in the problem with the complexity introduced by your solution”.

@Matthew: definitely agreed. I never meant to say that macros were the only hammer you have, only that not having it (or some other, equivalent tool) is stupid. The first rule of writing macros is *not* to use macros unless absolutely necessary, and I stand by that firmly.

Even C/C++ macros can do a lot if properly used, despite lacking the power of Lisp macros. Case in point, the assert() macro.

If even its own native speakers can’t communicate with it, it can’t even qualify for “lingua franca” status. The lingua franca that Sun created is actually the JVM, because it’s used by Jython, JRuby, Rhino, and others.

[…][…]…

And I am sorry if I offended the chinese language (such a beautiful language) by comparing it with the worst invention in programming languages, EVER.

Galo, sorry, I must disagree. Chinese *is* a lingua franca.

Lingua franca does not mean “anyone can understand it”, which is trivially false of all languages. It means “widely used beyond the population of its native speakers”, which Chinese definitely is.

– http://en.wikipedia.org/wiki/Lingua_franca#Chinese

Sure, English is also a lingua franca. Nothing about its definition says there can be only one. It’s an attribute of a language, not a contest between them.