Comment Static code verifications, anyone? (Score 1) 617
I am amazed nobody mentions the lack of static code checks in Lisp. There is this ongoing confusion about Lisp compilation: detractors say Lisp doesn't have compilation, then every pro-Lisper starts refuting that, "No, you moron, Lisp HAS compilation since ages".
Well, nobody really cares if it is interpreted, compiled, JITed, you name it, as long as it is fast. But most programmers (and most important, MANAGERS) care about compilation as a way to minimize run-time errors. Heck, they love it, they are obsessed with it. Look at the extremes people go with strong-typing, building strongly-typed collections wrappers over the standard collections classes in Java and C#.
While a code base with small incremental changes can do without static code verifications, lack of it makes team work so much harder. It feels absurd for modern programmers to use a language which cannot detect that a function is missing, unless you run the program. Or for which a string multiplication with a float is just fine.
Paul Graham tells you that bottom-up programming is the best. Well, in Lisp you only have bottom up programming, because every code fragment has to be tediously verified by running it, or you are going to mispel a symbol, miss some files not included in the project, or God knows what.
Add to that the extremely fragmented Lisp runtimes and development environments, and Lisp is disappearing.
Which is a terrible shame, since Lisp has so much power and potential. I love Lisp. I would dump C#, Java and C++ for a good Lisp any day now, because they lack any serious meta-programming capabilities. We have to repeat the same stupid work again and again, because you can't build truly reusable code in these braindead languages.
What we need for Lisp is a champion who would do for the Lisp concept what Java did for the C++ concept. Modernize it. Clean it. Implement it well and completely. Allow it to become the next-generation development system.
It is going to happen, sooner or later. The "Intentional Programming" meta-programming environment Microsoft Research was developing some time ago looks for all intent (pun intended) like Lisp. Like a poor's man Lisp in fact, because those guys never understood that they needed a true computation model (which Lisp has), they were just adding features by the bunch, like they did in Windows.
Don't hold your breath for this to happen.
Well, nobody really cares if it is interpreted, compiled, JITed, you name it, as long as it is fast. But most programmers (and most important, MANAGERS) care about compilation as a way to minimize run-time errors. Heck, they love it, they are obsessed with it. Look at the extremes people go with strong-typing, building strongly-typed collections wrappers over the standard collections classes in Java and C#.
While a code base with small incremental changes can do without static code verifications, lack of it makes team work so much harder. It feels absurd for modern programmers to use a language which cannot detect that a function is missing, unless you run the program. Or for which a string multiplication with a float is just fine.
Paul Graham tells you that bottom-up programming is the best. Well, in Lisp you only have bottom up programming, because every code fragment has to be tediously verified by running it, or you are going to mispel a symbol, miss some files not included in the project, or God knows what.
Add to that the extremely fragmented Lisp runtimes and development environments, and Lisp is disappearing.
Which is a terrible shame, since Lisp has so much power and potential. I love Lisp. I would dump C#, Java and C++ for a good Lisp any day now, because they lack any serious meta-programming capabilities. We have to repeat the same stupid work again and again, because you can't build truly reusable code in these braindead languages.
What we need for Lisp is a champion who would do for the Lisp concept what Java did for the C++ concept. Modernize it. Clean it. Implement it well and completely. Allow it to become the next-generation development system.
It is going to happen, sooner or later. The "Intentional Programming" meta-programming environment Microsoft Research was developing some time ago looks for all intent (pun intended) like Lisp. Like a poor's man Lisp in fact, because those guys never understood that they needed a true computation model (which Lisp has), they were just adding features by the bunch, like they did in Windows.
Don't hold your breath for this to happen.
