4 ms·

> It allows for the creation of real Domain Specific Languages and the real extension of lisp for a given problem domain. I think the problem was that until no

> It allows for the creation of real Domain Specific Languages and the real extension of lisp for a given problem domain. I think the problem was that until now you weren't giving any examples that explained how Common Lisp macros were giving one any greater ability to do this sort of thing than Ruby. One could say "Ruby allows for the creation of real Domain Specific Languages and the real extension of Ruby for a given problem domain," with only a slight alteration in what one considers "real". As such, saying something like "Ruby doesn't give you real foo" doesn't tell anyone anything about what CL gives you that Ruby doesn't. Also . . . I totally understand the desire to keep (non-earth-shattering) performance differences out of the discussion. After all, one doesn't use Ruby to write realtime systems in the first place. The performance of its implementations is simply not impressive. If I've chosen to use Ruby for something, I've already thought about the performance implications and decided that the performance penalties of Ruby as compared with something like C (or even Perl) are not limiting factors for the particular task I choose to tackle. As such, it makes sense to say "Please tell me about the syntactic and semantic value of Lisp macros, rather than the performance benefits of a Lisp implementation over a Ruby implementation." You did, however, get into a little more explanation this time, which is good. I just wanted to point out that your disbelief at the notion that someone didn't achieve enlightenment based on your previous comment seems a bit unwarranted.

KirinDave 17y ago
> Also . . . I totally understand the desire to keep (non-earth-shattering) performance differences out of the discussion. After all, one doesn't use Ruby to write realtime systems in the first place. The performance of its implementations is simply not impressive. If I've chosen to use Ruby for something, I've already thought about the performance implications and decided that the performance penalties of Ruby as compared with something like C (or even Perl) are not limiting factors for the particular task I choose to tackle. As such, it makes sense to say "Please tell me about the syntactic and semantic value of Lisp macros, rather than the performance benefits of a Lisp implementation over a Ruby implementation." Well also Lisp macros are more comprehensive. Ruby can only make DSLs that look mostly like Ruby. In Lisp, you can literally drop in a new parser if you really want to. In fact, the default PLT Scheme DrScheme distribution has an Algol implementation that is basically a macro set (with some special features). This allows you to inject whole new programming paradigms into your code. If you need a special kind of OO, you can easily create it. This is actually incredibly-hard-borderline-impossible to do in Ruby. Especially when you get into anaphora injection and complex closure manipualtion. > You did, however, get into a little more explanation this time, which is good. I just wanted to point out that your disbelief at the notion that someone didn't achieve enlightenment based on your previous comment seems a bit unwarranted. I had the same reaction this morning when a teenager asked me, "Okay so I get pointers. I get what they are. But they don't see good for anything in the real world." If you find this offensive or dismissive, I apologize. I am simply saying that I think you and I are operating at entirely different levels of abstraction, and bridging that gap is difficult for me.
- apotheon 17y ago
  > Well also Lisp macros are more comprehensive. Ruby can only make DSLs that look mostly like Ruby. In Lisp, you can literally drop in a new parser if you really want to. Much better. I haven't worked much (almost at all) with macros. How easy/difficult would it be to create a Ruby-esque DSL in Common Lisp? I'm just curious. > This allows you to inject whole new programming paradigms into your code. If you need a special kind of OO, you can easily create it. This is actually incredibly-hard-borderline-impossible to do in Ruby. Really? I dunno, I've seen (for example) some really interesting functional paradigm idioms implemented in Ruby, such as a lazy evaluation module. > Especially when you get into anaphora injection and complex closure manipualtion. I admit I haven't done "complex closure manipulation" much anywhere. I do use closures from time to time in my code, in a number of languages -- including Ruby. I'm not entirely sure how "complex closure manipulation" would look, actually, so I have no basis for comparison in that regard. I guess I'll just have to take your word for it. > I had the same reaction this morning when a teenager asked me, "Okay so I get pointers. I get what they are. But they don't see good for anything in the real world." If you find this offensive or dismissive, I apologize. I don't find it offensive or dismissive. I understand your perspective. I just don't think it helped anyone understand anything. I'm sure it all looks a lot more natural to someone who has done a lot more CL and Scheme coding, but it really doesn't to someone who has pretty much never touched a CL or Scheme macro in his life (or even for me, who has touched them a little). In fact, what Lisp macro experience I have is mostly limited to UCBLogo plus a light sprinkling of R6RS. > I am simply saying that I think you and I are operating at entirely different levels of abstraction, and bridging that gap is difficult for me. That's understandable. To some extent for my benefit, and even more so for the other guy's, I decided to explain some of where the gap lies -- so that you might find it easier to bridge. I hope.

KirinDave 17y ago

> Also . . . I totally understand the desire to keep (non-earth-shattering) performance differences out of the discussion. After all, one doesn't use Ruby to write realtime systems in the first place. The performance of its implementations is simply not impressive. If I've chosen to use Ruby for something, I've already thought about the performance implications and decided that the performance penalties of Ruby as compared with something like C (or even Perl) are not limiting factors for the particular task I choose to tackle. As such, it makes sense to say "Please tell me about the syntactic and semantic value of Lisp macros, rather than the performance benefits of a Lisp implementation over a Ruby implementation." Well also Lisp macros are more comprehensive. Ruby can only make DSLs that look mostly like Ruby. In Lisp, you can literally drop in a new parser if you really want to. In fact, the default PLT Scheme DrScheme distribution has an Algol implementation that is basically a macro set (with some special features). This allows you to inject whole new programming paradigms into your code. If you need a special kind of OO, you can easily create it. This is actually incredibly-hard-borderline-impossible to do in Ruby. Especially when you get into anaphora injection and complex closure manipualtion. > You did, however, get into a little more explanation this time, which is good. I just wanted to point out that your disbelief at the notion that someone didn't achieve enlightenment based on your previous comment seems a bit unwarranted. I had the same reaction this morning when a teenager asked me, "Okay so I get pointers. I get what they are. But they don't see good for anything in the real world." If you find this offensive or dismissive, I apologize. I am simply saying that I think you and I are operating at entirely different levels of abstraction, and bridging that gap is difficult for me.

apotheon 17y ago
> Well also Lisp macros are more comprehensive. Ruby can only make DSLs that look mostly like Ruby. In Lisp, you can literally drop in a new parser if you really want to. Much better. I haven't worked much (almost at all) with macros. How easy/difficult would it be to create a Ruby-esque DSL in Common Lisp? I'm just curious. > This allows you to inject whole new programming paradigms into your code. If you need a special kind of OO, you can easily create it. This is actually incredibly-hard-borderline-impossible to do in Ruby. Really? I dunno, I've seen (for example) some really interesting functional paradigm idioms implemented in Ruby, such as a lazy evaluation module. > Especially when you get into anaphora injection and complex closure manipualtion. I admit I haven't done "complex closure manipulation" much anywhere. I do use closures from time to time in my code, in a number of languages -- including Ruby. I'm not entirely sure how "complex closure manipulation" would look, actually, so I have no basis for comparison in that regard. I guess I'll just have to take your word for it. > I had the same reaction this morning when a teenager asked me, "Okay so I get pointers. I get what they are. But they don't see good for anything in the real world." If you find this offensive or dismissive, I apologize. I don't find it offensive or dismissive. I understand your perspective. I just don't think it helped anyone understand anything. I'm sure it all looks a lot more natural to someone who has done a lot more CL and Scheme coding, but it really doesn't to someone who has pretty much never touched a CL or Scheme macro in his life (or even for me, who has touched them a little). In fact, what Lisp macro experience I have is mostly limited to UCBLogo plus a light sprinkling of R6RS. > I am simply saying that I think you and I are operating at entirely different levels of abstraction, and bridging that gap is difficult for me. That's understandable. To some extent for my benefit, and even more so for the other guy's, I decided to explain some of where the gap lies -- so that you might find it easier to bridge. I hope.