So Smalltalk doesn't care for human-machine interface principles, but Java, "the best programming language", of course, does...
So you think that [obj keyword: param1 keyword: param2] is stupid and ugly. Well, that's your opinion. I don't want to tell you here how stupid and ugly I think your mom is, because that wouldn't be appropriate.
Let's just say, that I prefer one of the following two code samples:
Let's not argue about speed here, since most languages (Lisp and Java anyway) are fast enough for most applications.
Lisp isn't stuck in the '50s. It was invented AFAIK end of the '50s, evolved over more than two decades, in the form of different dialects, was then in 1984 merged into Common Lisp (with all the good and important features), got the first ANSI-standardized object system.
Lisp contains features from the '80s. Java's most advanced features are probably exception handling, object orientation, and garbage collection. The latter one *originated* in Lisp, and OO was implemented almost as in Java in Smalltalk 20 years before. I don't know about exceptions, but AFAIK '84 Lisp also can do that (I'm still somewhat new to the Lisp world).
It's funny that you say that everthing about Lisp except the syntax sucks: so object orientation is bad? A uniform data representation that is basically compressed XML is bad? Packages are bad? Keyword arguments are worse than Java's function overloading? Macros are bad? LOOP and FORMAT are bad? Do you even remotely know what you're talking about?
Sure, CAR and CDR derive from Lisp machine register names. So what? You can combine then to CADAR and CDDR, for instance. You can use FIRST and REST, if you prefer.
So Java is "the best programming language". Cool. Only that I don't think it's easy to grasp for computer newbies, which I happened to tutor. Java is unnecessarily complicated with its public static final stuff. All the ().[],;{} distract from what you actually want the program to do and makes nested expressions look dead ugly.
It has good APIs, but that's about all. You have to typecast everything out of containers, which defeats the purpose of static checking. Even cooler are boxing and unboxing of ints to Integers, so you can actually put them in Hashtables. Sure, this all changes in Java 5, but Java had these problems for 10 years! I'm not alone in saying that Java gets almost nothing right. *I* prefer coding in C, because I'm more productive there (less errors, less time spent debugging, better debugging facilities).
And that you can do macro stuff in Java: I doubt it! Can you easily integrate a yacc-parser in your Java program? How about alternative control structures, loops? C++ templates are unreadable and don't remotely match Lisp macros in power and elegance.
A group of Linux fans are sitting in a bar. One says, "I really like the performance of Linux #180." The next one replies "yes, but Linux #87's package system is better." Another one says "I really like Linux #298's lightweightness."
One guy says "What do you think about Linux #315's features?" One programmer starts looking totally impressed, stutters "wow, is thaaaat cool" and "man, I didn't know they could do that." All the other Linuxers start looking at each other. "Why? What's with Linux #315?"
If it creates just Flash, and not standard DHTML, why does openlaszlo.org say "These applications run on all leading Web browsers on all leading desktop operating systems from a single XML code base."??
Flash doesn't really run well (if at all) on most Unices.
BTW: "OpenLaszlo applications are written in XML and JavaScript" -- no thanks. I could write Mozilla apps if I wanted to, but I don't want to code JavaScript. And I don't want XML either, no matter how much of a cool new fad it is.
I'm not saying that you need to redefine stuff in a running image, but if you ever do, Lisp can do that, too.
I agree to your Smalltalk criticisms. It's kind of old. However: why does the Smalltalk syntax "just not make any sense"? Who defines what makes sense and what doesn't? IMHO Smalltalk syntax is much saner that Java or Python.
Smalltalk isn't slow, either. Maybe some implementations are. Java was slow for a long time too (arguably still is), and only after millions of programmers took it up, making Java maybe the most popular language in the world, and years of development, is Java faster...
I didn't say anything was God's language. I was saying that C dialects seem stuck in the '70s. That's my opinion. You don't even argue or refute that. BTW I'm not liking Lisp because it's on/. I'm discussing here on/. in this article because I like Lisp.
The (lack of) success of a language doesn't prove anything, by the way. Or do you think that Java is the best language around, just because it gets more/. coverage and has more programmers?
The Scheme macro artice you mention however, isn't too good an introduction, IMHO.
Scheme macros are one reason why I'm learning Common Lisp now;)
Lisp (unlike Scheme) isn't really about that much functional programming; it has a powerful object system and uses macros (for iteration for instance) extensively.
Emacs Lisp uses dynamic scoping, which is the Wrong Thing. Scheme and Common Lisp are more modern and cleaned this up.
Guile is one Scheme implementation. There are many more. The book discussed here is about Common Lisp, which is much more powerful than Scheme, Emacs Lisp, Java or most other languages.
Common Lisp isn't too fragmented (unlike Scheme, which I suppose you mean). It was standardized in 1984, the first ANSI standard with an object system (that's way more powerful than Java, btw).
Both Lisp and Scheme have some libraries, but Lisp tends to have more of them. You might want to take a look at the book!
By all means do. Compared to Emacs Lisp (hehe) Common Lisp is much more powerful.
Re:Good book, questionable language.
on
Practical Common Lisp
·
· Score: 2, Informative
You are clueless.
Common Lisp is lexically scoped, in fact it has closures (unlike Java). 1984 it was standardized with a really powerful CLOS (CL object system), in comparison to the puny Java object system 10 years later.
Lisp also *invented* garbage collection! Without it you'd still be using FORTRAN, not Java.
And the AI comment is simply stupid. Any language can be used to do AI; Lisp is simply used for it, because it's way more powerful than Java and other crap.
ML and Haskell are totally different languages, just like Lisp and Scheme, or C++ and Objective C.
In fact O-C is popular, even though it's way slower than C++. It's much more elegant, more readable and more powerful.
Re:This is not a troll, but a query...
on
Practical Common Lisp
·
· Score: 4, Interesting
Really? Which functional language has macros?
Which functional language lets you redefine stuff in a running image?
Haskell might be good for real functional programming, but in Lisp you don't do that much functional programming.
I learnt Scheme first, actually, and I'm not that impressed, just like Haskell didn't impress me that much. Scheme is a stripped down Common Lisp with nicer syntax, a weird macro system, and an emphasis on functional programming.
And I don't see in what way Common Lisp has old warts, compared to FORTRAN or C...
It was standardized in 1984, with an object system. That's not too bad, IMHO, considering that Java is just a worse smalltalk-80 and that other C dialects similarly seems to be stuck with '70s technology, compared to Common Lisp.
Imagine being forced to download a bunch of information YOU DON'T WANT just because you want to get updates from the site. Optional? Fine. Whatever.
That's why with NNTP you get only the headlines first. If you want you can see the content (like the article description). If you like the description, click the http link in the description and read the article in your browser.
(a) It doesn't matter *who* wants the features. What matters is that they aren't there.
(b) The *point* of the article (as opposed to the NNTP example) is that you can get old messages. Indeed some message boards have not only an RSS feeds for new posts, but also an NNTP feed.
If there were apps for NNTP to do what RSS apps do with RSS feeds, then the "logging in to usenet" would be as invisible as "fetching the RSS feed".
(c) If you click on a RSS button, you register with the feed (you tell your RSS app to monitor the feed I suppose). With NNTP you tell your newsreader instead. With email you have your browser auto-enter you mail-address (if your browser has that feature) and click submit.
Of course, as you said, comparing NNTP with RSS isn't really optimal.
An alternative to a static RSS feed that gets updated with every new article would be an RSS feed that you access with a HTTP query with parameter to tell it when you last checked. That way the server could give you a list of all new articles dynamically, even though it'd be slighly less efficient.
It would still use the RSS format (which is really ok), but solve the asynchrony problem.
But why am I even replying to you? Just because one specific person wrote down some criticism (instead of the whole world doing that), the article is bullshit anyway!
Maybe RSS is not meant for viewing older content, but I see no reason why I shouldn't use a uniform interface for all news.
When I'm traveling over the weekend, I don't want to scan all news sites manually, just becaues RSS only provides for the newest 20 articles or so.
An alternative to a static RSS feed that gets updated with every new article would be an RSS feed that you access with a HTTP query with parameter to tell it when you last checked. That way the server could give you a list of all new articles dynamically, even though it'd be slighly less efficient.
RSS *as a format*, instead of RSS as current usage, would be very capable of that, and I'd gladly use it. I don't think it would be any less simple than current RSS.
Especially in C++ you would rather use string classes instead of char pointers, so buffer overflows easily disappear. People only have to make the jump, port their char pointer code to use better string functions.
Why do you think that C strings use 1k per string? A string uses as much memory as you allocate for it. I typically use an input buffer, and when a string is completely parsed, copy it somewhere. A string of size 30 uses 30 bytes and 4 bytes for its length (or alternatively 31b for the typical C string + padding).
Well, any number of parameters can go into a function, but only one can be returned. That's braindead. Python, ML, Lisp can all do better.
If the parameters and return values are passed in registers or RAM (e.g. on the register-starved x86) is irrelevant to the question (just an implementation detail).
Slashdot Mirror
Would that be an alternative to /. ?
Gotta love HTML 3.2...
So you think that [obj keyword: param1 keyword: param2] is stupid and ugly. Well, that's your opinion. I don't want to tell you here how stupid and ugly I think your mom is, because that wouldn't be appropriate.
Let's just say, that I prefer one of the following two code samples:If you nest expressions it's even nicer:Let's not argue about speed here, since most languages (Lisp and Java anyway) are fast enough for most applications.
Lisp isn't stuck in the '50s. It was invented AFAIK end of the '50s, evolved over more than two decades, in the form of different dialects, was then in 1984 merged into Common Lisp (with all the good and important features), got the first ANSI-standardized object system.
Lisp contains features from the '80s. Java's most advanced features are probably exception handling, object orientation, and garbage collection. The latter one *originated* in Lisp, and OO was implemented almost as in Java in Smalltalk 20 years before. I don't know about exceptions, but AFAIK '84 Lisp also can do that (I'm still somewhat new to the Lisp world).
It's funny that you say that everthing about Lisp except the syntax sucks: so object orientation is bad? A uniform data representation that is basically compressed XML is bad? Packages are bad? Keyword arguments are worse than Java's function overloading? Macros are bad? LOOP and FORMAT are bad? Do you even remotely know what you're talking about?
Sure, CAR and CDR derive from Lisp machine register names. So what? You can combine then to CADAR and CDDR, for instance. You can use FIRST and REST, if you prefer.
So Java is "the best programming language". Cool. Only that I don't think it's easy to grasp for computer newbies, which I happened to tutor. Java is unnecessarily complicated with its public static final stuff. All the ().[],;{} distract from what you actually want the program to do and makes nested expressions look dead ugly.
It has good APIs, but that's about all. You have to typecast everything out of containers, which defeats the purpose of static checking. Even cooler are boxing and unboxing of ints to Integers, so you can actually put them in Hashtables. Sure, this all changes in Java 5, but Java had these problems for 10 years! I'm not alone in saying that Java gets almost nothing right. *I* prefer coding in C, because I'm more productive there (less errors, less time spent debugging, better debugging facilities).
And that you can do macro stuff in Java: I doubt it! Can you easily integrate a yacc-parser in your Java program? How about alternative control structures, loops? C++ templates are unreadable and don't remotely match Lisp macros in power and elegance.
A group of Linux fans are sitting in a bar. One says, "I really like the performance of Linux #180." The next one replies "yes, but Linux #87's package system is better." Another one says "I really like Linux #298's lightweightness."
One guy says "What do you think about Linux #315's features?" One programmer starts looking totally impressed, stutters "wow, is thaaaat cool" and "man, I didn't know they could do that." All the other Linuxers start looking at each other. "Why? What's with Linux #315?"
"I didn't know that one yet."
If it creates just Flash, and not standard DHTML, why does openlaszlo.org say "These applications run on all leading Web browsers on all leading desktop operating systems from a single XML code base."??
Flash doesn't really run well (if at all) on most Unices.
BTW: "OpenLaszlo applications are written in XML and JavaScript" -- no thanks. I could write Mozilla apps if I wanted to, but I don't want to code JavaScript. And I don't want XML either, no matter how much of a cool new fad it is.
No, the successor of the Formula Translator should be something like Formula Solver, Formula Writer, or Formula Inventor.
;)
Everything else is just more of the same
No, it will be EVEN BETTER than Java ;)
Yes, I'm not impressed either, although Guy Steele is a really capable guy. Let's wait and see what will come out of this.
Doing enough bad things to it that I still use C in preference?
So many people will go on coding Fortran.
I'm not saying that you need to redefine stuff in a running image, but if you ever do, Lisp can do that, too.
/. I'm discussing here on /. in this article because I like Lisp.
/. coverage and has more programmers?
I agree to your Smalltalk criticisms. It's kind of old. However: why does the Smalltalk syntax "just not make any sense"? Who defines what makes sense and what doesn't? IMHO Smalltalk syntax is much saner that Java or Python.
Smalltalk isn't slow, either. Maybe some implementations are. Java was slow for a long time too (arguably still is), and only after millions of programmers took it up, making Java maybe the most popular language in the world, and years of development, is Java faster...
I didn't say anything was God's language. I was saying that C dialects seem stuck in the '70s. That's my opinion. You don't even argue or refute that. BTW I'm not liking Lisp because it's on
The (lack of) success of a language doesn't prove anything, by the way. Or do you think that Java is the best language around, just because it gets more
Eat shit. Millions of flies can't be wrong!
Thanks for the information, but still I find Lisp's syntax and the combination of all its features the most convenient...
I did some coding in SML, but in the end I prefer imperative programming. Maybe I'll take another look at FP later.
Don't worry. Just wanted to clear things up...
;)
The Scheme macro artice you mention however, isn't too good an introduction, IMHO.
Scheme macros are one reason why I'm learning Common Lisp now
Lisp (unlike Scheme) isn't really about that much functional programming; it has a powerful object system and uses macros (for iteration for instance) extensively.
Oh, didn't know that.
Thanks.
You think Lisp and Scheme are out of ordinary??
What about Haskell, Forth, Self?
Or are you just rambling about all the parentheses?
I strongly suggest you go look at a *real* Lisp program, such as the examples in the discussed book.
Emacs Lisp uses dynamic scoping, which is the Wrong Thing. Scheme and Common Lisp are more modern and cleaned this up.
Guile is one Scheme implementation. There are many more. The book discussed here is about Common Lisp, which is much more powerful than Scheme, Emacs Lisp, Java or most other languages.
Common Lisp isn't too fragmented (unlike Scheme, which I suppose you mean). It was standardized in 1984, the first ANSI standard with an object system (that's way more powerful than Java, btw).
Both Lisp and Scheme have some libraries, but Lisp tends to have more of them. You might want to take a look at the book!
By all means do. Compared to Emacs Lisp (hehe) Common Lisp is much more powerful.
You are clueless.
Common Lisp is lexically scoped, in fact it has closures (unlike Java). 1984 it was standardized with a really powerful CLOS (CL object system), in comparison to the puny Java object system 10 years later.
Lisp also *invented* garbage collection! Without it you'd still be using FORTRAN, not Java.
And the AI comment is simply stupid. Any language can be used to do AI; Lisp is simply used for it, because it's way more powerful than Java and other crap.
What are you trying to say?
There are fast Scheme implementation too.
ML and Haskell are totally different languages, just like Lisp and Scheme, or C++ and Objective C.
In fact O-C is popular, even though it's way slower than C++. It's much more elegant, more readable and more powerful.
Really? Which functional language has macros?
Which functional language lets you redefine stuff in a running image?
Haskell might be good for real functional programming, but in Lisp you don't do that much functional programming.
I learnt Scheme first, actually, and I'm not that impressed, just like Haskell didn't impress me that much. Scheme is a stripped down Common Lisp with nicer syntax, a weird macro system, and an emphasis on functional programming.
And I don't see in what way Common Lisp has old warts, compared to FORTRAN or C...
It was standardized in 1984, with an object system. That's not too bad, IMHO, considering that Java is just a worse smalltalk-80 and that other C dialects similarly seems to be stuck with '70s technology, compared to Common Lisp.
Is the Redmond unemployment rate for engineers above the national average for engineers?
Or is the engineer unemployment rate higher than the non-engineer unemployment rate, nationwide?
Or is the local engineer unemployment rate higher than the national unemployment average (for non- and engineers)?
The world will probably never know.
No, they used the new, hot Software Engineering technique: Bug-free Software Engineering!
But I agree, it's sad that any company makes that claim.
Imagine being forced to download a bunch of information YOU DON'T WANT just because you want to get updates from the site. Optional? Fine. Whatever.
That's why with NNTP you get only the headlines first. If you want you can see the content (like the article description). If you like the description, click the http link in the description and read the article in your browser.
(a) It doesn't matter *who* wants the features. What matters is that they aren't there.
(b) The *point* of the article (as opposed to the NNTP example) is that you can get old messages. Indeed some message boards have not only an RSS feeds for new posts, but also an NNTP feed.
If there were apps for NNTP to do what RSS apps do with RSS feeds, then the "logging in to usenet" would be as invisible as "fetching the RSS feed".
(c) If you click on a RSS button, you register with the feed (you tell your RSS app to monitor the feed I suppose). With NNTP you tell your newsreader instead. With email you have your browser auto-enter you mail-address (if your browser has that feature) and click submit.
Of course, as you said, comparing NNTP with RSS isn't really optimal.
An alternative to a static RSS feed that gets updated with every new article would be an RSS feed that you access with a HTTP query with parameter to tell it when you last checked. That way the server could give you a list of all new articles dynamically, even though it'd be slighly less efficient.
It would still use the RSS format (which is really ok), but solve the asynchrony problem.
But why am I even replying to you? Just because one specific person wrote down some criticism (instead of the whole world doing that), the article is bullshit anyway!
Maybe RSS is not meant for viewing older content, but I see no reason why I shouldn't use a uniform interface for all news.
When I'm traveling over the weekend, I don't want to scan all news sites manually, just becaues RSS only provides for the newest 20 articles or so.
An alternative to a static RSS feed that gets updated with every new article would be an RSS feed that you access with a HTTP query with parameter to tell it when you last checked. That way the server could give you a list of all new articles dynamically, even though it'd be slighly less efficient.
RSS *as a format*, instead of RSS as current usage, would be very capable of that, and I'd gladly use it. I don't think it would be any less simple than current RSS.
I don't want to duplicate the article, so here's the link.
Especially in C++ you would rather use string classes instead of char pointers, so buffer overflows easily disappear. People only have to make the jump, port their char pointer code to use better string functions.
Why do you think that C strings use 1k per string? A string uses as much memory as you allocate for it. I typically use an input buffer, and when a string is completely parsed, copy it somewhere. A string of size 30 uses 30 bytes and 4 bytes for its length (or alternatively 31b for the typical C string + padding).
Well, any number of parameters can go into a function, but only one can be returned. That's braindead. Python, ML, Lisp can all do better.
If the parameters and return values are passed in registers or RAM (e.g. on the register-starved x86) is irrelevant to the question (just an implementation detail).