Slashdot Mirror
Wolfram Language Demo Impresses
theodp writes "The devil will be in the details, but if you were stoked about last November's announcement of the Wolfram programming language, you'll be pleased to know that a just-released dry-but-insanely-great demo delivered by Stephen Wolfram does not disappoint. Even if you're not in love with the syntax or are a FOSS devotee, you'll find it hard not to be impressed by Wolfram's 4-line solution to a traveling salesman tour of the capitals of Western Europe, 6-line camera-capture-to-image-manipulation demo, or 2-line web crawling and data visualization example. And that's just for starters. So, start your Raspberry Pi engines, kids!"
A picture is worth a thousand words, but most sets of 1000 words can't be succinctly described by a single picture.
Similarly, while I'm sure that you can write a few lines of Wolfram and do amazing things, I wonder how often you can set out to do an amazing thing and end up with a few lines of Wolfram. Maybe the answer is "pretty often", which would be wonderful. But I'm waiting to hear from some outside users.
So we don't count the lines of code behind the "FindShortestTour" function?
Did you solve the secrets of the Universe? Stephen Wolfram did. All is known because of Him, even though you could not hope to grasp what He knows. Show more respect and reverence.
It looks like a very nice library.
Doesn't really say very much about the power of the language at all, though.
As much as I would like to be impressed, what I see is quite underwhelming: a functional application language with some interface to "facts" and "databases" with a pattern matching engine might make some analysis easier but ... the principles of the language are mostly what you come to expect if you have seen lisp once or any modern functional language,e.g. haskell.
I can see it as being useful, but as another commenter pointed out, "FindShortestTour" is a library function (which might be handy), but definitely not an example of how concise the language might be; the same could be said about "EdgeDetect" or the like. The power of the language can be measured in how easily it can be extended or non trivial algorithms can be implemented ... not in how many functions are offered (even if this could be more convenient none-the-less).
It looks like Mathematica because Mathematica implements this language and is where it comes from. Historically the language developed ad hoc and now they have made an effort to standardize it into a "language".
W..w..W - Willy Waterloo washes Warren Wiggins who is washing Waldo Woo.
I first looked at the examples given in the article and said "harrumph this is mathematica". But then I watched his demo and I see what he's getting at. You could say this is just a really nice library but it's way more than the sum of it's parts. I think he's using the term Language not in the sense of "programming language", but rather in the sense that every real world speaking language like english, spanish contains an intrinsic model of the world itself and every part of a spoken language can be coupled to every other part. That is speaking has no incompatible interface between ideas does it? That question would never occur to you, but of course we have that problem with every programming library API.
SO he's talking about a Language for programming as much as a programming language. His accomplishment is to make a language of programming a programming language.
One of the great tricks he accomplishes is to combine symbolic programming and functional programming. I was somewhat surprised to notice that reactive programming actually falls out of that by accident. There's been a lot of spamvertising articles on Slashdot lately about the dogs dinner versions of Reactive programming for databases. Those are toys. Wolfram gets it right by not making it just fall out accidentally of two greater programming principles.
Decades ago I toyed with mathematica. The problem I had with is it was that the symbolics were nice but they let you easily create problems with permutations so large that it became incomprehensibly slow as your problem scaled. This if course was the users fault. I'm just saying that the power of the language gave me the power to be stupid. In a similar way APL with it's outerproducts instead of loops could easily use up all your computer memory in one command line without you even appreciating what had just happened. With procedural languages you had to think about how your algorithm was going to manage its own complexity and thus oddly worked better for scaling to complex problems.
It looks like what has happened is that mathematica --- now wolfram language--- has a lot more speed and wisdom about how to manage complexity and choose more wise approaches. SO perhaps that problem is solved more. But it's hard to say from the demo.
In any case that was a staggering demo.
Some drink at the fountain of knowledge. Others just gargle.
Am I supposed to be impressed by "a 4-line solution to a traveling salesman tour" when that 4 line solution calls a library function called "FindShortestTour()"?
That might be useful if your name is Martin Gardner, but...
No sig today...
Cramming 20 commands and 8 layers of brackets into one line doesn't make your programm an 'impressive 5-liner'. It, at most, makes a neat stunt by a mathematician in a proprietary programming language he invented himself. I'd be tempted to call it shitty programming.
Nothing to see here folks, move along.
No you miss the point. It shows that two things have been accomplished
first every command has an almost universal API for input and output letting you pipeline everything you do. try that with almost any normal library. it fails. now imagine achieving that across a language that is staggeringl comprehensive, deep and wide. it's a tour de force.
then imagine someone told you that, by the way, that API was also symbolic.
and wait it's also a functional programing
and reactive.
Some drink at the fountain of knowledge. Others just gargle.
You should watch the demo. At one point he enters a natural language expression "Show a blue dodecahedron and two red spheres" which pops up a shaded 3D image model of just that.
Some drink at the fountain of knowledge. Others just gargle.
The traveling salesman tour - is NOT a 4 line solution. By that definition I can write "Run Linux" and have a one line operating system.
Honestly, seeing that much power in a demo makes the hair on the back of my neck rise (and in the 'something vile beyond comprehension this way comes' sort of way, not the 'awe at technology indistinguishable from magic' kind of way).
If you can do extremely complex and powerful things with very, very, short commands, that suggests that all those commands have a lot of internal magic baked in, quite possibly including some might-as-well-be-nondeterministic guessing to paper over any ambiguity in commands, or in output from one command moving to be input for another.
In the context of a demo, where you can carefully test, and confine yourself to some highlights from the set of programs that are both cool and well behaved, fantastic. In the context of taking the language out into the wild, that sounds like every nightmare interaction with an unpredictable and opaque 3rd-party library that you'll never expunge from your nightmares....
and webscale ?
Ah, so it's like Lotus Notes, then.
Faster! Faster! Faster would be better!
"Show a pink cylinder with a purple sphere at the end with two pink spheres."
But that's not your job. You're not developing the language. You're just asking it to do things. Submit a bug report and pray to god that they fix it before you need to finish the project that you stupidly designed around this nondeterministic, amorphous pile of somewhat awesome but perhaps totally useless functions.
ftfy
captcha: formulas
The way this is setup isn't that that you code everything in natural language, rather it is just a shortcut to look up the correct formal language. Instead of searching/browsing documentation looking up the exact names of the functions you want and how to chain them, you just type what you want in natural language. If it interpreted you correctly, then great it saved you several minutes, and now you know the real syntax to use in the future. If not, well you only lost a couple seconds.
The idea of mixing natural language like this isn't so weird; the first step that most programmers would take in looking up documentation when they don't even know the name of the library the functionality is located in is to perform a natural language search on web browser, and then go from there. This just takes it one step further and streamlines the process, which is perfect for a interactive language.