Comments on: Erlang: A New Way to Program That’s 20 Years Old http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/ Tracking the Internet Evolution Wed, 09 Jul 2008 09:44:53 +0000 http://wordpress.org/?v=MU By: Multicore’s Not-So-Secret Problem - GigaOM http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-884474 Multicore’s Not-So-Secret Problem - GigaOM Thu, 19 Jun 2008 15:46:05 +0000 http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-884474 [...] are pushing Erlang as a potential solution to parallel programming, while those in the supercomputing industry are [...] [...] are pushing Erlang as a potential solution to parallel programming, while those in the supercomputing industry are [...]

]]> By: bad_article_ http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-863918 bad_article_ Mon, 03 Mar 2008 18:04:49 +0000 http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-863918 <p>"...He said it largely remains the same, but expressed a wish for Erlang’s concurrency model — lightweight threads that use message passing instead of shared memory — to be incorporated into his favorite modern languages, such as Java and Ruby..."</p> <p>So has he ever written code in a functional programming language or just imperative ones? which would be why he finds it "wierd". This guys is a flake.</p> “…He said it largely remains the same, but expressed a wish for Erlang’s concurrency model — lightweight threads that use message passing instead of shared memory — to be incorporated into his favorite modern languages, such as Java and Ruby…”

So has he ever written code in a functional programming language or just imperative ones? which would be why he finds it “wierd”. This guys is a flake.

]]> By: Web Worker Daily » Archive Open Thread: What’s Your Favorite Programming Language? « http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-848379 Web Worker Daily » Archive Open Thread: What’s Your Favorite Programming Language? « Sun, 13 Jan 2008 18:12:21 +0000 http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-848379 <p>[...] Meanwhile, Ruby on Rails doesn’t seem quite so hot this year as it was last January, Scala’s getting some laughs, and people have been wondering why Erlang’s so buzzy. [...]</p> [...] Meanwhile, Ruby on Rails doesn’t seem quite so hot this year as it was last January, Scala’s getting some laughs, and people have been wondering why Erlang’s so buzzy. [...]

]]> By: Milo http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-838797 Milo Wed, 09 Jan 2008 04:47:34 +0000 http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-838797 <p>Jon: "I take it you never heard MPI. It is only been taught in every super computer course since the late 90’s."</p> <p>I take it you are not a native English speaker. That's alright. In any event, I am a graduate student at a top 10 CS program. (I will assume it is alright to consider this a "super computer course".) We cover how to parallelize programs in the context of algorithms, but we don't necessarily use, or even teach, MPI per-say. I like Ocaml, and FP in general, but Erlang offers a new model for concurrency that <em>has been battle tested</em>. The latter is key!</p> Jon: “I take it you never heard MPI. It is only been taught in every super computer course since the late 90’s.”

I take it you are not a native English speaker. That’s alright. In any event, I am a graduate student at a top 10 CS program. (I will assume it is alright to consider this a “super computer course”.) We cover how to parallelize programs in the context of algorithms, but we don’t necessarily use, or even teach, MPI per-say. I like Ocaml, and FP in general, but Erlang offers a new model for concurrency that has been battle tested. The latter is key!

]]> By: Top Posts « WordPress.com http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-791613 Top Posts « WordPress.com Sat, 22 Dec 2007 00:02:08 +0000 http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-791613 <p>[...] Erlang: A New Way to Program That’s 20 Years Old Geeks everywhere got excited recently when they heard that SimpleDB might be based on Erlang. Why? Is Erlang the next […] [...]</p> [...] Erlang: A New Way to Program That’s 20 Years Old Geeks everywhere got excited recently when they heard that SimpleDB might be based on Erlang. Why? Is Erlang the next […] [...]

]]> By: Anne Zelenka http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-791472 Anne Zelenka Fri, 21 Dec 2007 22:43:08 +0000 http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-791472 <p>@Jonathan Allen: I didn't say C++ and Java <i>always</i> use shared memory for concurrency, I said <i>usually</i>.</p> @Jonathan Allen: I didn’t say C++ and Java always use shared memory for concurrency, I said usually.

]]> By: projectshave http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-790567 projectshave Fri, 21 Dec 2007 15:32:25 +0000 http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-790567 <p>Microsoft Research has an extension of C# called Sing# that uses message passing as their core concurrency technique. They used it to implement their research OS called Singularity. I can't think of anything in Erlang that can't be duplicated by a library in Java/.NET. The message passing part is easy. The reliability "features" fall out from their design techniques. I couldn't find anything in the VM that was superior to JVM/.NET.</p> <p>Check it out: http://research.microsoft.com/os/singularity/</p> Microsoft Research has an extension of C# called Sing# that uses message passing as their core concurrency technique. They used it to implement their research OS called Singularity. I can’t think of anything in Erlang that can’t be duplicated by a library in Java/.NET. The message passing part is easy. The reliability “features” fall out from their design techniques. I couldn’t find anything in the VM that was superior to JVM/.NET.

Check it out: http://research.microsoft.com/os/singularity/

]]> By: Jonathan Allen http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-789339 Jonathan Allen Fri, 21 Dec 2007 03:17:08 +0000 http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-789339 <blockquote> <p>Languages like C++ and Java work mainly by using sequential processing. They have the capability to run multiple pathways of execution at the same time — allowing computer instructions to run in parallel — but in order to do so, they usually use something called “shared memory.”</p> </blockquote> <p>I take it you never heard MPI. It is only been taught in every super computer course since the late 90's.</p> <p>http://en.wikipedia.org/wiki/Message_Passing_Interface</p>

Languages like C++ and Java work mainly by using sequential processing. They have the capability to run multiple pathways of execution at the same time — allowing computer instructions to run in parallel — but in order to do so, they usually use something called “shared memory.”

I take it you never heard MPI. It is only been taught in every super computer course since the late 90’s.

http://en.wikipedia.org/wiki/Message_Passing_Interface

]]> By: Bob Ippolito http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-788210 Bob Ippolito Thu, 20 Dec 2007 17:29:10 +0000 http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-788210 <p>Erlang works extremely well for us at Mochi Media, MochiAds has been using it from the beginning and MochiBot was ported to Erlang because it was so much easier to maintain and scale than the original Python implementation.</p> <p>The wide finder problem is kind of a red herring, Erlang works best with network I/O. It was designed for telecom, not batch processing from disk. There are of course efficient (and obtuse) ways to do disk I/O with Erlang, but the standard library doesn't do a very good job of it. On the flip side if you take Python (or Java or Ruby or damn near anything else) and try and make it do networking efficiently out of the box you'll lose in a similarly spectacular fashion.</p> <p>If any of you dying to learn Erlang, we're hiring ;)</p> Erlang works extremely well for us at Mochi Media, MochiAds has been using it from the beginning and MochiBot was ported to Erlang because it was so much easier to maintain and scale than the original Python implementation.

The wide finder problem is kind of a red herring, Erlang works best with network I/O. It was designed for telecom, not batch processing from disk. There are of course efficient (and obtuse) ways to do disk I/O with Erlang, but the standard library doesn’t do a very good job of it. On the flip side if you take Python (or Java or Ruby or damn near anything else) and try and make it do networking efficiently out of the box you’ll lose in a similarly spectacular fashion.

If any of you dying to learn Erlang, we’re hiring ;)

]]> By: Joe http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-788193 Joe Thu, 20 Dec 2007 17:23:57 +0000 http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-788193 <p>Erlang is probably the least academic language in existence. In fact, it is the <em>only</em> language that was created with the purpose of building scalable, fault tolerant systems in mind. Whichever language you're using now is probably far more academic than Erlang.</p> Erlang is probably the least academic language in existence. In fact, it is the only language that was created with the purpose of building scalable, fault tolerant systems in mind. Whichever language you’re using now is probably far more academic than Erlang.

]]> By: Jeffrey http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-788078 Jeffrey Thu, 20 Dec 2007 16:19:30 +0000 http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-788078 <p>When I need to do multithreading, it's a problem. Then an academic computer scientist says I have to learn an obscure new programming language to take advantage of it. Now I have two problems.</p> When I need to do multithreading, it’s a problem. Then an academic computer scientist says I have to learn an obscure new programming language to take advantage of it. Now I have two problems.

]]> By: Anne Zelenka http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-787989 Anne Zelenka Thu, 20 Dec 2007 15:33:28 +0000 http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-787989 <p>@Eddie: Relevant to whom? If you already knew about Erlang and what it offered and why people get excited about it -- maybe not. But not everyone who reads tech blogs is a programmer who stays up on the latest programming language buzz.</p> @Eddie: Relevant to whom? If you already knew about Erlang and what it offered and why people get excited about it — maybe not. But not everyone who reads tech blogs is a programmer who stays up on the latest programming language buzz.

]]> By: Eddie http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-787975 Eddie Thu, 20 Dec 2007 15:25:24 +0000 http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-787975 <blockquote> <p>The other language environment (Ruby, Python, Perl) are years away of what Erlang does offer.</p> </blockquote> <p>Who cares. Horses for courses. Ruby / Python / Perl were not designed with the same end goals as Erlang.</p> <p>I wonder if this posting about Erlang on GigaOm was really relevant?</p>

The other language environment (Ruby, Python, Perl)
are years away of what Erlang does offer.

Who cares. Horses for courses. Ruby / Python / Perl were not designed with the same end goals as Erlang.

I wonder if this posting about Erlang on GigaOm was really relevant?

]]> By: Anne Zelenka http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-787956 Anne Zelenka Thu, 20 Dec 2007 15:16:10 +0000 http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-787956 <p>@Mickael: I think what Tim meant by the language shows its age is that it's lacking some capabilities he finds indispensable for his purposes, such as regular expression support. Depending on what you're trying to do, it might or might not look old-fashioned.</p> <p>@wavell: I'd agree that the root of Erlang's "weirdness" is that it's a functional language -- immutable variables and use of tail recursion might look strange if you're coming from C or Java, for example, even if you learned about such things in a CS class.</p> <p>And yes, Erlang seems like a good way to learn not just about a different way of concurrent programming but also about functional programming (which itself is the basis for that different way of achieving concurrency).</p> <p>Very interesting that Matz is most interested in Erlang.</p> @Mickael: I think what Tim meant by the language shows its age is that it’s lacking some capabilities he finds indispensable for his purposes, such as regular expression support. Depending on what you’re trying to do, it might or might not look old-fashioned.

@wavell: I’d agree that the root of Erlang’s “weirdness” is that it’s a functional language — immutable variables and use of tail recursion might look strange if you’re coming from C or Java, for example, even if you learned about such things in a CS class.

And yes, Erlang seems like a good way to learn not just about a different way of concurrent programming but also about functional programming (which itself is the basis for that different way of achieving concurrency).

Very interesting that Matz is most interested in Erlang.

]]> By: wavell watson http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-787914 wavell watson Thu, 20 Dec 2007 14:52:46 +0000 http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-787914 <p>I'd have to say that Erlang is only 'weird' if you aren't familiar with functional programming languages. The 'weirdness' that you may be talking about is probably the immutable state of variables. The trade off is you lose most if not all side effects when you lose mutability. This can be a tremendously Good Thing when trying to debug because your code becomes extremely easy to write tests for. In any case, Erlang is a great way to become become familiar with functional programming if nothing else. BTW for the ruby lovers at rubyconf this year Matz (the creator of ruby) said the language he is most interested in right now (as far as stealing functionality from) is Erlang.</p> I’d have to say that Erlang is only ‘weird’ if you aren’t familiar with functional programming languages. The ‘weirdness’ that you may be talking about is probably the immutable state of variables. The trade off is you lose most if not all side effects when you lose mutability. This can be a tremendously Good Thing when trying to debug because your code becomes extremely easy to write tests for. In any case, Erlang is a great way to become become familiar with functional programming if nothing else. BTW for the ruby lovers at rubyconf this year Matz (the creator of ruby) said the language he is most interested in right now (as far as stealing functionality from) is Erlang.

]]> By: Bhasker V Kode http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-787896 Bhasker V Kode Thu, 20 Dec 2007 14:41:44 +0000 http://gigaom.com/2007/12/19/erlang-a-new-way-to-program-thats-20-years-old/#comment-787896 <p>Erlang's late entry is best explained by Joe Armstrong's threads : "Who cares if Erlang starts slowly - it was designed to start once and never stop - we have systems that have run for 5 years - a two seconds start-up time amortized over 5 years is not <em>too</em> bad. "</p> <p>Kepe Clicking, Bosky</p> <p>PS : btw, we're using erlang as well...</p> Erlang’s late entry is best explained by Joe Armstrong’s threads :
“Who cares if Erlang starts slowly - it was designed to start once and never stop - we have systems that have run for 5 years - a two seconds start-up time amortized over 5 years is not too bad. “

Kepe Clicking,
Bosky

PS : btw, we’re using erlang as well…

]]>