Posted by Sam on Aug 18, 2008 at 12:00 AM UTC - 6 hrs
Daniel Spiewak asks, "how do you apply polyglotism?"
To my knowledge, this sort of policy is fairly common in the industry. Companies (particularly those employing consultants) seem to prefer to keep the technologies employed to a minimum, focusing on the least-common denominator so as to reduce the requirements for incoming developer skill sets.
We're afraid of being eaten by the poly-headed polyglot monster.
Google avoids it by sticking to a limited set of languages. I don't work at Google, so I can't claim to know what their architecture looks like in terms of blending languages within applications. But the act of limiting languages does not in itself preclude polyglot programming. Choosing to limit yourself to one language, or one language per application, would do so. Likewise, choosing C# and Java as your languages would probably nullify most of the benefits. But the point is, just because you allow or encourage polyglot programming does not mean you let everyone choose whatever language suits them and then throw it all together in a melting pot, praying that everything works well together. You can have a method to sort out the madness.
In any case, what's going on with polyglotism, and how do we implement it?
What's going on
Back in 2002, Bertrand Meyer explained why the pursuit of polyglot programming is worthwhile, even though no one was using it
Let's start with the impolite question: Should one really care about multilanguage support? When this feature was announced at .NET's July 2000 debut, Microsoft's competitors sneered that it wasn't anything anyone needed. I've heard multilanguage development dismissed, or at least questioned, on the argument that most projects simply choose one language and stay with it. But that argument doesn't really address the issue. For one thing, it sounds too much like asserting, from personal observation, that people in Singapore don't like skiing. Lack of opportunity doesn't imply lack of desire or need. Before .NET, the effort required to interface modules from multiple languages was enough to make many people stick to just one; but, with an easy way to combine languages seamlessly and effortlessly, they may -- as early experience with .NET suggests -- start to appreciate their newfound freedom to mix and match languages.
Fast forward to the end of 2006, when Neal Ford was "beginning to see a time where even the core language (the one that gets translated to byte code) will cease its monoculture." And he sees it as more than the web's polyglot nature:
Applications of the future will take advantage of the polyglot nature of the language world. We have 2 primary platforms for "enterprise" development: .NET and Java. There are now lots of languages that target those platforms. We should embrace this idea. While it will make some chores more difficult (like debugging), it makes others trivially easy (or at least easier). It's all about choosing the right tool for the job and leveraging it correctly. Pervasive testing helps the debugging problem (adamant test-driven development folks spend much less time in the debugger). SQL, Ajax, and XML are just the beginning. Increasingly, as I've written before, we're going to start adding domain specific languages. The times of writing an application in a single general purpose language is over. Polyglot programming is a subject I'm going to speak about a lot next year.
Since Meyer's article, we have indeed seen an explosion of languages running on both "enterprise" platforms. Interest seems to have really grown recently. Because of that, it's getting easier to integrate languages. But as Neal Ford noted, we're still in the beginning (if you believe that paradigm is going to take hold).
How We Might Implement It
Like Neal Ford, Ola Bini seems to think polyglot programming is more than just the different technologies involved in a web application. He talks about how polyglotism might be implemented in layers within an application:
The first layer is what I called the stable layer. It's not a very large part of the application in terms of functionality. But it's the part that everything else builds on top off, and is as such a very important part of it. This layer is the layer where static type safety will really help. ...
Let's talk briefly about a web application, as I envision it using Ola's layers as a guide:
The second layer is the dynamic layer. This is where maybe half the application code resides. The language types here are predominantly dynamic, strongly typed languages running on the JVM, like JRuby, Rhino and Jython.
The third layer is the domain layer. It should be implemented in DSL's, one or many depending on the needs of the system. In most cases it's probably enough to implement it as an internal DSL within the dynamic layer, and in those cases the second and third layer are not as easily distinguishable. But in some cases it's warranted to have an external DSL that can be interacted with. A typical example might be something like a rules engine (like Drools).
In the stable layer, you might be talking about a banking application where transactions are performed in Java. You might have some tool set that performs important calculations, and which other applications depend on. It might be the parts of the application that really need to perform well.
In the dynamic layer, you might have code that integrates the static and domain layers. You might have code which glues together other stuff, or things which don't really turn into bottlenecks for the application, which aren't necessarily part of the application domain, but which don't require the ceremony of the stable layer. You're choosing to use a dynamic scripting language to increase productivity. You might define the DSL abstractions here, and use them in the DSL/domain layer.
In the domain layer is where you'll see the meat of your application code- it's where all your business object abstractions will reside, and they will in turn depend on the other two layers. It should consist entirely of code that uses business language.
Ola may have meant something else entirely, or just slightly different, but that's the way I've come to see it. The concept is still new, so there's nothing saying that vision of a "best practice" design will win out over ad hoc solutions that throw modules written in different languages together to form an application.
As far as low-level implementation specifics go, Ola Bini shared how he integrated Erlang and JRuby. Sean Corfield showed PHP and Ruby running within ColdFusion. Charlie Nutter has in-line Java for JRuby and of course there's in-line C/C++ for Ruby from Ryan Davis. Of course, most languages running on the JVM and/or .NET are fairly trivial to integrate with the libraries available to those platforms.
How I'm Using Polyglot Programming
In one of my jobs, we do a lot of work with Sharepoint. In fact, the bulk of the applications are built in Sharepoint: to manage documents, content, data and get all the integration with other Office products. However, Sharepoint is a beast. Often, doing something that would be trivial on another platform is an arduous task in Sharepoint. So we have to drop out of it.
In some cases, when we've needed the feature to reside within Sharepoint itself, we've opted to use C# within ASP.NET to create Sharepoint Web Parts. For the next one of these I come up against, I intend to look at using IronRuby and deriving from the appropriate parent classes.
In other cases, we're building stand-alone "sub-applications" that integrate with Sharepoint and Active Directory. For these, we have some utility classes written in C#, while the "sub-application" is built in Ruby on Rails. Soon, we'll be looking at running them on IronRuby, and taking further advantage of the .NET platform.
In my job with the bioinformatics lab at the University of Houston, things are a bit different. I'm building command line applications. After a machine takes photos of your DNA sample, there is an genome analysis pipeline from Illumina that analyzes the images, produces nucleotide sequences, and tries to align them to a reference sequence. That utilizes Perl, Python, and C/C++, though I don't know to what extent a in which "layers" each language does its job.
On my end, I'm using C++ to do the heavy lifting in analysis in proprietary libraries we've been producing to do our own analysis and post-analysis. When you're dealing with genome-size data sets, from 5 mega- to 3 giga-bases, even constant reductions in run-time can mean the difference of days of computation in algorithms with above-linear time complexity. But I'm also building scripts in Ruby that help piece it all together, and doing some analysis with it for linear-time algorithms. It's very ad hoc, in this instance.
Finally, my game development group plans to use C/C++ to forge our engine, while using a scripting language(s) to fill out the game details. As I understand it, much of the game industry already employs polyglot programming in this manner.
So that's how I'm doing it at the moment.
In 2008, the polyglot programming paradigm is still (at least in my opinion), a relatively novel concept. There aren't a lot of resources to show us how to do it the right way. As time moves forward and people publish their experiences, that should be expected to change.
On that note, do you have any stories about how you've done it, or any ideas on how you would if you were going to do it? Please feel free to share them - you'd be doing us all a favor.
Hey! Why don't you make your life easier and subscribe to the full post
or short blurb RSS feed? I'm so confident you'll love my smelly pasta plate
wisdom that I'm offering a no-strings-attached, lifetime money back guarantee!
Leave a comment
There are no comments for this entry yet.
Leave a comment