JRuby: What I Need to Know

I was recently asked, "As a developer, what kind of information do you need to be productive with JRuby?" Here are the thoughts that come to mind in
response to that question.

As always, when it comes to information delivery, the vital first step is to determine your audience. When it comes to JRuby, there are several categories, each with differing needs:

  1. Non-programmers who are learning JRuby as their first language.
    This group needs all the basics of computing, from algorithms to compilers, from the ground up. (That audience is generally served by schools, but if someone has a logical mind, it's fun to watch them learn how to program.)

  2. Java programmers who are coming at JRuby for the first time
    This group needs to know first, how a Java program would be translated into Ruby, so they learn the syntax differences. Then they need to know the idioms that make Ruby special. And they need to know how to access the Java classes they're so used to.

  3. Ruby programmers who want take advantages of the JVM.
    Ease of install, threading, and scalability probably top their interest list. But after answering the question "why do I want this", they'll be most interested in finding out which Java classes they really want to use, and how to use them to maximum advantage.

  4. Programmers of other languages.
    This is obviously a diverse group. A lot depends on which language(s) they're coming from. They won't need the basics that a non-programmer needs, but they'll need to find out how JRuby works, compared to whatever language they're coming from. So in reality, this is a catch-all category for a variety of groups. The most reasonable approach is for them to pick and choose from the material that is targeted at the other groups.

  5. Web developers who want to build and deploy web applications using JRuby on Rails
    This group needs to know how to set up a web server and deploy an application. Then they need to know how to build one. But since Rails is, in effect, a domain-specific language for web applications, Ruby-saavy developers don't necessarily know anything about it. (I happen to fall into that group, and would love a quick entry-path into the Rails world.) Similarly, Rails programmers often don't know much about Ruby. They need an entry way that shows them how to take advantage of Ruby's power.

Each audience has several types of (overlapping) information needs:

  • Razors: This is a frequently-overlooked category of information, but it's an important one. Whenever there are multiple ways to do something, you need a razor-- a way to decide which one to use (like Occam's Razor).

    Nowhere is that need more evident than in the open source world, where a plethora of projects and libraries compete for user attention. When looking at Ruby gems, for example, there are generally half a dozen libraries to choose from in any given category.

    Since such projects come and go rapidly, what's needed is a Wiki with an Amazon-style rating system--a system where people can create categories, add new entries, and rate them in various ways (for example, with respect to documentation, architecture, error messages, ease of use, flexibility, and exception handling. The first site that does that will draw a ton of traffic.

  • Basics: How to install the language. How to set up a development environment (directory structure, tools, and tests). How to define environment variables, set up JVM options, and run a script--both from the command line and from an IDE like NetBeans--and how to access Java classes and use Java interfaces from JRuby.

    But note that language precedence tables should not be included in that doc set. I was delighted to read that Tim Bray makes it a point of principle never to learn them. I couldn't agree more. After learning some 20 different programming languages, I can tell you that it would be impossible to keep them straight--and a program is a lot more readable when you put in the parentheses to make sure it does what you intend it to do.

  • APIs: Java APIs are the most thoroughly documented class libraries on the planet. They set a standard for usability, and I missed them immensely the moment I began programming in Ruby. So there is a serious need for Ruby API docs that match the quality and quantity of Java's.

  • Best-Practices Language Reference: Ruby lets you do things in many different ways. That's a weakness, in my mind. It makes it much harder to read a Ruby program, unless the programmer happened to be using the idioms you're familiar with.

  • Mini-Tutorials: I'm a big fan of tutorials--in small bites. I don't mind spending 20 or 30 minutes learning how to do something I'm working on. But I never have the time to sit down for a week or two and work through a huge tome. I admit: I would be a lot better off if I did. Chalk it up to a character flaw. I just don't. So I need tutorials I can use in bits and bites--especially one that matches my level of expertise and tells me exactly what I need to know, in the order I need to learn it.

    Fortunately, DITA makes it possible to deliver exactly that kind of customized tutorial, as explained in Build-to-Order Documents with DITA.) 

Comments:

Wouldn't it be useful to know, as a developer, if you write something that works with JRuby now, will it also work a couple years down? How about the financial arguments about why exactly will using JRuby create opportunities that didn't exist before, save money on the long run, or how will people starting to learn about it make the time to market shorter rather than longer?

I apologise if I'm being too serious during playtime.

Posted by Mikael Gueck on September 05, 2007 at 09:16 AM PDT #

Good questions, Mikael. They all come under the heading, "Why do I want this?" or "Why do I care?" The answers will of course be found in the documentation, in conversation with others, and by experience.

My own thoughts at this point are that Java classes have been remarkably backwards compatible over time, so using them gives you a leg up in that area. JRuby is anticipating some language change, I know. But that comes down to a matter of pace. I don't like a language that never evolves, nor do I like one that changes every week. But every 5-10 years or so seems like a reasonable time frame to me .

Then there is the matter of installation. When I installed Ruby on a multi-user server, which meant I couldn't use the default locations, all kinds of gyrations were needed to make it work, and to get gems to work. I had a lot of help, and I never did figure out what it was I did. The JRuby install, on the other hand, was a breeze. No problem at all. For single-user workstations that issue is probably a don't care, but for multi-user systems where the common directories are under IT control, it can matter a lot.

Then of course there is scalability and performance for Rails apps, yada yada. I'm no expert on those things, but they sound pretty appealing.

Posted by Eric Armstrong on September 06, 2007 at 04:07 AM PDT #

How do you approach the problem of buying a house for your family? Do you take a loan, buy a random house, and only after moving in try to figure out whether it meets your needs and if you like it?

There are many things that can't be known in advance, but there are two ways to approach the construction of something. You can either set up goals that you believe might be relevant to your expected customers, and then approach the construction project in such a way that you can gather reliable data on those goals so that you can make credible arguments based on them when you're selling whatever it was you constructed, or not.

There is nothing wrong with a person who constructs things just for the fun of it.

Which builder would you approach when looking for a house for your family?

That's how I approach the question of what major tools to use in whatever ventures I'm involved with. The wrong choice at the wrong time might cost my coworkers things that they are not yet prepared to part with.

So, will it work in a couple of years? What can I do with it I couldn't do before, can I save money, can I shorten time-to-market?

Posted by Mikael Gueck on September 07, 2007 at 11:06 PM PDT #

Post a Comment:
  • HTML Syntax: NOT allowed
Search

Archives
« April 2014
SunMonTueWedThuFriSat
  
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
   
       
Today