The Ruby Programming Language

bdelacey writes "In January 2008, just in time for Ruby's 15th birthday, O'Reilly published The Ruby Programming Language. The co-authors make a strong writing team. Yukihiro (Matz) Matsumoto created Ruby. David Flanagan previously wrote Java In a Nutshell and JavaScript: The Definitive Guide — he has a CS degree from MIT with a concentration in writing. Drawings are the work of Rubyist-extraordinaire why the lucky stiff and technical reviewers include well known Rubyists David A. Black, Charles Oliver Nutter, and Shyouhei Urabe." Read on for the rest of Brian's review.

The Ruby Programming Language
author	David Flanagan & Yukihiro Matsumoto with drawings by why the lucky stiff
pages	444
publisher	O'Reilly
rating	9/10
reviewer	Brian DeLacey
ISBN	0-596-51617-7
summary	A classic and comprehensive guide to Ruby.

According to the Preface, Flanagan and Matz modeled this book after the K&R "white book" — The C Programming Language by Brian Kernighan and Dennis Ritchie. Like the "white book", The Ruby Programming Language has a simple structure and provides complete coverage. Just as K&R served as the de facto standard for "C", The Ruby Programming Language will likely be seen as the most authoritative language book for Ruby. Flanagan and Matz provide the following guidance for their readers:

"Because this book documents Ruby comprehensively, it is not a simple book (though we hope that you find it easy to read and understand). It is intended for experienced programmers who want to master Ruby and are willing to read carefully and thoughtfully to achieve that goal. ... [T]his book takes a bottom-up approach to Ruby: it starts with the simplest elements of Ruby's grammar and moves on to document successively higher-level syntactic structures from tokens to values to expressions and control structures to methods and classes. This is a classic approach to documenting programming languages." (p. 17)

You'll read all about boolean flip-flops, duck typing, lambdas, maps, metaprogramming, reflection and patterns of rhyming methods (collect, select, reject, and inject!). You'll also learn about new features in Ruby 1.9, like fundamental changes to text for Unicode support and the introduction of fibers fo coroutines. If it's in Ruby, it's almost certainly in this book. Chapters flow together nicely, although some could even stand on their own as educational materials for a computer science course (e.g. Chapter 7: Classes and Modules covers object-oriented programming and Chapter 8: Reflection and Metaprogramming elaborates on concepts like hooks, tracing, and thread safety).

In Ruby programming, difficult tasks are typically not only possible but often easy. It seems the authors take the same approach in their writing. For example, the complex topic of Domain Specific Languages (DSLs) sometimes creeps into deep discussions involving Ruby. Flanagan and Matz describe it simply and clearly: "A DSL is just an extension of Ruby's syntax (with methods that look like keywords) or API that allows you to solve a problem or represent data more naturally than you could otherwise." (p. 296)

During Ruby's first ten years, nearly two dozen books were in print in Japan but very few were available in English. That changed in 2004 when the introduction of Ruby on Rails created momentum for the language. A flood of new books followed, including Programming Ruby (2004, 2nd edition), The Ruby Way (2006, 2nd edition), Ruby for Rails (2006), and Learning Ruby (2007).

Programming Ruby, with lead author Dave Thomas, is self-described as a "tutorial and reference for the Ruby programming language." The Ruby Way, by Hal Fulton, was intended to complement Programming Ruby. Fulton noted: "There is relatively little in the way of introductory or tutorial information." Ruby for Rails, by David A. Black, has a clearly defined audience: "This book is an introduction to the Ruby programming language, purpose-written for people whose main reason for wanting to know Ruby is that they're working with, or are interested in working with, the Ruby on Rails framework." Learning Ruby, by Michael Fitzgerald, is a 238-page survey for "experienced programmers who want to learn Ruby, and new programmers who want to learn to program."

Programming Ruby and The Ruby Way each weigh in at over 800 pages. The binding on my copy of The Ruby Way came unglued and split in the middle after a year of use. The Ruby Programming Language is a slim, more manageable 444 pages and, in contrast, is the only one to cover Ruby version 1.9. In general, this is a great example of "less is more". Informative text boxes are sprinkled across the book with brief highlights on key technical thoughts. The first chapter's text box on "Other Ruby Implementations" (e.g. JRuby, IronRuby, Rubinius) could, however, be expanded into a several-page discussion of Ruby's various interesting architectures. Inclusion of IDEs and development tools (e.g. Eclipse, NetBeans, and TextMate) might also be helpful. These topics would nicely round out Chapter 10: The Ruby Environment.

The Ruby Programming Language has excellent cross-referencing. Section signs () feel like embedded HTML links that enable you to easily follow your coding curiosity around the book. Or you can just read it the old fashioned way, straight through. As an example, Chapter 3: Datatypes and Objects has subheadings (e.g. 3.1 Numbers) and well defined sections (e.g. 3.1.3 Arithmetic in Ruby.) The page-footers, table of contents and index also provide efficient navigational aids.

Artwork at the "edge of abstract expressionism" is something you might expect from The New Yorker magazine, but a computer book? The Ruby Programming Language introduces readers to "the edge of graphite expressionism". Original "smudgy residue" pencil drawings by why the lucky stiff creatively start each chapter.The Beatles' album cover for Sgt. Pepper's Lonely Hearts Club Band sparked intrigue and investigations into coded messages with hidden meanings. The same could happen here.

In Words and Rules: The Ingredients of Language, author Steven Pinker asks a simple question: "How does language work?" When I think about a new programming language, I have the same type of question in mind: "How does this language work?" Flanagan and Matz provide the answers in outstanding fashion. The Ruby Programming Language should help seasoned programmers who want to master Ruby. In addition, there is enough structure and sample code for determined novices to begin their programming explorations. Better than any other, this book defines the language. It is a classic and comprehensive guide for Ruby and a great 15th birthday present.

One long-time Rails developer sent me an email with their first impressions of The Ruby Programming Language: "I have been finding the book very useful, and I'm glad I did get it sooner rather than later." Matz said "Ruby is designed to make programmers happy." It looks like similar design thinking went into this book.

Brian DeLacey volunteers for the Boston Ruby Group

You can purchase The Ruby Programming Language from amazon.com. Slashdot welcomes readers' book reviews -- to see your own review here, read the book review guidelines, then visit the submission page.

Why can't this book be free?

Why do computer software worth less than paper-printed books? Last I check, software codes do not destroy rain forest like books.

Re:Why can't this book be free?

It can't be free because the author wants to get paid for his work on it.

But they can make their money selling documenta.... oh wait.

(sorry. I *had* to heh)

Thanks for the review!

Thanks for the kind review Slashdot! Thanks, Brian!

You can browse the table of contents of the book and read the beginning of each section at O'Reilly's website [oreilly.com].

You can find another review of the book at rubyinside.com [rubyinside.com]

Re:

Well whilst the author is here, and while I'm part way through the book, let me ask a couple of questions. I should preface this by saying that I think Javascript the Definitive Guide made me hate Javascript all the more, while Java in a Nutshell and Java

Re:Thanks for the review!

Well whilst the author is here, and while I'm part way through the book, let me ask a couple of questions. I should preface this by saying that I think Javascript the Definitive Guide made me hate Javascript all the more, while Java in a Nutshell and Java Examples in a Nutshell seemed to exemplars of good programming books. So needless to say I've puzzled over David's writing over the last few years. Why do I have such mixed feelings about books by the same author? I still don't know but maybe the following questions will help.

Have you read the JavaScript book recently? The 5th edition is an improvement over the fourth... It is different than Java in a Nutshell--simply because it is a "definitive guide" rather than a "nutshell". This Ruby book is probably more like my JavaScript book than my Java books, so if you end up liking it well enough, maybe you'll give the JavaScript book another chance :-)

I was reading along today when I came to this: "Assignment to a constant that already exists causes Ruby to issue a warning. Ruby does execute the assignment, however, which means that constants are nor really constant." Now this is a bit of a surprising statement. Isn't this an elephant in the room? Shouldn't it get more of an explanation?

Frankly, I don't know the reason that constants are not constant. I could have pressed Matz to enlighten us on this point, but I suspect that the answer is not a simple one and would have been difficult to explain. In earlier drafts of the book I did actually draw attention to these rough spots in the language because they did, indeed, seem strange to me. As I spent more time with Ruby, however, I came to appreciate it more, and re-reading my drafts I felt I was being unnecessarily critical of the language I was documenting. So, as you surmise, I was left just being matter-of-fact about it.

A few pages further on, while discussing parallel assignment, we get this: "a,(b,(c,d)) = [1,[2,[3,4]]] # Parens: a=1; b=2;c=3;d=4". Now I can figure out what is happening and what the book is trying to explain. But at the same time the book seems to ignore a second elephant in the room. Why in the world would someone ever write just a difficult to comprehend statement? Is is a common Ruby idiom? If so might it not be wise to offer some explanation of why it's an idiom?

This is not a very common idiom, nor is it even very well known. The particular line of code you cite is, of course, extreme: I'm taking the destructuring parallel assignment syntax to an unreasonable level of nesting to test the reader's understanding of the concept. Its not a common idiom, but it is part of the language, so I document it. It actually seems pretty cool to me, not really an elephant in the room. In this case, were I writing a less formal book, I might have commented on how cool it is. In the same way, in a less formal book, I could have commented on how strange it is that constants can have their values changed.

And one last, sort of unrelated question: is anyone else disappointed by the drawings? When I read that the book would be illustrated by whytheluckystiff I thought that this would make a probably good book even better. But I find them very disappointing and far less visually interesting than what can be found on whytheluckystiff's web site.

Please keep in mind that we had to get these drawings through the internal bureaucracy at O'Reilly--we were asking a lot of the production and design teams to include them. I shouldn't speak for _why, but I think his goal for these drawings was to keep it mellow, and to be stylistically distinct from his work for the Poignant Guide [poignantguide.net].

Re:Thanks for the review!

There is a very good reason that constants are not constants which relates primarily to the way the Ruby is "compiled." I put "compiled" in quotes because as Ruby users know, there is actually no difference between compile time and run-time. There is, however, a phase where you load the code in from your Ruby files at which point the classes are defined through executable statements. Constants are usually defined in classes at this point as well.

If you couldn't redefine constants, the second time you loaded a file you would throw errors. The reason why you would reload files is in a running development environment such as a framework like Rails or Merb. I also have a web framework (currently private but possibly we'll open source later) so I know this issue pretty intimately.

To clarify, the way it works is this: You are viewing a web page through a Ruby framework. The page rendering code is in a class. You want to make a change to a class so you go to the file and edit it. This may include changes to a constant. When you view the page again (by hitting refresh), the Ruby framework checks to see if the file has changed. If the file has changed (e.g. the constant we changed), then the file gets reloaded. Although we don't want to be changing constants in production, the ability to change constants in development mode is valuable. The alternative is we have to restart the server. The fact that it throws a warning is actually a good compromise because it means it's not normal behavior, but it is acceptable in certain circumstances.

Re:

almiray, As its title implies, this book is strongly focused on the language itself. It includes one (long!) chapter on the core platform, but it hardly touches on the standard library. Since Ruby's testing frameworks are not part of the language or of

Another good book to pick up...

...if you're doing Rails apps is Advanced Rails [oreilly.com] by Brad Ediger. It's got a ton of helpful hints on all sorts of things - sessions, memcached, how Rails uses Ruby's dynamic features, how plugins work, how to do complex associations, details on REST, etc etc.

The nice thing is that he doesn't fool around with explaining the simple stuff that you know already if you've done even one Rails app; he gets right down to business. Of course there are always interesting gotchas [blogs.com], but this is a book every Rails developer should have.

collect, select, reject, and inject

Sounds like they should have listed Arlo Guthrie as co-author.

I bet mods don't get the reference and mark me down.

Ruby is

When will /. grok the fact that is well know in parser/grammar circles, that Ruby is a language with a problem so big it can only be considered fundamentally Borked. Slashdot used to be a place where most people where on the ball and a crude attempt like this to gain traction by leeching the good standing off K&R would have been spotted and binned.

This Ruby Grammar tree [flickr.com] shows part of the problem, that god like Primary token in the middle. Ruby requires an arbitrary look-ahead parser, an LL(k) parser which are notoriously problematic. LL(k) parsers are inefficient, difficult to implement and result in ambiguous semantics. The result is that the Ruby grammar requires the token analyser and parser to be tightly coupled, a code smell in most programs. Token lexical analysis is dependent on syntactic context and syntactic context is dependent on semantic information from dynamicaly typed local variables. This is lethal for parsers.

It is not possible to separate the lexical grammar from the language grammar, nor is it possible to properly describe the language grammar in a rigorous way. What defines a good language is that its semantics are well defined and consistent without reference to an underlying mechanisms, The fact that Ruby needs a LL(k) makes the semantics ambiguity, the parser implementation must make arbitratory assumptions about the semantics.

The Ruby syntax cannot be defined with by rigorous context-free grammar, it is ambiguous. This is lethal in Parsers, because two different parser implementations can attach different semantics to the same source code. YES, that's right a Ruby program can run differently on two platforms, even if they have correctly implemented parsers.

This cannot be fixed because this big problem with Ruby is the very thing that makes it attractive, the syntactic flexibility. That flexibility cannot be decoupled from semantic ambiguity, making it error prone to write and more importantly read.

Wirth said of good languages, a language is not so much characterized by what it allows to program, but more so by what it prevents from being expressed.
If you want a dynamic typed languages there are better choices with rigourous grammars.

Wirth was Wrong

Wirth said of good languages, a language is not so much characterized by what it allows to program, but more so by what it prevents from being expressed.

Wirth was wrong. At least for common definitions of 'good'. I dumped pascal for perl in the 90's, and

Re:Huh? Is this the "new" English?

Let me parse it for you.

Drawings are the work of Rubyist-extraordinaire "why the lucky stiff" [whytheluckystiff.net] and technical reviewers include well known Rubyists David A. Black, Charles Oliver Nutter, and Shyouhei Urabe.

The rest of it should be passable English.

Re:Trash

You're going to get modded as troll here, but I think there's been a huge backlash against Ruby lately (just read programming.reddit.com once in a while), and rather than blaming the language or the toolset itself I think the problem is that it was so excessively hyped (mostly because of Rails) that whenever people find a problem somewhere then the whole thing is declared as lacking.

Personally I don't like Ruby, but that doesn't mean the language is not good. It has some interesting syntactic sugar. And Rails didn't really invent anything new. MVC and ActiveRecord for example had been done to death before, but it packaged them in an attractive and simple little box with a nice glue language that made 80% of building a web app simple. It's always the remaining 20% that's hard, and I think Rails was not designed well enough to enable developers to jump the last hurdles. Much of the criticism I've read in the past few months about Rails is related to problems with how the public Rails API is designed and how too many developers are using runtime hacks to extend objects instead of having sane, language-specific inheritance/override methods for extension purposes. Again, nothing to do with Ruby the language, but you can't reall discuss Ruby without Rails because in the real world it's what currently drives its adoption. I doubt many people are writing desktop apps with it.

Let this be a warning to framework designers: your language, runtime and toolset better match the amount of hype you pile on your work. It's not good enough to have a cool language to go along. The popular Perl (Mason, etc), Python (Django, etc) and PHP frameworks prove it can be done.

Re:Trash

Much of the criticism I've read in the past few months about Rails is related to problems with how the public Rails API is designed and how too many developers are using runtime hacks to extend objects instead of having sane, language-specific inheritance/override methods for extension purposes. Again, nothing to do with Ruby the language

Actually, the absence of features for explict overriding (vs. implict overrides as the default) is something to do with Ruby the language (just like dynamic rather than static typing, etc. are inherent to Ruby as a language.)

OTOH, the complaints about "monkeypatching" are pretty much just a variation on the complaints about dynamic typing: "the language doesn't bind your hands enough, and someone else might do (or has done) something I don't like that was enabled by that."

Fine, look, everyone knows where to find Java and C if they need them. Beyond that, if you don't like the way the Rails team (or certain third-party plug-in writers) have used the power to reopen classes and objects, its not like there aren't plenty of other Ruby web application frameworks, ORM solutions aside from ActiveRecord, and libraries available for most uses in Ruby. Maybe the one library you really want was programmed by someone whose ideas about how best to use the language don't match yours, but that's possible with any language no matter what features it has. If you want code written to your specs, you've either got to write it yourself or pay someone else to do it for you, and complaining about a language because someone has used it to write a library using techniques you don't like isn't going to solve anything.

but you can't reall discuss Ruby without Rails because in the real world it's what currently drives its adoption.

I dunno, I find it pretty easy to discuss Ruby without Rails. Rails is certainly the framework that made Ruby really visible in the US (though I gather it was fairly popular in Japan even before Rails), but I think that as it is getting exposed, the size of the non-Rails Ruby community compared to the Rails segment is probably growing.

I doubt many people are writing desktop apps with it.

I would imagine that if there weren't quite a few people doing that, you would have fewer GUI toolkits for it. And, of course, its also useful for system scripting; not every programming task is about a big app, whether web or desktop.

Re:

You can use it to implement [soulsphere.org] the Intercal COME FROM [wikipedia.org] statement. No serious modern programming language should be without this feature.

Re:Coroutines

Call/cc can become useful in web applications where you have a batshit insane workflow. Seaside [seaside.st], the Smalltalk web framework, bets the farm on it. It's in other places like Perl's Jifty too.

Most of the web apps I write wouldn't benefit from this sort of treatment, so I haven't jumped in too far, but call/cc has its uses for sure.

Re:

Ruby doesn't scale. Discuss.

Ruby isn't lightning fast, but its fast enough for lots of things, and its easy enough to interface with C (if you're using the main implementation) or Java (if you are using JRuby) if you need to refactor identified bottlenecks

Re:Let's face it:

> As to scalability, I don't know of any particular problems with Ruby there.

Yup. It's strange - folks will post comments like "Ruby can't scale for a huge enterprise app!". The odd thing there is that for a huge enterprise app the opcode execution speed is probably not going to be the primary bottleneck. In fact, almost anything that reaches over a socket into a database is going to have bottlenecks that have nothing to do with how fast the Ruby interpreter can navigate an AST or set up a stack frame.

Re:

That's why you're supposed to use multiple Rails processes. Multiprocessing is functionally no different from multithreading, except for the fact that processes don't share memory with each other, so if 1 Rails instance crashes the others are still alive.

F

Re:Let's face it:

get RAD with Java.

These kids. In my day we used to get funky with Fortran and we liked it.

Re:Let's face it:

It was more an invitation to RTFM, but I just realized that Google isn't helping much, and I can't find the original articles. So here's the short version:

Performance is how fast an app runs for a given amount of hardware. This is usually measured in how fast it is with respect to a single core of a single CPU.

Scalability is how well the app behaves when you just throw more hardware at it. That is, if you give it another ten CPUs and five gigs of RAM, will it be able to take advantage of that?

Ruby performance sucks. It is impossibly slow, even compared to languages like Perl or Python. I make no excuses for that. Look at the performance improvements of rails 1.9, and you'll get an idea of just how bad 1.8 is.

But Rails scales very well -- in the simplest example, the default configuration for Rails nowdays is a mongrel cluster, even on a single machine. It's a really strange and perverse design where you put a load-balancing proxy in front of at least three Rails/Mongrel webservers, each listening on a different local port. And unless you've done something stupid, this will Just Work, right out of the box.

But as long as your database server can handle it, you can just throw another server behind that load balancer. A lot of work has gone into making load balancers and database servers scalable, and you can take advantage of that, if you really need to. (For awhile, Ruby is still going to be the bottleneck.)

What this means is that I probably would not write a performance-critical desktop app (like, say, a game) in Ruby. But for web apps, it makes sense. If you can write the same app twice as fast, with half the people, you can afford to throw four times the server resources at it -- and as a bonus, you got it up in half the time as your competitor, who thought scalability == performance.

Re:

There are more free resources for learning python.
http://docs.python.org/tut/
http://diveintopython.org/
http://www.swaroopch.com/byteofpython/
http://openbookproject.net/thinkCSpy/
http://www.greenteapress.com/thinkpython/
http://en.wikibooks.org/wiki/N

Re:Which to learn first: python or ruby?

Depends if you ever want to learn ruby or not?

Once you learn python you ain't gonna get past hello world in ruby. You will
sit there, look at the syntax and say what the hell do I want to learn this for.

Re:

Once you learn python you ain't gonna get past hello world in ruby. You will
sit there, look at the syntax and say what the hell do I want to learn this for.

I learned Python pretty much immediately before learning Ruby, and I personally find Ruby more attra

Re:

Are you saying it's impossible to write readable code in perl?

I'm not saying you're wrong but I don't think we can really be 100% sure until someone actually tries.