Ruby 1.9
The Next Ruby
Summarized using AI
The Next Ruby
by Bruce WilliamsIn the presentation titled 'The Next Ruby,' Bruce Williams discusses the upcoming Ruby 1.9 version, focusing on its significance and new features. He begins by sharing his history with Ruby, emphasizing the language's evolution from a fun programming tool to a viable career path since 2001. Williams highlights the vibrant Ruby community in Austin, where he resides, and the slow yet steady adoption of Ruby 1.9 in comparison to previous versions.
Key points in the presentation include:
- Ruby 1.9 Overview: Williams identifies Ruby 1.9 as a major evolution from 1.8.x, providing better performance and new features while highlighting its lack of strict backward compatibility.
- New Features and Improvements: The speaker discusses various new syntax and language enhancements in Ruby 1.9, including improved string handling with encoding support, the introduction of the Oniguruma regular expression engine, and changes to enumerables that now return enumerators by default.
- Migration and Risks: He emphasizes the importance of having tests before migrating to Ruby 1.9 since it introduces significant changes that may impact existing codebases. He references the experience of another developer who faced unexpected issues during migration and stresses that properly running tests can help identify these problems early on.
- Community Impact and Libraries: Williams notes that while certain prominent gems, like Rails, are already compatible with Ruby 1.9, many others might not be. He points out the need for community members to ensure gem compatibility with the new version.
- Additional Syntax Changes: The presentation also covers enhancements like proc literals, hash ordering by insertion, and native threading models in Ruby 1.9, making the language more powerful and efficient.
In conclusion, Williams encourages the audience to explore Ruby 1.9 but with caution regarding legacy codebases. He reiterates the importance of maintaining tests and awareness of gem compatibility before fully adopting the new version. Overall, the session serves as an insightful guide for developers looking to transition to Ruby 1.9 successfully.
00:00:06.319
Video equipment rental costs paid for by Peep Code Screencasts.
00:00:19.520
Um, okay. My name is Bruce Williams. I'm going to be talking about the next Ruby, which is Ruby 1.9.
00:00:25.119
First of all, to give you a little bit of background on me, that's my name in big red letters at the top.
00:00:32.640
I live here in Austin, Texas.
00:00:39.440
So, seriously.
00:00:44.800
As you can tell, we have a very large and vibrant Ruby community here.
00:00:50.559
With very small voices, the community here is growing at a pretty good rate.
00:00:56.559
It's nice to have all of you here.
00:01:02.079
So, you know, grab somebody from Austin if you're looking for a good restaurant or a good bar. There's plenty of both.
00:01:09.360
I've been a Rubyist since 2001, back when you didn't make any money doing it, and it was just a lot of fun.
00:01:14.880
Now you can make money doing it, and it's still a lot of fun.
00:01:20.479
So that worked out really well. I'm a language tourist.
00:01:27.040
What I mean by that is I'm a Rubyist by trade and it's my tool of choice, but I play with Erlang, Haskell, and whatever makes sense.
00:01:33.920
Ruby has a long lineage, and we owe a lot to many languages.
00:01:40.560
I think we should float around a bit and take a look at other languages as well.
00:01:46.799
If you had to distill my life into a series of URLs, those would probably be it.
00:01:52.799
That's where I blog when I blog, which is probably not as much as I should.
00:01:58.079
You can find me on GitHub there using my name.
00:02:04.320
I work at a company called Five Runs, which is here.
00:02:11.200
We have the Tune Up and the managed products, which you can see to the right.
00:02:17.760
Basically, I just annoy people and talk about what I'm eating for dinner and things like that.
00:02:24.959
My Twitter, I can't claim there's anything useful that comes out of that.
00:02:32.080
So why Ruby 1.9? Why care about Ruby 1.9?
00:02:37.440
How many people here are using 1.8.6 most of the time (MRI)?
00:02:43.920
Okay, keep your hands up. Now, if you're using 1.8.7, keep your hands up.
00:02:51.480
How about 1.9? How about 1.9 in production?
00:02:58.080
Dave, you don't count; you're an early, early adopter.
00:03:05.040
Actually, recently, oh yeah, okay Dave, I guess you do count, but only because you paid.
00:03:12.959
Ruby 1.9 recently did a poll; I don't know if any of you took part in it.
00:03:20.000
They found a pretty good percentage of people—about 10 to 15 percent—last time I looked.
00:03:25.040
Now the poll is gone, so I can't find it, which is perfect for me because I'm speaking about it today.
00:03:31.599
About 10 or 15 percent are using Ruby 1.9 on a regular basis.
00:03:37.840
Thinking back a little bit, Ruby 1.9 reminds me a lot of Ruby 1.7.
00:03:43.840
Back when everybody was using 1.8.6, there was 1.8.7, and people played in the world of 1.7 before 1.8 came out.
00:03:51.200
It's a little bit different now because people have businesses that rely on 1.8.
00:03:56.319
They're a little less willing to go forward willy-nilly and grab stuff because we're very much into robust enterprise software in this community.
00:04:03.840
However, Ruby 1.9 does a lot of good things for us.
00:04:09.280
We get a whole bunch of new syntax and language features.
00:04:14.560
As the next version, it's not strictly backwards compatible with 1.8.
00:04:21.040
There are definitely some issues when you're migrating code, which I'm going to discuss.
00:04:26.240
It definitely has better performance characteristics.
00:04:33.040
And of course, it has more bugs; that's just the way software works.
00:04:39.680
Although, it's probably easier to maintain these days.
00:04:45.199
What I am going to talk about is the new syntax and language features because that's what I really enjoy.
00:04:51.280
I am going to talk about migrating stuff over; I'm not going to talk about performance.
00:04:58.000
That's something that you can find loads of numbers about from the people who are far smarter than me.
00:05:03.040
1.9, as a whole, is much, much faster than MRI 1.8.
00:05:08.560
I'm not going to talk about the bugs that exist besides the ones that have immediately popped up to me.
00:05:14.240
Because I'm not psychic, I don't know where they are.
00:05:20.400
If I were, I would be helping Mats fix them, but that's not the case.
00:05:26.400
So, still on the thread of things I'm not going to talk about: I'm not going to talk about the other implementations of Ruby.
00:05:32.639
Most of which target 1.8, anyway, with the exception of Mac Ruby, which has the benefit of sitting on top of 1.9.
00:05:39.760
I'm going to be talking about YAR, so Yet Another Ruby.
00:05:46.800
Just the plain vanilla Ruby. Not that I'm not excited about those other implementations.
00:05:53.520
So clearly, don't go out and grab 1.9 right now; this is where you would go to grab it.
00:06:00.000
Ruby 1.9 is in Subversion these days. I think 1.8 is a branch as well that you can track.
00:06:06.960
It's a version if you want to get it; it's still changing on a regular basis.
00:06:12.240
Certainly more stable than it used to be.
00:06:15.840
The klaxon I guess was because it used to be very unstable.
00:06:22.080
You can get 1.9 directly from the downloads page or from Subversion, and that's essentially how you build it.
00:06:27.440
Make sure you build it with some type of program suffix or something like that so you don't get confused about what Ruby is on your system.
00:06:35.759
I use 1.9 as mine.
00:06:41.759
A little bit about the standard library: I apologize for this being a little out of date.
00:06:46.800
I've done this version of this talk a couple of times and this one didn't get updated very well.
00:06:52.640
So there may be some stuff here that is different, but in general what you have is RubyGems.
00:06:59.919
They are now part of the standard library. They're built-in Solar, Rake, JSON, Ripper, etc.
00:07:05.919
Some of these may seem a little crazy; you may not know what profiler is, for instance.
00:07:13.120
Secure Random for those of you that are Sran users is for you if you really want random.
00:07:19.840
Here's one for you to look at, and of course the HMAC digests.
00:07:25.120
CSV was replaced by the faster CSV implementation, which I believe we have someone to thank for.
00:07:31.200
That's a vast improvement, and in things that have disappeared, at least the last time I looked, SOAP, Wisdom and Base64 were removed.
00:07:36.800
So for Base64, you're going to take a look at pack and unpack.
00:07:42.800
Now we're going to talk about risk factors: parts of your code that you're going to have a problem converting over to 1.9.
00:07:49.760
Because you're going to do it eventually, and you should at least be looking at it.
00:07:56.480
The first thing I’ll say is that you need tests.
00:08:02.160
If you don't have tests now on your code, first of all, you should have tests on your code.
00:08:10.080
You have to have tests because people at conferences will annoy you to death if you don't.
00:08:16.640
VDD, BDD, BDD; that needs to be said at least three times in every talk.
00:08:22.800
Because it's new and cool and life-changing.
00:08:28.160
Actually, yeah, I like it as a concept, just not as a buzzword.
00:08:34.080
So yeah, write tests! James Edward Gray had a great quote in a Ruby talk thread many months ago.
00:08:40.160
He talked about when he was switching over Faster CSV to 1.9.
00:08:46.639
He talked about the fact that there are a lot of issues you didn't realize that were there.
00:08:53.680
Your tests really helped him find them.
00:08:59.680
That's something I keep coming back to when I've been converting some code over.
00:09:05.760
If I didn't have my tests, it would be a lot harder to find these things.
00:09:10.480
With Ruby, obviously, you're not going to find some bugs unless some branches of code are being executed.
00:09:16.160
So the idea is to get those branches to execute, and tests are the way to do that.
00:09:22.720
This is a really big change in 1.9: text processing.
00:09:28.639
If you're writing parsers, Faster CSV is a good example of this kind of stuff.
00:09:35.039
Regal parsers for instance, and Hpricot and Mongrel, etc., are good examples.
00:09:42.160
You've got the issue now where we have new encoding support.
00:09:48.560
Now when you're using the string sub-method and indexing into a string with a single index, you're going to get a character back.
00:09:55.039
You're not going to get a number back, which is what most of us expected.
00:10:00.480
That’s a thing we’ve brought up for years: yeah, when you index into a string, you get back a number.
00:10:06.000
So that's gone now; we don't have to have that conversation with people.
00:10:13.120
We need to convert over the code that relies on it.
00:10:19.760
Now you can actually manage different encodings for your actual source files.
00:10:25.679
This is kind of cool.
00:10:31.840
Thank you very much. I wonder who that is? I have a wife and two kids, and it's during my talk.
00:10:37.680
That's okay.
00:10:44.000
Okay, where was I? Thank you, dear.
00:10:49.280
Yes, encodings: we have these new encoding conventions, that's kind of an extra feature.
00:10:55.680
You certainly do need to know about string, ord, and various unpack methods.
00:11:02.080
Be familiar with unpack in general, and pack as well. They're very powerful.
00:11:07.680
Now we'll look at some examples of changes in Ruby 1.8 and Ruby 1.9.
00:11:12.800
Generally, in Ruby 1.8 to get what you see in Ruby 1.9, you do something like 0.1 or 0, 1.
00:11:20.000
A major difference there is that it's not backward compatible at all.
00:11:26.000
The recommended approach for instance, for those doing regal parsers is to unpack your string into numbers.
00:11:31.200
You can either convert one way or the other or use character or to pull stuff out.
00:11:38.000
Next, pulling the same kind of stuff out on the left-hand side, I'm only pulling out bytes.
00:11:44.480
So it doesn't really work nicely for me. But on the right-hand side, it actually pulls out the multi-byte character, which is nice.
00:11:50.800
This is obviously using some encoding trickery.
00:11:57.600
Now granted, actually typing this in TextMate with bi-directional text and trying to put an index on the end was a complete nightmare.
00:12:04.160
So I’ve taken the string and then taken the thing next to it and sat it down.
00:12:10.080
It looks like it should. TextMate gets a little confused by bi-directional.
00:12:16.000
For those of you who are dying to type Arabic, your time isn't here yet.
00:12:23.840
You see some unpacked stuff here with actually pulling out the Unicode multi-byte.
00:12:30.000
You know; use ri unpack to see all the crazy options.
00:12:36.000
If you do C-star, you get a bunch of crazy negatives for the Arabic word, so that's not very helpful.
00:12:41.920
Now this is another point: this should only affect you if you're insane.
00:12:48.680
There's this really neat trick, very clever, I always put clever in quotes.
00:12:55.040
Because it’s a synonym for stupid.
00:13:01.760
This kind of stuff, usually a bug, is when someone hasn't realized they're accidentally assigning to a variable that's outside the scope using a block argument.
00:13:08.000
You see some people do crazy things, like assigning an instance variable inside a block.
00:13:15.760
This shouldn't exist in any code you ever show anybody else.
00:13:22.320
And you see the output difference here; the very last loop is going to assign the last yield.
00:13:30.720
The block variable won't take place here; it’s going to shadow it.
00:13:37.120
It'll warn you if you're using Ruby -w, which you should be.
00:13:42.960
So you're going to want to name your variables in your blocks something different.
00:13:48.560
If there's anything here that...
00:13:54.000
So in a solution, just don't do that.
00:14:01.200
We're not going to talk about arrow lambdas just yet.
00:14:07.360
There's a way of declaring a variable as local so that you can use it inside without assigning outside.
00:14:13.680
Just pretend that little arrow is the lambda.
00:14:19.680
Inside of it, you can see you're reassigning d, but on the right side, you aren't because you're declaring it local.
00:14:26.160
But you will still get the shadow warning.
00:14:33.920
This is kind of cool; the principle of least surprises has changed.
00:14:40.960
Hash select now returns hashes, which is nice.
00:14:47.360
You don't get an array of arrays, which I used to just flatten and splat to a hash.
00:14:54.000
So you're creating an array and making a hash, which follows the principle that when you do a select it should return a hash.
00:15:01.200
You probably could remove code to migrate this code since you're generating a hash again.
00:15:08.000
You might be expecting arrays back and doing other things with it, so be careful.
00:15:14.160
You see, if you had a hash called conferences, you could select on it and pull stuff out.
00:15:20.560
Another issue is that the array does matter on select.
00:15:27.120
In the past, with one block argument, it would get both the key and value assigned.
00:15:34.000
Nowadays, it looks out.
00:15:40.000
So Sam Ruby actually originally did this talk at Scotland on Rails in Edinburgh.
00:15:44.640
That's another regional conference across the ocean.
00:15:50.160
I put out a PDF, and people looked at it on Ruby Inside.
00:15:55.680
Sam Ruby recently came out and did another talk; he's much more focused on string encoding.
00:16:02.080
This is a really good point he made: one of the main obstacles for us to decide if 1.9 is cool.
00:16:08.560
Especially in production applications, is the fact that we have all these gems.
00:16:15.680
It’s not like we're Red Hat Enterprise Linux where we have some kind of idea of maintenance status.
00:16:22.160
These gems are just sitting around; they might have 1.8 code that's not compatible.
00:16:28.960
If you get 1.9, install some gems, and some of your dependencies aren't working right, that's a real problem.
00:16:36.880
As a community, we need to do something about that.
00:16:41.680
We need to identify these gems.
00:16:48.560
Certain gems, for instance Rails, run fine on 1.9.
00:16:55.680
The dependencies run fine on 1.9.
00:17:01.760
What 37 Signals is using probably runs fine on 1.9.
00:17:07.680
That's great, but there's a lot of other stuff that you may be using.
00:17:13.920
Maybe you have some crazy list of gems that nobody outside your industry uses.
00:17:19.200
You're going to need to do something about that if you want to continue to use your software with 1.9.
00:17:24.400
So there's no easy solutions there.
00:17:31.760
Now, new features: let's talk a little bit about the mostly syntax changes.
00:17:37.200
Some of the encoding changes I don't want to go too deep into.
00:17:42.560
M17N—obviously, there's a lot.
00:17:48.560
I try not to put a lot of text on the slide, but here it is.
00:17:54.000
The main points here are that strings have encodings nowadays.
00:18:01.200
Your source files can actually have an encoding.
00:18:07.040
You can read things in encodings.
00:18:14.080
That encoding affects a lot of things: regular expressions, various methods on strings.
00:18:21.840
You can't just do .each on strings these days.
00:18:27.520
You have to iterate over characters in each line and each byte.
00:18:33.760
You can examine what the encoding is on the string; you also have different ways of setting what your encoding is.
00:18:40.000
You can use command line stuff from Ruby and also magic comments.
00:18:47.360
If you have the Emacs style comment because you're using X, you can use those.
00:18:54.000
I'm sure there's something for Vim as well.
00:19:01.760
If you're using TextMate, whatever else—there are basic versions that don't give hints.
00:19:08.640
Obviously, I already talked to you a bit about string sub and how to pull stuff out of a string.
00:19:14.000
It supports a lot of different encodings; this was a complete list a few months ago.
00:19:21.120
It might be different nowadays; it's just a lot of stuff here.
00:19:28.720
UTF-8 is obviously the big one for most people that want to do surface languages.
00:19:35.600
Being able to pull stuff out of files and do I/O stuff with different encodings—that is now possible directly.
00:19:42.000
You notice in the second example I'm using an open version of the new hash with a symbol.
00:19:48.560
It reminds you a little bit of keyword arguments, probably, or maybe a little bit of JSON.
00:19:54.240
This is a really, really cool feature.
00:19:59.680
It's one of the features that checks off that enterprise box that people pass around.
00:20:06.080
This is a good thing; it's something people complain about a lot.
00:20:12.000
We don't have support for Unicode, etc., and that’s been a valid concern for years.
00:20:19.360
It's been worked on, and we're going to continue to work on it.
00:20:26.160
This is really cool because if you're a text-processing person like I am, having good regular expressions is really important.
00:20:32.080
One of my main reasons for switching from Python to Ruby years ago was that Ruby's regular expressions were much more natural.
00:20:38.240
But it was missing some really, really cool features from newer regular expression engines.
00:20:45.440
Now, Oniguruma has been integrated into Ruby 1.9.
00:20:51.840
It provides cool things like named grouping, various look-aheads, and look-behinds.
00:20:57.920
It also supports encodings, which is obviously an important feature.
00:21:04.000
There are some differences with enumerables and the enumerators.
00:21:10.320
In Ruby 1.8, you have to require enumerators.
00:21:17.680
In 1.9, all enumerable methods will return an enumerator.
00:21:22.640
You can call next on that enumerator.
00:21:29.360
You can stop complaining about not having hash each with index.
00:21:38.560
This has been a constant complaint for years.
00:21:45.760
Now you have it; you can even add your own crazy things onto an enumerator if you like.
00:21:51.760
However, this does change things like each with index, so you have to be aware.
00:21:58.560
Dot instead of underscore. If you use RSpec, you're going to be confused where to use a dot and an underscore.
00:22:04.000
You'll probably figure it out just fine.
00:22:10.720
In enumerables, there are some other cool things you can do.
00:22:16.400
You’re familiar with symbols of Proc from Rails.
00:22:22.480
That's now in 1.9. Some like it, others don't—that depends on who you ask.
00:22:28.800
There's also something reduce, which is essentially the same as inject.
00:22:35.120
You can pass it either a symbol or a symbol to proc.
00:22:41.840
It flattens things down; you can think of this like dot sum.
00:22:47.680
There's lots of methods on enumerable that make sense to take a look at.
00:22:53.840
Some of which are here: cool things like take and group.
00:22:59.040
Group is now part of Ruby, which is nice.
00:23:05.440
Back in the old days, you only had sort, then got sort by, and now you have min by and max by.
00:23:11.680
You can do count, which I believe is essentially dot size.
00:23:17.760
So there are new features; we got take and drop, a way to quickly grab something out.
00:23:23.040
Hash changes: at least the last time I checked, hashes track their insertion order.
00:23:30.240
You can actually go over a hash in the order things were inserted.
00:23:36.320
This is probably familiar to those using PHP with associative arrays.
00:23:43.680
Now, you can see I'm putting stuff in, taking stuff out, and what you're putting in is staying ordered.
00:23:50.720
Tap is a little thing on an object; if you're familiar with Active Support, this is similar.
00:23:58.000
This is very similar to returning, and some say it's not great for performance.
00:24:04.880
Just be aware it's there, and it's kind of neat.
00:24:10.000
What it's doing is passing itself to that block; you can do stuff with that block.
00:24:15.680
This is kind of cool, especially for things where you're fiddling with an object in a method.
00:24:22.560
Returning that object back means you want the scope to be explicit.
00:24:28.960
This is probably the thing when people think of 1.9; it’s the new proc literal.
00:24:35.040
Forgive me if I use the words proc and lambda interchangeably, as they are interchangeable for me.
00:24:42.320
The main thing here is we actually now have something built into the parser.
00:24:48.640
It's flexible compared to using the goal marker type.
00:24:55.200
You can pass blocks to them; it's great.
00:25:01.520
You can also do things like default arguments, which is nice.
00:25:08.720
Also, you can declare things as local variables, which is experimental.
00:25:16.080
I thought I had another slide, but I guess I don't.
00:25:23.440
So simple changes: we’ve got proc; like I mentioned, there’s less sibling rivalry.
00:25:29.920
You know that Rails created this idea of using symbols for keys in hashes, etc.
00:25:37.440
Stuff coming in from parameters were strings.
00:25:43.920
You check them out; especially when you were indexing, it would all fall over.
00:25:49.760
Now you can actually pretend to be a string in certain ways.
00:25:56.560
I know the idea of symbols being frozen strings was flirted with.
00:26:01.920
That was removed due to performance concerns.
00:26:08.080
But now they pretend to be the same, though you can't use = to check them.
00:26:14.160
That would be a bit too much.
00:26:20.000
But triple equals works, so case statements work.
00:26:26.080
Threads have moved to a native threading model from a user threading model.
00:26:33.440
There are actually three different models if you open up thread.c.
00:26:39.760
You do things like that for some reason, like reading through C code.
00:26:45.680
One is user threads, one is native threads with a giant VM lock, and one with finer grain locks.
00:26:52.000
The one built-in now, I believe, is model 2.
00:26:58.560
I don’t read a lot of C in my leisure time.
00:27:05.760
Also, I’m not a VM guy, so I haven't dug in deep.
00:27:10.640
When 1.9 comes out, hopefully this Christmas, we’ll have all the locks down.
00:27:16.560
This is cool; it's nice to run Ruby, kick off a bunch of threads.
00:27:23.680
Performance should be much better; you can look at Top and see the thread count.
00:27:32.000
This has been another checkbox we needed.
00:27:37.040
Especially for people coming from the Java world, this is something they absolutely needed.
00:27:42.720
It's been something that has kind of hounded us.
00:27:47.680
Fibers are essentially semi-coroutines.
00:27:54.640
You can think of them as lightweight, user-level threads.
00:28:00.320
They're manually scheduled.
00:28:06.000
They're kind of neat, especially for I/O frameworks.
00:28:12.640
But this isn't something I generally do.
00:28:18.280
There are good blog entries to read on fibers.
00:28:25.760
Dave has a really good series of articles on fibers.
00:28:31.680
We're going to go into questions. We have about 15 minutes.
00:28:44.960
So questions?
00:28:50.560
I don’t think I can explain the history behind the tap method.
00:28:56.960
Mats may be willing to talk about that later.
00:29:02.560
The tap method seems to owe a bit of debt to returning.
00:29:07.840
So there seems to be that connection.
00:29:13.440
That reminds me of Smalltalk semicolons for chaining.
00:29:20.960
If you can put that up there, you can chain things onto the same object.
00:29:27.280
I can't say specifically why it's called tap. Maybe tapping into the chain?
00:29:33.040
It also reminded me of jQuery if I can go back but in a less gross manner.
00:29:41.760
I really like jQuery.
00:29:49.040
I can’t tell you the etymology of it, but from a usability perspective it owes a lot to returning.
00:29:55.200
There’s a good subset of people in the Rails community that use returning as a pattern.
00:30:02.480
So, yes?
00:30:09.040
Just give us your personal opinions on the changes that happened to lambda.
00:30:14.960
At first, I really hated it.
00:30:22.880
Partially because I’ve used Ruby long enough, seeing any change can be scary.
00:30:29.760
But I love lambdas; it’s my background.
00:30:36.560
At first, it took me back because it felt like a step away from clarity.
00:30:43.520
But knowing all the new, cool things means that there will be great use cases.
00:30:49.680
I'm hoping the Ruby community self-regulates with all this.
00:30:55.760
We can create a schedule of punishments for horrible lambdas.
00:31:01.960
I have disliked making things as obscure as possible to read.
00:31:07.840
I love elegant code.
00:31:13.920
The utility is also pretty significant.
00:31:20.960
One point I want to stress is that the syntax may feel less clear.
00:31:26.280
So, it’s okay to let it sit for a bit.
00:31:32.000
You can play with it; you might feel more comfortable after a while.
00:31:39.040
Does that kind of answer your question?
00:31:44.640
I think I do have time for two more questions.
00:31:49.440
Okay, so do we have another question?
00:31:57.520
If not, thanks for your time, and thank you Mats for being here.
00:32:03.040
Video equipment rental costs paid for by Peep Code Screencasts.
Explore all talks recorded at LoneStarRuby Conf 2008
+18