Parrot Babysteps 01 - Getting Started

Introduction

I want to explore basic installation and usage of Parrot. The usage will include output, simple variables, and input. PIR, the Parrot Intermediate Representation language, will be my language of choice. This is the same pattern as my other Babysteps tutorials, but the fact that PIR is a lower level language means that these are going to be some pretty big steps. I encourage you to walk along with me if you’re curious about PIR.

As always, I welcome suggestions from both novice and experienced Parrot hackers on how to improve this material for newcomers to Parrot.

My Setup

Linux

I’m running Ubuntu 10.10 with build-essential and Perl 5 installed. Perl 5 is installed by default on Ubuntu, and build-essential is easy enough.

$ sudo apt-get install build-essential

There may be other dependencies already installed on this machine that I do not know about. I tend to install many packages on my system, such as wget. Perl 5 and build-essential should do it as far as I know, though.

You will probably want a good text editor as well. Vim is my choice, but you’re welcome to use anything that makes you happy.

OS X

The iMac which rules our home is currently running OS X 10.6 with the Developer Tools and MacPorts. As usual, my text editor of choice for this platform is Vim. I have my essential toolkit including wget installed via MacPorts.

Windows

I’m using 32 bit Microsoft Windows 7 Ultimate on the other partition of my laptop. Hey, it’s better than Vista. I’m also using Vim to edit my files on this platform because I am a creature of habit.

Installation

You need to get Parrot before you can use it. The download page shows where you can get versions of Parrot appropriate to your system, including Windows packages.

Linux and OS X

I will be grabbing the latest stable source release.

$ cd ~/src
$ wget http://www.parrot.org/release/supported

Previous versions of these Babysteps were using the development releases of Parrot. Why did I switch? I have trouble keeping my articles up to date with the pace of Parrot development while also developing new material and working, so my focus moved to the slower “supported” cycle. I may switch back to using development releases if stable is horribly broken for something the Babysteps need. You’re welcome to use development. Most of the material will probably work as-is.

I’ve gone through a few installations of Parrot by now, and have finally noticed that bad things happen if you ignore your old installation. Cleaning up the important files for the old installation is a good idea. So, if you are upgrading:

$ sudo rm -rf /usr/local/lib/*parrot*

Of course, you better make sure that anything matching that name pattern is actually related to your old Parrot install before nuking them like this.

Next I’ll unpack, build, and test my download.

$ tar xfvz parrot-3.0.0.tar.gz
$ cd parrot-3.0.0
$ perl Configure.pl
$ make
$ make test

make test takes a little longer each time I install a new version of Parrot. They add tests in every iteration. Just think of those extra few seconds as proof that someone cares about the quality of software they’re writing for the world.

The Parrot authors can always use additional feedback about how well their baby works on different platforms. One piece of information that can be useful for them is a smoke report with test results telling them how well Parrot works on your machine. You can very easily do this on Linux and OS X:

$ sudo cpan TAP::Harness::Archive
$ make smoke
...
All tests successful.
Files=381, Tests=13848, 155 wallclock secs ( 2.36 usr  1.43 sys + 102.37 cusr 28.89 csys = 135.05 CPU)
Result: PASS

TAP Archive created at /Users/brian/src/parrot/parrot-3.0.0/parrot_test_run.tar.gz
Test report successfully sent to Smolder at
http://smolder.parrot.org/app/projects/report_details/14424
You can see other recent reports at
http://smolder.parrot.org/app/projects/smoke_reports/1 .

My smoke report and the list of recent reports are fun to look at in a nerdy sort of way.

There. Now I’ll install Parrot.

$ sudo make install

Oh, editor support is also available. Since I’m using Vim, I’ll need to make vim-install.

$ cd editor
$ make vim-install

This gives me syntax highlighting and template generation for new PIR files.

Windows

It appears to be even more straightforward to install Parrot on Microsoft Windows. The Parrot download page includes a link to an installer file.

parrot-win32 setup files

The stable setup file available currently is setup-parrot-3.0.0.exe. If you browse around, you will find a lot of interesting setup files. I won’t be exploring them right now, but I encourage you to check them out.

I will download the latest stable installer and run it, following all the defaults for now. That was easy. Don’t miss the CHM Parrot documentation in C:\Parrot-2.3.0\share\doc\parrot! It is good to have documentation handy when exploring strange new software.

I’m not sure if this warning is still true, since my recent reinstallation of Windows means I don’t have a previous Parrot install to overwrite. Still, it’s worth checking out.

When you install a new version of Parrot, the default behavior is to place it in the same location as your previous installation. This means 3.0 would sit in C:\Parrot-2.9.0 on my machine. That can cause trouble, since some important Parrot settings may be based on assumptions about where it was installed. The safest thing before upgrading Parrot on Windows is to uninstall old versions through the Control Panel, and to make sure that previous Parrot bin folders are not sitting on your Path. Check for that in Computer → System Properties → Advanced System Settings → Environment Variables. It gets easier to remember those steps after you’ve clicked them a few times. Do not install a new Parrot until you have removed the old one!

So - installing Parrot is easy, but I still want my Vim support. The installer was made with the reasonable assumption that Parrot users on Windows weren’t going to need support for their favorite Unix editors. Reasonable assumptions tend to break down around me, though.

If you’re not using Vim on Windows, you can just move on to the next section. As for the rest of us, let’s get to it.

Take your browser to the editor directory for the latest stable Parrot release.
Download the files that are important for Vim support:
- filetype_parrot.vim
- indent_pir.vim
- pasm.vim
- pmc.vim
- pir_vim.in
- skeleton.pir
Create a vimfiles directory in your home folder if you don’t already have one.
Using README.pod as your guide, move the support files into your vimfiles directory:
- filetype_parrot.vim → vimfiles\ftdetect\filetype_parrot.vim
- indent_pir.vim → vimfiles\indent\pir.vim
- pasm.vim → vimfiles\syntax\pasm.vim
- pmc.vim → vimfiles\syntax\pmc.vim
- pir_vim.in → vimfiles\syntax\pir.vim
- skeleton.pir → vimfiles\skeleton.pir
Open vimfiles\ftdetect\filetype_parrot.vim and execute the following:
- :4s/\.vim/vimfiles
- Or yeah, you can move to where it says .vim and manually replace that with vimfiles if that’s how you like to do things.

The steps may be a little more involved than just “click and run”, but it’s worth it for me. I like having Vim support.

Output

It is time to start writing code. Move to your projects directory and create a new directory for the Parrot Babysteps project.

$ cd ~/projects
$ mkdir parrot-babysteps
$ cd parrot-babysteps
$ gvim example-01-01.pir

When I’m connected to the iMac via ssh I settle for

$ vim example-01-01.pir

The steps are the more or less the same on Windows. First, open a command prompt by whatever means you’re comfortable with. I prefer typing cmd into the “Search programs and files” field of the Start menu. Then, move to your projects directory and create a new directory for the Parrot Babysteps project.

C:\Users\brian> cd \projects
C:\projects> mkdir parrot-babysteps
C:\projects> cd parrot-babysteps
C:\projects\parrot-babysteps> gvim example-01-01.pir

From here on, the differences between Windows and other platforms become small enough that I feel comfortable ignoring them until something really significant catches my attention.

We immediately see some benefit from installing editor support.

# Copyright (C) 2011 Parrot Foundation.

.sub 'main' :main
    # For Parrot developers.
.end

# Local Variables:
#   mode: pir
#   fill-column: 100
# End:
# vim: expandtab shiftwidth=4 ft=pir:

That looks like a heck of a lot for an otherwise empty file, but most of those lines are actually comments.

Comments

PIR comments start with a # character. Everything from that # to the end of the line is ignored by the parser.

# This is a comment

Comments are handy. They can explain why you chose a particular solution, concerns you have about the code, or pretty much anything else. The comments in our template serve to identify where the template file came from and a few directives for Vim to remember when editing this file.

Directives

There are only two important lines left when we ignore the comments.

.sub main :main
.end

Lines that start with a period (.) are directives, special instructions for PIR. There are a number of directives for Parrot, and it is possible to create your own. The .sub line is telling PIR that this line begins a compilation unit called “main” which serves as the :main compilation unit in the program. Compilation units are tasty named tidbits of code that get a particular job done. They seem roughly the same as subroutines in other languages, so that’s what I’m going to call them from here on. Less typing that way. Once you have created a subroutine, you can call it elsewhere in your code.

Subroutines can also have a tag after their name to add a little extra information. The subroutine tagged as :main is the main entry point for a PIR application - it’s where the program actually starts running. When you hand a .pir script to parrot, it directly executes the code contained in :main. If it can’t find a subroutine tagged with :main, then Parrot will directly execute the first subroutine in your code. It is a good idea to specify which is :main so you don’t have to rely on the default behavior.

The .end directive tells PIR that it has reached the end of this particular subroutine. Any .sub must be followed by .end eventually. Everything in between is the body of the subroutine - the stuff that gets executed when the subroutine is called.

We now basically know what a subroutine is, what it looks like, and how Parrot goes about figuring out which one to run first. We still don’t have anything to be executed, though. How about making this program do something?

Building The Program

# example-01-01.pir
.sub main :main
    say "Hello, world!"
.end

Finally - code! Fairly self-explanatory code at that. Well, almost. say is an opcode - a native Parrot command - for presenting values to the user. say won’t normally speak the phrase “Hello, world!” through your speakers. It will, however print the phrase on a line in your console.

$ parrot example-01-01.pir
Hello, World!
$

Variables

What if I want to change the greeting, so that it displays my name as part of the greeting? While simply rewriting the phrase does technically work …

# example-01-02.pir
.sub main :main
    say "Hello, Brian!"
.end

… I want something a little more flexible. I want to be able to define a name, and then print a greeting with the name. This will require adding a variable. We use variables whenever we want the program to remember something.

There are four types of things that Parrot can keep track of:

integers: Simple counting numbers, like 13 or -485
numbers: Decimal numbers like 1.3 or -1.3e+5
strings: text and phrases like “Hello, Brian”
Polymorphic Containers, or PMCs: A special type that basically refers to anything that’s not an integer, number, or string

Parrot has special rules for dealing with each of these types, so it is important to tell Parrot how you want your variable treated. One way to do that is with register variables. Registers can be thought of as variable buckets: one bucket for each type. Okay, maybe not a bucket. More like four rows of nicely organized variables, with each variable type in its own row.

Register variables are prefixed with the dollar sign $ and an uppercase letter indicating their type. The rest of the variable name is a whole number from zero on up. Here, let me show you:

# example-01-03.pir
.sub main :main
    $S0 = "Brian"
    $S1 = "Hello, " . $S0
    $S1 = $S1 . "!"
    say $S1
.end

Assignment to register variables is pretty straightforward, as you can see. You tell Parrot which position in which row is being assigned to on one side of the = symbol, and the value to assign on the other.

Oh, did you see what I did when creating $S1? I did mention that Parrot has special rules for handling each type of variable. One thing you can do with string variables is paste two of them together, making a whole new string. You use the . operator for that. But only for pasting two strings together. If you want more, you’ll need to do it on two lines like I did.

Back to register variables. No, it doesn’t really matter which numbers you use. I could have used $S9999 to hold the name and Parrot wouldn’t care. We just need to remember what purpose each variable serves.

Actually, that’s the main problem with register variables. Their names don’t tell you much about what purpose the variable serves. Sure, it’s easy enough to see that $S1 is a string, but will I still remember that it’s being used to hold the name once I’ve added a few hundred lines of code? Probably not. I’m getting flashbacks to programming BASIC in grade school, when I kept a cheat sheet of variable names handy.

Parrot is thankfully there for us with the .local directive. local allows us to create a readable name for our variable, asking Parrot to deal with the register side of things.

Creation of a local variable requires the .local directive, followed by type information and lastly a variable name.

.local <type> <name>

The type names match up to the Parrot types, of course:

int for integers
num for numbers
string for strings
pmc for PMCs

Variables created with the .local directive can be treated the same as a register variable.

# example-01-04.pir
.sub main :main
    .local string name
    .local string greeting

    name = "Brian"
    greeting = "Hello, " . name
    greeting = greeting . "!"
    say greeting
.end

You know, that line where we append an exclamation to our greeting is longer than it needs to be. The PIR provides special operators for when you are doing something to a variable and then assigning the result back to the same variable. That means we could describe “append an exclamation point to greeting” this way, too:

# example-01-05.pir
# ...
greeting = "Hello, " . name
greeting .= "!"
# ...

Input

We’ve covered a lot of conceptual ground already. What has it earned us?

$ parrot example-01-05.pir
Hello, Brian!

This is not impressive. We’ve actually increased the amount of code significantly compared to just saying “Hello, Brian!”. Splitting the name off into its own variable does lay the groundwork for a slightly more impressive task, though. Let’s make the program ask the user for his name when run.

# example-01-06.pir
.sub main :main
    .local string name
    .local string greeting
    .local pmc stdin

    stdin = getstdin
    name = stdin.'readline_interactive'("Please enter your name: ")
    greeting = "Hello, " . name
    greeting .= "!"
    say greeting
.end

We’ve only added a few lines, but these few lines have added a lot of functionality. The first thing you probably noticed was the pmc variable named stdin. PMCs are used in Parrot for complex types, like classes in other languages. They often have special subroutines called methods tied to them. Those methods can greatly simplify the creation of complex programs.

We use the opcode getstdin to figure out standard input and assign it to our stdin PMC. Standard input - widely referred to as stdin - is a FileHandle that describes where we expect user input to come from. That can be any number of places, but usually refers to the keyboard. The FileHandle method we care about most today is readline_interactive. readline_interactive prompts the user for input, and hands that input back to the program.

$ parrot example-01-06.pir
Please enter your name: Brian
Hello, Brian

Conclusion

We’ve done quite a bit today. We installed Parrot, created a simple script that prints a greeting to the console, then expanded our script to get the user’s name and incorporate the name into the greeting. It took less effort than I thought it would despite the fact that PIR is low-level compared to languages like Python and Perl 5.

The ability to handle user input makes it easier to create meaningful programs even with a bare minimum of knowledge, and can be very handy as we explore Parrot, the PIR syntax, and the many PMCs. From this starting point, we can go deeper into the Parrot and PIR documentation, and search the Web for Parrot tutorials.

Remember also that Parrot is released monthly. I do try to keep these pages up to date, but sometimes work and life demand more of my time, or I get distracted by bright shiny things. So don’t assume my version numbers are correct. Double-check them for yourself.

Most of all, have fun!