Here is the Hello World file I wrote:

1
2
3
4
5

IDENTIFICATION DIVISION.
PROGRAM-ID. HELLO.
PROCEDURE DIVISION.
    DISPLAY 'Hello, World'.
    STOP RUN.

Seems pretty easy, right? Well, yeah, no duh. It’s Hello World. I grabbed it from a Wikipedia article. It’s not exactly innovative genius. However, compiling it and linking it is a different story.

I tried at first using the ISPF foreground COBOL compiler. However, this was very non-obvious, and the documentation was lacking. All the documentation I found consisted mostly of how to do it through TSO, where it made mention that an ISPF interface did exist.

So then I went the JCL route.

IBM has some great documentation on how to do this. However, it is definitely not in the 900+ page Enterprise COBOL for z/OS Programming Guide. That was really just confusing. Here is the great documentation. This is what got me off the ground.

My JCL ended up looking like this:

1
2
3

//CLGP JOB
//CLG EXEC IGYWCLG
//COBOL.SYSIN DD DSN=HLQ.EXPERIME.COBOL(HELLO),DISP=SHR

It’s really pretty simple. The first line just declares the name of the job (CLGP), the next line executes the terribly named COBOL compile, link-editor, and run JCL procedure. IBM wrote it so you don’t have to. The last line just loads the dataset member that contains all of the COBOL code.

See? Mainframes and COBOL are THAT easy!

I am however, taking a Mainframes course at RIT this quarter. It promises to be relatively exciting. In it, we get to learn the intricacies of z/OS, and how to manage one of those beasts. That’ll be fun. Surprisingly, there’s a programming project towards the end of it (it’s surprising because it’s a Networking and Sysadmin course, we don’t program much). And since I’m learning how to administer and use mainframes, I thought to myself, “Hell, let’s do this project in COBOL.” So here I sit, trying to find a decent COBOL tutorial on the internet.

If anyone finds one, please send it to me. It’d be nice.

The interpreter was the easiest; it took me a morning to get that working albeit without support for nested parens. After that, I decided to write a text to Brainfuck converter, which (not surprisingly) turns text into executable Brainfuck code. Maybe if I get bored again, I could write a module to output the equivalent Perl or C code given some Brainfuck code.

This was my first real project using object oriented Perl. In all honesty, I like that Perl’s object system allows for many different types of objects (objects can be scalars, arrays, or hashes) and how flexible that is. Unfortunately, I didn’t get to exploit that flexibility in my code. It’s still cool though.

What I wrote is 2 Perl modules that make up the linked list that store data generated by Brainfuck code, a Perl module that interprets and runs the code, and a Perl module that translates text to executable Brainfuck code. I know I could have used a pre-built linked list, but I thought it was more fun to write my own.

The biggest challenge was definitely writing the translation code. I could’ve done it the easy way, and just output x amount of pluses for the character code of each letter, but I decided that was the lamest way to write Brainfuck. Instead, I looped through each character in the phrase, and added some nodes in my Brainfuck code that corresponded to the tens place of each character code. Then, when outputting a character, my Brainfuck code would change the data pointer to the node that was set to the tens place of the character code of the desired character, and increment or decrement it to the proper value.

A more ideal way to do it (to generate cleaner Brainfuck, at least) is to set each of those prep nodes to the nearest multiple of 20 for each character code, and then use whichever node is closest to the desired value. Maybe if I feel like doing more with this, then I’ll add that feature (probably not).

In addition to writing all this, I also made a GitHub account. My GitHub has all the above code, so you can clone the repo if you want to see what the code looks like (it’s actually really easy to read Perl).

This whole idea came from an idea on CSH. A guy here decided to publish a “Hello World of the Week” article in our newsletter, and to combat him I wrote a Brainfuck version of Hello World. This contest grew, and other people joined (albeit humorously) with such contributions as a boot loader that printed out Hello World, Hello World in White Space, and Hello World in Microsoft Word. My contribution (also in my git repo) generated valid Brainfuck code for the phrase Hello World, and then ran it through my interpreter (aka Hello World in 425ish lines).

So Google Wave is pretty awesome. It’s ridiculously buggy, but it works well enough if you aren’t using it in a production environment. I have yet to try out any of the multimedia stuff beyond embedding YouTube videos in Wave, but I imagine it works just as well as everything else does. I’ll probably talk more about that when I’ve played with it.

What I’ve really been playing with is the Wave API, in Python of course. The jist of what I’ve been reading is that the Java API isn’t as up to par as the Python API. Also, I hate Java, and I hate writing code in Java.

The Wave API is built off of Google App Engine, and has all the same limitations as App Engine. My roommate and I kinda half started working on an NNTP gateway to Wave. Unfortunately, the limitations App Engine sets makes this a little more difficult than we’d like (no sockets or anything of the sort).

The one thing I’ve come across in the API that I don’t like, is that you *must* include the entire Python Wave module in the app directory that you upload to App Engine. Not a huge issue, but definitely a bit of an annoyance. The only real problems I see are the possibility for people to modify the version of the Wave API they upload, and that the version of the Wave API they upload is outdated.

Another issue with using App Engine as the platform for all of the Wave code and such, is that you *must* use Python 2.5. App Engine doesn’t work with anything higher than that. So even though the Wave development site says you can use Python 2.5 or above, the above part is really a lie.

The good news: Google Wave won’t forever be reliant on App Engine. Sometime in the future, they will be allowing Apps from anywhere. This is will be awesome, and I can’t wait.

Oh, and I had other productive things I was gonna do this weekend, but they all got shot to hell when I got my Wave account.

Now you can wave me at ay1244@googlewave.com