Sargasso Sea

Well, that was fun...

I'm not sure what I hate more; good intentions, or round-to-its. I've learned a little about C# in the fourteen months since my last post. I've also learned a little about Java. None of it got recorded here, however, because... well, just because.

The title of this entry suggests I spent the time stuck in a morass of seaweed and lost ships. Mostly, however, I was extracting data from Lotus Notes so a certain Fortune 50 pharmaceutical company can retire its Notes/Domino platform. Come to think of it... Nah, that would be too easy.

I did have a couple of occasions to write little C# command-line utilities to clean up the data. They worked all right, I guess, but they were as ad-hoc as you can get -- basically procedural code stuck inside a single class. Unit testing? What unit testing? I don't gotta show you no unit testing!

I have another clean up project looming on the horizon and, since I have a little time, I'm trying to keep the horse out in front. I'm also thinking about doing it in Java so I can learn more about that language. The problem I'm having, though, is figuring out what and how to test.

The scenario is this: Twenty-some Lotus Notes application logbooks were extracted. The export placed the record data in a CSV file, one record for each document in the database. In addition, the exporter detached any files found with each Notes document in a subfolder named after the document's unique ID. Like this:

D:/App/Export/1234ABCD4D3C2B1AA1FE2DC9E34A9F25/MyDoc.doc

(The format actually makes sense; it uniquely ties each attachment to its source document.)

The problem is that there are over 123,000 files, spread across 70,000 subdirectories, all directly under the Export subdirectory, and the users want their files regrouped by region.

Conceptually, the application is simple. Create a list of the CSV files contained in the D:/App directory. For each CSV, extract the region ID from the CSV name, create a new subfolder with the ID as its name, then read in the CSV file, one line at a time. For each line that has a value in the fourth column (containing the full path and file name), create a string array from the value. Now, for each element of the array, retrieve the unique subfolder and file, move -- subfolder and file intact -- to the new regional subfolder, erase the original, rinse and repeat until clean. Oh, and since I'm paranoid about all things hardware, do a file check between the old and new files before deleting the old. (The I/O on this is going to be a BITCH!)

I guess the first test is to assertEquals("D:/App", path.name.toString()). We'll see how that goes.

Day One

I started a blog, some time ago, to log the voyages of the Fractal Sea upon the oceans of C# and Agile development. Shortly after setting out, however, the drunken navigator steered us directly into the βερμυδα Triangle! The ship and all hands were lost, save me. I survived by clinging to a coffin fancifully carved in ancient runes and totems by our savage bandersnatch harpooner.

Anyway, no vestige of that blog remains but I am still resolved to slay the frumious bandersnatch and learn what I can of C#, unit testing, and Agile development.

So...

Ahoy mates! There's a fair wind blowin', the skies are clear, and it's a beautiful day for hunting the bandersnatch. Let's go sailin'!

Point Class in C#

I originally intended to learn C#, and blog, while developing a fractal image generator. I figured a fractal generator would cover pretty much all the most used areas of application development: interface, text input, text output, algorithms for fractal generation, graphics output of the generated fractal, and, possibly, dynamic generation of the fractal during calculation. I soon landed myself in a bog, however, when I got stuck on just how I was supposed to translate somewhat esoteric (for me, at least) floating point calculations into iterative code. I got so distracted trying to figure it out the whole project came nothing. Nevertheless, in the end, I did learn something: RLC rule No. 1; Don't let the forest blind you to the trees.

Isn't the first rule of Agile development to break everything down into little chunks? Sure, I don't really understand the fractal formulas and I have little clue how to turn them into code, but there's a lot more to the application than the math engine. And even the math engine needs something to work on... something like a Cartesian point! Now, there's something I understand.

A Cartesian point, as a coding object, must have, at minimum, an X and a Y value, a constructor that takes no value, and a constructor that takes both an X and Y value. It also ought to be able to calculate its own distance from the origin. I'll talk later about what a fractal generator should do with this point but for now, here's a single tree I can work with.

As I seek help with my coding efforts from other bloggers and forums, I hope readers will come here to comment and help me develop as, well... a developer.