Logical Meltdown

I have been totally unable to accomplish any coding this week.  I’ve been rewriting this application at work and while I’ve made some decent progress, I haven’t gotten a damn thing done this week and it’s frustrating as hell!

Of course now that it’s Friday and I’m feeling the urge to start writing more code having to do with dongs.  What the fuck is wrong with me?  Don’t answer that.

I figured that by explaining some of what I’m struggling with, some magic will happen.  I need to blog more and work on my “explaining complex shit in a meaningful way” skills, so if this in any way interests you please read on.


The application I am rewriting is a MySQL database-driven app that has two interfaces:

  1. A command-line interface (CLI) utility written in Perl for direct manipulation (addition/removal of records) and querying (searching) of the database. Only engineers have access to this tool.
  2. A web application written in PHP for searching the database for non-engineers.  Mostly support and operations staff use this. This also provides a hacked up “API” that returns a list of comma-separated values (CSV) of matching records.
It’s something I wrote years ago using PHP & MySQL that is way more complex than I remember it and double-dog dare you more complex than it needed to be. It’s strange going back in time with my code, especially PHP (shame), and realizing how competent I was even then, but how unnecessarily convoluted I made some things.

There are a shit-ton of logical pitfalls in that code that have really thrown me off.  Being that we have a Perl CLI tool and a PHP web app, the much of the code is duplicated for each language. The tool was originally written in Perl with the web app piece tacked on as an afterthought.  At the time I was really into PHP’s native support for MySQL but was pretty much a novice so ended up making some downright lazy and bad choices.

With PHP and it’s bloated cousin Perl, there are so many ways to do the same thing that there really isn’t one right way to do things.  Typically that means there are that many more wrong ways to do them.  For example, horrible crap like this where we have PHP making a system() call to execute a Perl script from within index.php:

    <tr valign="top">
          system('perl body.pl column="' . $POST['column'] . '" value="' . $POST['value'] . '" netmask="' . $_POST['method'] . '"');
Couple that with my prior lack of regimented programming style because I originally learned to code from the back of a Cracker Jack box (aka the internet), and what I am working with is pretty much a nightmare.

“Why is that?”, you ask?  Well let me tell you!

First things first, the primary users of the app (aka the stakeholders) have been requesting some new features for a long time and with everything else going on it hasn’t really been a high priority for me.  Especially because it was written in Perl, which I have grown to hate passionately since I started using Python.

Next, the systems from which the CLI tool runs are being replaced, which means the app has to be moved to the new systems. We run a Red Hat offshoot distribution of Linux, so we use RPMs to manage what gets installed on machines.  What this means is that any Perl module dependencies need to be installed via RPM. With RPM and Perl there are always dependencies of your dependencies, and you can bet that some of those won’t be available in the repository. This opens a massive can of worms of having to download the module source and convert it to RPM.  For those of you who are familiar with Perl’s CPAN and RPM, you’re probably thinking “Just use cpan2prm!”. Right, if only it were that simple…

You see, the new hosts are running a 64-bit kernel, which means that the subset of available packages is even smaller because packages have to be specifically built (or rebuilt) for 64-bit installation. Even if it’s the same files inside of the package, it still has to be re-packaged for 64-bit.

So, I was left with the choice of either uprooting the tool from its current location, wrangling package dependencies, rebuilding packages, modifying the Perl code to shoehorn it into the new systems all the while overlooking the implementation of the requested features; or rewriting it using Python & Django.

Naturally, I chose to rewrite it because the thought of having to trudge through antiquated Perl hackery gave me diarrhea and I really don’t want to ever touch Perl or PHP again if I can help it. By rewriting the app in Python, I can implement the new features and bring the code into the present at the same time without having to go through bouts contemplating suicide. Annnddd… That leads us back to the nightmare that is Perl and PHP.

As I mentioned above, with Perl & PHP there are so many ways to do things that most of them are convoluted, confusing, hard-to-read, and just plain fucking wrong. With Python–and Django for that matter–there is usually only one right way to do things, and that way is pretty much always elegant.  Plus, one of the most beautiful things about Django is that it wholly and completely abstracts the database interaction from the source code.  That means not having to write complex SQL statements manually, and the less SQL I have to write, the better.

In meeting with the stakeholders on rewriting the application, we decided that all of the energy should be invested in replacing the web app with a real API with a RESTful interface. If you don’t know the basic premise of REST (which is a methodology, not a protocol), it’s that URLs should be easy-to-read, self-describing and beautiful. Beautiful URLs? WTF? How can a URL be beautiful?  I’m glad you asked!

The old PHP version of the API is very heavy on the query string, requiring GET variables to be passed in the URL itself which is not only cumbersome, but can get confusing and hard to read very quickly:


Compare that to a RESTful version of the same:


This is where Django excels because it encourages pretty URLs by making it easier to use pretty URLs than not to.

So, cutting this post off at the knees, I’ve been pounding on this for the past few weeks, learning a lot about how NOT to do things from my old code and how TO do things using Django as my guide.  I made a lot of progress, got the API working, and a basic search interface that returns stuff all using the same code.

But this week, I’ve been stuck.  I haven’t made any progress, have been getting really confused, and just generally bashing my skull into the keyboard.  I think I’m just fried from trudging through all of Perl’s curly braces, semicolons, ampersands, dollar signs… OMG THE SYNTAX!  KILL ME NOW!!

Besides, it’s Fried-day anyways, so I hope that just by spelling this all out here that I’ll be able to dig back into the code after the weekend with a renewed perspective.

p.s. Fuck Perl.

p.p.s. Fuck PHP too, but wrap that rascal. She gets around!

Positive Thinking is Awesome

This is me in 2  years if I continue to make posts like this.  I'm the one on the left.

This is me in 2 years if I continue to make posts like this. I'm the one on the left.

Recently I’ve been feeling pretty good about myself in a way that only a man can… In my pants. Ok just kidding, I always feel pretty good about myself there. But seriously, I made up my mind to focus on the positive things in my life instead of dwelling on the negative.

Despite the undertow of self-doubt I’ve made a decision to continue to feel good about my life. Fundamentally that’s really all there is to it. It comes down to making a simple decision: Am I going to be happy or not? And so my mind is made up. I am.

I insist on feeling good about what I’m doing with my life and to be okay with where I am at. I will no longer beat myself up for not doing the things I think I should be doing, or want to be doing. Like blogging. (Or not blogging.)

Getting to this point hasn’t been easy and this seems really gay now that I’ve posted it. I am grossly oversimplifying the whole process but whatever, just go with it.