The BileBlog

For the first time in a few years, I’ve found an article linked off of slashdot that manages to be both articulate and interesting. Once I recovered from the initial shock of such an anomaly, I found surprising parallels between the trials and tribulations of wikipedia and the open source community.

The article, for those who haven’t read it, discusses why the very openness and ‘community’ aspect of wikipedia ensures its own limited usefulness.

Open source suffers from the exact same problem. The very openness of the source, the low barrier to patch acceptance (in most cases), and the group oriented approach ensures the final results is somewhat of an incoherent jumble.

There are, of course, many exceptions. The exceptions are in cases where there is a strong gatekeeper who by default refuses community submissions, preferring instead to have the submitter sell their case successfully first.

I’m sure all open source participants/project owners like to think of themselves as qualified gatekeepers. Statistically though, the majority of the patches you receive will be mediocre, yet for most projects, the patch acceptance level is well over 50%.

Setting aside the gatekeeper issue, there’s also the spastic community aspect to further defecate onto the project in ways far beyond harmful patches.

User contributions and suggestions are a good example. Much like in politics, the average bumpkin hasn’t a clue on what’s best for them. They will kick up a fuss and demand features that are simple bad design, out of scope, irrelevant, or already covered through some other mechanism.

The examples of this sort of thing are everywhere. Jakarta clogging is a good one, where a tool has come about that nobody actually needs. JBoss is the same, with huge disparity between various modules. Some are written by people who have perhaps taken a java class or two, whereas some are written by people who’s sole ability seems to be to flail about helplessly in the general vicinity of a keyboard (UnifiedClassLoader 1-4).

To be fair, I’ll toss in a closed source example; IDEA. The more community-driven IDEA’s features are, the more end users are dissapointed. IDEA 4.0 for example was driven by users feedback and what they wanted. It’s a shining example proving that that particular set of end users is as stupid as Texans.

So open source (I realise I’m generalising, and there are plenty of projects that prove this wrong, but they’re the minority) ends up suffering from the same problems wikipedia does. The competency distribution curve makes sure of this.

How is closed source any different? Well, there’s a higher barrier to entry, and there’s a carrot/stick mentality that caters better to the fact that the average developer is a fuckwit. You do well, you get paid well. Do badly, and you’re punished. Being paid well lets you buy shiny things, which makes you happy. Not being paid will make you sad, and your significant other will mock you/become sad/point and laugh at your genitalia/starve. Beyond some dubious penis tugging, open source just doesn’t have the same cold hard incentive/disincentive approach. Submit a bad patch that gets committed by a desperate committer? No big deal, at worst some random joe schmo will think ‘well this code sucks’.

I do want to stress that there are projects that buck this trend, but the laws of statistics apply there too, they’re few and far between, with the majority sitting pretty in the middle hump of mediocrity.

It’s fairly impressive how unit testing seems to have grabbed a number of java developers by the balls, and made them sing in surprisingly high voices.

Of course, we all know the virtues of unit testing. It’s great, it lets you err…test your little unit of code. It lets you pretend you don’t need any QA engineers, and rewards you with nice green bars (or ascii dots, if you’re into that sort of thing) that seem to have a disturbing sexual appeal to far too many people.

What is surprising though is the religious aspect of it. For example, some comment on a blog a few weeks back had someone spouting shite like ‘well I’m not a zealot or anything, but I would never use code with no unit tests, it works by accident not design’. What’s particularly sickening about this is the fuckfaced disclaimer up front. It puts the whole thing right up there with such gems as ‘Well I’m no racist but those niggers…’ and ‘I’m no bible thumper but YOUR ALL BURNING IN HELL YOU GODDAM ABORTION LOVING BABYKILLERS’ (Note: The typo is deliberate).

Surprisingly, the developer world hobbled along quite successfully before junit. Huge portions of it keep on hobbling that way with astounding success. See, what people seem to have forgotten is that a unit test does not mandate junit. It’s the approach that matters. Coverage reports are there to make things more sexually stimulating, they’re NOT a fundamental part of the approach. You could have plenty of unit tests that are nothing more that main methods in your various implementation classes. That still has its value, and certainly is more useful than no tests at all.

The JUnit fetishists are the freakiest of them all. What’s impressive is that they’ve invented this whole new universe just to justify what a horrifically broken tool they’ve staked their careers on (yes thoughtworks fuckstains, I’m pointing and laughing at you, you dirty chozgobbling rumprangers). Need static initialisation for shared or expensive resources? Not possible in Junit! Therefore, it’s evil! if you need to do it you’re broken!

The same goes for anyone who ever expresses any need for something beyond one method one test. Why the religious zealotry here? We all write code every day that requires more than one method, so why can’t our tests function similarly? Why is it so unthinkable that there might be some element of good unit testing that are not fully captured by JUnit?

The sickest joke in all this is how badly JUnit itself is written. Just read some of the javadocs and inline comments, they make it fairly clear that a fairly liberal amount of drugs had been imbibed in the course of writing it all.

So I ask you, all of you who want nothing more in life than to bend over and have JUnit plug all your orifices, why? What do you get out of it? Why this sickening allegiance to a flawed, old, unmaintained, and dysfunctional tool?

Update

: In my fury at junit, I forgot to mention my shameless plug. If you’re a JCP member, vote for me in the EC elections! If you aren’t, join then vote for me. I’m the only nominee who is motivated purely by improving Java. Everyone else is there out of some business need or corporate agenda. Stick it to the man!

I am heartily sick and tired of javalobby’s endless crusade against all things Java. With friends like this, who needs enemies? It’s utterly revolting how those swine manage to go on and on and on against Java in so many different threads.

What’s astounding is that there’s actually an audience for that kind of fear mongering. All those people seem to secretly wish their jobs didn’t force them to use Java, and would allow them to wallow in .net filth instead.

Particularly bizarre is Rick Ross’ vendetta against Sun. Now, I don’t know who from Sun pissed in his coffee (and by the sound of it, has a rather oversized bladder and thus hasn’t stopped pissing for the last few years), but the guy has a very deranged and bizarre Sun fetish. He just can’t say enough mean things about that poor little company. Sun sucks, Sun doesn’t understand Java, Sun should , Sun hates developers, Sun ate my babies, Sun should stop molesting my pets, Microsoft will tug my penis but Sun will only point and laugh, and so on and so forth.

Don’t believe me? Let’s look over some threads on there shall we. We have a thread about how Sun has no faith in Java 5.0, as proven by the fact that the ‘latest’ download on java.com is 1.4.2. We have a discussion about how Java is going to die on the next version of Windows. Of course, it’s all not all anti java, we have some pro .net stuff too! To inject some variety, we also have the ‘come on guys, lets band together and promote interest in java’ type gibberish. It’s like there’s this horde of demented lobotomised children out there, who take what their fearless leader says a wee bit too seriously. Who can forget Rick Ross’ poignant offer to help NASA with their Mars ‘java stuff’ using nothing but the willpower and intellectual prowess of javalobby members? We laughed, we cried, but mostly we shuffled about feeling very uncomfortable and embarrassed. It’s a bit like when your senile uncle whips his willie at a wedding ceremony and brandishes it furiously at the groomsmen while tittering ‘I am a purple teapot, short and stout’. Funny for a split second, but very uncomfortable thereafter.

To its credit though, javalobby represents the masses perfectly. Endless stupid threads that somehow defy reason and all that is rational by generating comments that seem to outdo each other in terms of sheer ignorance.

The .net fetish is very disturbing, it has to be said. I avoid .net stuff like the plague (or SWT), so I have no idea how things look from that side of the fence. Do .net sites also go on and on about Java? It just seems like bad manners and a distinct lack of social grace. It’s not called randomlobbyforpartimejavamonkeyswhoalsodabbleindotnet, last I checked.