Hacker News Reader

Hacker News

Comment by progx | Hacker News Reader

Full thread

progx

If TDD is the best thing on this planet, why have we to discuss about it? Why did not everybody use it?

Answer: obviously, it is not the best thing on this planet.

Like "use the right tool to solve a problem", use TDD if it solves a problem.

View on HN

"If TDD is the best thing on this planet, why have we to discuss about it? Why did not everybody use it?"

That's a pretty easy logical fallacy. For instance, I've encountered developers who don't use version control. That doesn't imply that you shouldn't use version control because everyone doesn't.

if you use an IDE with local history support like PyCharms or Webstorm, you can view every single save and every line of code you changed. It works just like git diff.

You don't know what you're missing

Now what happens when you are working with one other person using only local history support? What about ten people? What if someone is a bozo programmer and screws up your code accidentally?

Do not generalize here, what if he is working alone?

How does "local history" interact with the other developers on the project?

Git is not just about git diff. For example, it gives you the ability to find which changeset introduced a regression by doing a binary search on the history and running tests automatically on each revision. The culprit could be a change made months ago.

To do so on an IDE with local history, you'd have to Ctrl-Z backwards all changes (not meaningful commits, but even small changes) linearly.

Oh come now, that's just an extra safety net.

Don't take IDEA local history as a replacement to version control, rather a supplement.

That seems like it probably is a passable solution for simple linear history (in certain, single-developer, contexts), but it doesn't sound like it offers much for branching, which I would think you would often end up wanting with most real software even in single-developer set ups.

> I've encountered developers who don't use version control. That doesn't imply that you shouldn't use version control because everyone doesn't.

His argument wasn't usage, it was controversy. Presumably, these developers you've encountered aren't announcing that version control sucks and you shouldn't use it. They're just not using it themselves, and you'd probably be right if you guessed that they wouldn't have a rational reason for avoiding it. Contrast this with TDD.

> Presumably, these developers you've encountered aren't announcing that version control sucks and you shouldn't use it.

I've seen people argue that. The fact that people argue something is bad doesn't mean that it is, in fact, bad.

There's a difference between a few people arguing and a controversy. Admittedly, it's a fuzzy difference, but you could also use the degree of controversy as an indication of just how off the idea is. With DVCS, not much, with TDD, you could conclude that the idea needs more development before you can really rely on it.

| There's a difference between a few people arguing and a controversy.

Only on average. We are fully capable of being irrational as large groups...

> you could also use the degree of controversy as an indication of just how off the idea is.

That's exactly the ad populum fallacy, so while you could do it, you probably shouldn't.

Not quite. Ad populum describes the position that a proposition is true based on its popularity. Having doubts of a proposition's truth is not the same thing.

Using the degree of controversy of an idea as a metric for "just how off the idea is" is exactly using the popularity of the proposition which is negation of the idea in question as a metric for the truth of that same proposition, and so is exactly the ad populum fallacy.

When version control was new and when dvcs's were new, there was a lot of controversy. And the period when something is 'new' and controversial can be quite long, but people tend to forget and history tends to repeat itself.

> Answer: obviously, it is not the best thing on this planet.

Or some people are stupid. The truth doesn't always win out.

> Like "use the right tool to solve a problem", use TDD if it solves a problem.

You could say that of anything; it doesn't actually help decisionmaking.

It has it merits and there's some pieces of code I would not want to write without it but those cases (at least of a front-end dev) are remarkably few.

"there is reasonable debate about an issue" != "this issue is inherently subjective."

Reply on news.ycombinator.com