Garbage Collection

Generally speaking, for systems software or critical infrastructure, never make the foundation of the language dependent on garbage collection.

Garbage-collected languages sound too good to be true. Programmers don’t need to be meticulous about memory management.

But if something sounds too good to be true, then—as that joke from The Neverending Story goes—it probably isn’t.

Just think about it: why the obsession with having someone else manage your garbage?

It’s such a difficult algorithm to get right, and thus 100% of implementations are not going to be right. It’s one of the reasons why C/C++ always wins as an implementation choice. I wish folks would be more honest about it. I lost a lot of productivity trying to use Mercurial, which was written in Python. It was a piece of literal dog shit compared to Git when it came to stability and performance. I want to go back and punch all the idiots that made the choice ambiguous just because they wanted to be polite. I joke - I am not a man of violence in reality and I would probably get my arse kicked if I tried, but metaphorically I feel the emotion.

So many programmers just get bored. They want to invent or learn a sexy new language. I might do that too, but I wouldn’t start from scratch. I would merely remove a lot of functionality from C++ to discourage people from using features I see as having no value. The best way to improve most systems is to remove functionality.

: