Hacker News Reader

Comment by ijustlovemath | Hacker News Reader

Another bit of cleverness not mentioned in the article is that assignment expressions always evaluate to the rvalue. So the expression `current->uid = 0` has the effect of making sure that entire conditional never actually runs (or at least, the return never runs), which means the overall behavior of wait4 doesn't change in an observable way. Very clever if you're trying to pass all of the existing tests

View on HN

But that should be something the compiler could catch. The expression is always false and the condition would never be executed. You usually get a warning for that. And if the compiler doesn't, linters do.

This is a common mistake, and I believe most linters have rules for that. And I don't think there is any situation where there is a good reason for code like this to exist. Either the expression is wrong, or it doesn't belong in a "if". You may get stuff like that in legitimate code with macro expansion, but again, it is not the case here, and from my experience, you get a warning anyways.

You didn't get a warning for that in 2003... or maybe with the pedantic flag. And even if it were the case, it would have been drowned by all the other unfixed warnings...

The only people using linters at that time was because it was forced by regulation (like automotive, aeronautics, ...)

> But that should be something the compiler could catch.

Today, certainly. My compiler even catches errors in the format strings to printf[1].

But back then? I doubt it, even with all the warnings turned up.

[1] Removing yet another common source of bugs.

The compiler does not, but tools like clang-tidy do, (bugprone-assignment-in-if-condition in clang-tidy) but that didn't exist back then.

Even now, and I'm pretty sure back then, the relevant warning in GCC is suppressed if the assignment occurs in parentheses, like it did in the diff (clang I believe has the same behaviour). So you would likely need a static analysis tool to flag up the behaviour, and those are quite noisy at the best of times.

Ohh, that is clever - unless someone writes a test for these two new lines, and finds that they never return -EINVAL.

Maybe the unit tests should test that the variables you intend to "not touch" do not in fact change. So record UID before and after the function. When I'm writing critical code like this, one gets a tingly feeling when typing in that variable name.

Did unit tests exist in 2003? I don't clearly remember when that idea came along, but comprehensive unit testing certainly was not standard practice 20 years ago... not in any organization I knew about at the time, anyway!

I believe unit tests existed. I had a test training in 2000 and things were pretty systematic already then. Not 100% sure whether the exact term was used then.

Edit: JUnit is from 1997. So the name was definitely in use in 2003. I attended a TDD tutorial before 2004 (don't remember the exact year). CI wasn't a thing yet, so you executed your unit tests manually. /edit

Do unit tests exist in the kernel today? There is some (or some would say a lot) of automatic testing for the kernel, but I don't remember seeing a single unit test.

Reply on news.ycombinator.com