Putting code with side effects into an assert is asking for trouble. Compile with NDEBUG set and the effects mysteriously disappear! Anything beyond an equality expression or straight boolean should be avoided.
I once spent several days debugging that same mistake. Stuff worked perfectly in tests but broke misteriously in production builds. Couldn't stop laughing for a few minutes when I finally figured it out.
I'm sorry, but what exactly is the problem with the code? I've been staring at it for quite a while now and still don't see what is counterintuitive about it.
An assertion can be arbitrarily expensive to evaluate. This may be worth the cost in a debug build but not in a release build. If all of assertions are cheap, they likely are not checking nearly as much as they could or should.
Possibly but I've never seen it in practice that some assert evaluation would be the first thing to optimize. Anyway should that happen then consider removing just that assert.
That being said being slow or fast is kinda moot point if the program is not correct. So my advisor to leave always all asserts in. Offensive programming.
This is just a symptom of a bad assert() implementation, which funny enough is the standard. If you properly (void) it out, side effects are maintained.
assert() is meant to be compiled away if NDEBUG is defined, otherwise it shouldn't be called assert(). Given that assert() may be compiled away, it makes sense not to give it anything that has side effects.
Abseil has the convention where instead of assert(), users call "CHECK" for checks that are guaranteed to happen at run time, or "DCHECK" for checks that will be compiled away when NDEBUG is defined.
Not just for functional programmers. Prints and other I/O operations absolutely are side effects. That's not running counter to the point being made. Print in an assert and NDEBUG takes away that behavior.
In C++ you should probably #include <cstdio> instead of <stdio.h> unless you have a good reason. And especially avoid #including both. <cstdio> provides the function std::getc(..) while <stdio.h> usually provides getc(..) as a macro.
htons(..) and related socket-utility names are also often macros, but I'm pretty sure there is not a std::htons(..) in the C++ standard, partly because 'htons' is not an attractive name. Since it's (sometimes) a macro don't qualify its namespace like ::htons(..).
A long time ago in the Microsoft C (and later C++) dev envs there were macros named "min" and "max", which I thought were terrible names for macros.
Yep. There's tons of others as as well. 16-bit x86 enjoyers will be happy to know there are `near` and `far` macros whose primary purpose in 2026 is to break my projection matrices. And of course every Win32 function that takes strings has a macro that resolves it to either the UTF-16 or ASCII variant, so your custom CreateWindow is now a CreateWindowA, tough luck buddy.
I usually wrap Windows.h in a header followed by 100 #undefs to contain the disease.
One of my favorite things from ATL/WTL was the _ASSERT_E macro which additionally converts the source expression to text for a better message to be logged
Yeah, but the macro system being so pitiful makes me long for one that allows something as magical as fiveam's is (https://github.com/lispci/fiveam/blob/e43d6c8e7da5a80d5c33e8...) instead of having to write special cases for unary and binary predicates.
Preprocessor is just doing text transformations on the sources.
It's not really something that can be fixed, other than moving away from the preprocessor and putting metaprogramming capabilities into the language itself (which C++ has been doing).
Template bloat, terrible compile errors, terrible debug build performance, 1 second of extra compile time per cpp file when you include ranges, and you can't step through it in a debugger.
[0] https://dev.epicgames.com/documentation/en-us/unreal-engine/...
That being said being slow or fast is kinda moot point if the program is not correct. So my advisor to leave always all asserts in. Offensive programming.
https://github.com/fiberfs/fiberfs/blob/7e79eaabbb180b0f1a79...
Abseil has the convention where instead of assert(), users call "CHECK" for checks that are guaranteed to happen at run time, or "DCHECK" for checks that will be compiled away when NDEBUG is defined.
https://github.com/abseil/abseil-cpp/blob/0093ac6cac892086a6...
https://github.com/abseil/abseil-cpp/blob/0093ac6cac892086a6...
`assert(vector.size() < 3)` is ridiculous to you?
But your meaning is clear. In an assert expression, don't call functions that might change the program/database state. Be as "const" as possible.
There are a few things like that, for example:
https://en.cppreference.com/w/c/numeric/math/isnan - isnan is an implementation defined macro.
https://en.cppreference.com/w/c/io/fgetc - `getc` may be implemented as a macro, but often it's a function.
htons(..) and related socket-utility names are also often macros, but I'm pretty sure there is not a std::htons(..) in the C++ standard, partly because 'htons' is not an attractive name. Since it's (sometimes) a macro don't qualify its namespace like ::htons(..).
A long time ago in the Microsoft C (and later C++) dev envs there were macros named "min" and "max", which I thought were terrible names for macros.
Yeah, this is still in windows.h unless you #define NOMINMAX
I remember having to guard against this in some inline code by surrounding the c++ calls with parenthesis, eg `(std::min)(a, b)`
I usually wrap Windows.h in a header followed by 100 #undefs to contain the disease.
https://github.com/fiberfs/fiberfs/blob/7e79eaabbb180b0f1a79...
In this case, the ability to see the actual values that triggered the assert is way more helpful.
It's not really something that can be fixed, other than moving away from the preprocessor and putting metaprogramming capabilities into the language itself (which C++ has been doing).
But I agree, fewer special tricks is better and that includes the preprocessor.