C++ Moment

devilish666@lemmy.world to Programmer Humor@programming.dev – 24 points –
32

Meanwhile Rust: you might get an error at line 45 word 3 because it assumes variable foo is an int32 but it could be (whatever else idk), let's not compile this before you correct this by changing line 43 in this specific way. Here's the before and after code snippets so you can just copy-paste the fix.

Man I fucking love the Rust compiler. Easily the most understandable and useful error messages I've ever seen.

Have you seen Elm's error messages? They were what inspired Rust to have its error messages.

I like how Elm error messages are like the compiler talking to you as a person.

In my IDE there us even a button for accepting the compilers recommend fix. This is only possible because the error messages and recommendations are that good.

Hm. Rust sounds better and better every time I hear something new about it.

JavaScript: error: undefined is undefined or some nonsense like that. Sorry to repeat the old JavaScript bad, but I really hate debugging JavaScript!

Someone needs to be introduced to gdb...

have fun without those debug symbols

Why wouldn't I have debug symbols in the software I'm developing?

And what happens when you release it?

If you want the same traces as Java and python in the meme, you leave them, if you don't you strip them. Or you ship them separately. You decide, like a big boy.

Have the user compile it without debug symbols to save space. If the user has a problem they can just recompile it with debug symbols and see what went wrong with gdb.

It's been a minute since I used C/Cpp but if you compile with debugging symbols and using gdb give you info like in Java? At least the location of the crash.

And then you realise the program doesn't crash when compiling with debug symbols 😢

Then it's time to have a closer look at how your concurrent threads are behaving and where you missed a sync point or mutex.

That's when you break out valgrind because you certainly are using uninitialized memory.

And much more, it tells you each operation it goes through, where it is in the code, what's in the registers and more.

but with a core dump you can just load it up and see the state of the process when it crashed...

I'm trying to remember the last time I actually had a core file. I think core dumps have been disabled by default on Linux since at least 2000.

They are stored in the system log and thus rotated automatically to save storage. At least in Arch.

I use Arch, BTW.

But it returned 139! That's a start even without a debugger!

Story time. Back at uni I had a c++ subject. Me being lazy as fuck I didn't attend many classes and let alone do the practicals during the semester. Exam time comes around. I realise I can't cram in a whole semester's learning in a week. Luckily it's open book exam. Big brain time, I print the whole c++ documentation to take into the exam. I frantically page through the hundreds of pages in my lever arch file looking for answers. I pretty much copy and write example code to questions. Very sad when I failed.

gdb> break before it crashes
gdb> record full
gdb> continue
(segfault)

gdb> set exec-direction reverse

I have segfaulted in all those languages.

The code editor I had to use for Java once didn't give me anything like that.

Meanwhile for C you can just use gdb, it's great!

And C/C++ are like that by design. Compiled languages were new and the developers were afraid additional checks would decrease performance. It was certainly performant in racing toward a crash.

And C/C++ are like that by design. Compiled languages were new and the developers were afraid additional checks would decrease performance.

If you have a credible citation showing that was what guided K&R's decisions, I think you should post it.