Worthwhile websites?

t. PHP PRO

C was fine compared to the alternatives when it was invented, but it hasn't kept up and just isn't a very good language. Can you justify this?
tedunangst.com/flak/post/to-errno-or-to-errorr

this is a myth

holy fucking shit you guys are so gullible aren't you?

...

Fixed link:
tedunangst.com/flak/post/to-errno-or-to-error

I don't mind being able to do something in multiple ways. It's having the standard library (i.e. the language itself) do something in multiple incompatible cryptic ways that pisses me off. Do I need to use the return value? If so, should I check for 0, -1, or something else? Can the error return value also be returned when nothing goes wrong (atoi)? Do I need to pass a pointer to the function to record the error? It's a mess of incompatible hacky solutions, and it's not merely possible to use multiple of them, you're practically forced to use multiple of them. Consistent error-handling in a style that fits C is not impossible, just look at how Go does it.

zombo.com
somethingawful.com
voat.co
reddit.com

C was designed to be minimal language to allow people to implement C compilers in assembly.

Or in other words, it's the brainfuck of systems languages.

People should probably stop bringing it up, since it's an extremely predictable thread derailer. There are far too many delusional idiots out there who still worship it as the pinnacle of languages or some other such bullshit.

It's not at all fair to compare C to brainfuck, because C was actually quite practical to use 20 years ago, especially when the overhead of garbage collection and such would have been too much. And lot of microcomputer operating systems back then didn't even have memory protection either, because that was also too much overhead (and one of the reasons AmigaOS was so efficient).
But now C does have some pretty heavy downsides though given all the undefined behavior that's handwaved as something for the programmer to just deal with. And that's not really safe in today's environment where almost everything is connected 24/7 to global Internet. Sure you can run OS with "mitigations" but that's just addressing the symptoms, like bailing out water from a sinking ship. When the ship is old and springing leaks, it's time to redesign from scratch. That includes also ditching the 40 year old Intel architeture.
But I'm not holding my breath, so instead my attention is devoted to 80's computers and languages like BASIC, assembly, Forth that are still fun for me. Everything modern is too depressing to consider, and I'd rather just not have anything to do with it at all.

You can think what you want about C. It'll still be my preferred language. The problem is that people think they can just learn C, without knowing anything about machine architecture. That's insane. When you know C, you know what the compiler will output.

I think the best way for people to learn any language is to code, compile, debug/disassemble and do over until you understand. The real problem is people who think that programming is just about learning a syntax.

>preterhuman.net/
here have phrack.com instead

Still wrong. Knowing machine architecture doesn't give you shit all for writing conforming code. As soon as you switch compilers, compiler versions, or port your code to some new trendy machine, your codebase will be potentially full of security vulnerabilities. Even on a fixed target with only one compiler, you can't guarantee non-confirming code will be safe.

So I've heard. Also you can do basically anything you want because you're the only person who knows what you are doing.

Oh really? Then what is your solution?

Oh of course, a Rustfag. Okay, opinion discarded.

He was joking.

Your argument is bullshit. If I take your program and poke holes in it, it's full of holes!

No.

Right, no one says that. You also write your code conforming to a C standard. If you want a compiler that cleans up for you, when you don't know how to code, then that's your problem not anyone else's.

You try to build a case where I would be using a faulty compiler that introduces bugs or blindly compiling code that is not portable on a platform it is not portable to. If I started being so headless I wouldn't be following my own advice would I?

Error error! Non-confirming code!

thank you C

FTFY

Totse was never good. I wish I could get back the time I wasted on that pos website.

This isn't about cleaning up for me. Writing conforming C/C++ is an order of magnitude harder than any other mainstream PL (not to mention you have to pay $100 to read the spec for ANSI C). Also the environment and libraries used matters as well. For example I bet you couldn't write a conforming signal handler that merely sets a variable (got_interrupt) which is polled by the main thread.
No, it is in fact everyone else's problem, and not mine, because I've been using memory safe PLs for the last 10 years (I write C too but not for anything high assurance). Most C code does not conform to anything other than some de-facto version of C people believe to exist.

I fail to see that C is the problem. It seems like bad programmers are.

The core of my argument was that you shouldn't just code and cross your fingers. You should disassemble and analyze the result. I see people asking stupid questions all the time that should have been answered by a debugger.

I agree with you that we (obviously) have issues with quality software and that it appears in plain sight with low level programs done in C. I however blame lazy and ignorant (as in opposite of curious) programmers.

To get better software we need better programmers and my proposal is to obsessively use debuggers, disassemblers and reverse engineering to teach programming.

If you're not curious enough to take stuff apart and put it back together you're never going to amount to more than a mediocre mind.

If people are insecure about their C programming skills, the compiler often offers a lot of flags that will protect you, just like any "safe" language claim to do.

C and Unix were made by bad programmers.

hack.org/mc/texts/classic-multics.pdf

I'm not saying C is the problem, I'm saying what's needed to write C properly. Dissassembly isn't enough, you need to know the C spec. I wrote assembly and reverse engineered compiled C/C++ for years before writing C and it didn't help much.
Not that I know of. There's valgrind and stuff like that but that wont save you.

It's not about the programmer. C, just like Unix, wasn't designed to be good, just "good enough". It was always a compromise.

Totse was always pozzed, Seth.

stop spamming this bullshit, just look at the url you cant market dickwad

...

What is exploitable code?

Jim P?

you're blaming C shortcomings that make sense in the context of C on C++ which aims to be backwards compatible with C. if you don't like how the standard library handles errors then make a wrapper that handles it in the way you want or write your own function

Nothing in tedu's rant/flak page mentions C++. It's all about straight C.

In GCC there is the -fsanitize flags and many other instrumentation options.

gcc.gnu.org/onlinedocs/gcc-6.3.0/gcc/Instrumentation-Options.html#index-fsanitize_003daddress-947

Also on the clang docs page it's obvious at a quick glance that C programmers are not blind, to the cases that need extra safety precautions.

clang.llvm.org/docs/

I personally don't understand the argument for new "safe languages", when you can opt-in for safety instrumentation with C too.

I'm not blaming C++ for anything. I don't even know C++.

I could fix it with tons of wrappers, but that doesn't justify C's shortcomings. I'd be turning it into another language if I did that.

The only sites you should ever need on a regular basis are chans and RMS's blog.

How is it different? It's going to be compiler instrumentation features in any language. It's not possible to write a language that will catch runtime memory errors at compile-time.

C as a default does not add instrumentation (I like that). Other languages as a default add instrumentation (Okay for me if it's easy to disable).