/tech/ - Technology

Only people who are terrible programmers complain about C.

>>537659

So, a growing segment of OSS then apparently anon?

>>537660

I might have to become a 9front user just to avoid all this shit.

>>537672

Well it is objective facts - more people are idiots than otherwise.

>>537682

A programmer didn't make that decision - HR did

Someone average with C++ can shit out software to do basically anything, and there's so many people who can do it that they're cheap. Note how all of these companies need bespoke and proprietary software - it's probably all as bad as you see in those CS grad threads. Shit buggy code that'll lock your brakes and disable your horn.

>>537659

Only CS grads circlejerk about C.

>>537691

And yet many large projects decide to use it instead of c++

>>537687

I missed CERN and NASA. They probably have higher standards because they release their code as OSS, iirc

>>537695

Oh wow, they use it over the only language worse than it.

>>537700

Sorry, that was disingenuous. Shit like Javascript still exists.

>>537700

>>537703

What are you even trying to say?

>>537708

Anon, I- I love you

>>537660

>mozilla

>gnome

>the entirety of the github userbase

Yes.

>>537748

>ur stupid

Nice argument.

>>537753

Well, it was the initial argument anyway.

>>537652

>implicating you need anything other than assembler and Lisp

Everything else is just a meme.

I know you are all perfect programmers who have memorized the C spec and compiler quirks from head to toe, and can read and fully understand any >3000 LOC project in just a few minutes, but you must understand not everyone is a supreme glorious aryan 240 IQ übbermensch like you, and other inferior subprogrammers will eventually write some errors in their programs or be unable to understand some projects due to their complexity and there is nothing you can do about it. Moving away from C is for the better, as inferior minds should not be allowed to touch something as powerful as C.

Although C++ has even more vulnerabilities than C, but hey, moving towards a slightly more abstract programming language is a start.

>>537695

>Not knowing why they use it in the first place

Wow, what a fag. I don't know how the fuck you pretend me to take you seriously regarding C with your expert knowledge about the language.

>>537840

Java is not as bad as /tech/ makes it out to be, but I will never understand how the fuck haven't D or even Go successfully replaced it yet.

>>537652

You can't just put up a giant piechart and say one language is more important than the other.

Jews are only 2% of the population of the US, yet they completely dominate the country.

It works this way everywhere else, too.

What's being programmed in C and C++, and Javascript, for that matter.

Guys, is C good for a first language?

>>537855

Yes and no.

Yes because it is simple enough and will teach you about low level programming, which you will need whenever you need to make something minimally useful.

No because you will have to fight the limitations of the language, and that will consume time that will not be spent thinking about algorithms, which is what programming actually is. Furthermore, you won't get to learn more interesting paradigms such as OOP or functional programming.

I'd say start with Python to learn the basic concepts of programming, then C to learn the ugly side of programming.

>>537860

Doesn't learning Python teach you bad habits?

>>537863

Why should I learn lisp over C?

>>537864

Python is a language for bad/lazy programmers.

>>537867

Still would rather apps be written in python than shit like C# or Java though.

>>537866

It's easy to pick up, while still powerful, and SICP is a god tier book.

>>537855

C is probably not the best language to learn fundamental computer science concepts. Perhaps Lisp, Haskell, or Python would be better as first languages.

You should learn C and Assembly, together, at some point. Perhaps not as a first language, but soon after you program for a year in your first language. While you do this, you should also read a book on how computer hardware components work.

>>537864

Python is basically "English: Programming Language edition". It feels more or less like reading and writing actual sentences, which is why it has replaced Pascal as the "pseudocode turned actual code" language.

It teaches you bad habits if you learn Python and The Python Way(TM), but not if you learn programming through Python. Just stop when you understand what's an object supposed to be, then move on to a more useful languages.

>>537870

SICP is about scheme, how different is that from lisp?

>>537874

Scheme is to Lisp as C is to C++.

>>537874

Lisp is not a language, but a family of languages. All of them are based around lists and lots of stupid parenthesis.

As for Scheme Lisps, they tend to be much simpler and reduced than other Lisps, like Common Lisp (which is what most people understand when casually talking about "the Lisp").

>>537863

>guise, you should totally learn this meme language that will never be relevant outside of academia circlejerking

Might as well learn COBOL and SAS at that rate.

>>537886

How powerful is scheme compared to common lisp and emacs lisp? Is it just as powerful?

>>537873

What bad habits are these?

>>537896

Writing functional software.

>>537897

That doesn't explain much.

>>537904

It does once you've spent enough time on /tech/.

>>537905

All I am hearing is people who don't know how to pick the correct tools for a job.

>>537652

If C is so terrible, why are kernels written in C rather than languages like C++ or Java or Python?

>>537920

Because unfortunately Unix was the OS that caught on and C is meant for Unix.

Windows' kernel is written in C++ instead of C.

>>537920

Explain how a kernel written in Python would work.

>>537929

Not him but if you used assembly for the more low-level parts it could easily be done. It would also be unbelievably slow.

>>537875

backwards, but yea

>>537920

muh_legacy, and a defacto yet accidental standard ABI in C.

>>537920

BTW, neither Java nor Python would be suitable as kernel languages, though C and C++ are.

>>538243

>C++

>http://www.boost.org/doc/libs/1_59_0/libs/geometry/doc/html/geometry/design.html

I would go to the ends of the earth to avoid your kind.

>>537896

I was mainly talking about "pythonic code", which often means "abuse import because there is a library somewhere on the Internet that somewhat does what you want to do but not really". I get it that the NIH syndrome makes projects slower, but you should learn to implement some things yourself instead of half assing everything just because you don't want to write much code and would rather use something that is barely related to what you want to do.

However, in general, interpreted languages simply won't prepare you for compiled languages, which is what makes most of the computing world work. Things like declaring global variables inside a branch inside a function are stupid things to do that can't be done in most compiled languages, but that are perfectly possible to do in scripting languages because they have nothing to predict during the compile phase. Sure, you can try to avoid these practices and be a great Python programmer, but no tutorials will mention what's stupid and what's not, just what you can do with the language.

Anyway, if you ever want to do anything complex, you will have to abandon the idea of using scripting for everything altogether. The thing scripting has is that it tries to abstract from you as much as possible, which makes it easier and faster to code but much slower during runtime. Like, Lua's tables are great and shit, but they will never be as efficient as a fixed size, correctly stored array. If you want to do simple webpage generation scripts it will be great, but if you want to make a 3D videogame engine, where you must take profit of shit like cache hits, it will fall flat.

Mind you, this is a problem with all scripting languages and not exclusively Python.

>>537870

>SICP is a god tier book

Stop with this meme, there is nothing of value to learn from SICP in the year 2016. Just learn python, you'll learn far more from that than you could from ISCP.

>>538265

SICP as a general introduction to programming book isn't that good, but I gotta admit it's useful if you want to learn how to write interpreters.

>>538268

It's awful. No one does things they way they do today. You'd have so much to unlearn.

>>538501

>year 2020

>no programmer under 30 can understand autotools

>all programmers over 30 have been culled to prevent the spread of wrongthink

>no maintenance can occur

>an attempt to rewrite autotools in javascript on node.js is begun

>year 2030

>all software on Linux has been unmaintained since 2020

>>537660

Is this a serious question, or sarcasm? I can't tell.

The answer is obviously yes.

>abomination language

>C++

>Javascript

Pick three

>>538567

As far as I can tell the question was meant to imply that "a growing segment of OSS" are terrible programmers. Presumably to make the user he's responding to question that assumption which is implicit in his own post.

>>537660

I hate that being "open source" is just a trend to follow now.

What is a good language to learn and stay with that can do most of what you need (software, kernels, etc)?

>>538511

autotools is even worse than this javascript hipster shit

at least I can ignore webshit

>>538581

>kernels

>anything but c and asm

>>538508

is K&R still relevant?

>>538581

C is a good language for OS development. I would never ever use it for software development, though.

>>538598

Just to clarify, is the book the C Programming Language, 2nd edition, still relevant today?

>>537652

I mostly like C and I want to become proficient in C because of its relevance and the importance of knowing your shit in order not to create a security nightmare.

However that doesn't mean I want to see C used everywhere. I rejoice from seeing less and less inadequate uses of the C language. Most application software should never touch something like C: it's a burden on the programmer's part, it makes development slower and its benefits on speed and memory usage are of no use to applications that would be much better served by a not-so-stupidly-slow memory-safe and garbage-collected language.

>>538618

IMO it's the best book for learning C up to an intermediate level, but it won't teach complete beginners the basics of programming.

You'd still want to take a look at something else to keep up with modern standards.

>>537652

Until ocaml replaces C++, there is still work to be done. It's a great start, though. Truly 2016 will be a fantastic year.

>>537855

scheme or lisp is better as a 1st language.

>>537659

Have fun with pointers faggot.

>>537868

This 100%. For local code projects I go from Shell > Python > C++

If I can't do it in shell scripts I use python. If python doesn't fit the bill I use CPP.

>>537855

Python. It will teach you structure.

>>538680

>not FORTRAN

>>538511

I will still be under 30 in 2020.

I understand autotools.

Your argument is invalid.

>>538256

All that overengineering is perfectly justified in a library that is supposed to deal with multiple coordinate types efficiently.

Not a fan of template overuse though.c'est

>>538835

If you want to write Haskell, write Haskell.

>>538750

Fortran has long-standing issues. It's a much better first language than C, but scheme or ocaml are the best first languages because they allow you to focus on programming (which is what you're trying to learn) instead of banging your head against a wall because you haven't mastered a shitty language before you started learning the shitty language.

>>538911

>it's appropriate uses*

>>538501

>pic

kekd

I love C++ and C. They are my two favorite languages, but don't fool yourselves, C++ is a better language in almost every single way.

Even on a basic level, C++ has better, safer typing, a preprocessor that is part of the language, better scope and name handling, actual generic metaprogramming, and reference types.

Literally the only better things about C are better ABI definition (as in, it actually has one), no weird name mangling, intercompiler portability, better linking definition, a light/no runtime, an easier standard library, fewer language features to learn, and better platform portability.

C++ frequently even runs faster than C. I use C when I have to target shit like AIX and HP-UX at work (because so many programs choke when you try to use GCC and their libraries are compiled with XLC or something), but other than that, sepples is way easier to improve and maintain, not to mention keep readable.

>>538911

Modern embedded development is actually done in C++, no longer C (or ASM).

The C language itself is a massive issue. It's much younger than many languages that were way safer, way faster and way higher-level when they were developed. The only reason why C took over is that computers worked more like smartphones, in that there was no restriction, initially, on what hardware and software could be run, because the barrier of entry was so low, so if someone came up with a "C computer" (that was bell labs), and then pushed everyone aside M$-style, i.e. not on merit but on money and name recognition (as they did), superior technology can easily be squashed. Nobody in the world liked unix machines when they were first introduced. They couldn't be fixed on-the-fly like lisp machines could, they were buggy as shit unlike lisp machines, and they were slow as shit compared to lisp machines (because lisp machines used special lisp-focused hardware and C compilers at the time were garbage). To this day, people try to take more and more features from BCPL/lisp/ML in their shitty C-inspired languages because of the damage C caused to the programming world.

>>538925

C's ABI is not defined in the standard at all, just like C++. The difference is that everyone used a compatible ABI so everyone could use every software, while everyone wanted to become the de facto C++ compiler and thus decided on an incompatible ABI on purpose. Due to this, the C ABI has a de facto standard, but it is not formalized in any way.

All C++ essentially is at its core is just a set of features for C. On a desktop or any modern system, you'll need those. Sure, on an embedded device or something special, C is a lot better of a choice ,since you probably won't need all that stuff, but the fact of the matter is you end up just using a bunch of libs. You may as well just use C++ since it's implemented in a common and highly audited way.

>>538929

C makes certain assumptions about an ABI easier though. You can't assume any specific name mangling patterns about C++, where C linkage generally requires little or no mangling.

>>538930

No it's not.

C++ is very similar to C in some basic ways and offers some fundamental compatibility, but it isn't well treated as a superset or an extension of C. It's far too different on every level.

C++ is its own language. It's not "C with classes and templates" like a lot of people assume. C code compiled with a C++ compiler frequently fails to compile, fails to run, or occasionally runs and operates differently than it would if compiled as C.

>>538935

C++ expert here. No, it's C with classes, templates, and exceptions. Most C++ projects prohibit most C++ features to make it more like C with classes.

>>539067

>Most C++ projects prohibit most C++ features

:(

Having only worked on small projects either by myself or with one other person, I don't have much experience of this. What kind of thing is usually banned?

>>539081

Nothing, that faggot obviously never worked on a real C++ project. Go look at the unreal engine 4 source, it's a great example of a large, professional C++ project.

>>539067

It's C with classes, namespaces, exceptions, templates, different type rules, different linking rules, name mangling, different argument rules, different naming rules, different scope rules, different parameter declaration and definition rules, and runtime type introspection.

Huh, almost like it's a completely different language that only shares the same base syntax or something.

>>539081

RTTI and exceptions used to be banned everywhere due to horrible performance until a couple years ago. You'll find they're still banned in many large projects or have just recently been allowed. Qt and Android (see: NDK) are two examples. Variadic templates are often banned due to no control of inlining without using compiler specific hacks (you wouldn't believe the problems until you've seen them, your mind will be blown). "boost-like" constructs and metaprogramming are banned frequently because they have a massive impact on compile time and make stack traces incomprehensible. Anything to do with streams are banned in almost every large project as they make internationalization impossible and are probably the stupidest language feature ever created.

>>539176

For an example of simple C++, you can take look at the Perforce C++ client API. When I was using it, about 5 years ago, it was just basic "C with Classes" or some such; and didn't even use the std::string class (maybe because the library is a few years older than the C++ standard?).

>>538863

or just pick Rust

http://rustbyexample.com/generics/impl.html

>>539557

Rust is SJW bullshit

>>538927

Are you suggesting that lisp should be a first language one should learn?

>>539571

Yes.

>>539605

why?

>>539428

A lot of Windows C++ libraries avoided using std::string because of the problems Windows had with passing them across a DLL boundary and the massive fuck up with wchar_t (especially on Windows). It might still fuck it up but I've not done Windows development in some years.

>>539693

they still do, messagebox is a massive example of this fuck up

>>539671

No boilerplate (java etc.), no shooting yourself in the foot due to the language being underdefined (C), consistent and extremely powerful syntax (unlike python, ruby, etc.), optional typing in a lot of dialects so you can add that on when you're ready (although I still think forced types with inference is better, it does a similar job), truly multiparadigm so you can try any paradigm from lisp instead of jumping languages, and the macros allow implementing whole new paradigms on-the-fly (I implemented HM on top of a type-less scheme once using only macros).

>>539671

Install Gentoo.

>>539712

Lisp is a terrible beginner language as it doesn't interoperate with anything, can't make anything normies would care about, requires rare tools in poor condition, its community died 60 years ago so there is little help available, and teaches incredibly bad practices stemming from being too far removed from the hardware. Javascript is a better beginner language in every way.

>>539887

You're only considering in the context of industry. The community didn't die 60 years ago, that's a gross exaggeration. It's evolved quite far, with Clojure being a reasonably popular modern implementation.

Help is available from generations of literature given its status as one of the oldest still in use languages, as well from Scheme being the language upon which SICP was originally based.

And why would a computer scientist/programmer give a damn about what normies care about? Writing formal descriptions of procedures isn't a normie activity.

>>539929

>The community didn't die 60 years ago, that's a gross exaggeration

stackoverflow only gets a couple common-lisp questions a day. It's been dead for ages. I'd hate to have a difficult question I need answered.

>And why would a computer scientist/programmer give a damn about what normies care about?

Some people like to get shit done rather than wank. If you can get shit done you can do things of relevance to normies and make money.

>>538835

No, boost's absurd templating and metaprogramming do not deal with anything efficiently. Smart pointers are almost the only parts of that mess that were salvageable.

>>539929

Enjoy your deprecated meme language.

>>540098

>this is what ctards believe

My sides!

>>540128

No, faggot. I'm a tard++ with life experience. But feel free to rewrite a kernel module using list_head with boost::intrusive to show the class how much easier, safer, faster to compile, smaller in size, and more concise boost can make code. Spoiler: it will be worse in every possible way despite being almost the simplest datastructure imaginable.

>>540158

>I'm retarded and nobody can stop me!

Why do I even bother with /tech/ anymore?

Is there any niche programming language nowadays that will land you a good job easily?

>>540168

How about you do what I told you to do. Show us the code.

>>538927

>. The only reason why C took over is that computers worked more like smartphones, in that there was no restriction, initially, on what hardware and software could be run, because the barrier of entry was so low, so if someone came up with a "C computer" (that was bell labs), and then pushed everyone aside M$-style, i.e. not on merit but on money and name recognition (as they did), superior technology can easily be squashed

Completely wrong; the reason why C beat out so many other higher-level languages is because it had far superior performance. That's a huge deal when the computers people had to work with still had memory measured in KB and CPU clock speeds measured Mhz.

>>540249

C was the slowest piece of shit ever made when it was released. It was significantly slower than lisp or forth or fortran for a solid decade. You don't even have the slightest clue. Fuck off.

>>540356

Nothing has ever been slower than lisp. A garbage collection sweep used to take so long you could go watch a TV show.

>>539887

Apparently you haven't seen QuickLisp.

The community is alive and well, and CFFI just works.

You need to unlearn your myths.

>>539887

Apparently you haven't seen QuickLisp.

The community is alive and well, and CFFI just works.

You need to unlearn your myths.

And fuck your flood detected when I get a 502 and it doesn't post

>>540400

My sides!

>>540176

Probably COBOL.

In my experience the programming language doesn't get you the job - or maybe I just know too many

>>539887

ECL allows Common Lisp to talk to C, while ABCL does the same for Java. Plus there's the CFFI. Common Lisp was my second language (after learning the basics of Perl), and by far it has been the most fun I've had programming.

>>540412

>the community is alive and well

>but it's so small you haven't seen it

>>540745

Every shortcoming of C++ is a direct consequence of the existence of C.

>>540745

Not to mention the fucked up way of doing things in the newer standards.

>>538927

It's just an anecdote, but I'm an embedded programmer and have used C++ for about 2 years back in the 1990s; C for the other 13. Everyone's definition of embedded is different: I'm currently working with 8 Kb of RAM and the company's coding standard prohibits dynamic memory management, which I think is close to impossible in C++ (think of all those temporary objects created just to evaluate an expression).

>>540745

>Every shortcoming of C is also true for C++

>Implying C++ is a super-set of C

Not anymore. Both languages are now free to have their own faults and fuckups.

>>537843

The main problem with java is the devs. Java is easy, so people don't take it seriously and only memorize a few commands before coding, making barely-working codeblobs.

>>541360

That problem is at least twice as bad in python.

>>538696

being this retarded to not understand pointers

I don't know where to start. Javascript? Python?

>>541388

lisp

assembly

or c

>>541390

>lisp

>assembly

Good choices.

>or c

Fuck no.

OCaml, though.

>>541388

1. Leave /tech/, too many people here have language waifus

2. Go to Wikipedia and look up programming languages. Just try a couple of them out, stick with whatever feels right to you.

>>541403

> Thinks learning a language is like test driving a bicycle.

Damn, son, you're retarded.

>>541472

Isn't it? I'd just take some code I wrote and re-implement it in the new language to get a feel for it.

>>541400

>using lisp unironically

>>541476

The syntax is usually insignificant and that's really the only thing you can get a feel for ata glance. The heart of a language is its philosophy and its community, and it is impossible to learn that unless you are already heavily invested in the language.

>>541477

What is Autocad?

>>537837

what is formal verification?

Although to be fair, Ada is better for that (which is its only redeeming feature whatsoever.)

>>537659

How's that OpenSSL going faggot?

>>541482

autocad is deprecated

>>541548

>blaming C for the OpenSSL guys tying themselves up in a billion platform-specific conditionals and rewriting their memory allocator to make finding these bugs nearly impossible

learn to program

>>541548

OpenSSL is poorly coded. It wouldn't matter what language it is written in, the guys who code OpenSSL are terrible programmers.

f I recall correctly, a lot of their code can also be described as an IFDEF maze. I never checked, just what I heard.

>>541479

>The heart of a language is its [...] community

That sounds awfully cancerous, anon.

>>541482

Autocad is dead like LISP. Everyone who isn't too old to learn something new is using SolidWorks.

>>541765

Bad community = bad library ecosystem and bad documentation and bug response discipline.

>>541707

>I'm retarded therefore I'm right!

Not how it works.

>>541677

>blaming devs for using the only mechanism the language allows

>blaming devs for bugs when the bugs wouldn't exist in any other language

OK kid

>>541827

>blaming devs for using the only mechanism the language allows

No, they wrote their own, rather than using the one the language provides, or even using one provided by any common library like tcmalloc or jemalloc.

>blaming devs for bugs when the bugs wouldn't exist in any other language

You can achieve retardation of "writing my own terrible memory manager" levels in any language. You can even write your own memory manager in Java, given enough tenacity and stupidity, without even having to resort to JNI.

>>541834

I lol'd. Posts like these are why I don't visit /tech/ anymore. People who know what they're talking about never post anymore, only retards who couldn't even tie their shoelaces, spouting /x/-tier nonsense as if it were a god-given truth.

>>541841

Nice argument. I've been contributing to the discussion and here you are just throwing FUD everywhere and waving your superiority complex around.

If you're going to argue, argue. If not, fuck off and stop shitposting.

>>537691

Hskell...

>>541332

>C++ is a super-set of C

I didn't imply that.

C++ isn't memory safe by default, just like C, regardless of whether it is a C superset or not.

Well, he actually fucked off and stopped shitposting. The absolute madman.

>>537863

I'm going to blindly do what you say as LISP seems interesting and I need to learn a language.

Thanks anon.

>>546215

But first, you must install Gentoo.

>>541707

It's true. Here's a general overview.

http://www.openbsd.org/papers/eurobsdcon2014-libressl.html

Unfortunately I can't find the article about its poor explicit_bzero implementation. The problem with regular bzero is that compilers can optimize this function out, which is bad if you needed it to clear sensitive data from memory. Several platforms have a variant that is designed to not do this. OpenSSL wrote its own, but not only is it an ugly function, they also screw up in its usage (passing sizeof pointer instead of the size of what was being pointed to).

>>541827

OpenSSL barely even use the language. Whatever can be rewritten is rewritten. malloc, snprintf, explicit_bzero.

>>546239

Oh, it was this one.

http://www.viva64.com/en/b/0178/

So yeah, marvel at that function.