Thread 105885207

>mommy!!! I want to build a highly concurrent program but learning to use C correctly is sucky-wucky, I need a language that handles all that icky process scheduling for me at the expense of being slower than molasses ;_;
If you use this it’s even more embarrassing than using jeetlangs like java or python.

>speaking like a female child to make a point
embarrassing, you're a 20/30s year old grown man. Act like it.

Looks like someone got filtered hard.

>>105885207 (OP)
>Highly concurrent
I present to you: The uncontested champion of concurrent programming.

>>105885220
I’m a woman thoughbeit and I exclusively use C and Rust

>>105885225
You got filtered by thread local arrays and locks? Yes you did.

>>105885278
Nice and you are very valid. Just out of interest, were you born with a vagina?

>>105885207 (OP)
As a member of the Brahmins, I can confirm that all functional languages are only used by the Dalits.

>>105885312
Yes, but because of a birth defect it was covered up by a penis and testicles until I had them removed.

>>105885207 (OP)
>build a highly concurrent program
>use C
pick one

>>105885207 (OP)
The computer should make life easier

>>105885597
Erlang itself is written in C, meaning Erlang’s concurrency is ultimately C’s concurrency, just dumbed down for söydevs

feel free to write all of your apps in hand optimized assembly the rest of us will be making systems that actually work

>>105885782
C is itself written in Assembly, meaning C's concurrency is ultimately Assembly's concurrency, just dumbed down for söydevs

>>105885812
shut the fuck up, tranny

>>105885207 (OP)
erlang is dead and buried as nokia and ericsson are migrating to C++ because it's just a better language anyways
source : me, I've worked on this

>>105885938
I worked on erlang for a while.
Not being able to change variable values takes some getting used to.
> what about efficiency?
You won’t find any, no.

>>105885938
>are migrating to
Is it 2002? They migrated decades ago.

>>105885266
Less baby-tier than Erlang but still baby-tier. Imagine letting a GC handle memory for you. What’s next, letting it handle your sexual duties to your wife?

>>105886713
It's a good job I dont have a wife then isn't it.

>>105886736
ESLs are not allowed in this thread.

>>105885207 (OP)
I did my capstone project in erlang. It's a neat language and I would like to see how it handles web backend stuff, since all I've done in the backend is either some php stuff or java/kotlin with spring.
Erlang is just a nice and concise language. Never see it talked about here though, let alone irl (outside of college).

>>105885986
>Not being able to change variable values takes some getting used to.
Really? I guess it could depend on what kind of work you're doing, but in web, almost everything has already headed in the "functional" direction. At work, we very rarely mutate a variable, there's just no need.

>>105885358
How does it smell now?

>>105886883
>Using the most dysfunctional, laughing stock ecosystem as an indicator of where serious programming is headed

>>105887216
It's where all the buzz is.

OP failed an elixir interview and is seething (and indian)

>>105887339
If by "buzz" you mean "low IQs", then you're right.

>>105885207 (OP)
>>105885278
ywnbaw

>>105888226
Those 2 are often comparable, yes.

>>105885207 (OP)
If its so great why cant they post high score on techempower benchmarks?

>>105885782
it's called higher level programming, dumb cnile

>>105886713
I use only sticks and stone as a real MAN(ape)

>>105886713
YWNBAW

>>105885207 (OP)
>but learning to use C correctly is sucky-wucky
You're either a Russkie or an African being paid by them.

>>105885207 (OP)
Everything is slow compared to C. That said, i saw some comparisons and Go is definitely faster than erlang based on latency, but I was told that there are cases with parallel programming at scale where erlang beats it.

From what I've read the main advantage of erlang and elixir is that they scale well and backends written in them are less likely to fall apart under high load.

>>105885207 (OP)
Elixir is actually really nice language
> it's slow!!!11!
yeah, you know there are other metrics to distribute sw, like fault tolerance and such you code monkey

>>105886883
Javascript/webdevs totally don't jump on the 'new thing is best' bandwagon every 2 weeks right? FP has some nice concepts but if you want to do right by customers and provide a performant solution then it's doa. FP is generally slow as shit, especially forcing JS to be like a functional lang (it's .foreach() and .map() etc can be 5x slower in some cases) and every time you see a functional lang perform like an imperative lang, it is because the compiler was able to rip apart all the functional shit and replace it with loops and in place mutation.

>>105893497
>Fault tolerance
Skill issue. Write better code (hard to do in a shitty language with dynamic typing).

>>105886713
>the computer can clean up unused memory automatically but he does it himself
>he does the computer's job
>for free

>>105893646
>Skill issue. Write better code
Fault tolerance is not a matter of code correctness you absolute mongoloid

>>105885597
Have you heard something called “operating system”?
https://pages.cs.wisc.edu/~remzi/OSTEP/

>>105893635
>it is because the compiler was able to rip apart all the functional shit and replace it with loops and in place mutation.
and this is bad because....???
if i get the superior code clarity and expressiveness of functional programming without the cost because the compiler can deal with the plumbing for me then I'm all for it. we should do more of that instead of insisting on doing the plumbing manually because muh real programming or whatever. remember the purpose of writing code is to make things and get things done. otherwise we'd all just be manually toggling bits and jerking off to imperceptible speed improvements

>>105885207 (OP)
reminder that erlang is better at handling bits than C, the supposedly "low level" language

>>105893710
ah yes, I will build my own to enable highly concurrent programming
thanks for the tip!

>>105893646
A C program that does what Erlang can do will end up just being a shittier, ad-hoc reinvention of Erlang.

>>105893646
A cnile trying to write their own concurrent message passing supervisor model would be hilarious.

>>105893870
What does that even mean?

>>105885285
lmfao
you have no understanding of BEAM's scheduler

>>105885807
Barely anyone uses Erlang or Elixir. Even Ericsson barely uses it anymore. There are far more highly concurrent systems written in C and C++.

>>105895032
to expand on this:
- processes never share memory: no pause-the-world GC, a failed process cannot affect the state of another process, cache respectability, natural representation of horizontally-scaled programs
- processes run only a configured number of operations before yielding to another process: no hogging of CPU time by any process
also, Elixir is one of the nicest languages for describing complex algorithms and data transformation pipelines. the standard lib includes a fuckload of things, including directed graphs and the ability to set a memory capacity on any process (useful for things like running sandboxed code).
Elixir's Phoenix and Ecto libs are wonderful for interacting with humans and databases.

>>105895129
Wow, that's so cool, it does what every single OS can do with 0 configuration.

>>105895148
no

>>105895158
You're actually retarded.

>>105895176
am I?

>>105895190
Yes you are.

>>105895057
ericsson stopped using erlang because they didn't want to have to pay people to maintain it then went back to using it again in some products because it worked so damn well.
erlang is still highly relevant in its niche, just because (You) haven't worked with it doesn't mean no one else has

>>105895215
so you're telling me that a stock Linux configuration can run hundreds of thousands of processes all synchronizing their efforts to deploy a soft-realtime program in a manner where message passing is abstracted such that remote processes are interacted with in the same way that local processes are?

>>105895246
Yes.

>>105895251
no, it can't. Stop telling bullshit
not him btw

>>105893793
You know webdev includes backend and DB too?
We care about performance a lot, and I've spent months on just optimizations. Not once did I go through and remove maps and replaced them with a faster raw loop.
Why not? Because those are hardly an issue. The biggest issue is always network speed (especially with our customer base). So front to back communication, DB queries, and back to back communications were the forefront of all optimization efforts.
Made the overall usage of the webapp faster by 10x (100x in some smaller areas).
Changing FP code over to traditional imperative style would never give those speed increases.
Micro optimizations are only stressed by retards that literally can't architect code.

>>105895271
Meant for >>105893635

>>105895264
The real answer is it doesn't need to, because it's not irrelevant pajeetware.
Also IP protocol is older than erlang, and most definitely older than you, pipsqueak.

>>105895292
>Also IP protocol is older than erlang, and most definitely older than you, pipsqueak.
Are you actually retarded? The IP protocol? What are you talking about?

>>105895292
>The real answer is it doesn't need to
But you have to when your application must be distributed over network, dumbo

>>105895271
>internet speed is the problem
Rajesh, I could walk in, firmly shake hands with your manager, tell him about protobuf and why it's better than json, and get you fired in the same minute.

>>105895357
Network speeds haven't been a problem for years, maybe they are in india...

>>105895337
You think the technology is the problem? We already gzip everything. When I say internet speed, I should've mentioned its more of a bandwidth problem since we only have so many instances handling 10s of thousands of users. We were having tons of timeouts because of traffic overload.

>>105895370
You know that internet speeds vary drastically across the US, especially with the military, not to mention all the international bases.

>>105895390
>>105895406
You know, I assumed white country + datacenter ownership + hundreds of millions users globally, not some pajeet rentoid vps problems. I'm sorry for everything I posted, I concede, I truly wouldn't know how to deal with this, because I'm white and would be willing to invest into real hardware with white skin and working brains so the problems you're having cannot happen to me.

>>105895441
Literally everyone i work with is white.
I think the real issue here is that you don't work at all and have no clue about how the real world works.

>>105895441
>he hasn't set up synchronous replication across several datacenters in different states for a Postgres instance that needed fault tolerance

>>105895370
Internet speed would be a problem if were not for caching/replication/clever distributed algorithms
You would need 10s of seconds to open the "nearest" website on the fastest commercial connection available on the planet

>>105895219
No they stopped using it primarily because it has absolutely awful performance and their routers and switches were getting absolutely blown out by their competitors who almost exclusively use C and C++.

>>105893635
compilers take functional code, transform it into imperative machine code, then CPUs take imperative machine code and try transforming it into a functional stream of microops so the cpu can fill as many instruction pipelines as possible and run them concurrently

you've never bothered looking at modern CPU architecture and it shows. even the PowerPC 970 from yesteryear is designed like this

>>105895491
Latency isn't speed, rajesh.
>>105895455
You're literal bottom feeder telling me that your basedware struggles with tens of thousands of users. That can be served by one stinkpad.

>>105885207 (OP)
I don't think you understand how distributed systems work. Erlang is the undisputed champion of distributed systems despite being so ugly.

>>105885266
It's not. It doesn't do anything particularly different from all the other imperative programming languages. Even fucking Java already caught up.

>>105889248
Notice that the stinky smeller BEAMer shills ITT ignore this post. Results talk.

>>105894960
erlang is inherently concurrent because everything is just a process based around message passing.
it has what's called a Supervisor model which is just a hierarchy of processes managed by another process.
https://www.erlang.org/doc/apps/stdlib/supervisor.html

>>105893635
> Functional programming
> JavaScript
JavaScript is not a functional programming language, that's why it's slow. It doesn't have optimizations. If you want real functional programming you use Haskell, not JavaScript, or any imperative programming language pretending to be functional. If I'm going to use one of those the minimum I accept is Scala. Everything else is just pretend.

>>105895543
You really don't understand anything, do you?

>>105895527
The only compilers that can reliably optimize functional code are GHC and Rust via LLVM and many of those optimizations are only possible because of mutability tracking. Ocaml can do a small number of optimizations with Flambda but that's unsafe in many cases. None of the popular JITs will optimize even basic functional combinators.

>>105895631
In the case of GHC it's also because it splits pure from non-pure. Rust can't do that.

>>105895673
GHC has a lot of degenerate cases though where fusion will not happen and it results in it being slow as shit.

>>105895742
Yeah but it's constantly being optimized. It's good enough to compete in performance while not allowing fuckery with mutable state. This is specially important with concurrency because optimal concurrency doesn't have locks or shared state, which makes virtual/green threads possible. You can carefully do the same thing in imperative languages but it's more likely you're going to fuck up or it's suboptimal. Erlang does this sort of trickery too but in a different way with messages and Akka copied.

>>105895627
I understand that bandwidth isn't a problem when you're under 100k users.

>>105895520
you have no idea what you're talking about go back to your containment general
>>105895615
FP for frontend webshit can have decent performance if you use a language that compiles to JS, it's much faster than hand written functional style JS

>>105895190
Hey look its you! I wondered where you went, you spam your shitty lisp code that does nothing but play with a few hash maps and pretend to be clever. Worried you 41%'d yourself because you got a pretty big beat down last time you shown your head. Glad you didn't.

>>105895963
QRD? I always loving dunking on elixir dweebs

>>105895963
if you had a bitch, I would rail her and then have you eat my cum off her face

>>105896053
thank mr skeltal

>>105896053
If you had a wife, you wouldn't be making up these situations that will never happen.

>>105896053
eat a sandwich you zombie retard no wonder your brain is atrophied

>>105895883
I would be happy to show all of our statistics if I was allowed to.

>>105896108
You aren't allowed to, because everyone's sides would go up into orbit and blow up this entire solar system if we saw.

For me, it’s clusterFORHT.

>>105895900
Erlang is slow as shit. The Erlang JIT is absolutely pathetic and is decades behind V8, the CLR, the JVM and even SBCL. Deal with it retard.

>>105896212
>and even SBCL
lisp weenie fails to understand why languages get used for things, what a shocker

>>105896222
The only valid justification is incompetence.

Elixir? More like He-Licks-Her!

>>105896222
There is no reasonable justification for Erlang or Elixir. You're probably going to scream about parallelization/concurrency while continuing to ignore the fact that most decent languages can achieve far better performance even when restricting themselves to single threaded synchronous code. It gets even worse for Erlang when you consider that basically every serious popular language now has support for immutable data structures and actor frameworks that go far beyond what OTP offers. Distributed Erlang has also always been an insecure joke.

>>105896313
you are still completely ignoring the unique runtime in the BEAM
perhaps you are too fucking retarded to grasp the nuances of those architectural decisions

>>105895543
>Latency isn't speed, rajesh.
It's the same thing from user's point of view ameritard

>>105896334
> muh processes
> muh otp
> muh ets
There is literally nothing interesting about BEAM and that's why it has failed to capture basically any marketshare.

People who only see the concurrency aspect of BEAM are the real jeets.
Show me your C program that can spawn tasks that can be horizontally scaled at will while having a built-in crash recovery.

>>105896371
Show me your reliable supervisor tree in a non-trivial application.

>>105896370
you are not pointing out the correct things, try again

>>105896385
Oh, so you can't do that in C without essentially recreating the entire BEAM? Cool, gotcha, brownie.

>>105896398
>brownie
Dynamically typed languages are exclusively for browns.

>>105896398
Filthy Dalit

>>105896371
Isn't this easily achieved with a bash script + your os of choice?

BEAM’s main strength is fault tolerance, which has been completely nullified by Kubernetes. I’m smart and post on hacker news and I’ve shipped plenty of production code so I know what I’m talking about.

>>105896484
Embarrassing.

>>105896501
Well you can automatically restart crashed processes using bash and you can easily spawn whatever the fuck you want in bash. The os is a far more powerful process manager than your programming language. Seems like someone is angry that BEAM is a meme.

>>105896498
BEAM's fault tolerance is mostly a myth.

>>105896513
you are retarded

>>105896568
Ad hominem doesn't disprove me. I win, you lose. Beam == meme.

>>105896599
Oh and those dubz proves it. Flawless victory.

>>105893823
>reminder that erlang is better at handling bits than C, the supposedly "low level" language
So is Common Lisp.

>>105896498
>his service requires external build tools to poll, ensure stability and requires looking at 5 dashboards to ensure everything werks
ok i just deploy my one elixir codebase and close the laptop

>>105896513
>one task crashes
>entire process goes down
>10s of thousands of clients disconnect because of a single client causing a segfault
Great job, go back to writing FizzBuzz.

>>105896623
You aren't serving tens of clients.

>>105896657
It's about the principle you drooling nigger, it does not matter if you have 2, 100, 1000, or 10000 clients, as long as a single NULL pointer dereference in your outdated boomer language can take down the entire process and cause all of the clients to get disconnected, C is not a viable alternative to Erlang where the software requirements call for it.

>>105896615
Sounds like a skill issue to me.

>>105896702
I'd rather have 500% faster software that crashes and use my increased profits to pay for the errors. This is what every successful business does.

>>105895129
>including directed graphs
When did they add DAGs to the standard lib?
I had to write my own DAG implementation a decade ago for a job

Guys, stop trying to convince retards like OP that elixir is good.
I'd rather these fags stick to C++ and other meme languages

>>105897351
https://www.erlang.org/doc/apps/stdlib/digraph
idk when it was added

>>105897529
oh its part of OTP
i thought it might be something new in Elixir

>>105896847
You're so braindead it's not even funny.

>>105893823
Bitwise operations are Erlang’s worst area. You are a nocoder

>>105897373
>meme langs like C++
C++ is the second least meme lang in existence behind only C.

>>105897699
>he doesn't know about pattern matching on bits
lmao

>>105897752
>running a virtual machine to do bit operations
LOL, erlang is trash and you're all shit programmers

>>105898556
>not running a virtual machine
https://github.com/phoenixframework/phoenix_live_dashboard/

also all of you performance faggots are forgetting that you can write C/Rust NIFs when actually needed

>>105898780
>library for a framework used for a language written in another language that compiles to a VM that is written in C
VS
>write it in C
>>105898838
Yeah but they’re extremely limited. Anything longer than 1ms can kill the BEAM VM. Real number crunching on huge datasets will never work in BEAM.

>>105898853
>what is a dirty NIF
bro get your head out of your ass and actually build something

>>105898902
Still has to yield more frequently and still substantially less efficient than just writing it in a real language to begin with.

>>105885207 (OP)
agree. yet another proof that the >muh FP language retards are useless

>>105898838
All of you "performance doesn't matter" faggots forget than in reality people will use node.js then upgrade to C++ without ever touching this turd.

>>105897717
C and C++ share same spot, that's why many people correctly intuit C/C++ and they are correct to call it that, however they cannot explain why that is, and their incompetence creates a false gap.
Optimally written C++ absolutely BTFOs C, but optimality requires discarding your shackles of ignorance and knowing a set of all target architectures, kernels, and usecases. It requires writing quite a bit of inline assembly, disabling rtti and exceptions, and most important of all, not using the godawful libc/libstd++. Most of you aren't competent enough to do that, but it's not C++'s fault.

>>105897717
The High Priests of a low cult.

>>105896782
dont tell me you actually enjoy kubernetes

>>105898838
NIFs are one of the more annoying things with erlang, its C interop unfortunately is not good, which is why it languished in the desktop space when it should have dominated.