Thread 107152975

Within the next two years, all AMD and Intel cpus will have BUILT IN memory safety at the hardware level. Which means all C/C++ code will have the same level of safety as Rust code.

https://community.intel.com/t5/Blogs/Tech-Innovation/open-intel/ChkTag-x86-Memory-Safety/post/1721490

Complimentary video:
https://www.youtube.com/watch?v=XJQ3XGkGrgc

Also there's Cheri, which is even more safe than Chktg, but it is for Risv/Arm:

https://cheri-alliance.org/wp-content/uploads/2025/03/CHERI-basics.pdf

>>107152975 (OP)
I trust RUST just about as much as i trust jewduke to be correct about anything.

>>107152975 (OP)
Too late chud, Rust™ has carved its own niche (writing colorful skins for traditional command line utilities) and it's not going anywhere. U mad?

Between this and Fil-C rust is officially over

>Fil-Meme
lol

>>107153655
lol
the "sanitizer" tards are much bigger of a meme tbf, and this is a variant of that tardism.
of course, none of them code or know any low-level shit. in the real world, one distro (chimera) tried to build with something as small as cfi by default, and discovered the harsh reality quickly. lmao.
https://github.com/search?q=repo%3Achimera-linux%2Fcports+disable+cfi&type=commits

>>107152975 (OP)
As a trans person and a member of the lgtbq++ community I disapprove of this technology.

>>107153722
lol
I tried compiling a few of my programs with Fil-Meme and they just crashed a lot, despite running with absolutely no problems normally. It really doesn't like unaligned accesses (not a security bug) or custom allocators (because it has a garbage collector).

>>107152975 (OP)
sounds neat, but why wait until runtime when you can know at compile time whether there are any memory safety issues?
the linked article is a bit sparse on details. i wonder if the proposal also deals with things like race conditions

>>107152975 (OP)
>Same level of safety
Assuming all your allocs are from malloc and you aren't using massive alloc pools for your shit. Memory tagging isn't a panacea. Correctness is the only solution and Rust is closer to correctness via its model of borrows than anything else while still being actually fast and no forced runtime.

>>107153610
>shitty runtime checker that kills performance
>ending rust
Delusional.

>>107154178
that doesn't even come close to covering it. lol. when it comes to memory, it's not just about the where (location) and when (lifetime). the what (type) is even more important.
this shit can only cover the where and when, and not even all of it (e.g. doesn't cover synchronization). but it doesn't cover any of the what.

>>107154393
Ya. Exactly. It's materially better than nothing if you still have shitty software everywhere, but I don't buy the overall hype.

>>107154446
>It's materially better than nothing if you still have shitty software everywhere
even then, all the shit software will just crash, BUT IT WILL CRASH EFFICIENTLY.
lol

>>107153547
Communist shitskin detected.

>>107152975 (OP)
>memory safety at the hardware level
You still need to manage that memory as unsafe in C++ unless they update the language, lol.

Rust is a good modern language combining type safety, functional and OOP style, and concise syntax for fast development. It will not go anywhere.

Rust is under Mozilla and will continue to be used there as C# is used around Microsoft.

>>107154666
>Rust is under Mozilla
what products does mozilla offer these days?
firefox has 2% market share
what else do they do?

>>107152975 (OP)
good while rust can be fun the community/cult behind it really makes me want the language as a whole to die in a fire.

>>107154666
>Rust is under Mozilla
didn't know that, since when?

>>107152975 (OP)
>BUILT IN memory safety at the hardware level
is this just a long-winded way to say more hardware backdoors?

>>107152975 (OP)
So the program will just segfault?
That IS an improvement.

However, I'd much prefer a language where the program doesn't segfault when it doesn't have to, by ensuring the memory accesses are valid.
That's where Rust delivers. There and in developer experience. All those type hints actually help you avoid all sorts of bugs if you know what you're doing.

why did they name the language Rust?
Everybody knows that rust destroys things

>>107152975 (OP)
Hardware is going to cripple your low level dreams. That's my prediction at least... C was a language I enjoyed using but I think a lot of the reason was autistic reasons like "feeling" like a retro boomer in the 2000s programming The Sims 1.

When I use a modern language like Go the level of productivity is really shocking. It just feels like everything you want to accomplish is mere seconds from your fingertips... I don't get to LARP that I'm programming Rollercoaster Tycoon or that the year is 1998, but the software does benefit. Especially how insanely cross platform all the functionality is. It does so much shit to the filesystem and works identically on Linux, Windows, whatever... I think this is going to be the future for a lot of reasons... Hardware is going to allow for it and implement things like this into the actual physical devices themselves.

>>107155067
>When I use a modern language like Go the level of productivity is really shocking.
what have you produced in go so far?

jew eceleb tricks to get views no thanks lunjew

>>107152975 (OP)
You forget about Mojo, the up-and-coming memory safe language designed by Chris Lattner, creator of the LLVM compiler Rust is based on, and Swift. His new compiler, which has a whole new level of tricks, is the basis for Mojo.
> I don't like the proprietary license
Rustfags who don't give a shit about GPL can suck our collective dicks. Because Mojo has a killer feature that is going to destroy the Rust community: it's Python.
Everyone who knows Python will be able to become Mojo programmers. The whole purpose of Mojo is to replace C and C++ functions with memory safe equivalents, specifically functions related to AI programming.
Rust is fucking finished.
> https://en.wikipedia.org/wiki/Mojo_(programming_language)
> https://en.wikipedia.org/wiki/Chris_Lattner

>>107154666
Didnt Rust split from Mozilla?

>>107155131
I have shown it before but I think a lot of Kick, YouTube, Rumble, shows etc. are simply dramatically better than real TV. So this program will automatically download, organize, and insert metadata into every video and put it into my media folder so it shows up the same as my TV shows etc do.

Some of the things it does:

1. Manipulates specified metadata fields with various parameters (replace strings, copy/paste to/from other fields, prefix date tags in a given format like yyyy-mm-dd), append, whatever.

Also manipulate filenames the same, can be based on metadata matching. Templating possible. Like "{{date}} - {{fulltitle}}" will use the metadata fields.

2. Selective output folders based on metadata matching, strict contains filters or lax (e.g. must have X, or any of X, Y, or Z). So if the title has dog in it, it will go into my video/dogs subfolder... Also accepts templating like {{channel_name}}/{{year}}.

3. Filtering for operations, so each channel can have multiple URLs possibly, you can set operations for specific channel URLs, meta matches.

4. Automatic hardware acceleration. Support for various codecs, h264, hevc, whatever. Falls back to software if codec is incompatible with the given hardware. It calls FFmpeg under the hood but there's logic involved to make the command as it uses different codecs for hardware. Libx264 is software, you have to do something like libx264_vaapi for VAAPI hardware enabled acceleration etc.

5. Scrapes the web for missing metadata. It can get the date of upload for a video for sites ytdlp can't. Screenshot is how I use it because it ensures alphabetical ordering if you do year month day.

6. Config files usable. Various formats, toml, ini, yaml, whatever you prefer.

And now as said I added a bunch of thumbnail logic.

I am building the web frontend using TailAdmin (I am not going to learn HTML etc, so template). Then I will release it for people to use in Docker or Podman.

>>107152975 (OP)
The biggest issue with c/c++ is not memory safety. The biggest issue is undefined behavior making it basically impossible for any c/c++ program to ever be considered fully correct. Every c/c++ program in the world has thousands of hidden bombs ready to explode at the whims of the compiler devs. Compiler devs have fucked us hard, in their pursuit of micro-benchmarks, they ruined our entire foundation of computing.

>>107154666
Rust doesn't do OOP
> and that's a feature

>>107155292
Here is the TailAdmin template I started. Because this is Go, it can server the web page and also generally interact with the web end very trivially. JSON tooling is in the language directly. Everything is built straight into the language... And everything works on every platform. It works exactly the same on Windows when I call the .exe as in Linux, because in Go whatever you do with the file system, it is already built in. UserHomeDir works for whatever, temp directory path, it can get from any system, os.Rename will work on everything.

>>107155346
Oh and then it notifies your services when a new video is downloaded so the library updates. Go works flawlessly with SQL.

CREATE TABLE IF NOT EXISTS notifications (
id INTEGER PRIMARY KEY,
channel_id INTEGER NOT NULL REFERENCES channels(id) ON DELETE CASCADE,
name TEXT NOT NULL,
channel_url TEXT,
notify_url TEXT NOT NULL,
created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
UNIQUE(channel_id, notify_url)
);
CREATE INDEX IF NOT EXISTS idx_notification_channel ON notifications(channel_id);
CREATE INDEX IF NOT EXISTS idx_notification_url ON notifications(notify_url);

Say something like this, Go has tooling inbuilt to query and write to the database. And then trivially scan the database entries into a given struct because you can tag them with `db:"column_name"` and unpack straight into it. It doesn't require imports for this, again it is built into the language.

>>107155346
sounds good but last time I tried to run a go program on my laptop from 2023 it said bad system call
are you exaggerating a bit perhaps?

>>107155293
>Bake a static analyser into the compiler
>Endlessly shill it as the next big thing
>Still no sane bootstrap path
Ok retard sure whatever it's totally the evil *other* compiler devs who are wrong

>>107155199
Based and Mojo-pilled.
Its borrow checker is less of a pain than Rust's, too, and it's built from the beginning to generate GPU code.

Chris started off saying nice things about Rust, but he's more recently said that he hopes to take it down, or compete with it. Be a replacement.

Rust is a mess. Its type system isn't complete or safe, it has big holes (bug #1), and it's f*ing ugly. And its culture is that of trannies.

>>107155390
I've run the program on my server for over a year. I have created bugs sometimes (because originally the program was JUST for me I AM the "staging" etc) but as everyone knows Go is the "if err != nil" language (or 'if err := function(); err != nil { return err }' shorthand) and any panic has a full stack trace with functions and lines and whatever. Paired with the simplicity of the language and instantaneous build time (only the initial build takes a minute or so, after that they are immediate), they can be fixed right away.

There are some other things the language is just really good at, and the tooling is there in the language to handle it. Password and encryption stuff, of course that's also all built in, I have to use that to save people's passwords into the database.

>>107155478
most, if not all, distro rustc packages were bootstrapped from mrustc many years ago, tech illiterate zogduke tard.

>>107155497
Rust requires importing crates for just about everything. Rust does not lead to maintainable projects that contributors can easily get on board with. Watch how much this ends up mattering. Rust is designed for large teams of developers to create software for a company with. So each developer can maintain their bit of code. This is not something you throw onto Github and expect to get a bunch of serious contributors.

Higher level languages lend themselves to solo projects because anyone can understand the code and contribute new features immediately. And you can immediately understand what they have added in seconds because high level languages are simple and legible.

>>107155615
Like hell they were. Debian still "bootstraps" from precompiled shit

>>107154393
>it doesn't cover any of the what
AFAIK these allocators do assign different tags to different types to prevent some type confusion bugs though that can't cover e.g. differently typed fields within a struct.

>>107155630
Rust is relatively friendly for outside contributors, certainly compared to C++. The tooling is unified and there aren't many competing subsets and that takes away a lot of friction.
I jumped into a Rust project once just for a small fix and then kept contributing bigger and bigger features and cleanups until I'd touched every file in the project. Scaling up went very naturally since it was laid out just like 90% of other Rust projects.
The strong static checks are nice too, reduces the number of things you have to be paranoid about during reviews. Even for logic bugs, enforcing that functions are called in the right order, ensuring that all cases are covered.

>>107152975 (OP)
Why is this board so obsessed with this meme language, and not with languages that have actual potential like Huawei's Cangjie?

>>107155713
where did the first debian rustc compiler come from, genius?
that's what we're talking about.

>>107156216
>AFAIK these allocators do assign different tags to different types
allocators (well, global ones) don't have that info to be able to do anything in that regard.

So what happens on use after free?

>>107152975 (OP)
>Which means all C/C++ code will have the same level of safety as Rust code.
Rust safety is about stopping the bugs in source code before it's released, basically preventing the problem from happening at all. It's good that x86 is finally getting some safety features (even though some CPUs had it in the 60s and Lisp machines had something much better in the 80s), but that doesn't prevent bugs like Rust and other better languages do. It just makes the damage from bugs less harmful. If C/C++ didn't exist, we could have been using something like that 40 years ago, and Lisp machines did have something better 40 years ago.

>>107154771
>Everybody knows that rust destroys things
sciencelet, rust is a byproduct of destruction that has already happened.

>>107154156
>why wait until runtime when you can know at compile time whether there are any memory safety issues?
Yeah, that's like suddenly deciding around 14yo that you're actually a girl, and not a boy--oh, wait.

>>107152975 (OP)
>safety
they locking it down. to keep you out. only people with correct thoughts can do computer. soon no computer for you, you suspect

>>107152975 (OP)
I just like the syntax

>>107157583
Mind broken retard

>itt:
>my meme will outmeme your meme

I just don't care about safety. let wagecucks worry.

>>107152975 (OP)
The fact that you made this post shows you're a retarded nigger. I have nothing more to add.

>>107158231
It's just a trannylanger pretending to be a not-trannylanger sowing drama because otherwise trannylang goes ignored, unnoticed and avoided, just like actual trannies irl