I’m starting to think that feds are spamming the “TAX” idea as a distraction. If you genuinely believe that “taxing the rich” will get us Communism then you need to read more. “The Principles of Communism”, “Wage Labour and Capital” and “The Communist Manifesto” are good starting points.

It’s his Pixelfed Account on the pixelfed.social Instance: https://pixelfed.social/Yogthos

One time I asked DeepSeek for guidance on a more complex problem involving a linked list and I wanted to know how a simple implementation of that would look like in practice. The most high level I go is C and they claim it knows C, so I asked it to write in the C language. It literally started writing code like this:

void important_function() {
    // important_function code goes here
}

void black_magic() {
    // Code that performs black magic goes here.
}

I tried at least 2 more times after that and while it did actually write code this time, the code it wrote made no sense whatsoever. For example one time it started writing literal C# in the middle of a C function for some reason. Another time it wrongly assumed that I’m asking for C++ (despite me explicitly stating otherwise) and the C++ it produced was horrifying and didn’t even work. Yet another time it acted like the average redditor and hyper focused on a very specific part of my prompt and then only responded to that while ignoring my actual request.

I tried to “massage” it a lot in hopes of getting some useful information out of it but in the end I found that some random people’s Git repos and Stackexchange questions were way more helpful to my problem. All of my experiences with LLMs have been like this thus far and I’ve been messing with them for 1+ years now. People claim they’re very useful for writing repetitive or boiler plate code but I am never in a position where I’d want or need that. Maybe my use cases are just too niche lol.

A few days ago I was configuring some software where it’s difficult to find good documentation about so I decided to ask DeepSeek. I described what I’m trying to do and asked if it could give me an example setup so I can get a better understanding. All it did was confidently make shit up and told me things that I already knew. And that’s only the most recent example. I have yet to find LLMs be a useful tool.

That’s only been my experience with software that depends on many different libraries. And it’s extra painful when you find out that it needs hyper specific versions of libraries that are older than the ones you have already installed. Rust is only painless because it just downloads all the right dependencies.

Some old software does use 8-Bit ASCII for special/locale specific characters. Also there is this Unicode hack where the last bit is used to determine if the byte is part of a multi-byte sequence.

I always have to lol whenever I see a lib being very serious about some comically overblown made up shit. It’s like a kid trying to tell me they saw a big scary monster in the woods.

There are a couple websites that use it. Others that come to mind are http://leningrad.su/ (no HTTPS) and https://sovnet.su/

Sent me a message from 2 different accounts. It’s just spam mail.

This reads like it was written by some LLM.

Enable journaling only if needed:
tune2fs -O has_journal /dev/sdX

Don’t ever disable journaling if you value your data.

Disk Scheduler Optimization
Change the I/O scheduler for SSDs:
echo noop > /sys/block/sda/queue/scheduler
For HDDs:
echo cfq > /sys/block/sda/queue/scheduler

Neither of these schedulers exist anymore unless you’re running a really ancient Kernel. The “modern” equivalents are none and bfq. Also this doesn’t even touch on the many tunables that bfq brings.

Also changing them like they suggest isn’t permanent. You’re supposed to set them via udev rules or some init script.

SSD Optimization Enable TRIM:
fstrim -v /
Optimize mount settings:
mount -o discard,defaults /dev/sdX /mnt

None of this changes any settings like they imply.

Optimized PostgreSQL shared_buffers and work_mem.
Switched to SSDs, improving query times by 60%.

No shit. Who would’ve thought that throwing more/better hardware at stuff will make things faster.

EDIT: More bullshit that I noticed:

Use ulimit to prevent resource exhaustion:
ulimit -n 100000

Again this doesn’t permanently change the maximum number of open files. This only raises the limit for the user who runs that command. What you’re actually supposed to do is edit /etc/security/limits.conf and then relog the affected user(s) (or reboot) to apply the new limits.

Use compressed swap with zswap or zram:
modprobe zram echo 1 > /sys/block/zram0/reset

This doesn’t even make any sense.

Firefox isn’t much better in that regard. Bootstrapping alone takes many minutes, which is no surprise since their build system is not only huge but also written in Python of all things. Even if you use gcc instead of clang to compile the corporate spaghetti, the tons of rust in the codebase will make up for all that time gained lol. On my 2014 Xeon workstation where a generic Linux Kernel takes less than an hour, FF takes 6+ hours to build.

I’m aware of FreeTube but unforutnately it’s yet another Electron App, which is just Chromium w/ NodeJS. I host my own personal Invidious instance for YouTube.

The current “web” is utter trash. It’s so complex that compiling any browser that supports at least most web standards from source takes a lot more time and memory than compiling the entire Linux Kernel or even the mesa userspace drivers. The only way to “fix” the web is to throw it in the dumpster and start over. A web browser should only be a document viewer that can load and display interactive documents from a remote location and nothing more.

Web search is shit anyway

Yeah. It’s getting less and less usable even with FOSS frontends. I’m working on my own web search off and on. I’m so sick of wrangling with search engines just to get useful results for my query.

I used to use Waterfox for a while until they got bought by System1 (a yankee-based ad company). Librewolf is probably the best option. GNU Icecat is also pretty good if you’re on Linux.

I’ve given up on Element a long time ago. I’ve been bouncing between Nheko and NeoChat since those are the most mature Matrix clients that aren’t bloated webapps.

What bugs me more is encryption still is kind of a mess across devices and clients. Also I hate how there still isn’t any alternative Matrix Server that’s not Synapse or Dendrite and isn’t abandoned, doesn’t suck, is relatively fast and supports at least most of the Protocol. Can’t even really blame people because I’ve tried writing my own Matrix Server and Client before but I eventually gave up because the protocol is what I call “A JSON clusterfuck”. Why can’t the protocol be as simple as IRC? Why does it always have to be JSON over HTTP?

Interesting feature, I had no idea. I just verified this with gcc and indeed the return register is always set to 0 before returning unless otherwise specified.

int main(void)
{
    int foo = 10;
}

push   %rbp
mov    %rsp,%rbp
movl   $0xa,-0x4(%rbp) # Move 10 to stack variable
mov    $0x0,%eax       # Return 0
pop    %rbp
ret

int main(void)
{
    int foo = 10;
    return foo;
}

push   %rbp
mov    %rsp,%rbp
movl   $0xa,-0x4(%rbp) # Move 10 to stack variable
mov    -0x4(%rbp),%eax # Return foo
pop    %rbp
ret

What a load of bloat. NetBSD has a whole guide on diskless booting various machines, none of this shit is new.

Unless your machine has error correcting memory. Then it will take literally forever.

Your CPU has big registers, so why not use them!

#include <x86intrin.h>
#include <stdio.h>

static int increment_one(int input)
{
    int __attribute__((aligned(32))) result[8]; 
    __m256i v = _mm256_set_epi32(0, 0, 0, 0, 0, 0, 1, input);
    v = (__m256i)_mm256_hadd_ps((__m256)v, (__m256)v);
    _mm256_store_si256((__m256i *)result, v);
    return *result;
}

int main(void)
{
    int input = 19;
    printf("Input: %d, Incremented output: %d\n", input, increment_one(input));
    return 0;
}

That’s literally what I’m saying; It’s fine as long as there wasn’t any unwritten data in the cache when the machine crashes/suddenly loses power. RAID controllers have a battery backed write cache for this reason, because traditional RAID5/6 has the same issue.