Hot take:
Don't make a <service>2shell exploit, repo, tool name, etc... if it is NOT a full-chain RCE that affects the latest version of that <service> using a standard/common/default configuration. Anything else makes you look stupid. Sorry, not sorry.
@bigmag@bardia_heydari جدی چرا pirate نمیکنید. خیلی راحتتر و با کیفیتتره که. من هنوز کامل مهاجرت نکردم از اسپاتیفای ولی واقعا برنامه مزخرفی داره که ابتداییترین ویژگیهای یک پخشکننده موسیقی رو نداره.
@vasilios_s If the connections must stay alive and politely asking the clients to reconnect is not an option, the only way I can think of is to extract and move their TCP state from the old kernel and transfer them to the new one and route the packets to the new container.
A couple of years ago I had an interview with Cloudflare. The moment the interview started I apparently got so stupid that my IQ dropped by ~50 points. Packet headers had a magic number field and I couldn't even figure out what its purpose was. This still haunts me to this day😭
In the late '90s I was a huge 3dfx fanboy. For a few years, they were so far ahead of everyone, making all other video cards look like jokes. I had a Voodoo 1, 2, and 3, and was excited for the Voodoo 5 with its promised full-screen anti-aliasing (FSAA). (I was in high school / starting college, a big gamer, and doing some OpenGL graphics programming as a hobby project.)
The day the first GeForce was announced, though, I immediately recognized that 3dfx was toast. Like, immediately. Why?
3dfx had been entirely focused on maximizing fill rate. That is, given a set of textured triangles that are *already* specified in 2D screen space, the hardware would rasterize them as pixels. It was up to software to get the triangles into screen space in the first place, which mostly involves applying some matrix math to every vertex. But 3dfx figured there are way more pixel than vertices, and focusing on more pixels was the way to go to improve quality. With enough pixels they figured they could literally render every frame at 4x resolution and then down-sample to make the edges smoother (FSAA). The Voodoo 5 was a monstrous brute force design whose top-end version had four copies of the chip in order to render those four pixels at the same time.
The GeForce, on the other hand, went ahead and moved vertices into hardware. The result? It could render an order of magnitude more triangles, which meant surfaces could be smoother. It was immediately obvious that this had a far bigger impact than FSAA, both from the screenshots and from what I'd been learning about graphics programming.
A friend of mine (who still attends my LAN parties today) was staying with us at the time. I remember going over to his computer while he was out, loading up the screenshots from the announcement, and leaving a little note in a text editor about how I'd switched allegiance.
Of course, games did not get the benefit automatically. Engines had to be rewritten to move their vertex handling into hardware, whereas 3dfx's brute force and FSAA would immediately apply to existing games. This is why 3dfx's bet was the easy one to make, and Nvidia's so much harder.
But Nvidia has continued making these bets, over and over, ever since. They have repeatedly built for markets that didn't exist yet, investing in stuff that would take many years to pay off. CUDA when nobody was asking for it. Ray tracing when nobody was asking for it. Almost every generation of Nvidia hardware has something qualitatively new.
Jensen calls these "zero-billion-dollar markets".
Doing this requires leadership. The natural state of a big company is to focus on whatever incremental improvements their biggest customers are asking for, because those are the easiest decisions to justify. It works great for a while, then fails. That's the Innovator's Dilemma. And even executives who have read the book find it hard to internalize. Even executives hand-picked by the CEO who has internalized it often cannot themselves internalize it. It's weirdly hard to find leadership that gets it, and no amount of bottom-up innovation can make up for it.
And that's why I sold my Nvidia stock this morning.
the more i watch people flail around desperately looking for yet another hype-cycle to avoid disciplined responsibility the more i realise that ownership-first development is the way.
Any Telegram message can now become an app.
⚡ Downloads instantly (less than 4 KB) 🎨 Interactive and natively rendered 💸 Payments and viral mechanics built in
🛠️ Build your own. More info and docs for devs: https://t.co/v6hbiQErCB
Today I asked an LLM to read my blog and analyze the personality of the author based on the writings. ngl It was quite interesting and even fun to see a machine analyzing me.
Few months ago, I acquired my own ASN.
I published a blog post explaining the steps I took to set it up, configure BGP sessions, and begin advertising my own IP prefixes to the global routing table.
You can read it here: https://t.co/4iXkBR4ko4
@ph4r5h4d قربونت. بیشتر منظورم این بود که وقتی پاسپورتم رو بگیرم دیگه نمیخوام دوباره برم یه کشوری مثل آمریکا که ویزا و اقامتم وابسته به کارم باشه و از صفر شروع کنم و دستم زیر ساطور کارفرما باشه چون فشارش برام زیاده.
@ph4r5h4d من خودم از اون دسته بودم (البته حالا حالاها مونده به پاسپورتم) ولی الان میبینم دیگه از لحاظ روانی تحمل زندگی روی ویزا رو ندارم :)
احتمالا برم انگلیس که فرصتهای شغلی بیشتر و بهتری در اختیارم باشه.
AI bros do not want you to read assembly! 🤡 Here is what you are missing if you listen to them.
Welcome to the case where dumb C++ code wins over a "smart and optimized" code. Here are two implementation of an ECS helper that assigns an id for a component by type. Both approaches work in my game. The first ("dumb") version is what I have had for a while.
But look! It has an extra if check and stupid manual assignments. Surely, we can simplify it with a post-increment! Oh, wow, we are so smart. 🤡 Lets do that and compare the generated assembly for both.
Surely, the smart case will have less code, be faster, be awesome, and let us feel how smart we are!