husband, father, shadowy super coder - innovating @hatchet_dev - @uwaterloo CS '21
Amsterdam, The Netherlands
- Tweets 1,791
- Following 647
- Followers 233
- Likes 19,592
gone are the days you'd go to jwt[.]io for decoding jwts
use this instead:
jwtd () { jq -R 'split(".") | .[1] | @base64d | fromjson' <<< "$1" }
31
what you have described is “aloo keema” and I have had it since childhood
I’m once again reminding you to chop up some potatoes, cook with ground beef, onions, and garlic. Add veggies and whatever spices you want.
Once complete will have created a meal that’s high protein, fiber, and micronutrients while tasting delicious.
68
best people, best location, best company, best offsite
Here's a look behind the scenes at our team offsite in Cape Cod last week! ⛵️🚲🦞🎄🌊
(Not shown here: @igor_kupczynski trying a lobster roll for the first time, loving it, and then ordering a second, third, fourth, etc.)
1
3
136
This is why Docker takes so much RAM on your computer (it really depends on the host OS).
On macOS and Windows, Docker needs to create Linux VMs to operate on. Containers need a Linux kernel so Docker Desktop runs a full Linux VM (based on Apple's Virtualization Framework or WSL2/Hyper-V on Windows). That VM needs memory allocation and for the majority of it, the host OS sees the contiguous memory fully allocated even though the guest VM might not be using it fully.
If this was not enough, there are a few other things that can make it a memory hogger. The Linux kernel inside the VM has page cache build up (think dentry, inode, build context, volume I/O). This is generally reclaimed inside the VM due to causes like memory pressure and cache drainage. Docker Desktop UI itself (Electron, etc.) adds its own overhead. In most cases, you can easily change the maximum memory allocated to containers using the Docker Desktop settings.
If you are using Docker on Linux, it is going to be relatively cheap. This is because on Linux we are simply depending directly on namespaces, cgroups, processes and such.
1
109
Megan created a better version of the Chrome dino game for our offsite hackathon. So I trained Jev to play it for me. The high score is about 36k 🍌
1
4
262
A few interesting learnings from trying to run the C SDK Hatchet worker on the TRMNL:
1. Initially, there was a lot of TLS handshake "fatal alert from peer" errors. This happened every time we tried connecting to Hatchet Cloud. After debugging for a while, we found the root cause - our endpoint is TLS 1.3 only and every prebuilt Arduino core for the ESP32-C3 ships mbedtls with TLS 1.3 compiled out. The fix was to build the Arduino core manually as an ESP-IDF component so that we own sdkconfig. That let us turn TLS 1.3 on ourselves and the handshake succeeded.
2. We tried to have Claude implement the C SDK with hand-rolled HTTP/2 HPACK along with protobuf. The idea was to not have any dependencies and to have the worker runtime as small as possible. The code that skips unneeded HTTP/2 frames was using the reply buffer as scratch space (quite common to do in the embedded world to optimize data ingestion and reuse unused free space). Due to small size of the buffer, the server's closing bytes overwrote the reply before we parsed it and led to data corruption. We had to give that path its own local buffer. Embedded devices and their constraints require a lot of buffer mapping such as this.
Eventually we gave up on this idea and the C SDK because having the gRPC network stack implementation done natively on the ESP32-C3 was just not viable. Glad we have the new serverless operator support arriving in Hatchet very soon!
Some more reading on mbedtls if you're interested: docs.espressif.com/projects/…
We got these @useTRMNL e-ink displays for the Hatchet hackathon, and naturally @mnafees and I wanted to get these running as Hatchet workers.
Initially we tried to get a vibe coded C SDK running. This worked, but the esp32-c3 processors only have 400kb of RAM, so we were limited to spawning tasks with payloads <25 chars.
So instead we started playing around with a new feature in Hatchet called operators, which lets you turn anything that can run an HTTP server into a hatchet worker. This isn't natively supported on TRMNLs, but modifying the firmware to add /heartbeat and /trigger endpoints was very easy.
Obviously, the next step was sending base64 images, and then playing Bad Apple at around 1 fps. Had to do some more firmware shenanigans to disable "full refresh" to get below .3fps, but it ended up working pretty nicely.
1
3
237
just use @atuinsh
109
this is crazyyy
We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics.
The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra.
The problem concerns whether the description of smooth three-dimensional fluid motion modeled by the Navier-Stokes equations can break down. It has remained unresolved for roughly 90 years.
40
I started following the Boost C++ libraries account on X, and my God! So much hostility against the language it works on 🤣
32
how have I not known about this!?
en.wikipedia.org/wiki/World_…
32
My first programming language was C++ and I was 11 years old when I learnt it. People have different reactions when I say this. But it has been the best decision for my professional career and for my systems thinking.
Time and again it has proven to be the catalyst in deciphering details when no other peer of mine was even interested to dig deep into why something was failing. Just by having the insight from reading and writing years of C++ and ASM, you are light years ahead in decrypting problems based on first principles.
I chose to attend @UWaterloo specifically because it is a C++ inclined university. They teach you C, C++, ASM and all projects are based on those programming languages (with options for different languages accordingly).
Google researcher @lauriewired warns the shift from C++ to Python in universities is creating programmers who don't understand how computers actually work:
"When I was in college getting my CS degree, my year was one of the last years that was using C++ as the core language for teaching students. Actually they were switching over to Python of all languages."
"One of the biggest issues that I have is that a lot of these higher level languages are really abstracting away what is happening under the hood. If you're not aware of how the computers work in general, then you're gonna have problems when it comes to debugging."
"Starting with a language like C++ gives you the ability to do these really high level abstractions, but it also gives you the ability to customize your memory management and get really close to the underlying machine. Especially if you're trying to work in high performance computing, game development, or high frequency trading."
"At C++ conferences, there's almost this fundamental anger response when it comes to mentioning something like C or C++. People start complaining about the complexity of it."
@Google
109
every time I see some code in Python, my eyes roll automatically
what a sad programming language to exist
and to think that universities are teaching students python as the sole programming language nowadays boils my blood
39
