Gregory Estrade

8.6K posts

Gregory Estrade banner
Gregory Estrade

Gregory Estrade

@Torlus

Coder, reverse-engineer, hardware hacker.

Toulouse - France Katılım Temmuz 2009
1.7K Takip Edilen1.2K Takipçiler
Gregory Estrade retweetledi
Will Crichton
Will Crichton@tonofcrates·
Love this figure from Google's "pipes for SQL" paper. Matching syntactic and semantic order is a great heuristic for a well-designed API.
Will Crichton tweet media
English
36
171
1.9K
118.3K
Gregory Estrade retweetledi
Tock
Tock@yvan_theriault·
Enfin les vacance 😬
Tock tweet media
Français
295
194
1.2K
113.6K
Gregory Estrade retweetledi
ian @ Dangerous Prototypes
ian @ Dangerous Prototypes@dangerousproto·
Mobile phone SIM cards and bank IC cards respond to an asynchronous Answer To Reset command over a half-duplex UART. Pump a continuous clock signal into the clock pin. Bring the reset pin low and then high. The ATR response comes out of the IO pin...
ian @ Dangerous Prototypes tweet mediaian @ Dangerous Prototypes tweet mediaian @ Dangerous Prototypes tweet media
English
12
241
2.5K
292K
Gregory Estrade retweetledi
Aurélie Vache 🥑🐳🦸‍♀️✏️
En 2019 je devenais speakeuse a @DevoxxFR pour la première fois avec un duo sur Docker, Kubernetes et Istio. Cette année j'ai encore la chance de faire un duo et sur un sujet encore différent. Ce n'est peut être rien pour vous mais pour moi ça veut dire beaucoup 😊
Aurélie Vache 🥑🐳🦸‍♀️✏️ tweet media
Français
2
4
60
2.1K
Gregory Estrade retweetledi
Greg Linares (Laughing Mantis)
Greg Linares (Laughing Mantis)@Laughing_Mantis·
One of my favorite things to do to online pedos in the 90s who were idling overnight in AOL or IRC chat rooms was to send them a message with the following text and watching them disconnect: +++ATH0;ATDT^19004469090;\T0 IYKYK
English
35
20
435
88.3K
Gregory Estrade retweetledi
Arnaud Carré
Arnaud Carré@leonard_coder·
I released my bitmap converter demo tool. One key feature is the brute force color quantization algorithm, running on GPU for speed. (used when color reduction is needed, or for Amiga HAM format) github.com/arnaud-carre/a…
Arnaud Carré tweet media
English
6
29
85
6.4K
Gregory Estrade retweetledi
J-F DEL NERO
J-F DEL NERO@jfdelnero·
Plus de 40 ans après sa fabrication, double résurrection du télétel de l'@assomo5 : Replacement des nappes internes, puis après une nouvelle panne, replacement d'un pont de diodes et d'un jeu de condensateurs chimiques. Il s'agit là de la machine démarrant l'aventure du Minitel.
J-F DEL NERO tweet mediaJ-F DEL NERO tweet mediaJ-F DEL NERO tweet mediaJ-F DEL NERO tweet media
Français
1
8
38
4.4K
Gregory Estrade retweetledi
Abhinav Upadhyay
Abhinav Upadhyay@abhi9u·
Just announced LPython - a high performance compiler for Python with multiple backends. LPython is a novel, fast, and retargetable Python compiler that can compile type-annotated Python code to optimized machine code. It offers several backends such as LLVM, C, C++, WASM, Julia, and x86. LPython features quick compilation and runtime performance, as demonstrated by the benchmarks in their blog. LPython also offers Just-In-Time (JIT) compilation and seamless interoperability with CPython. It is designed to be compatible with existing Python libraries and tools. LPython is released as an alpha version, meaning that users may encounter bugs when using it. One of the key features of LPython is its ability to generate optimized machine code from type-annotated Python code. It uses the Abstract Semantic Representation (ASR) shared with LFortran, which allows for independent optimizations at the ASR level. The goal of LPython is to produce a compiler that runs exceptionally fast and generates fast code. LPython provides several backends, including LLVM, C, C++, and WASM. It uses a multi-phase compilation process, transforming input code into an Abstract Syntax Tree (AST), then into an Abstract Semantic Representation (ASR), which preserves all semantic information. The ASR undergoes several ASR-to-ASR passes, where abstract operations are transformed into concrete statements. Finally, the final ASR is sent to the selected backend to generate code. LPython supports Ahead-of-Time (AoT) compilation, which allows users to compile their Python code to optimized machine code before execution. It also supports Just-In-Time (JIT) compilation, where functions can be decorated with @lpython to enable JIT compilation. Currently, only C backend is supported for JIT compilation, but support for LLVM and other backends is planned for the future. LPython offers interoperability with CPython, allowing users to access any library implemented using CPython. This is achieved through the @pythoncall decorator, which allows calling functions in CPython libraries. Users can also access libraries like NumPy, TensorFlow, PyTorch, and Matplotlib using LPython. The benchmarks provided in the blog demonstrate the performance of LPython in various scenarios. For example, LPython's JIT compilation is compared to Numba on tasks like summation of elements in an array, pointwise multiplication of arrays, insertion sort on lists, and the Dijkstra shortest-path algorithm. LPython shows competitive performance in all these scenarios. LPython looks like a promising project to aid applications which require high performance, such as AI. Mojo is also adopting a similar approach, however they are not only building a compiled Python, but also extending it to offer systems programming constructs as first class features. It would be interesting to compare the two. Here's the full blog post covering their alpha release: lpython.org/blog/2023/07/l…
English
22
177
975
211.8K
Gregory Estrade retweetledi
Ivan at Wallarm / API security solution
Just in case you missed this. This is probably the most critical API exploit this quarter. ESPv2 allows malicious requests to bypass authentication if both conditions are true: The requested HTTP method is not in the API service d…lnkd.in/eQnsc-85 lnkd.in/eFXc__T4
English
3
144
394
59.5K