Derek Perez’s Post

Hello everyone! Today, I’d like to share a piece of technology that can be incredibly helpful for those dealing with unexplained performance issues. It’s called Coz—a profiler for native code (C/C++/Rust) that uses causal profiling to uncover optimization opportunities often missed by traditional profilers. Unlike typical tools, Coz predicts the direct impact of optimizing specific lines of code on overall throughput or latency, offering insights into where improvements will yield the most measurable performance gains. Coz profiles serial, parallel, and asynchronous code without requiring instrumentation for concurrency or library calls. This approach allows developers to measure the precise causal effect of optimizations, providing a reliable way to focus efforts where they will have the greatest impact. Have a great day! https://lnkd.in/dRH3vpXk

Here are the few limitations of using Supabase Edge functions you need to keep in mind: * The Pro plan imposes a bundle size limit of 10 MB. This limit can be easily exceeded when installing libraries with substantial dependencies. * Supabase Edge Functions currently support only the Deno runtime. The Deno ecosystem is still maturing, which means some libraries available in Node.js are not yet usable in Deno. * Edge Functions have a default execution timeout limit of 60 seconds. * Deno's immaturity, particularly in comparison to Node.js, introduces new paradigms that may slow down developers who are already accustomed to Node.js's specific features and workflows. * Running functions locally can result in vague and non-descriptive error messages due to compilation or container issues. Developers must develop a systematic troubleshooting process, primarily involving commenting out code and adding console logs, as there is no known way to debug Deno functions with step-through debugging. * Certain NPM packages available for Node.js may not exist for Deno. In some cases, a Deno-compatible dependency may exist but with a different API or functions compared to the Node.js versions. This may require implementing an interface to standardize the usage of these libraries when sharing code.

Oh the irony of one of the trolls of the flutter community posting comments on my flutter substack post on BlueSky about state management.....yeah the same one that tried almost 2 years ago to gas-light state management solutions performance.... The reason why we like immutables is that they not alone help state management but make the API more robust in that any data put into an immutable model has to flow into it via having the reference copied...i.e. it makes the data flow from model to view or view-model in one direction . Compilers in general like immutables as most compilers no matter the language treat them is const objects which saves memory, not to mention preventing memory leaks. #flutter

Dynamic Code Flexibility with importlib and getattr 💥💥 GET FULL SOURCE CODE AT THIS LINK 👇👇 👉 https://lnkd.in/e_mYhBcP Dynamic code flexibility allows developers to create programs that can adapt to changing requirements and environments. Two powerful tools in Python for achieving this flexibility are importlib, which provides a way to dynamically import modules, and getattr, which allows accessing attributes of objects dynamically. This combination enables developers to write code that can be modified at runtime to accommodate different scenarios. By leveraging importlib and getattr, developers can create more flexible and scalable systems. For instance, they can write code that can dynamically load plugins or extensions, or create systems that can adapt to changing data formats or protocols. This flexibility is particularly useful in systems programming, scientific computing, and data analysis, where code often needs to be modified or extended to accommodate new or changing requirements. To further reinforce your understanding of dynamic code flexibility, consider implementing the following suggestions: * Create a simple plugin system that can dynamically load and call different plugins. * Write a program that can parse and process data from different file formats. * Implement a system that can dynamically adapt to changing network protocols or data formats. Additional Resources: No additional resources provided. #stem #Python #DynamicCodeFlexibility #importlib #getattr #PythonProgramming #ProgrammingLanguage #SoftwareDevelopment Find this and all other slideshows for free on our website: https://lnkd.in/e_mYhBcP #stem #Python #DynamicCodeFlexibility #importlib #getattr #PythonProgramming #ProgrammingLanguage #SoftwareDevelopment https://lnkd.in/eNgy5ucW

What is the difference between APIs.json and the api-catalog IETF draft? Two projects are being built publicly that have complementary goals. On the one hand, APIs.json aims to "create a file format that can represent a collection of APIs." On the other hand, api-catalog's goal is to "facilitate the automated discovery of a Publisher's public API endpoints." Learn what these two projects are and how they can complement each other. https://lnkd.in/dTGbngGA #API #catalog #discovery

“The most recent reaffirmation happened when we tried to add telemetry to the instance for on-premise organizations. Erlang provides a standard telemetry API that many packages use to report telemetry information, including packages like Phoenix (the web framework) and Ecto (an ORM to access databases). Without a single line of code, we had all that information available, and with just a few lines, we had it converted and served to Prometheus via a /metrics endpoint. Mind-blowing.” #Elixir #Erlang #Engineering https://lnkd.in/dm5Mh9Br

🚀 Excited to share my latest article on LinkedIn: "Mastering Connection Pooling in TypeORM: Why Releasing Transactions Matters". If you work with TypeORM or are interested in efficient database management, this is a must-read! Dive into the importance of releasing transactions and optimizing connection pooling for smoother operations. Check it out and let me know your thoughts! #TypeORM #DatabaseManagement #ConnectionPooling https://lnkd.in/gvanr5C4

Building in Open Source means nothing without new contributors coming together to make a project better. Laravel Rector has had a couple fixes recently. @MrPunyapal firstly fixing some of the string to class options in the LARAVEL_CONTAINER_STRING_TO_FULLY_QUALIFIED_NAME set. Also to @its_hans_thomas for fixing an issue with the ApplyDefaultInsteadOfNullCoalesceRector where it would go wrong and try to set an exception throw as a default value. We're always looking for new contributions, even if it comes in the form of new ideas! The PRs mentioned for those looking to learn more: https://lnkd.in/eKtiZGCY https://lnkd.in/eeyxQ_fj

My friend Vaughn Vernon posted a meme about Complexity is what people wants, and that reminded me of a thought I had a few days ago... Being a Developer nowadays is no longer about writing code, designing good/elegant solutions, but a matter of fighting to "Run Systems". Code is written in 3 languages with 5 frameworks, 12 executables, packaged in Docker, deployed on Kubernetes (requiring 3 Kubernetes Operators), interacting with 4 database systems, 2 messaging systems, 3 caching systems, centralized logging system, a data lake, storing archives in S3 and to top it off....requires a testing system that takes 3 man-years to make useful. And that is just the backend... New UI frameworks being bolted on to UIs that already have 2-3 now-outdated frameworks stuck from before that nobody will ever remove. Ever more complex inter-dependent systems that no one completely understands and probably wasting 99+% of the total computing power actually required for the problem at hand. Have we lost all our collective marbles? Is it mass-psychosis going on?

🚀 Introducing Prompt Fuse - A New Tool for Managing and Versioning LLM Prompts! Envision having all your LLM prompts organized, tested, and tracked automatically in one place! With Prompt Fuse, prompt engineering just got a major upgrade. This open-source tool brings clarity and efficiency to developers by offering intuitive prompt versioning, individual testing capabilities, and a seamless interface to experiment, all in one platform. 🛠️ Key Features Include: - Automatic version tracking to capture every iteration - Integrated LLM support for quick setup and experimentation - Interactive testing interface for text and image prompts - Performance metrics to keep you informed - Project-based organization to bring structure and scalability "Prompt Fuse may seem like a simple prompt management tool today, but it represents something much bigger—a foundational step toward making prompt development as disciplined and collaborative as software engineering itself. Just as Git revolutionized code version control, we believe Prompt Fuse will evolve from basic prompt organization to an essential platform for AI development. This is just the beginning." 📚 Learn more: GitHub: https://lnkd.in/dMkCCXQC Detailed Article: https://lnkd.in/dvd_Tp7v Note: This project is still in (v 0.1.0 - beta) so bugs might arise. #OpenSource #AI #PromptEngineering #LLM #Python #React #Developer #MachineLearning