Antonio Angelino

💥 Wow. What an inspiring webinar with Valentina (Cupać) Jemuović 🤖 on AI in TDD! A big thank 🙏 you to Valentina for sharing her experience on using Copilot for TDD. 👉 Key takeaways: 1- Creating failing tests: This is the task with the highest cognitive load in TDD, and it requires a lot of back-and-forth prompts to get the right tests. The generated tests often had quality issues, such as not challenging poor requirements or failing to express good interface design. 2- Writing code to make the test pass: Copilot performed ok but often needed multiple attempts. 3- Refactoring step: Copilot provided refactoring ideas but also required very specific guidance when performing challenging refactoring (which is why you need to be intentional about your coding practices - 🚀 go Packmind). Overall, Copilot falls short in the same areas as humans, which is not surprising. The 'red step' is particularly challenging for developers; it's where you decompose the requirements and drive the design. This is typically where Valentina spends the majority of her time coaching teams on TDD. Recording available in comment 👇 #TDD #AI #Copilot

41

11 Comments

One of my mentors in the MSFT SQL Server team once said: Him - "Do you know the difference between a brilliant, young developer and a brilliant, seasoned developer?" Me - "No, what is the difference?" Him - "The brilliant, young developer has not yet learned to fear complexity." I've seen that simple truth proven over and over. As the story below illustrates again, simple engineering is the best engineering.

4

1 Comment

Serverless vs Containers: I have the answer right away. It's not one or the other. You can build the same thing both ways. Let's revisit two projects from 5 years ago: ↳Invoice Automation: Generating and processing invoices. ↳FinTech: Complex with many dependencies. We had experience with serverless. We had experience with containers. Invoice Automation went serverless. Results: ↳ Easy scaling ↳ Low costs during quiet times ↳ Devs focused on features ↳ 40% under budget ↳ Less operational work ↳ Simple cloud service integration ↳ Faster feature releases ↳ Occasional cold start issues Despite the success, we chose containers for FinTech. FinTech container results: ↳ Consistent performance everywhere ↳ Easier debugging and testing ↳ Met compliance needs ↳ Better runtime control ↳ More portable between clouds ↳ Suited for long processes ↳ Predictable long-term costs ↳ Needed more setup and maintenance Serverless: Faster and cheaper deployment. Containers: Better control and consistency. Lesson: Choose tools based on needs, not hype. We struggled with one e-commerce project with serverless. Not because of serverless, but because our team could not adapt it. Avoid chasing trends. There is no silver bullet. Sorry. Find what suits you. Consider your team's skills. Your turn: What's your experience? Seen similar outcomes? Share in the comments! #Serverless #Containers #TechDecisions ---- 🔥 Want more #CloudComputing tips? 👉 Follow Sanjeev Thapa for weekly insights! 💬 Thoughts? Questions? Comment below! 🚀 Found this helpful? Share it!

21

26 Comments

In this episode, Daniel Palma goes into what makes Estuary Flow unique in the world of generic streaming platforms. Check it out! and subscribe to the channel for more byte sized sessions about streaming data processing.

13

1 Comment

Insightful talk "The Myth of the Modular Monolith" by Eileen M. U. is now available! Check this out - https://lnkd.in/dK_3_mZu Eileen delves into the appealing yet often flawed concept of a modularized Rails monolith. She highlights that while this approach might seem like a perfect solution, it frequently fails to address deeper architectural, operational, and organizational issues. I couldn't agree more. A modular monolith isn't a silver bullet, but it does help steer us in the right direction. I think there are two architectural concepts that should be strictly followed to achieve the goal of manageable modular monolith. 1. Modules and their boundaries. Thanks to Shopify, we have a tool for that in the Ruby world. Packwerk gem allows us to define modules and plays like an analyzer that highlights dependency violations between your modules. You can read retrospective about this tool here - https://lnkd.in/dtWCRPjr 2. Module interface. A robust module facet is needed to hide all internals from external users and provide a reliable and concise interface. Unfortunately, Ruby lacks a built-in mechanism for this. There was an attempt with packwerk at some point, but it didn’t get off the ground. Nevertheless, modular monolith unlikely to be truly manageable and comfortable to work with without module interfaces. The closest example of modules and interfaces is Java/Kotlin. In this world we just can't build an application if we break the module interface and try to use things hidden from external use. When we start working with interfaces, we inevitably have to deal with interface types and signatures. Interfaces cannot really be relied upon without signatures. Fortunately, in Ruby, we have RBS and Sorbet. RBS will eventually become the default option for signatures. By the way, RubyMine already works well with RBS/Sorbet and highlights signature mismatches in your code preventing bugs in production. So take a look. Regarding large existing Rails applications that self-move towards a mash of code. Well, there really don't seem to be any easy solutions here. Being huge, a Rails application is cramped, leaving no easy solutions from a technical point of view. Indeed, it seems that only gradual refactoring and changing the culture and priorities within the product can improve the situation. However, I think it's potentially possible to implement an approach in Rails that will force developers to build a reasonable architecture and not mix things up. This will inevitably lead to a slowdown of Rails development velocity, which contradicts the main feature of the framework and is unlikely to ever be implemented. After all, only a few Rails applications survive to the point where we have to solve architectural problems rather than business problems. Tell me about your experience with packwerk and modularisation in the comments! Interesting to see your thoughts! #RailsWorld #Rails #Ruby

4

Stateful Orchestration Engines. Thinking and coding in Blue, Red, and Maroon Functions. And my second of two posts published today. After academic thoughts in the first post, this one focuses on applying the ideas of distributed consensus in the most pragmatic way I can imagine today: by designing software to be executed under the control of a general-purpose stateful orchestration engine. https://lnkd.in/gVJ7tkj4

9

4 Comments

Recently https://lnkd.in/dCFVh_wf has passed the 800 stars in Github 😄 If you are looking for a faster implementation of OpenAI Whisper, whisper-ctranslate2 is a cli (which supports the same arguments as OpenAI original one) and it is build on top of faster-whisper. It allows you to access a faster implementation of Whisper by switching cli the with no effort.

16

1 Comment

Here's how I scrape anything and send that data anywhere. In this case, how I automatically scrape Amazon deals and create a post on FB about it. And I only these 2 free nocode tools. - Automa: To scrape data from websites - Make.com: To do something with that data Here's how it works: 1. First the Automa workflow "goes" to a link (the Amazon deals list) 2. Then, it clicks on a product and extracts the title, price, image, etc. 3. Afterwards, that data gets sent to Make.com 4. Here the data is cleaned up if needed 5. And finally it creates a post on a FB page with the product info Of course, once the data is in Make you can do whatever you want with it. This is just one of the many use-cases. And I actually recorded a step-by-step guide on how to create this exact automation. Just comment "bazinga" and I'll send it over. (We need to be connected)

11

4 Comments

I’m still a Rust neophyte, but this is an interesting post about async and complications it introduces in a work-stealing runtime where any task can be scheduled on any core at any await point. Things are much simplified if you assume a “share nothing” architecture where once scheduled onto a thread, the task (and its descendants) is guaranteed to never migrate off that thread. Reminds me a lot of the Seastar C++ framework used in ScyllaDB.

8

Many of us are working on exciting projects but only few succeed. During my attendance to two European Ruby conferences in June, I had the opportunity to ask top Ruby developers this crucial question: why do software projects fail? These are some great insights they shared on common pitfalls, unexpected challenges, and lessons learned from project failures . Watch my full video on my YouTube channel for more in-depth insights and discussions: youtube.com/@tillcarlos

6

Unpopular opinion: It's not the software devs who need convincing. Selling devs on TBD, TDD, CI/CD, and ensemble programming, is a waste of time. We need instead to focus on continuous feedback. We need to sell it to the higher ups. To the people who call the shots. Because in the end, the aim of these tools and processes is to decrease the time it takes for us to collect valuable feedback from customers. Once the c-level people are sold on the merits of fast feedback, software devs will quickly learn how to write and deliver quality code.

6

4 Comments

☕ The Coffee-Driven Code Review Manifesto: A Sleep-Deprived Guide to Excellence [Everyone talks about clean code, but let's be honest - most of our best reviews happen after that third espresso shot] 🌅 6 AM: First Commit Code looking pristine Tests all green Documentation spotless Reality: Haven't written a single line yet ☀️ 9 AM: The Morning Stand-up Revelation "Wait, did I just promise to deliver this by EOD?" Frantically opens IDE Types while drinking coffee Spills coffee on keyboard Git commits anyway 🌮 12 PM: The Lunchtime Code Review PR Comment: "Interesting choice of variable names. Is 'pizzaTime' really the best name for a payment processing function?"  Me: "I was hungry when I wrote that..." 🌙 5 PM: The Final Push git commit -m "Fixed everything (hopefully)"  git push origin please-work  git prayer --recursive 📊 Our Revolutionary Metrics: Bugs Found vs Coffees Consumed (1:1 ratio) Time Spent Reviewing vs Time Spent Making Memes About Code Number of "LGTM" Comments That Actually Meant "I Trust You” 🎯 Pro Tips for Sleep-Deprived Reviews: If the code works on the fifth try, it's probably production-ready Comments like "Magic. Don't Touch." are perfectly acceptable documentation If you can't explain your code, blame it on the keyboard 🤔 The Real Code Review Checklist: Does it work? No one knows why it works But it works Ship it! 💭 Question for fellow night owls: What's your record for most consecutive hours staring at a PR before realizing you were looking at the wrong branch? #CodingLife #TechHumor #SoftwareEngineering #DeveloperLife #CodeReview #CoffeeDriver Development #TechMemes

2

Probably I am not the only one that tries to find some spot here and there to do a bit of reading (specially if you have 80 tabs open with "Save for later" articles 😅 ). In the IT field, being able to "catch up quickly", "learn quickly" and "solve quickly" is seen as a must to stay in the sector for the long run. To be honest, it is overwhelming. The constant focus on "quickly" can make us forget that real learning and problem-solving takes time and effort. This has deeply bother me for quite sometime and even make me feel unproductive, blocked to finish writing / creating things I had in mind for a while because I am already late for it. But when I take a step back and look at everything I've actually done, I realise I have been doing way more than I give myself credit for (specially given the amounts of hours a day has). It's just hard to see that when you are struggling in the grind. And honestly? It’s exhausting and confusing 🫠 It was a good timing then that Today I ended up with this article from the Hacker Newsletter: https://lnkd.in/dBKG7v7K and something resonated there: "I want to know everything, immediately, quickly. Since this is not humanly possible, I end up doing nothing." "You find yourself in a quest for productivity, feeling productive because of the quest, but not really doing any productive work." "Difficulty concentrating, lack of focus, inability to understand what I read, stress, anxiety. I see these kinds of complaints from many people lately, and in my opinion, our biggest common ground is digital content consumption." And avoiding, like the author, to go against that article, I also do not have an advice, but just leaving the article there for your own reflection on this. Slowing down, taking a break, and letting yourself process what you already know can be more valuable than rushing to consume the next thing. 🌱 Give it a read if this resonates with you. Ask yourself what would slowing down look like for you? I'm trying to figure out for myself right now

36

Ready to take your first steps in TON (The Open Network)? If you're a developer looking to dive into the TON ecosystem, a great starting point is creating a smart contract that displays "Hello World." The TON network uses the FunC (Functional C) language, which is optimized for smart contracts in its blockchain environment, offering scalability and efficiency. Here's how you can do it: 1.- Set up your development environment - Install TON-CLI and FunC by following the official documentation. - Use a code editor like VS Code. - Set up a TON wallet and ensure you have enough funds for gas fees. 2.- Write your 'Hello World' contract Check out the example code in the image below ⬇️ to create your first smart contract. 3.- Compile and deploy Compile your contract using FunC. Deploy it on TON with TON-CLI. 4.- Verify the deployment Use blockchain explorers like TONScan (https://tonscan.org/) to find your contract's address and verify its status. 💡 Pro Tips for TON Developers - Explore the official FunC documentation to build more advanced projects. - Use tools like TON Labs IDE to simplify debugging. - Optimize your contracts to reduce gas costs and improve efficiency. Are you already working on TON? Share your projects, ideas, or questions in the comments! #TON #BlockchainDevelopment #SmartContracts #FunC #Web3 #BlockchainDevelopment

9

Llama 3.3 70B - Everything you need to know 🔷 WHAT IS NEW? 🟢 Llama 3.3 (70B parameters) delivers the same performance as Llama 3.1 (405B parameters) 🌱 🟢 Sets New Standards for Text-based Use Cases: Multilingual Dialogue (English, German, French, Italian, Portuguese, Hindi, Spanish, Thai), Math, Knowledge. 🟢 Output Modalities: Multilingual Text and Code 🔷 HOW? 🟢 Uses Grouped Query Attention Mechanism, innovations in online reinforcement learning and post-training optimization 🔷 SO WHAT? 🟢 Llama 3.3 delivers top-tier performance at a fraction of the cost compared to Llama 3.2 (405B) 💸 🔷 INTERESTING FACT 🟢 Training the model resulted in 11,390 tons of CO2eq emissions (~2.77 million car rides ❗) 🔷 HOW TO GET IT 🟢 Available for use on Meta’s website and Hugging Face (Links below) #AI #GenAI #Llama3 #MachineLearning #Efficiency #Sustainability #Innovation #TechLeadership #Meta #HuggingFace #EnterpriseAI

19

3 Comments

Daily Dose of D3FEND: File System Link 🛡️ Overview The maze of digital pathways grows ever more deceptive under the necromancer’s hand. Once, File System Links were trustworthy signposts, guiding users and systems to their rightful destinations—whether direct routes like Hard Links or shortcuts like Symbolic Links. But the necromancer, weaving his power through the system, has learned to twist these links, turning safe passages into traps. With a simple misdirection, the paths you trust most can lead straight into danger. Today, we follow these twisted links to uncover how they’ve been turned into tools of subversion. 🔍 What is a File System Link? A File System Link is a digital pointer, directing your system to a file or directory. Hard Links are direct references to the same data, like multiple doors leading to the same room. Symbolic Links (or Soft Links) act as shortcuts, pointing elsewhere, making access more efficient. 🦠 Threats to Watch Out For 1️⃣ Shortcut Modification (T1547.009): The necromancer's hand is most dangerous when unseen. Through Shortcut Modification, he alters Symbolic Links and startup scripts, redirecting trusted actions to execute malicious code. What looks like an innocent shortcut can now point to a dangerous executable, running at system startup without detection. It’s like swapping signposts in the maze, ensuring you walk straight into his trap. 2️⃣ Symbolic Links (Soft Links): The necromancer’s mastery extends to Symbolic Links, which, when exploited, allow him to redirect not just users but critical system processes. Attackers use symbolic links to elevate privileges or access protected files, creating backdoors to the heart of your system. This isn’t just a shortcut gone wrong—it’s a rerouting of core operations to benefit the necromancer’s schemes. 3️⃣ Hard Links and NTFS Links: Hard Links give the necromancer a different tool—persistence. By creating multiple direct paths to critical files, attackers can modify or replace system files without detection. Even if defenders try to remove malicious code, the Hard Links remain, allowing the malware to resurface. It’s like building hidden tunnels in the maze that allow the necromancer to infiltrate and re-infect your system repeatedly. In NTFS systems, these links are nearly invisible, making them difficult to detect or erase. 🔥 Closing Thoughts The necromancer’s strength lies in his subtlety, twisting File System Links into traps that deceive systems and users alike. Whether through Shortcut Modification, hidden Symbolic Links, or the concealed paths of Hard Links, he turns trusted routes into dangerous deceptions. To navigate this maze safely, you must remain vigilant—scrutinize every shortcut and question every path. Dive deeper into File System Links: https://lnkd.in/ea52QtKZ #Cybersecurity #D3FEND #FileSystemLink #InfoSec

1

What a context parameter is + why we need to update how they work (how to follow the user pain) Last week we launched support for 6 new workflow blocks including things like “navigation block” and “agent block”. And now we have 2 major updates: 1. You can now define parameters for them with the ubiquitous {{ parameter }} notation 2. You can now understand that you can define parameters with that notation by clicking a handy little + icon You might be wondering.. why did we launch this feature? Well .. get ready for scrappy startup story time. When we first launched our workflows feature, we realized pretty quickly that users need to be able to define their own parameters to generate any value out of it (obvious). But we had no way for them to reference attributes within these parameters. Think of cases like.. you want to loop over every invoice you find extract from within a page. We quickly shipped this thing called “context parameters” that allowed you to build a configuration to reference them Then we followed Garry Tan's advice. Ask a user to use your product and watch them build it Every single user we watched went through the same hell: “how do I reference these parameters?” “why is it so confusing?”. It was painful. So.. we fixed it. We launched the ability for people to define these parameters using the ubiquitous {{ parameter }} notation Then we followed Garry's playbook again. Asked users to use this. And suddenly.. they’re all like “hey how do I add parameters?”. Wow. Turns out no one reads little hover texts to understand how to do it So.. we added a little + button on the side of the text box as a way to prompt the user to click on it. And it worked! Other workflow automation tools (ie Zapier) had the same pattern, and so people found it intuitive to navigate this change Lesson learned here: Ask users to use your new UX and if it makes you cringe … fix it!

14

2 Comments

🚀 OpenAI DevDay San Francisco is today! 🎉 But let's manage our expectations: 📌 This is a developer-only event 📌 No livestream available 📌 Videos will be published after the event 📌 Focus on API updates only 📌 No major model announcements expected All of this has been confirmed by the OpenAI development team in recent days. So, what API updates are you hoping to see today? Don't forget, there are two more DevDays coming up: 🇬🇧 London 🇸🇬 Singapore #OpenAI #DevDay #AIDevelopment #TechEvents

11

4 Comments