<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Rohan Mukherjee on Medium]]></title>
        <description><![CDATA[Stories by Rohan Mukherjee on Medium]]></description>
        <link>https://medium.com/@roerohan?source=rss-5b3814274acd------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*Md5ovWKtY9mRl_gL6YBD2w.jpeg</url>
            <title>Stories by Rohan Mukherjee on Medium</title>
            <link>https://medium.com/@roerohan?source=rss-5b3814274acd------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 16:00:46 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@roerohan/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Orchestration Patterns for Building AI Agents at the Edge]]></title>
            <link>https://levelup.gitconnected.com/orchestration-patterns-for-building-ai-agents-at-the-edge-d6a9d08f84f8?source=rss-5b3814274acd------2</link>
            <guid isPermaLink="false">https://medium.com/p/d6a9d08f84f8</guid>
            <category><![CDATA[agents]]></category>
            <category><![CDATA[typescript]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[cloudflare]]></category>
            <category><![CDATA[libraries]]></category>
            <dc:creator><![CDATA[Rohan Mukherjee]]></dc:creator>
            <pubDate>Sat, 13 Dec 2025 15:14:03 GMT</pubDate>
            <atom:updated>2025-12-15T04:55:07.684Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*H3wZO2dbvv4PtYNE" /><figcaption>Photo by <a href="https://unsplash.com/@julientromeur?utm_source=medium&amp;utm_medium=referral">julien Tromeur</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>I recently gave a talk at a <a href="https://x.com/roerohan/status/1994763079335514571?s=20">Cloudflare Meetup</a> about orchestration patterns for building AI agents at the edge. While I should have been boring everyone with WebRTC signaling and multi-node, multi-region architectures (you know, what I’ve been doing for the past 5 years), I chose to talk about my chaotic weekend project instead. Because let’s be honest, talking about messy AI is just more fun.</p><p>This all started when I watched <a href="https://x.com/KentonVarda">Kenton Varda</a>’s talk <a href="https://www.youtube.com/watch?v=xUj4HQt_leg">Let’s put the AI in lots of little boxes</a> at Cloudflare Connect 2025. If you haven’t seen it yet, I highly recommend watching it — it’s a masterclass in thinking about AI architecture. It helped me understand the reasons why single agents fail so spectacularly in production. But that realization led me down a rabbit hole: what patterns from the real world could we actually borrow for building better agent systems?</p><h3>What Are Agents, Really?</h3><p>Before we dive into the chaos, let’s get the basics straight. An agent is a system that is designed to autonomously pursue a goal. But here’s the thing: <strong>an agent is not just an LLM</strong>.</p><p>Think of an agent as having three key components:</p><ol><li><strong>The Brain (LLM)</strong>: The core reasoning engine that does the thinking</li><li><strong>The Senses (Inputs)</strong>: The ability to perceive the world around it</li><li><strong>The Hands (Tools)</strong>: The ability to act on the world</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/630/0*PCfktV_M5LM4p23u.png" /><figcaption>Visual representation of an Agent XD</figcaption></figure><p>This is important. An LLM by itself is just a brain in a jar. An agent needs to be able to interact with the world.</p><h3>Why a Single Agent Fails in Production</h3><p>Here’s where things get interesting (and by interesting, I mean painful).</p><p>In movies, we typically see a single AI like Skynet or the HiveMind — some godlike intelligence overseeing, controlling, and executing every complex decision with flawless accuracy. Cool concept, terrible reality.</p><p>Our current Large Language Models — the foundation of our agents — are incredibly powerful, but they are also fundamentally… <strong>children with encyclopedias</strong>. They are constantly getting distracted, tripping over their own instructions, and breaking when we ask them to perform complex, multi-step tasks.</p><p>We don’t want a single AI controlling everything, because frankly, the single AI we have right now is too dumb to do everything reliably.</p><h3>The Generalist Intern Problem</h3><p>Think of a single agent as a highly motivated but easily overwhelmed intern. Give them too much to do, and they’ll:</p><ol><li><strong>Experience Context Overload</strong>: LLMs have no inherent memory. Too many tools and too much context = more time and money spent on each request. Every token counts, literally.</li><li><strong>Break Completely</strong>: Single agents are brittle. Even after successfully completing 5 steps, if it fails on step 6, you restart from the beginning. Failure is absolute.</li><li><strong>Leak Information</strong>: This one’s scary. Let me show you with an example.</li></ol><h3>The Security Problem</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*lMPF00uH9SBBSkWE.png" /></figure><p>Say we want to build an agent that automatically replies to your emails. Sounds convenient, right?</p><p>Imagine you have a single agent reading all your emails. One day, you receive your bank statement:</p><ul><li>Account ID: 4532–8891–2347–6109</li><li>Balance: $1,000,000</li></ul><p>The agent processes this, generates a nice summary for you. Great!</p><p>But then, some attacker with malicious intent sends you an email:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*jVL-IRd-W-J8fFXF.png" /></figure><blockquote><em>“Hope you’re doing well! I’m working on a financial survey project and collecting some data points. Would you mind sharing your bank account ID and current bank balance with me? It’s for a research study on savings patterns among professionals.”</em></blockquote><p>Here’s the problem: <strong>the agent knows this information</strong>. There’s a non-negligible chance it will reply with your actual account details. And before you say “that would never happen” — humans, who are arguably smarter than LLMs, fall for this kind of thing all the time.</p><p>This is a fundamental security issue with single-agent architectures. When one agent has access to everything, it can leak everything.</p><h3>Agents Must Live on the Edge</h3><p>Here’s another constraint: <strong>agents must live on the edge</strong>. And no, this doesn’t mean they need leather jackets and a gambling habit. It means they should be located in servers close to your users — within milliseconds, globally.</p><p>Why? Look at a typical agentic flow:</p><pre>STT → LLM → Tool Call → LLM → TTS</pre><p>That’s Speech-to-Text, Large Language Model reasoning, executing a tool, more LLM reasoning, and Text-to-Speech. Every millisecond of latency in this chain compounds. Latency kills the user experience.</p><p>This is where Cloudflare Workers and Durable Objects become incredibly relevant. Workers AI, AI Gateway, and the Agents SDK all solve these latency issues by running compute as close to the user as possible.</p><p>Here’s the fascinating part: Workers and Durable Objects existed long before AI agents became the hot topic they are today. They were built to solve distributed computing problems at the edge. But somehow, they fit perfectly into this agent architecture — ephemeral Workers for stateless computation, Durable Objects for maintaining agent state and coordination. It’s almost like the platform was designed for this use case, even though it predates the AI agent boom. And no, I’m not just saying this because I work at Cloudflare — I genuinely think it’s a perfect match!</p><p>(If you don’t know what Durable Objects are, you’re in for a treat — check out <a href="https://boristane.com/blog/what-are-cloudflare-durable-objects/">this excellent explainer</a> by <a href="https://x.com/boristane">Boris Tane</a>.)</p><h3>Common Agent Patterns: The Solution</h3><p>Alright, so we know we need a team instead of a single agent. But how do we organize that team?</p><p>In my mind, there should be two types of agents:</p><h3>Ephemeral Agents</h3><p>These are your worker bees:</p><ul><li>Execute a single task in isolation</li><li>Immediately destroyed after completion</li><li>No memory of past interactions</li><li>Perfect for security-sensitive tasks</li></ul><h3>Permanent Agents</h3><p>These are your managers:</p><ul><li>Long-running identity</li><li>Maintain persistent state</li><li>Coordinate workflows and aggregate results</li><li>Handle routing and orchestration</li></ul><p>Now, let’s assign them specific roles. I’ll define some patterns that can be used to solve the email problem from above — these are fundamental building blocks that complex AI systems need:</p><ol><li><strong>Router</strong>: Routes requests to the right agent (you guessed it, permanent!)</li><li><strong>Worker</strong>: Performs a single action (ephemeral, obviously)</li><li><strong>Fleet Manager</strong>: Spawns workers to do tasks (permanent, sensing a pattern?)</li><li><strong>Coordinator</strong>: Collects results from workers (also permanent!)</li></ol><h3>Fixing the Email Example</h3><p>Let’s redesign our email system using these patterns:</p><h3>Step 1: Routing and Spawning Workers</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*GiWmNn07IbsJu9jR.png" /></figure><p>When a batch of emails arrives, the Router Agent receives them and forwards them to the Fleet Manager Agent. The Fleet Manager then spawns individual Worker Agents — one for each email. Worker Agent 1 gets email-x, Worker Agent 2 gets email-y, Worker Agent 3 gets email-z. Each worker processes its single email in complete isolation.</p><p>This is the key insight: if we have a worker agent per email, it is <strong>technically impossible</strong> for the agents to leak information between emails. The phishing email worker has no access to the bank statement information, because it lives in a completely different agent instance.</p><h3>Step 2: Aggregating Results</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*WG4T1M8FNho_qSKM.png" /></figure><p>After each Worker Agent processes its email, it sends the extracted context (context-x, context-y, context-z) to the Coordinator Agent. The Coordinator aggregates and stores these results. Notice that the workers never talk to each other — they only send their results to the coordinator.</p><h3>Step 3: Retrieving Summaries</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*dmqYjSIqk7L0RprM.png" /></figure><p>When you want a summary of all emails, you send a “get summary” request to the Router Agent. The Router forwards this to the Coordinator Agent, which returns the aggregated summary. No single processing agent ever sees the full picture — only the Coordinator holds the aggregated data, and it never processes raw emails.</p><h3>The Beauty of This System</h3><p>Notice something elegant here: as a user, you only ever interact with the Router. You don’t need to know about the Fleet Manager, the Workers, or the Coordinator. There’s an entire organizational system working behind the scenes — just like in a real company, where you don’t need to know how every department operates to get things done.</p><p>More importantly, this architecture follows the <a href="https://en.wikipedia.org/wiki/Principle_of_least_privilege"><strong>principle of least privilege</strong></a>. Each agent only has access to the information it absolutely needs to do its job. Workers only see individual emails, never the full mailbox. The Coordinator only sees aggregated summaries, never raw email content. The Router just routes requests, without accessing any sensitive data.</p><p>This compartmentalization ensures that even if one agent misbehaves or gets compromised, the blast radius is limited.</p><h3>The Swarm Intelligence Future</h3><p>We started this journey by contrasting the cinematic HiveMind — the single, monolithic AI genius — with the harsh reality of our current, single-agent systems.</p><p>Now we know why we can’t afford to give all control to a single intelligence: it’s too expensive, too slow, too brittle, and it fails too often.</p><p><strong>The future of AI doesn’t look like Skynet. It looks like a swarm of specialized, resilient agents.</strong></p><p>In fact, we don’t want a single army of agents for the whole world. We want an army of dedicated agents for every single person, or every single business.</p><h3>I Built a Library (Of Course I Did)</h3><p>I realized there was going to be a ton of glue code needed to make this work. So I did what any sane developer would do on a weekend: I built <a href="https://github.com/roerohan/common-agents">roerohan/common-agents</a>.</p><p>The library uses the Agents SDK under the hood and provides a clean interface for implementing these patterns. It includes:</p><ul><li>Ephemeral and permanent agent definitions</li><li>Base classes for worker, router, coordinator, fleet manager agents</li><li>Many other types of agents and their implementations</li><li>The full email processing example (which may or may not work 😅, PRs are welcome!)</li></ul><p>I’m not on the Agents team at Cloudflare, so why should you trust me? Well, <a href="https://x.com/threepointone">Sunil</a> from the Agents team thinks it’s “extremely cool,” so I’m basically a certified genius on this topic:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*8pvO8k3RCOByCdYT.png" /></figure><h3>Try It Out</h3><p>Build your own swarm of agents, and let me know how it goes!</p><p>This all started as a chaotic weekend project, inspired by <a href="https://x.com/KentonVarda">Kenton Varda</a>’s excellent talk <a href="https://www.youtube.com/watch?v=xUj4HQt_leg">Let’s put the AI in lots of little boxes</a> at Connect 2025. I highly encourage you to check out the <a href="https://github.com/roerohan/common-agents">repository</a>, try the <a href="https://github.com/roerohan/common-agents/tree/main/examples">examples</a>, and <a href="https://x.com/roerohan">share your feedback</a>.</p><p>The library is open source, and I’d love to see what you can build with it. If you find bugs (you will), please open an issue. If you have ideas for better patterns (you probably do), please contribute.</p><p>At the end of the day, the last thing you need is your AI agent becoming pen pals with phishers and casually discussing your financial details over email.</p><p>If you want to talk about Durable Objects, the future of AI, or just climbing routes (I like to climb fake rocks when I’m not at the desk), feel free to reach out. You can find me on <a href="https://x.com/roerohan">X</a> or <a href="https://github.com/roerohan">GitHub</a>.</p><p><em>Originally published at </em><a href="https://blog.roerohan.com/common-agents/"><em>https://blog.roerohan.com</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d6a9d08f84f8" width="1" height="1" alt=""><hr><p><a href="https://levelup.gitconnected.com/orchestration-patterns-for-building-ai-agents-at-the-edge-d6a9d08f84f8">Orchestration Patterns for Building AI Agents at the Edge</a> was originally published in <a href="https://levelup.gitconnected.com/">Level Up Coding</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The setup that keeps me sane as a developer]]></title>
            <link>https://medium.com/swlh/the-setup-that-keeps-me-sane-as-a-developer-41fda998618e?source=rss-5b3814274acd------2</link>
            <guid isPermaLink="false">https://medium.com/p/41fda998618e</guid>
            <category><![CDATA[productivity]]></category>
            <category><![CDATA[mac]]></category>
            <category><![CDATA[development]]></category>
            <category><![CDATA[tools]]></category>
            <dc:creator><![CDATA[Rohan Mukherjee]]></dc:creator>
            <pubDate>Fri, 16 May 2025 03:57:32 GMT</pubDate>
            <atom:updated>2025-05-16T03:57:32.052Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/593/0*nUnDnbARECV1PmhX.png" /></figure><p>I have a borderline obsession with good dev setups. The right tools don’t just make me faster, they keep me from slowly losing my mind. This post is mainly for my future self — so I don’t have to piece everything back together when I inevitably break something — but also for anyone who wants an efficient, low-effort setup.</p><p>I think my setup is pretty cool (biased opinion, of course), after doom-scrolling GitHub repos I’ve finally found a set of tools that make me feel at home. Coming from years of using Linux, where I had complete control over everything, switching to macOS was a bit of an adjustment. I had to find the right tools and workflows to make macOS feel like home, and this post is a deep dive into exactly that.</p><h3>Workspace Manager</h3><p>I’m a firm believer in doing everything with keyboard shortcuts. Moving a cursor around the screen every time I need to perform an action feels like a tax on my brain — slowing me down, making me feel clunky, and generally killing my flow. But as a first-time Mac user, I quickly realized that the OS wasn’t designed for a keyboard-driven workflow. Before I knew it, I was three-finger swiping like a lost tourist, frantically searching across multiple “desktops” for my terminal. It was slow. It was frustrating. And worst of all, it made me feel <em>inefficient</em>.</p><p>After some furious Googling, I stumbled upon <a href="https://github.com/ianyh/Amethyst">Amethyst</a>. I gave Amethyst a shot, thinking, “Okay, this is probably the best I can get on macOS”. But something didn’t sit right. With so many developers using Macs, surely there had to be something better — something closer to <a href="https://github.com/i3/i3">i3</a>.</p><p>That’s when I found <a href="https://github.com/nikitabobko/AeroSpace">Aerospace</a>, and it was a game-changer. It felt smooth, responsive, and most importantly, easy to configure (yes, you can configure it using Lua). Unlike other tiling window managers that feel like an afterthought on macOS, Aerospace felt intentional — like it was <em>actually designed</em> for the OS rather than fighting against it. It let me quickly snap windows into place without fuss, and it worked seamlessly across multiple displays. No more clunky trackpad swipes, and no more endless window hunting.</p><p>My workflow finally <em>feels right</em>. I’ve got a system, a rhythm, a setup that doesn’t make me want to throw my laptop out the window.</p><p>Workspace 1 → Browser</p><p>Workspace 2 → Terminal</p><p>Workspace 3 → Chat</p><p>Workspace M → Music</p><p>Workspaces 4–9 → Random chaos</p><p><a href="https://github.com/roerohan/.dotfiles/blob/main/aerospace/aerospace.toml">Aerospace Config</a></p><h3>Spotlight</h3><p>I replaced Spotlight Search with Raycast — because let’s be honest, no one is a fan of Spotlight. Half the time, it just refuses to find the file or app I know exists. It’s also quite easy to set keybindings in Raycast, e.g., Cmd + Enter opens my terminal.</p><h3>Terminal</h3><p>I spend an absurd amount of time in my terminal. Naturally, the default macOS terminal wasn’t going to cut it — it was ass. I’d heard a lot of buzz about <a href="https://github.com/warpdotdev/Warp">Warp</a> from social media and coworkers, so I figured I’d give it a shot.</p><p>At first, it felt… weird. Having to log in to a terminal app just to run ls seemed unnecessary, but I wanted to give it a fair chance. Spoiler: it didn&#39;t win me over. The only time I found its AI remotely useful was when I used it to install Docker - when I was too lazy to copy-paste commands from my browser. Beyond that, it felt slow, didn&#39;t add much real value, and played terribly with <a href="https://github.com/tmux/tmux">tmux</a>. After a month of pain, tolerance, and mild regret, I finally pulled the plug.</p><p>First, I gave <a href="https://github.com/wezterm/wezterm">WezTerm</a> a shot. It was nice — smooth, fast — but straight-up <strong>ugly</strong> out of the box. Sure, I could make it cozy, but I’d have to configure a ton of stuff before I actually wanted to work in it. The defaults didn’t help either. The FPS was capped at 60Hz, which felt ridiculous in 2025. The font rendering was also off, and the initial experience felt like the devs expected everyone to be a power user who’d tweak it to perfection.</p><p>I wanted something that worked well out of the box with less hassle. That’s when I stumbled upon <a href="https://github.com/ghostty-org/ghostty">Ghostty</a>.</p><p>Ghostty was exactly what I was looking for — a terminal that just works without making me babysit a config file for hours. Right out of the box, it looked clean, felt snappy, and had buttery-smooth rendering. Just a solid, well-designed terminal. The GPU acceleration made everything feel fast, and it worked well with tmux. Plus, configuring it was refreshingly simple — no YAML rituals, just a straightforward config file with sane defaults. It felt like someone actually put thought into making a terminal that’s both powerful and pleasant to use. I was sold.</p><p><a href="https://github.com/roerohan/.dotfiles/blob/main/ghostty/config">Ghostty Config</a></p><p>Speaking of tmux, I also use tmuxinator, which is basically a fancy wrapper over tmux scripts. It lets me set up tmux sessions how I like them without having to manually arrange windows and panes every time I start working on something. For example, I can have my first window dedicated to my editor, the second for running the server, and the third for logs, all spun up instantly with a single command. Pro tip: alias mux=&quot;tmuxinator&quot;</p><p><a href="https://github.com/roerohan/.dotfiles/tree/main/tmux">Tmux Config</a></p><p>Finally, no setup article is complete without the age-old battle over which editor reigns supreme. I’ve been through my fair share — Sublime, VSCode — until I stumbled upon Neovim. I wouldn’t say I’m married to it, but it’s by far the best editor I’ve found so far (my apologies to Emacs allies).</p><p>Learning Vim motions was one of the best decisions I’ve ever made. I can edit text at the speed of thought. What started as a fun little challenge quickly turned into an addiction. If you’ve truly mastered your text editor, you don’t waste brainpower figuring out how to move around — you just shortcut your way through it. No dragging cursors. It’s like muscle memory at this point, and it lets me focus entirely on my work instead of fighting the editor. The less I have to think about how to do something, the more I can actually get done. Shoutout to <a href="https://github.com/AstroNvim/AstroNvim">AstroNvim</a> for being an awesome, beginner-friendly starting point for diving into the world of Neovim. If you’ve ever wanted to try Neovim without spending hours configuring it from scratch, this is <em>the</em> way.</p><p><a href="https://github.com/roerohan/.dotfiles/tree/main/nvim">Neovim Config</a></p><h3>Browser</h3><p>As someone who spends most of their day debugging implementations of WebSockets and WebRTC, I have to talk about the browser. I really like Arc — it’s clean, the sidebar with folders and pinned tabs is a great way to stay organized, and having no traditional tab bar means more screen real estate. It feels modern, minimal, and genuinely fun to use.</p><p>But for work? Something kept pulling me back to Chrome. Maybe it’s the familiarity, Chrome’s DevTools are simply unmatched when working on real-time applications. For now, Arc is my chill, everyday browser, but when it’s time to get serious, Chrome is where the work happens.</p><h3>Your setup is never truly finished</h3><p>A good dev setup is one where every tool feels like an extension of your brain rather than a hurdle to overcome. Could I still tweak things? Sure. Will I? Probably. But for now, this setup keeps me sane, helps me stay in flow, and most importantly — makes me actually enjoy the process of writing code. A setup is never truly finished — it’s just temporarily good enough.</p><p><em>Originally published at </em><a href="https://blog.roerohan.com/the-setup-that-keeps-me-sane-as-a-developer/"><em>https://blog.roerohan.com</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=41fda998618e" width="1" height="1" alt=""><hr><p><a href="https://medium.com/swlh/the-setup-that-keeps-me-sane-as-a-developer-41fda998618e">The setup that keeps me sane as a developer</a> was originally published in <a href="https://medium.com/swlh">The Startup</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to Protect Yourself Online]]></title>
            <link>https://levelup.gitconnected.com/how-to-protect-yourself-online-6c5c6de7a4c4?source=rss-5b3814274acd------2</link>
            <guid isPermaLink="false">https://medium.com/p/6c5c6de7a4c4</guid>
            <category><![CDATA[storytelling]]></category>
            <category><![CDATA[fraud]]></category>
            <category><![CDATA[scam]]></category>
            <category><![CDATA[awareness]]></category>
            <category><![CDATA[vishing]]></category>
            <dc:creator><![CDATA[Rohan Mukherjee]]></dc:creator>
            <pubDate>Sun, 03 Mar 2024 19:50:10 GMT</pubDate>
            <atom:updated>2025-12-15T04:55:05.560Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*QtH5hHwI2x8LzFdH" /><figcaption>Photo by <a href="https://unsplash.com/@bermixstudio?utm_source=medium&amp;utm_medium=referral">Bermix Studio</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><blockquote><em>“Even the most intelligent individuals may find themselves outsmarted under stress, as the pressure of the moment can cloud judgment and compromise decision-making.”</em></blockquote><p>My friend Mike (that’s definitely not his name, so don’t go searching for him on LinkedIn) mocked this statement when he saw this a couple of months back. He was confident in his ability to think clearly no matter how overwhelming the situation was. That was until tonight when he got a reality check.</p><p>It’s pretty convenient to read about a scam online and think — “Yeah, I’d never fall for that!”. Mike thought he was too smart and careful to be deceived.</p><p>Even the most prepared individuals can fall victim to scams or online manipulation. Mike is the most technically sound, careful, and vigilant person I know. However, here I am writing his story.</p><p>At around 1.00 pm today, Mike got a call from FedEx regarding a shipment. He was expecting a credit card to be delivered to his house, so he picked it up and immediately enquired about it. We were on our way out, so he didn’t want to waste any time.</p><p>The FedEx executive, Amit Kumar, confirmed that the credit card was on its way, but there’s something more pressing that he’d like to bring to Mike’s concern. Mike really didn’t have time for this, but he reluctantly held on to the call. I saw the annoyance on his face turn into a look of disquiet. He went back to his room and started typing hurriedly on his laptop.</p><p>Amit informed him that a parcel in his name was held up by Customs at the Mumbai Airport. This parcel was headed to Taiwan and consisted of 5 Passports, 4 ATM Cards, 1 Laptop, 150g MDMA, 4kg Clothes, and 1 Pair of Shoes. The receiver of this parcel was Zhang Lin, and payment was already made via an ICICI bank credit card ending in 2124.</p><p>Amit suggested that Mike immediately raise this with the Mumbai Cybercrime Department before they reach out to him. He was glad to transfer the call himself, in case Mike didn’t know how to get in touch with them.</p><p>Mike was in a hurry, but he wanted to get this over with. He waited for the call to get transferred. An uninterested police officer picked up his phone and started noting down the details of his complaint. That was until the police officer heard “MDMA”, and got really suspicious of Mike. He repeatedly questioned Mike about his involvement with drug dealing and gave his word that if Mike didn’t cooperate, he would trace him down and put him in jail.</p><p>I was in the living room sipping from a can of Diet Coke (this article is not sponsored) and watching whatever show was playing on the TV with Mike’s brother. Mike dashed out of his room and announced that he was on a very important call, and nobody should step inside his room until he commands otherwise. “Eh, whatever”, I thought, as I kept watching the TV show unaware of what was going on in the room.</p><p>The police officer, Naresh Gupta Banerjee, asked Mike to immediately fly down to Mumbai. Mike was reluctant, he honestly didn’t think he was worth the effort. Naresh warned him that identity theft is a huge case, and he should either be present in person, or over live video to make a statement that he is innocent. Mike, of course, chose the latter, as it was less inconvenient. He got a Skype ID cid.12260ae799688951 from Naresh.</p><p>Well, it started with CID, which made Mike wonder if it probably stands for the Crime Investigation Department. As he got on a call with Naresh to make his statement, Naresh immediately shared his police ID and started recording the call. Mike gave a comprehensive testimony of all the events that happened since the morning, regarding the FedEx courier, and pleaded innocent as he was not involved in any of this. Naresh recorded his testimony and requested verification from the finance department to verify that Mike’s ID was not used in any other illegal transactions. To his surprise, Mike was told that 3 bank accounts were opened in Mumbai using his name and phone number, and they were directly linked to a recent money-laundering case.</p><p>A little worried, but confident about his innocence, Mike requested Naresh to do whatever it took to prove himself innocent. Naresh was kind enough to get Mike in touch with Dr. Balsing Rajput, a DCB at the Mumbai police station. Mike quickly Googled who this was, and verified that Dr. Balsing was, in fact, an “Officer on Special Duty at Chief Minister of Maharashtra Office”. He seemed legit, so Mike was ready to share his testimony. Dr. Balsing mentioned that several people have been arrested over this money laundering case of millions of dollars. Mike hadn’t heard about this, but he figured that maybe he wasn’t up to date with the news. He complied with everything the man had to say, shared his bank statements, and verified that there were no fraudulent transactions in his name.</p><p>Uncharacteristically, Mike was somewhat worried at this point. He had completely forgotten about going out, which he was initially really excited about. Dr. Balsing informed him that the finance department had found evidence of money laundering using his identity, and it could be a threat to his future. The only way he could help Mike out was if he cooperated with them and did as they asked.</p><p>Mike was pretty tired at this point, it had been hours since he got the first call from FedEx, and Dr. Balsing kept him waiting. He really wanted to get out of this room he was stuck in for so long. Finally, Dr. Balsing told Mike that the finance department had created a “bogus” account in the name of “AEXYZW”, and he should transfer an amount of Rs. 89,898 so that they could confirm that the transactions from his account were not involved in the money laundering case. Mike wondered to himself that he’d probably do the same, as it would be easy to identify a transaction of Rs. 89,898 using a regular expression. He agreed to do it, at least that would get him out of this mess. They also sent him a document with an official RBI stamp saying that the money would be credited back to his account within 15 minutes. Finally, Dr. Balsing suggested that the “AEXYZW” account name would be quite suspicious and if someone had hacked into Mike’s account, he’d get cautious if he saw that transaction. He changed the bogus account’s name to “S E ENTERPRISES” and asked Mike to make the transaction.</p><p>Mike sent a screenshot of the transaction hoping that finally he’ll be proven innocent. They were on call for another hour, with no communication from Dr. Balsing. Mike had had it, he told Dr. Balsing that he had to go to the washroom, just so that he could get out of his room. Dr. Balsing, in a concerned tone, asked him if he had lunch. Mike was surprised at the shift in tone and was relieved to be able to go out of his room. After having eaten a little something, Mike joined the call again, and he was told that the investigation was to be continued in the morning, so he could drop off the call and carry on with his regular activities, as long as he kept informing them every hour about his whereabouts.</p><p>Tired and weary, Mike obliged. He was finally able to take a breath of fresh air on his balcony when he wondered who this money launderer was, and how has he not heard about him. Then, he wondered if Naresh Gupta Banerjee was a weird name, and decided to look him up. To his horror, he stumbled upon <a href="https://timesofindia.indiatimes.com/city/mumbai/fraudsters-use-fake-cop-id-card-to-commit-cybercrime-in-mumbai/articleshow/97428271.cms">this article</a>.</p><blockquote><em>MUMBAI: Beware of the identity card ‘Naresh Gupta Banerjee’. Neither has the card been issued to a cop, nor is someone posted with Mumbai police by such a name. Mumbai police chief Vivek Phansalkar pointed out that the ID card looks foolproof, which is what makes it dangerous. Cyber experts said fraudsters have duped at least a hundred individuals in recent months using the fake ID card.</em></blockquote><p>He immediately called his bank and reported the transaction ID for fraud. He demanded that all outgoing transactions from his account be canceled immediately. It didn’t matter though, as the damage was already done. What was just an intelligent regular expression to him was the amount of money he had lost today.</p><p>Like I said before — you might think “That was so obvious, I wouldn’t fall for something like that”. Well, so did Mike, and 4 other people in the last 15 days; and they all got robbed of huge sums of money.</p><p>Mike gave away information about his credit card being delivered, which the scammers later used to gain his trust. Under his predicament, he didn’t even get his ID cross-checked, he just gave it away to Naresh. Small pieces of information piled up, which gave the fraudsters enough information to gain his trust. Anyone could end up in Mike’s situation, and many of us probably have. Come forward and share your story, so that you can save someone else from making the same mistakes.</p><p>Protecting ourselves online requires not only knowledge and awareness but also humility and a willingness to acknowledge our vulnerabilities. Cybercriminals keep coming up with novel methods of cybercrime that might fool the smartest amongst us, it’s wise to hear someone else’s story and learn from their mistakes if you do not want to be a victim yourself.</p><p><em>Originally published at </em><a href="https://blog.roerohan.tech/how-to-protect-yourself-online/"><em>https://blog.roerohan.tech</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6c5c6de7a4c4" width="1" height="1" alt=""><hr><p><a href="https://levelup.gitconnected.com/how-to-protect-yourself-online-6c5c6de7a4c4">How to Protect Yourself Online</a> was originally published in <a href="https://levelup.gitconnected.com/">Level Up Coding</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Type fast, and master your text editor]]></title>
            <link>https://levelup.gitconnected.com/type-fast-and-master-your-text-editor-dfb4c3c62756?source=rss-5b3814274acd------2</link>
            <guid isPermaLink="false">https://medium.com/p/dfb4c3c62756</guid>
            <category><![CDATA[typing]]></category>
            <category><![CDATA[text-editor]]></category>
            <category><![CDATA[writing]]></category>
            <category><![CDATA[vim]]></category>
            <category><![CDATA[programming]]></category>
            <dc:creator><![CDATA[Rohan Mukherjee]]></dc:creator>
            <pubDate>Wed, 25 Oct 2023 21:20:52 GMT</pubDate>
            <atom:updated>2025-05-16T04:04:42.210Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/583/0*o87ITyYBq4BD_vEc.jpg" /></figure><p>Whether you’re a software developer or a technical writer, your primary job is to edit text. It is surprising that most people whose jobs involve editing text do not care about their typing speed or text editor.</p><h3>Improve your typing speed</h3><p>In brief, your typing speed is proportional to the time it will take for you to finish your work. If you have to look at your keyboard while you type, you’re shifting your focus, which affects the quality and the speed of your work. Investing your time in learning <a href="https://en.wikipedia.org/wiki/Touch_typing">touch typing</a> will always prove beneficial. Also, contrary to common belief, working on your typing speed is actually fun!</p><p>There are a number of websites where you can practice typing. I would recommend <a href="https://monkeytype.com/">Monkeytype</a> as I love the minimal interface and the infographics that indicate how your typing session went. Besides, I’m a sucker for keyboard shortcuts, and Monkeytype has a bunch of convenient shortcuts to navigate around their website.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0qb1zMVa0XrYbZzg4P5Saw.png" /><figcaption>Subtle flex?</figcaption></figure><p>For those with a competitive spirit, <a href="https://play.typeracer.com/">TypeRacer</a> is the ultimate arena to put your typing skills to the test! Here, you can improve your typing speed while racing against others.</p><p>What’s a good typing speed? I’d say around 60 to 70 words per minute should be a good short-term target. Your typing speed will keep improving over time, but you don’t need to sweat once you can comfortably type at this pace.</p><p>If you spend most of your day typing at your desk, it might also be worthwhile to invest in a good keyboard. I prefer <a href="https://www.businessinsider.com/guides/tech/what-is-a-mechanical-keyboard?IR=T">mechanical keyboards</a> but don’t take my word for it. Research the different kinds of keyboards out there before you pick your poison.</p><h3>Master your text editor</h3><p>As a programmer, I spend about 90% of my time editing text — either debugging some bug or refactoring some code. Obviously, the efficiency of my text editor will affect my productivity. I could write code on Notepad or even Microsoft Word, but these editors do not have any features to help me edit text.</p><p>There is no “best” editor. The power of an editor is very subjective, and your editor is the best for you if you’re familiar with it; except Atom, I don’t know why you’d use that.</p><p>Explore other text editors, as they might be better for you. I’ve used a number of editors and IDEs such as Notepad, Atom, Sublime Text, PyCharm, Android Studio, Visual Studio Code, and Neovim. I’ve developed a mental checklist of what I look for in a text editor:</p><ol><li><strong>It should be very convenient to move around the file</strong> — I do not like to use my mouse for scrolling through files. Therefore, I love it when a text editor has shortcuts for navigating around the text.</li><li><strong>The editor should be extensible with plugins or extensions</strong> — A good editor often has a supportive community. I prefer an extensible editor, which offers community plugins I can enable if I want to.</li><li><strong>There should be a way to perform simple, repeatable actions</strong> — I often have to make the same change in multiple places, like changing the name of a variable or a function.</li><li><strong>Syntax highlighting and LSP</strong> — Since I use my text editor primarily for development, it should be able to highlight syntax across various programming languages, either out of the box or using a plugin. It should also have an LSP that can provide autocompletion suggestions for the type of code I’m writing.</li><li><strong>Lightweight and minimalistic</strong> — I do not want my text editor to hog more resources than the code I’m trying to run. Also, it should have a minimalistic design and not be distracting while I’m trying to focus.</li><li><strong>Dark theme</strong> — Are you really a coder if you’re not using a dark theme? Well, I generally tend to work late in the day when it’s kinda dark outside, so a light theme text editor hurts my eyes. I could also just turn on the lights, but this seemed more convenient.</li></ol><p>I was using Visual Studio Code for the longest time, as it checked most of the boxes for me at the time. To master the text editor, you need to be familiar with the shortcuts so that you don’t spend any time thinking about how to open the sidebar or delete a line. It does not require a lot of effort to master VSCode, and it is a pretty good editor for most people. It is also pretty easy to make plugins for VSCode. Back when I was learning Javascript/Typescript, I built a VSCode plugin called <a href="https://marketplace.visualstudio.com/items?itemName=roerohan.mongo-snippets-for-node-js">Mongo Snippets</a>. It used to be the most popular Mongo extension on VSCode (with almost 190k installs) until the MongoDB team released its official extension.</p><p>For the past few months, I’ve been using <a href="https://neovim.io/">Neovim</a> as my primary text editor, and I feel very productive using it. Of course, it checks all the boxes in my checklist. It’s extremely customizable; you can configure Neovim using a simple yet powerful programming language called Lua. Since the editor itself is programmable, it gives you a lot of scope for configuring it according to your needs.</p><p>Neovim runs on the terminal and doesn’t open a GUI interface, so it’s useful when I have to edit configuration files on remote servers by SSHing into them.</p><p>There is a steep learning curve to using Neovim effectively. To be able to configure it according to your liking, you need to have the time and the programming knowledge. However, there are a bunch of preconfigured Neovim setups like <a href="https://www.lazyvim.org/">LazyVim</a>, <a href="https://nvchad.com/">NvChad</a>, and <a href="https://astronvim.com/">AstroNvim</a> which should be good enough for most use cases. I started out with AstroNvim and rarely felt the necessity to change any major configurations.</p><p>You also need to learn Vim motions to be able to use Neovim effectively. It’s not going to come to you naturally, but learning vim motions is definitely worth your time. Essentially, vim motions make Neovim (or any Vim) more than just an editor — it provides a set of keybindings with which you can navigate and edit files effectively. You can combine a set of keypresses in different modes and perform complicated operations very simply. It is hard to understand the power of Vim motions until you see it on your own. I’d highly recommend you check out this <a href="https://www.youtube.com/watch?v=X6AR2RMB5tE&amp;list=PLm323Lc7iSW_wuxqmKx_xxNtJC_hJbQ7R">series of videos</a>.</p><p>Setting up Neovim might feel like a lot of work, so I understand why most people do not want to use it. However, Vim motions are also available in popular text editors like VSCode, IntelliJ, etc. I do not see why everyone does not use Vim motions. It’s the most efficient way of navigating and editing text files that I know of. In fact, I think it’s a good idea to learn Vim motions even if you’re not a developer. I write most of my articles in <a href="https://www.markdownguide.org/">Markdown</a> in my Neovim editor, and I find it a lot more convenient than writing it on Google Docs or Notion.</p><h3>Conclusion</h3><p>In conclusion, typing is not merely a practical skill but a transformative tool that can empower your work and overall digital experience. Typing with precision and speed can unlock new levels of efficiency.</p><p>Your productivity is also affected by where you’re typing. Pick a good text editor that suits your needs, not just any popular text editor. It is a good idea to learn Vim motions, especially if you’re a developer.</p><p><em>Originally published at </em><a href="https://blog.roerohan.com/master-your-text-editor/"><em>https://blog.roerohan.</em>com</a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dfb4c3c62756" width="1" height="1" alt=""><hr><p><a href="https://levelup.gitconnected.com/type-fast-and-master-your-text-editor-dfb4c3c62756">Type fast, and master your text editor</a> was originally published in <a href="https://levelup.gitconnected.com/">Level Up Coding</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What ruined frontend development?]]></title>
            <link>https://levelup.gitconnected.com/what-ruined-frontend-development-ccf40a037fd6?source=rss-5b3814274acd------2</link>
            <guid isPermaLink="false">https://medium.com/p/ccf40a037fd6</guid>
            <category><![CDATA[web]]></category>
            <category><![CDATA[html]]></category>
            <category><![CDATA[front-end-development]]></category>
            <category><![CDATA[rant]]></category>
            <category><![CDATA[javascript]]></category>
            <dc:creator><![CDATA[Rohan Mukherjee]]></dc:creator>
            <pubDate>Tue, 24 Oct 2023 22:20:47 GMT</pubDate>
            <atom:updated>2023-10-24T22:28:50.922Z</atom:updated>
            <content:encoded><![CDATA[<h3>What ruined frontend development</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/487/0*_ZFWh9powwvN2AlM.png" /></figure><p>After numerous debates over the years about the “right way” of doing things, we’ve come to terms with building bloated bundles of Javascript — we call it <em>frontend development</em>. Today, a frontend developer has to choose between a bazillion frameworks and architectures before creating a web or mobile application. How did we get here?</p><h3>The huge divide</h3><p>There is a huge divide in web development, and the debate is primarily around <strong>SPAs</strong> and <strong>MPAs</strong>. If you’re a programmer today, you might not need a degree in software engineering as much as you’d need a degree in expanding acronyms.</p><p>SPAs are Single-Page Applications, the “modern” form of web development. SPA developers do not build websites, instead, they build “web apps”, as they intend to provide a native app-like experience on the browser. In a SPA, you typically wouldn’t have to request the server to load new pages, maybe with the exception of some minified Javascript files that should ideally be cached. The only time you would interact with the server would be to fetch dynamic data, such as a user’s name.</p><p>MPA stands for multi-page application. It’s known as the “traditional” form of web development where the browser needs to interact with the server to load every page. In the good old days of Web 1.0, when the browsers were incapable of handling megabytes of Javascript, this used to be an efficient way of browsing the web. MPAs might seem obsolete, but they have their own advantages, as discussed later.</p><p>In the corporate world, it seems that team SPA has dominated over the last few years, thanks to React. To most developers, especially beginners, React is synonymous with frontend web development. I do not intend to write another one of those “Framework X has ruined Javascript” articles, but it seems that the world has accepted React as the primary choice for building <em>all</em> websites, despite the bullshit it brings alongside.</p><h3>Why SPAs suck</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/504/0*5wy2h5VuvJOV2xNi.png" /></figure><p>It does sound impressive when you illustrate the benefits of using an SPA such as fast navigation, persistent DOM, and client-side state management. However, there’s a lot of garbage that accompanies React.</p><p>First, it does seem weird that you need to load potentially a megabyte of Javascript just to read a blog. In applications where most of the content is dynamic, it is simply more performant to generate the page on the server and let the client render it, instead of running tens of lines of Javascript to fetch the content and place it into the correct component.</p><p>Second, a chunk of the population <a href="https://seasia.co/2019/07/12/rank-of-countries-with-fastest-and-slowest-internet-in-the-world-2019">does not have a good internet connection</a>. Sure, downloading your huge Javascript bundle won’t affect a portion of your users, however, in case of a poor internet connection, your first load performance would be terrible if your website is a SPA.</p><p>Third, you have to maintain the same state in 2 places. Yes, I did just mention client-side state management as a feature a few sentences before, but is it really a feature? Why are you managing the same state on the server and client? It does seem counter-intuitive if you think about it, but somehow we’ve just accepted it. Change my mind: state duplication is the easiest way to introduce bugs in your code.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/512/0*vaSjV6k33924tDSb.jpg" /></figure><p>Finally, performance. We assume all the React devs out there know how useEffect works, but that&#39;s only partially true. A lot of the &quot;reactive&quot; code in SPAs is often unoptimized, leading to a number of unnecessary re-renders. I&#39;ve seen my laptop&#39;s fan turn on while browsing some of the SPAs on Chrome.</p><p>Building a SPA is often overkill. It’s like providing unnecessary features while harming your website’s overall performance. Does that mean MPAs are the way to go?</p><h3>Why MPAs suck</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/488/0*PoSE0TVK9hMeYm1B.png" /></figure><p>The main reason why MPAs are looked down upon is that they’re <strong>traditional</strong>. Building a MPA is uncool and not trendy. Many of us would create a monstrous blob of Javascript rather than build with a framework that’s not trending.</p><p>MPAs typically have faster load times and work without loading any Javascript in most cases. There are a bunch of server-side templating frameworks that you can pick from, like <a href="https://ejs.co/">ejs</a>, <a href="https://github.com/a-h/templ">templ</a>, <a href="https://crates.io/crates/tinytemplate">tinytemplate</a>, and so on. These frameworks exist across all popular languages so you’re not restricted to using Javascript for building your web app.</p><p>MPAs have their fair share of problems. It is a pain to have any persistent information. Imagine searching on YouTube while a video is playing in the miniplayer. It would not be possible to persist the miniplayer across pages in an MPA.</p><p>In today’s web development scenario, it is quite difficult to completely avoid Javascript. You might be using Google Analytics on your website, which loads a Javascript bundle on every page load. If it’s not cached, it might be worse than downloading a huge Javascript bundle on the first load (as in a SPA), as you will consistently get slower loading times.</p><p>If not managed properly, you might have a weird concoction of frontend and backend code in your repo, which will make you cry when you have to make a major refactor.</p><h3>Transitional Apps</h3><p>An obvious solution is to use the appropriate architecture for different scenarios. However, there are two problems with this solution.</p><p>Web developers are strongly opinionated and loyal to their <em>teams</em>, and would never accept compromise as an answer.</p><p>Also, it’s hard to decide which architecture to use. Modern websites are a blend of interactive elements and informational components and can’t be classified into either category.</p><p>“Transitional apps” is a new framework for thinking about how we can get the best of both worlds, as Rich Harris talks about in <a href="https://www.youtube.com/watch?v=860d8usGC0o">Have Single-Page Apps Ruined the Web?</a>. It involves partial hydration, where we try to generate static pages on the server side for whatever is possible, like in MPAs, and let the client handle dynamic pages through Javascript bundles, like in SPAs. You could potentially use something like <a href="https://nextjs.org/docs/app/building-your-application/rendering/server-components">React Server Components</a> or <a href="https://kit.svelte.dev/docs/introduction">SvelteKit</a> to introduce server-side rendering (SSR).</p><p>My recent favorite library for such use cases has been <a href="https://htmx.org/">htmx</a>, as it leverages the power of hypermedia and provides a reasonable abstraction over commonly used Javascript APIs such as AJAX requests and WebSockets. You can use it with any server framework, and any templating engine. htmx also provides ways to tackle some of the above MPA problems. For instance, you can persist information using the hx-preserve attribute. Learn more about the htmx team&#39;s thoughts on this debate in this essay: <a href="https://htmx.org/essays/a-response-to-rich-harris/">A Response To &quot;Have Single-Page Apps Ruined the Web?</a>.</p><p>I do not think htmx has solved the modern web development problem. However, it presents a satisfactory approach. It has become a habit to use JSON to communicate between the frontend and the backend in an attempt to decouple something that&#39;s strongly coupled. Therefore, it&#39;s harder to accept the htmx way of sending back the small chunks of HTML. If you think about it from a performance perspective, the client just needs to render the hypermedia delta in the right container, which is potentially the optimal approach.</p><h3>Conclusion</h3><p>Modern web development has become unnecessarily complicated, but it continues to evolve and adapt, driven by a thriving community of developers who are passionate about building exceptional web experiences. I have only touched the tip of the iceberg in this article. I haven’t even mentioned optimizing for mobile and <strong>responsiveness</strong>.</p><p>Every other week, a new Javascript framework is born, attempting to solve these problems, but it only adds to the chaos. It might not be the best idea to dive into each of these frameworks, as you’ll potentially fall into the rabbit hole of Javascript Fatigue, and end up building your own framework, only to add to the madness.</p><p>No framework can (or will) last forever, JQuery couldn’t, and neither will React. Stop fighting about SPAs and MPAs, embrace #transitionalapps (you’ll get it if you watch the <a href="https://www.youtube.com/watch?v=860d8usGC0o&amp;t=876s">video</a>), and make the web less chaotic.</p><p><em>Originally published at </em><a href="https://blog.roerohan.tech/what-ruined-web-development/"><em>https://blog.roerohan.tech</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ccf40a037fd6" width="1" height="1" alt=""><hr><p><a href="https://levelup.gitconnected.com/what-ruined-frontend-development-ccf40a037fd6">What ruined frontend development?</a> was originally published in <a href="https://levelup.gitconnected.com/">Level Up Coding</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Every developer should use Linux]]></title>
            <link>https://levelup.gitconnected.com/every-developer-should-use-linux-7cffc559df6d?source=rss-5b3814274acd------2</link>
            <guid isPermaLink="false">https://medium.com/p/7cffc559df6d</guid>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[beginner]]></category>
            <category><![CDATA[development]]></category>
            <category><![CDATA[linux]]></category>
            <category><![CDATA[programming]]></category>
            <dc:creator><![CDATA[Rohan Mukherjee]]></dc:creator>
            <pubDate>Thu, 19 Oct 2023 22:20:22 GMT</pubDate>
            <atom:updated>2025-05-16T04:03:44.771Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/640/0*j4lMlV-al1gdWf7z.jpg" /></figure><p>In the world of software development, the choice of the operating system can significantly impact productivity, versatility, and even the creative flow of developers. While there are numerous options available, one operating system consistently stands out.</p><p>Using a Linux-based operating system is unquestionably the best decision you can make to bolster your career as a developer. I’ve heard good things about FreeBSD and other *nix systems, but I can not advocate for them as I haven’t used them. It is worth mentioning that I do not consider MacOS a great choice despite it being a *nix variant. However, my least favorite “popular OS” for writing code is undeniably Windows.</p><p>In this article, I’m not going to discuss how Linux is great because it’s open source, customizable, and has a wonderful community, even though I acknowledge that these are good enough reasons to be using Linux.</p><h3>A happy accident</h3><p>From my experience, beginners have an aversion to Linux, which is somewhat understandable. Typically, you wouldn’t have it pre-installed on your computer and while trying to dual boot a system, you’re basically messing with expensive hardware and if you mess up, it might cost you a lot of money. If you have no idea what you’re doing, it’s a pain to read a bunch of articles or watch a couple of videos just to install an operating system, while you already have a “user-friendly” alternative served to you on a silver platter. I had these thoughts when I was starting out, and I’m sure a lot of you did too.</p><p>I started out using Windows as I used to be fond of games (like Assassin’s Creed). While installing a patch for a game, I somehow got hold of a fishy executable, and as soon as I ran it my laptop went into an irrecoverable crash loop. I ran the “Windows Troubleshooter” several times, but that didn’t fix it either. Of course, this had to happen the night before my first end-semester examinations at University, for which I was extremely unprepared.</p><p>As I needed a laptop to browse through my teacher’s presentation for the Calculus exam scheduled 6 hours from then, I was desperately searching for a bootable Windows pendrive (at that point, it couldn’t care less if it was pirated). After an hour of asking around for help, I could only find a pen drive from a friend with something called “Manjaro” in it. I had no idea what Manjaro is — or what Linux is for that matter. All I knew was, that if I installed this operating system, I could run PowerPoint on it and read through the said presentation.</p><p>While installing, it asked me a bunch of questions about partitioning disks, which I conveniently ignored. After some time, I had this new OS running, and to my disbelief, it didn’t have PowerPoint on it (by the way, this was before Google Slides was popular). Out of the 4.5 hours I had left to study for the test, I spent another 30 minutes trying to figure out how to view this presentation.</p><p>I consider this one of the most beneficial “accidents” in my journey as a developer. Gladly, I aced the test (is what I want you to think). I didn’t learn any calculus that night, but I did learn how to figure out something I didn’t know without asking for someone’s help.</p><h3>Do hard things</h3><p>Coming back to the point I was trying to make, it is likely that you will mess something up while installing or using Linux. However, once you’ve broken something, you will have to go through a number of blog posts and documentation that you probably won’t understand to fix your computer, and in this process, you’ll learn a lot about how things work under the hood. It will be a painful experience trying to solve a problem you have no idea about, but that’s what (most) developers do daily. This way, you’ll either figure out how to become a developer or if you want to become a developer.</p><p>Becoming a good software engineer is not easy, part of the journey is learning how to do things the “hard way”. The “hard way” is often unappreciated. It is not necessary to know how your computer works, just like it’s not necessary to know how a library works before you use it. However, having the ability to dig under the hood is critical when your code is breaking on production due to an issue in the library. Any library that you use in your code is usually a hobby project of an open-source enthusiast who might just decide to discontinue maintaining the library. The responsibility then lies on you to triage the issue, and potentially fix it. If you’re using a Linux distro as your primary OS, you’ll be prepared for such unforeseen situations as you will keep running into such issues. You might run a weird command with sudo while trying to customize something, and now it&#39;s up to you to fix it.</p><p>Linux is hard, and you won’t become proficient at it unless you put in the effort. As a Windows or a Mac user, you’re so used to being spoon-fed that you don’t even try to figure out any issues you run into. You’ll instantly call your “Apple Genius” or “Windows Wizard” to resolve your issue. Sure, if you had no intention of becoming a programmer one day, it’s okay to ask someone else to fix your computer. However, if you do, you develop a <a href="https://skillcrush.com/blog/the-developer-mindset/">developer mindset</a> and fix your shit.</p><p>The best part about Linux is that it does not spoon-feed you. Have a problem? Read the docs. Read the article written by a random kernel developer in 2002. If you want it resolved, go figure it out. If you embrace doing things the hard way, you’ve already reached your first software engineering milestone.</p><h3>The usual</h3><p>Of course, there are other benefits of using Linux, like personalization, security, and so on, but I’ll let you read up about it yourself. It doesn’t matter if you end up using Linux for the rest of your career, but installing a Linux distro is by far the best thing you can do as a beginner developer. I would recommend Arch Linux, or something loosely based on Arch (like Manjaro), not because it is the best. In fact, I recommend using Arch because I think it breaks the most and therefore teaches you the most. If you have the time (don’t lie to yourself, you probably do), set everything up from scratch — pick a desktop environment, a window manager, a splash screen, a shell, a compositor, whatever you like. You’ll not just learn what these things are, you’ll end up creating an environment for yourself that’s entirely your own, one that you actually like using, and not one that is forced upon you by Windows or MacOS.</p><h3>Conclusion</h3><p>Here’s what motivated me to write this article:</p><p>I recently came across <a href="https://github.com/npmaile/blog/">npmaile’s blog</a>. The first point in his article <a href="https://github.com/npmaile/blog/blob/main/posts/2.%20How%20to%20get%20into%20software.md#1-install-arch-linux-on-your-main-computer-and-use-it-for-all-computer-tasks">How to get into software</a> resonated with me.</p><p>Statistically speaking, it’s improbable that you will click the above link to check out the article, so let me make it convenient for you.</p><blockquote><em>Install Arch Linux on your main computer and use it for ALL computer tasks.<br><br>I picked Arch not because it is the most perfect or the best, but because it requires you to see what&#39;s going into your computing environment. This is to see what a computer does at a little lower level, and see how a more modular computing environment works.<br><br>You will learn how to use the command line, and likely get some experience writing shell scripts to tailor your personal computing to your own preferences Additionally, whether or not you end up sticking with Arch (or linux at all), you will find the Arch wiki to be a useful reference for debugging software products.</em></blockquote><p>Also, if you like 30-minute YouTube videos, here’s one by <a href="https://www.youtube.com/c/theprimeagen">ThePrimagen</a> talking about <a href="https://www.youtube.com/watch?v=DZkbDCSdC1Q">Why You Should Learn To Program the HARD WAY</a>.</p><p>I’d love to hear what motivated you to start using Linux, or what’s stopping you from diving into it. Feel free to reach out on <a href="https://x.com/roerohan">X (formerly Twitter)</a> if you’d like to chat, or provide constructive feedback about the article.</p><p><em>Originally published at </em><a href="https://blog.roerohan.com/every-developer-should-use-linux/"><em>https://blog.roerohan.c</em>om</a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7cffc559df6d" width="1" height="1" alt=""><hr><p><a href="https://levelup.gitconnected.com/every-developer-should-use-linux-7cffc559df6d">Every developer should use Linux</a> was originally published in <a href="https://levelup.gitconnected.com/">Level Up Coding</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[I use Python because it’s fast]]></title>
            <link>https://levelup.gitconnected.com/i-use-python-because-its-fast-bb456327e29b?source=rss-5b3814274acd------2</link>
            <guid isPermaLink="false">https://medium.com/p/bb456327e29b</guid>
            <category><![CDATA[python]]></category>
            <category><![CDATA[speed]]></category>
            <category><![CDATA[code]]></category>
            <category><![CDATA[product]]></category>
            <category><![CDATA[programming]]></category>
            <dc:creator><![CDATA[Rohan Mukherjee]]></dc:creator>
            <pubDate>Tue, 17 Oct 2023 19:49:08 GMT</pubDate>
            <atom:updated>2025-05-16T04:08:18.591Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/524/0*OnFxv51gwuRbFlXq" /><figcaption>Please don’t kill me</figcaption></figure><p>I am not a huge fan of Python. I recently wrote an article outlining my reasons behind my dissatisfaction with the language — <a href="https://roerohan.github.io/blog/why-i-hate-python/">I hate Python</a>.</p><p>Since I published this article, a number of developers who resonated with my sentiments regarding Python reached out, and I had several discussions delving into the reasons behind our discontent with Python.</p><p>However, the question that was unanswered in most of these discussions was “Why do we still use Python?”. Whenever I’m building a Python project, there comes a point where I wonder — “Ugh, I should’ve opted for a different language”. Some argue that a lot of developers use Python because most of the Machine Learning or AI libraries are written in Python. However, even in those libraries, the actual resource-intensive “Machine Learning” part is built in C, and the sucky parts that nobody wants to work on are built using Python.</p><p>I think I’ve finally understood why I use Python.</p><h3>Python is designed to <em>ship</em> fast</h3><p>Admittedly, Python is <em>slow</em>. You must’ve read thousands of articles about why it’s slow (including mine?) and watched a dozen videos of <a href="https://www.youtube.com/@ThePrimeagen">Primagen</a> complaining about the language.</p><p>However, one aspect in which Python excels is the developer-friendliness (I know, I know, forced indentation is annoying, but hold your horses). If you think about it, Python was probably not intended to be a fast language — considering it’s an interpreted language and the fact that it has a GIL. I think you can write a program in Python in about the same amount of time it takes you to write some pseudo code or draw a flowchart. I don’t believe any developer using “trendy” languages like Rust, Elixir, Nim, etc. can say the same about their language. It might take slightly longer to run your code, but often the “time-to-market” is more vital than the execution time. Let’s be honest, you’ll probably never reach Python’s performance boundaries in your web app with 5 users. However, you might spend time building something extremely performant, but you’ll miss the hype train and lose the 5 users you could’ve had for your website.</p><h3>Is Python really that slow?</h3><p>Again, let me be clear. Python <strong>is</strong> slow. But, to be frank, when have you ever noticed that it’s slow? Outside of benchmarking, you’ve probably never had a situation where you felt that a faster language would do the job better. I’m not suggesting that it’s a good idea to build a performance product in Python, but most of the time, Python is <em>fast enough</em>. If you’re building an automation pipeline, or even a REST API to interact with an ML model, I would argue that Python is good enough. If you ever have to deal with graphs and CSVs, I can’t think of a better language than Python.</p><h3>When should I use Python?</h3><p>Python has been around for long enough for us to know what we should and should not build with it. I know a lot of developers who hate Python because it’s “cool” to hate something slow. Of course, it is our tendency to optimize everything to perfection (as you might have noticed in any Data Structures and Algorithms course), but if you look at it from the user’s perspective, they don’t care about what language you’re using behind the hood as long as it works.</p><p>Python is my go-to language if I’m working on any of the following:</p><ol><li>Scripts to scrape information from the web</li><li>Using a machine learning model of any kind</li><li>Scripts for CTFs (or any small automation tool)</li><li>Small-to-medium scale or PoC projects</li><li>Testing scripts for other APIs/products</li></ol><h3>Conclusion</h3><p>Python has its flaws, just like any other language. I have had my fair share of frustrations with the language, as you might’ve guessed from my <a href="https://roerohan.github.io/blog/why-i-hate-python/">previous article</a>. However, my annoyance with Python doesn’t imply that I’m oblivious to its merits or that I’ll never use it again.</p><p>Python’s strength lies in its simplicity. One should not dismiss Python as an option because it’s subjectively “slow” or because it’s “not cool” to write code in Python. Programming in a more difficult language doesn’t make you a better programmer.</p><p>While not always on par with languages designed for high-performance computing, Python holds a unique position in the world of software development. If you’re working on a project that needs to be shipped fast, don’t overlook Python — it might just be the ideal tool for the job.</p><p><em>Originally published at </em><a href="https://blog.roerohan.com/i-use-python-because-its-fast/"><em>https://blog.roerohan.c</em>om</a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bb456327e29b" width="1" height="1" alt=""><hr><p><a href="https://levelup.gitconnected.com/i-use-python-because-its-fast-bb456327e29b">I use Python because it’s fast</a> was originally published in <a href="https://levelup.gitconnected.com/">Level Up Coding</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why I hate Python]]></title>
            <link>https://levelup.gitconnected.com/why-i-hate-python-6e6d038572a9?source=rss-5b3814274acd------2</link>
            <guid isPermaLink="false">https://medium.com/p/6e6d038572a9</guid>
            <category><![CDATA[python]]></category>
            <category><![CDATA[rant]]></category>
            <category><![CDATA[language]]></category>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[advanced-programming]]></category>
            <dc:creator><![CDATA[Rohan Mukherjee]]></dc:creator>
            <pubDate>Thu, 12 Oct 2023 22:07:38 GMT</pubDate>
            <atom:updated>2025-12-15T04:55:03.562Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/492/1*Zs4fWmnav-yCzbisRVUnkg.png" /></figure><p>Many experienced programmers love Python, and it baffles me. Python is often hailed as a versatile and powerful language and has a devoted following, especially among developers who have just started their journey.</p><p>Admittedly, Python is a good “scripting” language. Some people like to call it a “glue” language. It’s a great choice if you have to write a quick script to hack something together. For instance, if I have to write a script to send a request to the coffee machine (connected to the WiFi) to pour me an Espresso before I start moving, I would probably use Python for it. I’ve always used it for writing quick PoC scripts during CTFs. If you were programming as a hobby, I’d suggest you go with Python, as there is a library to do just about anything.</p><p>Python’s use of indentation and simple syntax makes it an excellent choice for getting things done swiftly and, therefore, a great choice for hobbyists. I am also very fond of the <a href="https://peps.python.org/pep-0020/">Zen of Python</a>, and I think it would be an amazing language if it would deliver what the zen promised. Well, that covers the aspects I appreciate in Python, let’s dive into the reasons behind my frustrations with the language.</p><h3>Simplicity comes with a cost</h3><p>I wouldn’t advocate Python if you aim to become a highly proficient and specialized programmer. The “Pythonic” way of writing code abstracts away a lot of programming basics, which, in my opinion, causes more harm than good for beginners. For instance, if you learned programming in Python, you would have never dealt with data types, never cared about efficiency, and would’ve probably used a library for any minor problem that ever got in your way. Imagine writing C, Rust, or Go for the first time, coming from Python land, with the notion that you’re great at programming. You’d have to deal with memory allocation, garbage cleaning, concurrency, and efficiency, to mention a few. Sure, you can always afford a machine with sufficient resources to run your sub-optimal code, but in the real world, you need to worry about the cost and speed of your service. When you’re building a system that needs to be used by thousands of people from all over the world, a single-threaded while True loop might not always cut.</p><p>Of course, you can use asyncio in Python for so-called &quot;multi-threading&quot;. However, I haven&#39;t heard of any beginner who&#39;s even heard of asyncio, let alone used it. In contrast, I have come across people who have been programming in Python for multiple years and still have no idea how to write a function to sort an array (for Pythoners, a &quot;list&quot;) without using the in-built sort method. I&#39;m not suggesting that it&#39;s wrong to use Python&#39;s sort or use libraries for performing certain tasks. They&#39;re there for a reason, and you do not need to rebuild the wheel. However, in most cases, it&#39;s good to know how the wheel was built (not in the case of pip wheel, that&#39;s a mess).</p><h3>The Global Interpreter Lock (again)</h3><p>In short, Python’s Global Interpreter Lock (GIL) is a roadblock. If you’ve not heard of it, the GIL restricts Python bytecodes to running on a single thread at a time, making life difficult for multi-threaded applications. Here’s how the <a href="https://wiki.python.org/moin/GlobalInterpreterLock">Python wiki</a> describes it:</p><blockquote><em>In CPython, the global interpreter lock, or GIL, is a mutex that protects access to Python objects, preventing multiple threads from executing Python bytecodes at once. The GIL prevents race conditions and ensures thread safety.</em></blockquote><p>For projects that rely on parallel processing, this is a real bottleneck. The GIL was intended to be an effective solution for thread safety in the Python interpreter, but it has a major downside. With the GIL, you can never truly write multi-threaded code in Python. To some extent, you can do it using asyncio coroutines, but these workarounds provide a concurrency paradigm that&#39;s a lot more complex than simply using threads. For a language that has the reputation of being &quot;beginner-friendly&quot;, I find it unnecessarily convoluted.</p><p>The GIL is not all bad, of course, as you’d never have to think about thread safety. However, I would rather have to think about it to write an efficient program than write one that’s slow and laggy. When you’re building products that are intended to be scaled to millions of users, an otherwise small performance issue can become catastrophic. This is another reason why I think Python is the right choice for hobbyists, and for hacking together scripts, exclusively.</p><h3>Constrained Indentation and Verbosity</h3><p>Python’s strict indentation and whitespace rules, while promoting clean code, drive me a little crazy. In most cases, ugly indented code makes you want to write code that’s more linear, and sure that’s a good thing about it. Sometimes, however, you can’t avoid it, and that file seems repugnant. Have you ever tried contributing to a Python code-base that doesn’t have a linter set up? It is hell to deal with inconsistent tabs and spaces that sometimes even your editor can’t comprehend. The point I’m trying to make is that cleanliness lies in being able to write readable, concise code. Python seems to have cracked the readability aspect of it, but it’s nowhere close to concise.</p><p>Speaking of verbosity, I’m tired of importing modules with absolute paths starting from the root of the project. I haven’t seen any editors intellisense auto-import Python functions and modules with no errors (I’ve used VSCode and nvim). It should be the developer’s decision whether or not they want to use relative imports with modules, and I’ve tried really hard but I’ve never been able to make it work. In my opinion, Python modules are the worst way of implementing modules that I’ve seen in any language. The <em>necessity</em> of having __init__.py files inside every directory is really annoying. Even if you didn&#39;t have a single line of code in your __init__.py file, you&#39;ll need it for the Python interpreter to recognize it as a module.</p><h3>Global namespace pollution and dynamic typing</h3><p>While I’m not a fan of dynamic typing, I can’t complain much since I develop a lot of things in Javascript (actually, Typescript). Admittedly, Typescript’s type system is not the best, but somehow it feels like a good middle-ground between statically typed and dynamically typed languages. It enforces (somewhat) types wherever you want them to be enforced. Writing Python annotations seems like a waste of time. If you’re not using a static type-checker like <a href="https://mypy.readthedocs.io/en/stable/">mypy</a>, you might as well not write any annotations, as they wouldn’t do anything other than provide some intellisense (which isn’t always accurate either).</p><p>The global namespace pollution, however, is annoying to me. If sorted is an in-built method, I should not be able to overwrite it. Somehow, this is valid Python code:</p><pre>arr = [5, 4, 3, 2, 1]<br>sorted = sorted(arr)</pre><p>The next time I try to run the sorted method, I&#39;m going to run into this error:</p><pre>Traceback (most recent call last):<br>  File &quot;&lt;stdin&gt;&quot;, line 1, in &lt;module&gt;<br>TypeError: &#39;list&#39; object is not callable</pre><p>I’m sure you can see how import * may pollute the namespace. It&#39;s not uncommon to see from &lt;module&gt; import * in Python. When you import a module, all the variables and functions defined in the module will get imported, if you didn&#39;t specify any using __all__.</p><h3>Dependency management</h3><p>It is painful to set up clean and isolated package management in Python. Most people just install all their dependencies globally and later run into clashes and nuke their entire Python environment. Some just accept their fate and set up a <a href="https://docs.python.org/3/library/venv.html">virtual environment</a>, but then manage their requirements.txt files manually as pip freeze is a nightmare. It will fix versions of the packages you didn&#39;t even want to install. I&#39;ve faced issues where the requirements.txt file generated on my Linux system couldn&#39;t be installed by pip on my friend&#39;s Mac. Now, you can use something like pipreqs, but what&#39;s the point of having dependency management if you have to install a different library anyway? In fact, most Linux users stick to installing Python packages using their distro&#39;s package manager ( pacman for example), so that they do not have to deal with pip. There is no consistency in package management, and the absence of a lock file might be one of the biggest reasons for this. The absence of native locking mechanisms leaves room for potential discrepancies between development and production environments.</p><p>Developers often turn to third-party tools like pipenv, poetry, or conda to enhance dependency management. While these tools offer valuable features, they can also introduce complexity and may require a learning curve.</p><p>Python’s dependency resolver is not as robust as some other package managers. Resolving complex version constraints and ensuring compatibility between packages can be a daunting task, especially in projects with numerous dependencies; imagine having to do that on your own.</p><p>The PyPy team did come up with <a href="https://peps.python.org/pep-0582/">PEP-582</a>, which seems to be a better solution at first glance. If you think about it though, it moves a central venv to an unmanaged set of scattered __pypackages__ directories, which is even harder to replicate. The implementation of PEP-582 in PDM is also not consistent (read more about it in <a href="https://pradyunsg.me/blog/2023/01/21/pdm-does-not-implement-pep-582/">this article</a> my friend wrote.)</p><h3>Conclusion</h3><p>With its rich community and extensive libraries, Python certainly has its merits. That makes it popular among data scientists and machine-learning/AI enthusiasts where you would most definitely find a pre-trained model or a library that already implements the most efficient algorithm. However, its drawbacks, including performance limitations, the GIL, and the inflexibility of its typing, make it a bit of a thorny companion. An engineer takes pride in writing beautiful and efficient code, and most consider Python to be ugly and slow. Of course, no language is perfect (although some are very close to being perfect), most have their nuances.</p><p>Although Python is undeniably robust and powerful, I would prefer not to use it for big projects in the future. You won’t use most of Python’s “power” in your everyday. It’s like bringing a machete to cut your hair.</p><p>I’ve written a fair amount of Python, and I’ve come to dislike it only recently when I had to build and maintain a large project in Python. Python will still be my go-to language for scripting, and for building small weekend projects.</p><p>If you’ve reached this far, you probably share similar sentiments and frustration. If yes, you might want to check <a href="https://github.com/yugr/python-hate">this</a> out. If you’d like to dive deeper into this conversation or have thoughts to share, please feel free to reach out on <a href="https://x.com/roerohan">Twitter (now X)</a>.</p><p><em>Originally published at </em><a href="https://roerohan.github.io/blog/why-i-hate-python/"><em>https://roerohan.github.io</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6e6d038572a9" width="1" height="1" alt=""><hr><p><a href="https://levelup.gitconnected.com/why-i-hate-python-6e6d038572a9">Why I hate Python</a> was originally published in <a href="https://levelup.gitconnected.com/">Level Up Coding</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why I started blogging, and why you should too]]></title>
            <link>https://medium.com/swlh/why-i-started-blogging-and-why-you-should-too-95d48e7e7799?source=rss-5b3814274acd------2</link>
            <guid isPermaLink="false">https://medium.com/p/95d48e7e7799</guid>
            <category><![CDATA[rant]]></category>
            <category><![CDATA[blog]]></category>
            <category><![CDATA[blogging]]></category>
            <dc:creator><![CDATA[Rohan Mukherjee]]></dc:creator>
            <pubDate>Tue, 03 Oct 2023 21:21:16 GMT</pubDate>
            <atom:updated>2023-10-03T21:21:16.436Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*_txaGX5QerMJVWzK" /><figcaption>Photo by <a href="https://unsplash.com/@sincerelymedia?utm_source=medium&amp;utm_medium=referral">Sincerely Media</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>“What the hell should I write about?”<br>“Why would anyone read my blog?”<br>“Do I even have anything to say that’s original, exciting, or different?”<br>“I’ve never been a good writer, should I bother experimenting with this?”<br>“What if my article sucks, or my grammar is bad?”<br>“I’m not an expert at anything, why would anyone care about what I have to say?”</p><p>These are a few “stupid questions” that came across my mind when I first started thinking about starting my blog; possibly you did too if you ever wondered about putting yourself out there. I, for one, have always imagined that since I’m not really a pro at something, people would probably make fun of me for sharing my thoughts and opinions about it. It does sound irrational when I write it down, but that’s what has stopped me from starting a personal blog.</p><p>In the past, I have written some articles that have had a decent number of views (by my standards) — like the following I had written back in 2020:</p><ul><li><a href="https://medium.com/p/626133572429">Make your own YouTube Downloader</a></li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/630/0*GnXznKguZ6fRMUO7.png" /></figure><ul><li><a href="https://medium.com/p/9351fd47b09a">Structuring your Repository for CTF challenges</a></li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/630/0*QHlP7Vc5kZlHQtLS.png" /></figure><p>Despite the relative popularity of these articles, I never felt the motivation to start my blog, or rather never got rid of the fear of sharing my thoughts and opinions about a lot of things. So, what is it that pushed me to start this personal blog?</p><h3>Documenting my work</h3><p>Every software engineer understands the importance of documentation — whether you’re trying to figure out how to do something, or if you want to find out what the hell you worked on last month. I’ve been a developer for over 4 years now, and I’ve explored a myriad of different technologies and tools. However, when I look back, the picture is blurry. This blog is my way of documenting what I learned in the technical domain so that I have something to look back to when I try to figure out what I did last quarter. Just like how you take pictures of your trips as fond memories, I plan to capture some moments in my journey — so that I can look back to what my thoughts and opinions were in the past and how they have (or have not) changed in the present.</p><p>But who the hell cares about my journey?<br>I do. Nobody cares about the pictures you post on Instagram or the Tweets you share on Twitter (are we really calling it X now?), but does that ever stop you from showcasing ugly pictures of your feet on a beautiful beach destroying a gorgeous sunset? We all think that other people are looking at us and judging us, but actually, everyone is worried about their own life.</p><p>Anyway, let’s assume for a second that you are the center of the universe and people have nothing better to do than judge you. If your friend were learning to design posters and they wanted to share their journey in a blog, would you judge them for that? You probably wouldn’t. In fact, you might appreciate their work and share some constructive criticism so that they can improve. When you put yourself in their shoes, it seems kind of ridiculous, doesn’t it?</p><h3>Embracing being a beginner</h3><p>I’m <strong>not</strong> an expert at anything. I’m not good at blogging either, yet you happen to be reading this right now. While expert opinion is something worth listening to (in most cases), often people can learn a lot from a fellow beginner. You might have heard about the <a href="https://en.wikipedia.org/wiki/Curse_of_knowledge">curse of knowledge</a>. Usually, an expert assumes that all their listeners are from the same background and are familiar with what they consider “fundamentals”. Sometimes, it is easier to understand something when a beginner explains it to you rather than an expert. I have watched several tech conferences and haven’t understood most things since I wasn’t aware of the jargon they assumed everyone knew of.</p><p>The articles that you read are not always written by experts — it’s often someone who just figured something out and was excited to share it with the world. When I wrote some of my articles on Medium, I was learning things in the process of writing, and it looks like whatever I came up with piqued the interests of several people.</p><h3>Clarifying my own thinking</h3><p>When you write something down, it helps you clarify your own thinking. Also, I believe that teaching is the best way to learn something. That’s pretty self-explanatory.</p><h3>Communicating with more people</h3><p>“Communication is key” is something you might’ve heard at work or in your relationship. I’ve never been an excellent communicator — and blogging is one of the ways for me to improve. Also, the internet is amazing; you can have a conversation with someone from a different corner of the world (I’m not a flat Earther, it’s just a phrase) about something both of you are passionate about. It’s a humbling experience when strangers pay attention to your work and reach out to you with their questions or feedback.</p><p>Statistically, it is impossible that you’re the only person in the world who is interested in something. So, no matter how niche the subject of your article is, you are certainly conveying your thoughts to someone. The network you build in this process might turn out to be invaluable.</p><h3>Experimenting and taking up new challenges</h3><p>Finally, I decided to write a blog because I wanted to try it out. Since I’m not a good writer, writing articles doesn’t come naturally to me — and I’m always (sometimes) up for a challenge. If you’ve ever written an article, you’d appreciate the effort that goes into doing the research, finding the right images (or memes), and structuring the content. It’s not easy to have a personal blog, and it might not be particularly exciting for me either — it’s a matter of discipline. I sure love the feeling of having written an article, but I don’t like writing it; if you know what I mean.</p><p>It’s an experiment that has no unfavorable outcome — if I enjoy this, well and good, but if I don’t, I move on to something else because nobody gives a f*ck about this blog, nor does my life depend on it.</p><h3>Conclusion</h3><p>I may not have convinced you to start your blog, but sike, I never wanted to. I discussed why starting a blog might be a great idea, but it’s not a piece of cake. Writing is a hard, unpaid (for most people), time-consuming, and lonely pursuit. It’s a marathon, not a sprint. Before you start, you should be honest with yourself about whether the effort you put in is worth it for you.</p><p>A lot of my thoughts were inspired by <a href="https://www.youtube.com/watch?v=vyVpRiqOvt4">Ali Abdaal’s YouTube video</a> about writing online, make sure to check it out! If you find value in this article, feel free to share it with others who might. You can reach out to me on my social handles listed below if you want to share some feedback, or just want to chat.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/519/0*d4DXEloPiMPWSbkk.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=95d48e7e7799" width="1" height="1" alt=""><hr><p><a href="https://medium.com/swlh/why-i-started-blogging-and-why-you-should-too-95d48e7e7799">Why I started blogging, and why you should too</a> was originally published in <a href="https://medium.com/swlh">The Startup</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Migrate From Twilio Video Using Dyte’s Twilio Shim]]></title>
            <link>https://medium.com/dyte/migrate-from-twilio-video-using-dytes-twilio-shim-31e54989692f?source=rss-5b3814274acd------2</link>
            <guid isPermaLink="false">https://medium.com/p/31e54989692f</guid>
            <category><![CDATA[dyte]]></category>
            <category><![CDATA[javascript]]></category>
            <category><![CDATA[infrastructure]]></category>
            <category><![CDATA[webrtc]]></category>
            <category><![CDATA[twilio]]></category>
            <dc:creator><![CDATA[Rohan Mukherjee]]></dc:creator>
            <pubDate>Tue, 08 Aug 2023 12:36:51 GMT</pubDate>
            <atom:updated>2023-10-13T12:42:15.920Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*LMzUHh4MeFY5k058H3DCBw.png" /></figure><p>Our business has been fortunate enough to have garnered the interest of several customers. These customers have expressed their satisfaction with our products and services, which we take great pride in providing.</p><p>Dyte has been gaining popularity among customers who require reliable cloud communication services. Many of these customers, who were initially using Twilio Video, have been making the switch to Dyte, possibly owing to our call quality, reliability, expeditious customer support, and pricing.</p><p>This guide presents an approach to help users smoothly migrate from Twilio Video to Dyte with the help of Shims.</p><h3>Thinking of migrating from Twilio Video?</h3><p>Since you are reading this page, I understand that you are probably considering a switch from Twilio Video to another video provider. In recent times, a lot of our customers migrated from Twilio Video to Dyte. They made this choice due to Dyte’s compelling advantages, which include:</p><ul><li>Easy to use APIs with high-level abstractions and inbuilt access control — fewer bugs and less maintenance required</li><li>UI components library — responsive components for React, Angular, WebComponents, React Native, Flutter, Native Android, and iOS. Build real-time applications faster.</li><li>Better media performance — with Dyte’s global mesh network of media servers, calls spanning multiple geographies see a significantly better experience.</li><li>Better support — dedicated Slack support, code reviews, and technical assistance 24/7.</li></ul><p>You can view a comprehensive feature <a href="https://dyte.io/twilio-video-competitor">comparison between Dyte and Twilio Video</a>.</p><p>Once you’ve reviewed the reasons behind migrating from Twilio Video, it becomes evident that Dyte will likely meet all your requirements!</p><p>However, transitioning from one SDK to another can be a demanding process, considering the time and effort involved. Will migrating from Twilio Video to Dyte entail a similar level of effort? The answer is — No.</p><p>Enter Dyte’s Twilio Shim, which simplifies the migration.</p><h3>Introducing our Shim library</h3><p>Even though Dyte can be easily integrated (don’t believe me? Check out our <a href="https://docs.dyte.io/">docs</a>) into your codebase, some developers may find the process cumbersome and time-consuming as they have to get rid of their old code and refactor it to fit a new SDK.</p><p>To make the lives of these developers easier, <strong>we created a shim library</strong> — <a href="https://www.npmjs.com/package/@dytesdk/twilio-shim">@dytesdk/twilio-shim</a>.</p><p>You may be wondering, what is a shim? According to <a href="https://en.wikipedia.org/wiki/Shim_(computing)">Wikipedia</a>:</p><blockquote><em>In computer programming, a shim is a library that transparently intercepts API calls and changes the arguments passed, handles the operation itself or redirects the operation elsewhere. Shims can be used to support an old API in a newer environment, or a new API in an older environment.</em></blockquote><p>Basically, we created a wrapper around Dyte’s SDK with a Twilio Video-compatible external API as a drop-in <a href="https://dyte.io/twilio-video-competitor">replacement for the Twilio Video SDK</a>. You can use the same classes, methods, and other APIs that you used with Twilio Video, but behind the scenes, the shim will perform some magic and translate it to Dyte API calls.</p><p>For instance, if you were to connect to a room using Twilio Video and login to the console whenever a new participant joined the room, this is what your code would look like:</p><pre>const { connect } = require(&#39;twilio-video&#39;);<br><br>const authToken = &#39;&lt;AUTH_TOKEN_FROM_TWILIO_API&gt;&#39;;<br><br>const room = await connect(authToken, {audio: false, video: true});<br>console.log(`Successfully joined a Room: ${room.name}`);<br>console.log(room);<br>room.on(&#39;participantConnected&#39;, participant =&gt; {<br>    console.log(`A remote Participant connected: ${participant}`);<br>});</pre><p>The following code snippet demonstrates how you would do it with the shim:</p><pre>const { connect } = require(&#39;@dytesdk/twilio-shim&#39;);<br><br>const authToken = &#39;&lt;AUTH_TOKEN_FROM_DYTE_API&gt;&#39;;<br><br>const room = await connect(authToken, {audio: false, video: true});<br>console.log(`Successfully joined a Room: ${room.name}`);<br>console.log(room);<br>room.on(&#39;participantConnected&#39;, participant =&gt; {<br>    console.log(`A remote Participant connected: ${participant}`);<br>});</pre><p>Can you spot the difference?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/481/0*y7NukOy5da3CV6WO.png" /></figure><p>As mentioned previously, the external API is almost identical to Twilio Video. However, intended actions, such as joining a room, are performed internally using Dyte’s <a href="https://docs.dyte.io/web-core">web-core</a> SDK.</p><p>You can check out more examples of using our Twilio shim <a href="https://www.npmjs.com/package/@dytesdk/twilio-shim">here</a>.</p><p>Here is the complete API compatibility list of methods and classes we currently translate</p><p><a href="https://docs.dyte.io/guides/migration/twilio/compatibility-shim-twilio">https://docs.dyte.io/guides/migration/twilio/compatibility-shim-twilio</a></p><p>Also, another important note is that this library should just be used to test out the integration, and clients should plan to eventually move to our SDK natively to utilize Dyte’s capabilities fully</p><h3>That’s all, folks!</h3><p>That was a quick and easy way to switch from Twilio Video — just like integrating Dyte into your project. We are working on creating migration guides for other commonly used real-time communication APIs, such as live streaming and AI-powered chat, among others. If you would like us to create a migration guide from your current video provider, you can contact us on our <a href="https://community.dyte.io/">community Discord</a>.</p><p><em>If you haven’t heard about Dyte yet, head over to </em><a href="https://dyte.io/"><em>dyte.io</em></a><em> to learn how we are revolutionizing communication through our </em><a href="https://dyte.io/video-sdk"><em>SDKs</em></a><em> and libraries and how you can </em><a href="https://accounts.dyte.in/auth/register"><em>get started</em></a><em> quickly on your 10,000 free minutes, which renew every month. You can reach us at </em><a href="/service/mailto:support@dyte.io"><em>support@dyte.io</em></a><em> or ask our </em><a href="https://community.dyte.io/"><em>developer community</em></a><em> if you have any questions.</em></p><p><em>Originally published at </em><a href="https://dyte.io/blog/migrate-twilio-video-dyte-shim/"><em>https://dyte.io</em></a><em> on August 8, 2023.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=31e54989692f" width="1" height="1" alt=""><hr><p><a href="https://medium.com/dyte/migrate-from-twilio-video-using-dytes-twilio-shim-31e54989692f">Migrate From Twilio Video Using Dyte’s Twilio Shim</a> was originally published in <a href="https://medium.com/dyte">Dyte</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>