<?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 Nainesh D on Medium]]></title>
        <description><![CDATA[Stories by Nainesh D on Medium]]></description>
        <link>https://medium.com/@nainesh.dd03?source=rss-12f72a0bdc01------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*M1jN1IpzziKk20DA</url>
            <title>Stories by Nainesh D on Medium</title>
            <link>https://medium.com/@nainesh.dd03?source=rss-12f72a0bdc01------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 09:22:45 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@nainesh.dd03/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[Want to Learn Docker? I Wish Someone Had Told Me This Years Ago]]></title>
            <link>https://medium.com/@nainesh.dd03/want-to-learn-docker-i-wish-someone-had-told-me-this-years-ago-427ef995a43d?source=rss-12f72a0bdc01------2</link>
            <guid isPermaLink="false">https://medium.com/p/427ef995a43d</guid>
            <category><![CDATA[docker]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[docker-swarm]]></category>
            <category><![CDATA[docker-image]]></category>
            <category><![CDATA[dockerfiles]]></category>
            <dc:creator><![CDATA[Nainesh D]]></dc:creator>
            <pubDate>Thu, 05 Nov 2026 20:16:22 GMT</pubDate>
            <atom:updated>2026-11-05T20:16:22.084Z</atom:updated>
            <content:encoded><![CDATA[<blockquote><em>A Docker for beginners guide built on the official docs, with a small project to make it stick</em></blockquote><blockquote>I wish someone had told me this before I spent years using Docker without understanding it.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qpKItHReadvBV6YP4aJrxg.png" /></figure><p>I had built many projects with Dockerfiles, and they all ran, yet I could not explain how a container reached my browser or what really happens when one stops. So I built a small tool to stop typing the same commands by hand, and it taught me the fundamentals I had skipped. This post is the guide I wish I had on day one. If Docker has ever felt too big or too technical to start, it is for you. You will learn what Docker is, why it exists, what containers and images are, the basic Docker commands, how Docker networking works, and the easiest way to keep going.</p><h3>Why learn Docker from the official documentation?</h3><p>One piece of advice before we touch a single command. If you want to learn Docker, or anything new in tech, go to the official documentation first. An AI agent gives you the exact answer you asked for. Reading the documentation, or browsing around manually, gives you more. While you search for one thing, you run into related ideas, options, and warnings you did not know to ask about. That hidden knowledge is what turns an answer into understanding.</p><p>Docker’s <a href="https://docs.docker.com/get-started">get started section</a> is written for people who are new, and this post is built mostly on the official docs and links back to them, and where the docs do not cover something, I explain it in plain words. Keep it open in a tab as you read.</p><h3>What is Docker and what problem does it solve?</h3><p>Start with the problem, because it explains everything else. An app works on your laptop and breaks on a colleague’s machine or on a server. The code is the same. The difference is everything around it, such as the operating system, the language version, and the installed libraries.</p><p>Before containers, a common answer was the virtual machine. The <a href="https://docs.docker.com/get-started/docker-concepts/the-basics/what-is-a-container/">docs describe a VM</a> as an entire operating system with its own kernel, hardware drivers, programs, and applications. It works, but every VM carries all of that.</p><p>Docker’s answer is the container. The same docs call a container an isolated process with all the files it needs to run, so it does not rely on anything pre-installed on the host. Share a container, and everyone gets the same thing working the same way, and the one on your laptop works the same way in a data center or in the cloud. Because containers share one kernel, you can run more applications on less infrastructure.</p><p>In one line, Docker lets you build an app once and run it the same way anywhere. The docs describe it as <a href="https://docs.docker.com/get-started/docker-overview/">an open platform for developing, shipping</a><a href="https://docs.docker.com/get-started/docker-overview/">, and running applications</a> that separates your applications from your infrastructure, so you can deliver software quickly.</p><p>Now, the short history, as told in <a href="https://www.oreilly.com/library/view/docker-up/9781492036722/ch01.html">Docker Up and Running</a> from O’Reilly.</p><ul><li><strong>March 15, 2013.</strong> Solomon Hykes, founder and CEO of dotCloud, introduces Docker in a five-minute lightning talk at PyCon in Santa Clara. Only about 40 people outside dotCloud had tried it so far.</li><li><strong>Soon after.</strong> The project is open-sourced and published on GitHub for anyone to use and contribute to.</li><li><strong>Within a year.</strong> Almost everyone in the industry has heard of Docker, though many are still working out what it is and why people are excited.</li><li><strong>Today.</strong> Docker Desktop is available for Mac, Windows and Linux, as the <a href="https://docs.docker.com/get-started/docker-overview/">Docker overview</a> notes.</li></ul><h3>What is a Docker container? Images, containers and registries explained</h3><p>Docker is an open platform for developing, shipping, and running applications, and it packages an app and everything it needs into a unit called a container. Docker comes with a lot of vocabulary, but only a few ideas matter at the start, and here they are in one place.</p><ul><li><strong>Image.</strong> A read-only template with the instructions and files needed to create a container. Read more in <a href="https://docs.docker.com/get-started/docker-concepts/the-basics/what-is-an-image/">What is an image</a>.</li><li><strong>Container.</strong> A runnable instance of an image, running as an isolated process on your machine. Read more in <a href="https://docs.docker.com/get-started/docker-concepts/the-basics/what-is-a-container/">What is a container</a>.</li><li><strong>Registry.</strong> A place that stores images, and Docker Hub is the default public one. Read more in <a href="https://docs.docker.com/get-started/docker-concepts/the-basics/what-is-a-registry/">What is a registry</a>.</li><li><strong>Docker client.</strong> The docker command you type in the terminal, explained in the <a href="https://docs.docker.com/get-started/docker-overview/">Docker overview</a>.</li><li><strong>Docker daemon.</strong> The background service, called dockerd, that does the real work, also covered in the <a href="https://docs.docker.com/get-started/docker-overview/">Docker overview</a>.</li><li><strong>Docker Desktop.</strong> An app that bundles the daemon, the client, Docker Compose and more, described in the <a href="https://docs.docker.com/get-started/docker-overview/">Docker overview</a>.</li></ul><p>Start with the container, since it is the heart of everything. The docs describe it as a runnable instance of an image that is relatively well isolated from other containers and from your machine. Under the hood, Docker uses a Linux kernel feature called namespaces to create that isolated workspace. You can create, start, stop, move, or delete a container with the Docker CLI or API. Remember those verbs, because they come back later.</p><p>Here is how the pieces work together. When you type docker run nginx, the client sends your request to the daemon. If the image is not on your machine, the daemon pulls it from a registry. Then it creates the container and starts it. One command, and every term in the list above has played a part.</p><h3>Basic Docker commands for beginners</h3><p>The best way to understand a container is to run one. Install Docker Desktop with the <a href="https://docs.docker.com/get-started/get-docker/">official install guide</a>, open a terminal, and try these in order.</p><pre>docker run -d --name my-nginx -p 8080:80 nginx<br>docker ps<br>docker stop my-nginx<br>docker ps -a<br>docker start my-nginx<br>docker stop my-nginx<br>docker rm my-nginx<br>docker images</pre><p>Here is what each one teaches you.</p><ul><li>docker run pulls the nginx image if you do not have it, creates a container and starts it, as the <a href="https://docs.docker.com/get-started/docker-overview/">Docker overview</a> describes. The -d flag runs it in the background and --name gives it a name you can use later.</li><li>docker ps lists the running containers.</li><li>docker stop stops the container without deleting it, which is why docker ps -a still shows it.</li><li>docker start brings the same container back.</li><li>docker rm removes a stopped container for good.</li><li>docker images lists the images on your machine. The nginx image is still there after the container is gone.</li></ul><p>That is the whole life of a container. The docs list the same verbs, since you can <a href="https://docs.docker.com/get-started/docker-overview/">create, start, stop, move or delete a container</a> with the Docker CLI or API. One catch is worth knowing early. When a container is removed, any changes to its state that are not stored in persistent storage disappear.</p><h3>Docker networking and port mapping explained</h3><p>A container is isolated. That is its strength, and it is also why networking confuses newcomers. By default a container on Docker’s bridge network does not expose its ports to the outside world, and the daemon blocks access to ports you have not published, as the <a href="https://docs.docker.com/engine/network/port-publishing/">port publishing docs</a> explain.</p><p>So you tell Docker which door to open. That is what publishing a port does.</p><pre>docker run -d -it --name my-nginx -p 8080:80 nginx</pre><p>Read -p 8080:80 as machine port first, container port second. Nginx listens on port 80 inside its isolated space, and Docker forwards port 8080 on your machine to it. Open localhost on port 8080 and you reach the container. Leave out the -p and nginx still runs, but its port is not published, so nothing outside the Docker host can reach it.</p><p>Three more facts from the <a href="https://docs.docker.com/engine/network/">networking overview</a> are worth knowing early.</p><ul><li>The bridge network is the default driver, and Docker lists others such as host, none, and overlay.</li><li>On the default bridge, containers reach each other only by IP address. Once you create your own network, they can use container names too.</li><li>Publishing a port opens it to the outside world by default. Put 127.0.0.1 in front, as in -p 127.0.0.1:8080:80, if only your own machine should reach it.</li></ul><h3>Dockerfile, data persistence and Docker Compose explained</h3><p>Once the basics feel comfortable, three more ideas complete the beginner picture.</p><p><strong>Images and Dockerfiles.</strong> An image is built from a Dockerfile, a simple file of instructions. Each instruction creates a layer, and when you rebuild an image, only the layers that changed are rebuilt, which is part of what keeps images small and fast. The docs cover this in <a href="https://docs.docker.com/get-started/docker-concepts/building-images/understanding-image-layers/">Understanding image layers</a> and <a href="https://docs.docker.com/get-started/docker-concepts/building-images/writing-a-dockerfile/">Writing a Dockerfile</a>.</p><p><strong>Persisting data.</strong> Containers forget. Remove one and its changes disappear unless the data lives in persistent storage, which the docs explain in <a href="https://docs.docker.com/get-started/docker-concepts/running-containers/persisting-container-data/">Persisting container data</a>.</p><p><strong>Docker Compose.</strong> When an app needs several containers, such as a web app and a database, <a href="https://docs.docker.com/get-started/docker-concepts/the-basics/what-is-docker-compose/">Docker Compose</a> lets you run them together. It is a brilliant shortcut, and far more useful once you know what each container is doing.</p><h3>A simple Docker project for beginners</h3><p>So I built <a href="https://github.com/nainesh-builds/docker-container-handler-tool">Docker Container Handler</a>. It is a small web app written in Python and Flask with plain HTML and CSS, and the repository also has a terminal version for people who prefer the command line. It talks to the Docker installation on your own machine, and every option runs one real Docker command.</p><ul><li><strong>View images</strong> runs docker images. It shows that images are templates stored on your machine.</li><li><strong>View containers</strong> runs docker ps -a. It shows that containers have states and that stopped ones still exist.</li><li><strong>Create a container</strong> runs docker run. An image becomes a container, with a published port.</li><li><strong>Start a container</strong> runs docker start. An existing container can be started again.</li><li><strong>Stop a container</strong> runs docker stop. Stopping is not the same as removing.</li><li><strong>Delete a container</strong> runs docker rm. A container must be stopped before it is deleted.</li></ul><p>To connect a button to a command, you have to understand what the command needs. Creating a container meant asking for a container name, a machine port, a container port, and an image. I could not build that form without understanding port publishing. Deleting meant respecting the order: stop first and remove second.</p><p>You do not need to copy mine. Build your own tiny version in any language you like. The tool is simple on purpose, and to try mine you need only Python, Docker Desktop, and Git.</p><h3>How to learn Docker as a beginner, step by step</h3><p>Here is the path I would follow if I were starting today. Nothing in it needs prior experience.</p><ol><li>Install Docker Desktop with the <a href="https://docs.docker.com/get-started/get-docker/">official install guide</a>.</li><li>Read what a <a href="https://docs.docker.com/get-started/docker-concepts/the-basics/what-is-a-container/">container</a> and an <a href="https://docs.docker.com/get-started/docker-concepts/the-basics/what-is-an-image/">image</a> are in the docs. A few minutes each is enough.</li><li>Run your first container by hand. Start nginx, publish a port, open it in your browser, then stop it, start it again, and remove it.</li><li>Break the networking on purpose. Forget the -p flag, see what happens, then read the <a href="https://docs.docker.com/engine/network/port-publishing/">port publishing page</a>.</li><li>Automate those commands in a small script or app, the way my handler does.</li><li>Only now move on to Dockerfiles and Docker Compose, with the official <a href="https://docs.docker.com/get-started/tutorials/run-an-app/">Build and share a containerized application</a> tutorial as your guide. You will know exactly what they are doing for you.</li></ol><p>Two habits make this stick. Type the commands yourself instead of pasting them, and keep the documentation open next to your terminal.</p><h3>What is Gordon, Docker’s AI assistant?</h3><p>Docker now ships an AI assistant of its own called Gordon. According to the <a href="https://docs.docker.com/ai/gordon/">Gordon documentation</a>, it is available in Docker Desktop and through the docker ai command in the terminal. A request can be as simple as this.</p><pre>docker ai &quot;show me logs from my nginx container&quot;</pre><p>A few facts are worth checking in the docs before you try it.</p><ul><li>Gordon is part of Docker Desktop, which the docs describe as available for Mac, Windows, and Linux. The Gordon page does not tie it to one operating system.</li><li>You need Docker Desktop 4.74 or later at the time of writing, plus a signed-in Docker account.</li><li>It is enabled by default for signed-in users, but Business organizations need Docker Support and an administrator to switch it on.</li><li>You can also ask Gordon questions from the icon on any page of docs.docker.com, which is free and needs no account.</li></ul><p>Here is how it connects to everything above. Gordon asks for your approval before it acts. If you have never typed docker stop or docker rm yourself, how do you know whether to approve? Building the handler taught me what each command does, so I can read what Gordon proposes and judge it. Without that, I would be pressing approve and hoping for the best.</p><p>AI assistants are a great accelerator. They are a poor substitute for understanding.</p><h3>Common Docker questions, answered</h3><p><strong>What is Docker?</strong> Docker is an open platform for developing, shipping, and running applications. It lets you package an app with everything it needs into a container and run it the same way anywhere. See the <a href="https://docs.docker.com/get-started/docker-overview/">Docker overview</a>.</p><p><strong>What is a Docker container?</strong> A container is a runnable instance of an image, running as an isolated process on your machine. See <a href="https://docs.docker.com/get-started/docker-concepts/the-basics/what-is-a-container/">What is a container</a>.</p><p><strong>What is a Docker image?</strong> An image is a read-only template with the instructions for creating a container. See <a href="https://docs.docker.com/get-started/docker-concepts/the-basics/what-is-an-image/">What is an image</a>.</p><p><strong>What is the difference between a Docker container and a virtual machine?</strong> A virtual machine is an entire operating system with its own kernel. Containers share the host kernel and run as isolated processes, so you can run more applications on less infrastructure. The comparison is in <a href="https://docs.docker.com/get-started/docker-concepts/the-basics/what-is-a-container/">What is a container</a>.</p><p><strong>What is Docker Hub?</strong> Docker Hub is the default public registry where Docker images are stored and shared. See <a href="https://docs.docker.com/get-started/docker-concepts/the-basics/what-is-a-registry/">What is a registry</a>.</p><p><strong>What does </strong><strong>-p 8080:80 mean in Docker?</strong> It publishes port 80 inside the container to port 8080 on your machine. See <a href="https://docs.docker.com/engine/network/port-publishing/">Port publishing and mapping</a>.</p><p><strong>What is Docker Compose?</strong> Docker Compose lets you run an app made of several containers together. See <a href="https://docs.docker.com/get-started/docker-concepts/the-basics/what-is-docker-compose/">What is Docker Compose</a>.</p><p><strong>What is Docker Desktop?</strong> Docker Desktop is an app that bundles the Docker daemon, the client, Docker Compose, and more, and it is available for Mac, Windows, and Linux. See the <a href="https://docs.docker.com/get-started/docker-overview/">Docker overview</a>.</p><p><strong>What is Gordon in Docker?</strong> Gordon is Docker’s AI assistant, available in Docker Desktop and through the docker ai command. See the <a href="https://docs.docker.com/ai/gordon/">Gordon documentation</a>.</p><p><strong>What is the easiest way to learn Docker?</strong> Start with the official documentation, run containers by hand, then automate the commands in a small project of your own, as described in the step-by-step section above.</p><h3>Start small</h3><p>Docker looks big from the outside. Underneath, it is a handful of ideas, namely an image, a container, a registry, a port, and a network. Learn those from the documentation, build one small thing with your own hands, and the rest will follow.</p><p>The code for my handler is on <a href="https://github.com/nainesh-builds/docker-container-handler-tool">GitHub</a>. Its improvement list still includes container logs, restarting containers, and deleting images, so you are welcome to build a better one.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=427ef995a43d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How Databricks Deletes a Single Row From a 10GB File Without Rewriting It]]></title>
            <link>https://medium.com/@nainesh.dd03/how-databricks-deletes-a-single-row-from-a-10gb-file-without-rewriting-it-9f842a2750d7?source=rss-12f72a0bdc01------2</link>
            <guid isPermaLink="false">https://medium.com/p/9f842a2750d7</guid>
            <category><![CDATA[sql]]></category>
            <category><![CDATA[data-engineering]]></category>
            <category><![CDATA[etl]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[databricks-sql]]></category>
            <dc:creator><![CDATA[Nainesh D]]></dc:creator>
            <pubDate>Sun, 23 Aug 2026 23:10:15 GMT</pubDate>
            <atom:updated>2026-08-23T23:10:15.704Z</atom:updated>
            <content:encoded><![CDATA[<blockquote>A question that stopped me mid-interview, and the answer that changed how I think about Delta Lake</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_y8z85ONjpUjIPqgqL54YA.png" /></figure><p><strong>Quick Summary</strong></p><ul><li>Rewriting a 10GB Parquet file to delete one row used to be the only option, and it’s expensive</li><li>Deletion vectors mark rows as deleted in metadata instead, without touching the file</li><li>Your SELECT queries stay exactly the same. Deleted rows just disappear on their own</li><li>The real cleanup happens later, on your schedule, using a completely different command</li></ul><p>Curious what that looks like in actual SQL, and what happens under the hood the moment you run DELETE then keep reading.</p><h3>The Question</h3><p>“You have a 10GB Parquet file. You only need to delete a handful of records. What’s the most efficient way to do it?”</p><p>That was the exact question. My instinct said rewrite the file, since that’s how deletes have always worked in a lake. That instinct was outdated. The real answer is a <a href="https://docs.databricks.com/aws/en/delta">Delta Lake</a> feature called <a href="https://docs.databricks.com/aws/en/tables/features/deletion-vectors"><strong>deletion vectors</strong></a>, and once you understand it, the old approach looks needlessly expensive.</p><h3>Why Deleting a Single Row Used to Be Expensive</h3><p>Parquet files are immutable once written. There’s no way to edit a row in place. So historically, removing even one record out of millions meant three steps.</p><ol><li>Read the entire file into memory</li><li>Filter out the row you want gone</li><li>Write a brand new file to replace the original</li></ol><p>For a 10GB file, that’s a full read and full rewrite, just to drop a few rows. At scale, across thousands of files, this becomes a genuine performance and cost problem.</p><h3>What a Deletion Vector Actually Is</h3><p>A deletion vector is a small piece of metadata, stored as a compact bitmap, that records which row positions in a Parquet file should be treated as deleted. The Parquet file itself is never touched at delete time. Instead, Delta Lake writes this small vector and updates the transaction log to point to it.</p><p>The result is a soft delete. The row still physically exists in the file, but every query engine that reads deletion vectors will skip it automatically.</p><h3>Turning It On</h3><p>Deletion vectors need to be enabled per table.</p><pre>CREATE TABLE sales_data (<br>  id INT,<br>  amount DOUBLE<br>)<br>TBLPROPERTIES (&#39;delta.enableDeletionVectors&#39; = true);</pre><p>On a table that already exists, run this instead.</p><pre>ALTER TABLE sales_data<br>SET TBLPROPERTIES (&#39;delta.enableDeletionVectors&#39; = true);</pre><p>There’s one thing worth knowing before you flip this on in production. Enabling deletion vectors upgrades the table’s protocol version, and older clients that don’t support deletion vectors can no longer read the table. Check your client compatibility first.</p><h3>Deleting Rows Normally</h3><p>Nothing changes in how you write the delete statement, which is the elegant part.</p><pre>DELETE FROM sales_data<br>WHERE id = 540644;</pre><p>Under the hood, this doesn’t touch the Parquet file. It writes a small deletion vector and commits a new entry to the transaction log recording which row to hide.</p><h3>Does Reading the Data Change Too?</h3><p>No, and that’s the whole point. You run the same query you always would.</p><pre>SELECT * FROM sales_data;</pre><p>Delta Lake resolves the current state of the table at query time, applying the deletion vector automatically before returning results. The deleted row simply never shows up. There’s no flag to set, no filter to add.</p><h3>How Do You Confirm a Deletion Vector Actually Exists?</h3><p>You don’t query it directly with SQL, but you can verify it two ways.</p><p>Start by checking the table’s operation log with <a href="https://docs.databricks.com/aws/en/sql/language-manual/delta-describe-history">DESCRIBE HISTORY</a>.</p><pre>DESCRIBE HISTORY sales_data;</pre><p>You’ll see the DELETE operation logged, without a corresponding full file rewrite.</p><p>Then check the file-level detail with <a href="https://docs.databricks.com/aws/en/tables/operations/table-details">DESCRIBE DETAIL</a>.</p><pre>DESCRIBE DETAIL sales_data;</pre><p>And if you list the table’s underlying storage path directly, you’ll find a separate file sitting next to your Parquet files, with “deletion_vector” in its name and a .bin extension. That&#39;s the actual soft-delete marker.</p><h3>When Does the Data Actually Get Removed?</h3><p>The deleted rows still exist physically in the old Parquet file until you choose to clean up. This is a deliberate, separate step, not automatic. Run <a href="https://docs.databricks.com/aws/en/sql/language-manual/delta-optimize">OPTIMIZE</a> to rewrite affected files.</p><pre>OPTIMIZE sales_data;</pre><p>For a more targeted rewrite that touches only the files containing deletion-vector changes, use <a href="https://docs.databricks.com/aws/en/sql/language-manual/delta-reorg-table">REORG TABLE ... APPLY (PURGE)</a> instead.</p><pre>REORG TABLE sales_data APPLY (PURGE);</pre><p>Follow that with <a href="https://docs.databricks.com/aws/en/sql/language-manual/delta-vacuum">VACUUM</a> once you&#39;re past the retention window to physically remove the old files from storage.</p><pre>VACUUM sales_data;</pre><p>This last part matters if you’re deleting for compliance reasons, like a GDPR right-to-be-forgotten request. A deletion vector alone is not a physical purge. You need the REORG and VACUUM combination to actually erase the data from disk.</p><h3>Why This Is a Bigger Deal Than It Sounds</h3><p>One design choice removes one of the most wasteful patterns in working with large immutable files. Deletes and updates that used to mean rewriting gigabytes of data now cost a fraction of that, and reads stay correct the entire time, with zero change to how you query.</p><p>I didn’t know deletion vectors existed until I got asked about them directly. Now I can’t look at a Delta table the same way again.</p><p><strong>Source</strong> <a href="https://docs.databricks.com/aws/en/tables/features/deletion-vectors">https://docs.databricks.com/aws/en/tables/features/deletion-vectors</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9f842a2750d7" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>