<?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 Luke Barwikowski on Medium]]></title>
        <description><![CDATA[Stories by Luke Barwikowski on Medium]]></description>
        <link>https://medium.com/@lukebarwikowski?source=rss-395d59f878ff------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*7KJSd9vtcnphSELx.</url>
            <title>Stories by Luke Barwikowski on Medium</title>
            <link>https://medium.com/@lukebarwikowski?source=rss-395d59f878ff------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 08:59:10 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@lukebarwikowski/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[WhatNow — A Technical Writeup]]></title>
            <link>https://medium.com/@lukebarwikowski/whatnow-a-technical-writeup-ad771e8e493e?source=rss-395d59f878ff------2</link>
            <guid isPermaLink="false">https://medium.com/p/ad771e8e493e</guid>
            <category><![CDATA[full-stack-development]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[python]]></category>
            <category><![CDATA[microservices]]></category>
            <category><![CDATA[redis]]></category>
            <dc:creator><![CDATA[Luke Barwikowski]]></dc:creator>
            <pubDate>Sun, 27 Oct 2019 22:37:18 GMT</pubDate>
            <atom:updated>2019-10-27T22:37:18.489Z</atom:updated>
            <content:encoded><![CDATA[<h3>WhatNow — A Technical Writeup</h3><p><a href="http://whatnow.co/">WhatNow</a> is a peer support platform for recently injured athletes. It sets out to provide a space where people who are struggling with their mental health can find someone to talk to who might understand what they are going through. It does so by creating two different signups, one for people who need to talk to someone, and one for people who are willing to offer support for their peers and then matching people together from those lists.</p><h3>How WhatNow Works</h3><p>I wanted WhatNow to have as small of a barrier of entry as possible. Asking someone to continually engage with an app, or even download an app is a large commitment. I decided that the signup form on the website would be the only interaction with my website, and the rest of the interactions would all be through SMS to make the barrier to get help as low as possible.</p><p>To interact with the platform you must first go to our website:<a href="http://www.whatnow.co/"> www.whatnow.co</a>. From there you’ll be greeted with our mission statement, and then you’ll be asked if you are looking for someone to talk to or if you are willing to offer support. You can see more about the process on<a href="http://www.whatnow.co/talk"> www.whatnow.co/talk</a> or<a href="http://www.whatnow.co/volunteer"> www.whatnow.co/volunteer</a>. Basically for those who are injured, they sign up and confirm that they are willing to give their number for someone and wait for someone to text them. For those who volunteer, they are asked if they wanted to be in the rotation when someone new signs up for the platform we ask the first volunteer in line if they are available and if they are we send them the contact information of the injured person for them to talk to.</p><h3>Well That Sounds Straightforward</h3><p>The concept sounds conceptually easy at first, however, this became a much more interesting and challenging problem to solve than I originally had thought. This is due to a host of issues. Some of the problems were associated with text messaging (I essentially wrote a Twilio Bot), and some of the problems were associated with the actual product (it turns out there were a lot of edge cases that needed to be considered).</p><p>· When our platform receives a text message from someone how do you know what to send back?</p><p>· You need to give the volunteer the option to accept a match</p><p>· What if a volunteer doesn’t reply at all?</p><p>· We should let volunteers match with someone again after their first match, but we should give them a break.</p><p>· How can we make this all scale?</p><h3>The Architecture</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/720/0*c3WtkqNeganSAMA2" /></figure><p>When I design architecture I always design it to be as decoupled as possible. If a section of code doesn’t need to be on the same server, I will always separate them and make them separate services. I decided to go with a task-based queue architecture for WhatNow, using Redis to store my queues. All of these are in production right now running on the same server, however, I could very easily split up each of these services into their own services and run multiple copies of them at the same time.</p><h3>The Frontend</h3><p>I chose Wordpress as my frontend. You can read more about the decision as to why in my article<a href="https://medium.com/@lukebarwikowski/why-web-developers-should-still-use-wordpress-95736ed82b5"> Why Full-Stack Developers Should Use Wordpress</a>. In summary, Wordpress has a reputation for being bloated but the advantages it offers (being able to rapidly update designs, publish content, SEO tricks) outweigh those disadvantages in my opinion. And with good caching strategies, you can mitigate the bloat and deliver a pretty quick site. This was especially true in this use case where the only interaction with my backend was two forms.</p><p>If you are changing how your website looks frequently, it is best to use a tool that allows you to easily change how your website looks. The other added advantage was that I wanted to do a lot of content marketing by writing blog posts, so having a CMS was key</p><h3>The Backend</h3><p>The backend of the application is split into several services: the REST API, an SMS Service, a Coordinator Service, and a bunch of Task Queues. The glue behind everything is Redis. That is where the majority of the data is being saved, and how each service speaks to each other.</p><h3>The REST API</h3><p>The REST API is written in NodeJS and ExpressJS. It hosts the Webhook that Twilio talks to us through, and also 1 route at /api/u/ where our frontend can get/create new users.</p><h3>The SMS Service</h3><p>This was more challenging than I thought it would be initially. I explain it more in-depth in this article I wrote: <a href="https://medium.com/@lukebarwikowski/how-to-create-a-simple-twilio-conversation-bot-6e22d87bbfc2">How to Create a Simple Twilio Conversation Bot</a>. However, in summary, the SMS Service is a task-based queue running on Redis and Python. We have 2 types of tasks our queue can receive, we can ask our service to 1) send a message, or 2) receive and process a message. In order to do that we have to keep track of things like what we sent that phone number we are receiving a text message from last, we have to think hard about the way we store our messages, and we have to set up webhooks. I highly recommend reading the linked article, it goes over a lot of good stuff.</p><h3>The Coordinator and Queues</h3><p>The last piece of the puzzle is the coordinator service. This is the service that initializes and manages all of our task-based queues. In production right now I only have one instance of this coordinator service running, and it runs every type of task-based I wrote on this one instance. However, I have wrote the service so that we can have multiple instances of this service running concurrently, and we can control which of our queues run on each instance of the service. This means we could have multiple coordinators each running one of our queues. This could scale massively without any code-rewrites. Decisions like this are why it is so important to write code designed to scale from the beginning.</p><p>Now for our queues. These are the queues we have running right now and they are all associated with a different event that is created in our system. The names of them should be self-explanatory.</p><ul><li>createInjuredPerson</li><li>confirmInjuredPerson</li><li>createVolunteer</li><li>confirmVolunteer</li><li>createMatch</li><li>pendingMatchExpires</li><li>matchCooldown</li><li>removeVolunteer</li></ul><p>Each of these queues is based on either a what I call a ‘TaskQueue’, a ‘PriorityQueue’, or an ‘ExpirationQueue’.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/5bba2d2ecf986a1eac2e5c1b02ea14e0/href">https://medium.com/media/5bba2d2ecf986a1eac2e5c1b02ea14e0/href</a></iframe><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/93f15a6b08671b4c01c2d2b364d9d925/href">https://medium.com/media/93f15a6b08671b4c01c2d2b364d9d925/href</a></iframe><p>Though, most of these queues also extend those queues in some way to add their own rules as they operate slightly differently. Each instance of a Queue also needs to override the respond() function.</p><p>Here is the implementation of the createVolunteer Queue, it’s shockingly simple after all of this setup.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/5f2e801e751685d1514d96e4a7a88a48/href">https://medium.com/media/5f2e801e751685d1514d96e4a7a88a48/href</a></iframe><p>This is the sign of well-written code. In the file I call tasks.py that defines all my tasks, you only see the functional rules of the product. Someone who doesn’t code could look at this file and tell you what each task does.</p><h3>Pending Matches</h3><p>It is also important to note that I had to introduce the concept of pending matches. How the product works is that anytime someone signs up (or someone is introduced back into the volunteer rotation, or a pending match expires) we check and see if we can create a match. To do that we send the volunteer a text message and ask them if they are available for a specific person. Before the volunteer says yes though, they are in a state where there is not a match, but we still need to wait and give them time to respond. They are in a pending match. I created the idea of Expiration Queues for this purpose. We poll the expiration queues seeing if there is a task ready for us just like the other queues, except this type of task means that there is an item in the queue that has been in the queue longer than the allowed time we specified.</p><h3>Conclusion</h3><p>This was a much larger and difficult project than I originally had imagined. This project was the perfect use case for using Redis Queues though, so it was great practice for that. If you stumble upon this article and want a closer look at the code, let me know!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ad771e8e493e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to Create a Simple Twilio Conversation Bot]]></title>
            <link>https://medium.com/@lukebarwikowski/how-to-create-a-simple-twilio-conversation-bot-6e22d87bbfc2?source=rss-395d59f878ff------2</link>
            <guid isPermaLink="false">https://medium.com/p/6e22d87bbfc2</guid>
            <category><![CDATA[python]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[twilio]]></category>
            <category><![CDATA[microservices]]></category>
            <dc:creator><![CDATA[Luke Barwikowski]]></dc:creator>
            <pubDate>Sun, 27 Oct 2019 20:03:19 GMT</pubDate>
            <atom:updated>2019-10-27T23:52:26.028Z</atom:updated>
            <content:encoded><![CDATA[<p><a href="http://www.whatnow.co/">WhatNow</a> interacts with users through text messaging. I had to write an entire Twilio bot that could guide users through signup and then also ask them questions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/750/0*6PNpGXEBV9zQ-ob1" /></figure><p>This was more challenging than I thought it would be initially. Not only is this slightly challenging technically, but there are also many functional requirements that you may not think of when developing this on your own. Planning this well is key.</p><h3>The Black-Box</h3><p>When writing services, it is important to think of them as a black box. Our service takes in two types of tasks, 1) sendMessage and 2) recieveMessage. Then sometimes it sends data to another one of our services. When you are designing a whole system it is crucial to look at each piece just like this and to agree upon a language in which all of your services will talk to each other.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/720/0*UbxLgeTlBcK0ipXz" /></figure><p>I created 2 Redis Queues for this service, ‘ReceivedQueue’ and ‘SendQueue’. Pushing to either one of these queues is how other services will interact with our service. They are the inputs to our BlackBox.</p><h3>Message Structure</h3><p>To be able to send and receive text messages to a user, and have a coherent conversation or flow of actions we need to ask ourselves these questions:</p><p>· If we receive a message from the user is that because we asked them a question or asked for a response?</p><p>· What did we ask them?</p><p>· Does that response make sense?</p><p>· Does that trigger any action?</p><p>· Do we send them anything next?</p><p>The way we store the actual messages matters. We do not want to hard code messages. Instead, we want to separate them in their own file or perhaps later in a database, I have put them in their own file messages.py. We can smartly store our messages in a way that all of the above questions are answered for each message. This is infinitely more maintainable Here is an example of the way I stored my messages.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/7b522c5bf0dbda514e47b5b6fbeb418c/href">https://medium.com/media/7b522c5bf0dbda514e47b5b6fbeb418c/href</a></iframe><p>We store all the messages in on big dictionary, where the message’s name is its key. “GreetVolunteer”, “default”, and “confirmVolunteer” are the names of three messages I have shown as examples. “body” is the actual text that gets sent to the user. When we send a message we tell our service send(number, MESSAGES[“GreetVolunteer”]), and it will send the body of this message to the user.</p><p>When we send a message to a user, we normally send them a message because we are expecting a reply. We can create trees of possible flows by through this array “options”. Each option is a possible way that a user could answer the text we send them and the according to action and response that we will respond with. An option consists of “responses”, all valid responses a user can send us that will trigger the action. A “reply”, the name of the next message we will send them, and a “publishResult”, this will trigger an action and send data to our other services. There is also a “default” option that will send a message if the user doesn’t reply in a way we expect.</p><h3>Twilio Integration</h3><p>Full Twilio integration is made of 2 parts. We need to be able to both send and receive text messages.</p><h4>Sending Messages</h4><p>Sending messages through Twilio is easy in Python. It is as easy as importing the library. However, we want to send messages from our store of messages. We create a function that takes in a dict with the keys “number” and either “message” or “message_id” so that we can send a message from our message store, or send any message we want.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/7e8b37197f315f961f3f287f1e342e46/href">https://medium.com/media/7e8b37197f315f961f3f287f1e342e46/href</a></iframe><h4>Setting up a Webhook</h4><p>Twilio requires you to create a webhook to receive messages. Webhooks sound intimidating at first, but they’re not complicated at all if you already understand how REST API’s work. A webhook is essentially an API route you create that a third-party can interact with. In my case, I already had a REST API running on NodeJS and ExpressJS, so I simply added one more route onto that. But if you can use whatever you like. You need to do some setup for this to work, check out the <a href="https://www.twilio.com/docs/usage/webhooks/sms-webhooks">Twilio SMS Webhook Documentation</a>.</p><p>I added a new route to my REST API /sms/</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/55d1a430aa4198c48b912e3dfc0f46de/href">https://medium.com/media/55d1a430aa4198c48b912e3dfc0f46de/href</a></iframe><p>And here is that Sms.recieve function…</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/b15c0e749e7db34dbe50ddaba0fe4c53/href">https://medium.com/media/b15c0e749e7db34dbe50ddaba0fe4c53/href</a></iframe><p>Really simple code. When we receive a message from our webhook we then communicate with our SMS Service via the recievedQueue to tell it that we received a message with a given body and from the corresponding number.</p><p>Note that to test this locally you’ll need to use a service like <a href="https://ngrok.com/">ngrok</a>, as Twilio will not work with a webhook that is running on local host. Ngrok is a service that will put your api running locally on a server so that you can test locally on your machine.</p><h3>Last Question Caching</h3><p>When we receive a message from our Twilio Webhook, we are given no context behind that message. We only are given the phone number, and what they said to us. How do we make sense of this? When we receive a text message from a phone number, we can assume that they are either 1) interacting with our platform for the first time or 2) responding to a text message that we sent them. Our platform sends the first text message, so we don’t need to account for case 1. But for case 2, we need to figure out what text message they are replying to then. To do this need to keep track of the last message we sent to a user that we expect them to reply to. Luckily we keep this structure in our message structs. Whenever we send someone a message where our message struct has the “expectsReply” filed out, we need to store that we sent that message to that number.</p><p>This is the perfect use case for a Redis hash where the key to the hash the number of the user and the value is the last message we sent them. I wrote a couple of functions to handle my caching.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/8fe68b15c118f2b28f852dbd4119edfc/href">https://medium.com/media/8fe68b15c118f2b28f852dbd4119edfc/href</a></iframe><p>And we should wrap our sendSMS function in a new function sendMessage where we update the cache accordingly.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/5c6a4b8e5da693b9ff98776867d5a667/href">https://medium.com/media/5c6a4b8e5da693b9ff98776867d5a667/href</a></iframe><h3>Putting It All Together — Handling Replies</h3><p>Now we have all the puzzle pieces and we just need to put them together to handle replies. When we receive messages from Twilio we need to</p><ol><li>Check our cache and see what message we last sent our user</li></ol><p>If we didn’t send them a message, send them our default message (“We weren’t expecting a message from you!”)</p><p>2. Grab the “expectsReply” schema from our messages JSON for the last message we sent our user</p><p>3. Match what the user sent us to the list of options in the corresponding reply schemas</p><p>4. If we matched the user’s reply to a replySchema then we need to</p><ul><li>Clear the last message cache of that user’s phone number</li><li>Take Action (send that data left we specify in the replySchema)</li><li>Send the next message</li></ul><p>5. If we didn’t, we need to ask them to send us a message again.</p><p>Here is that in code:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/120b3a23673ce39c14e37c5524bd0ea4/href">https://medium.com/media/120b3a23673ce39c14e37c5524bd0ea4/href</a></iframe><h3>Conclusion</h3><p>To deploy this I just have the python running on my server using nohup. Since this service is taking in inputs you can easily run multiple instances of this service as long as you properly distribute the last question cache as well.</p><p>Creating a Twilio conversation bot isn’t too hard of a task to undertake. If you look at the code, there isn’t that much code there, and it should be straightforward enough to understand.</p><p>Check out the full code base here <a href="https://github.com/lbarwiko/twilioconversation">on my Github</a> (it is lacking the webhook though, but you can find plenty of documentation on that elsewhere!)</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6e22d87bbfc2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What Getting Injured Taught Me About Myself]]></title>
            <link>https://medium.com/@lukebarwikowski/what-getting-injured-taught-me-about-myself-dc061e8d8cf9?source=rss-395d59f878ff------2</link>
            <guid isPermaLink="false">https://medium.com/p/dc061e8d8cf9</guid>
            <category><![CDATA[knee-surgery]]></category>
            <category><![CDATA[mental-health]]></category>
            <category><![CDATA[injury]]></category>
            <category><![CDATA[identity]]></category>
            <category><![CDATA[lessons-learned]]></category>
            <dc:creator><![CDATA[Luke Barwikowski]]></dc:creator>
            <pubDate>Sun, 28 Jul 2019 02:39:32 GMT</pubDate>
            <atom:updated>2019-07-28T02:42:14.749Z</atom:updated>
            <content:encoded><![CDATA[<p>In the fall of 2017 I returned to school after having had the most amazing summer of my life where I had been going climbing, surfing, and hiking as much as I possibly could. I found so much joy in these activities, and I made these activities become a large part of my identity. When I was introduced to people I was introduced to them as a climber, as a surfer, or as ‘really adventurous’. These had become the core of who I was. I continued to ride this wave through my time back at school and things seemed amazing. These activities weren’t as easily accessible in Michigan, but I still took any opportunity I could to stay fit and train for when I had the opportunity to partake in them again. I had become obsessed.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/0*RxeQ4jupmNLzaEz_.jpg" /></figure><p>Then it happened. I was playing volleyball with friends and I went up to spike a ball. When I landed, it didn’t feel right. I took a break, and I expected it to feel better again in 5 minutes and then I could keep going back at my usual pace. But that 5 minutes turned into the rest of the night, and the rest of the night turned into the next week. And it wasn’t getting better. I knew something had to be wrong with my knee, and the thought of that terrified me.</p><p>The next month and a half involved a lot of different nurses wrongly diagnosing what would eventually be identified as a bucket handle tear in my meniscus as an LCL sprain. That month and a half was really hard because an LCL sprain is supposed to get better with time, but my knee never did. This was a period that involved a lot of false hope and missed expectations. When I finally got the diagnosis that my meniscus was torn and that the only way I would ever feel better was an invasive surgery, there was a sense of relief that I finally knew what was going on, but a new overwhelming sense of fear that I may not ever be the same person I was before my meniscus tear.</p><p>I had built my identity around my athletic ability. My athletic ability and the activities it allowed me to do where what gave me my confidence, where I felt at home, and how I relaxed. My athletic ability introduced me to new friends, it’s how I got the attention of the opposite sex, and it’s what led me to cool places and new experiences. When I tore my meniscus I had all of that taken away from me. I not only lost a piece of my meniscus, I lost a large piece of myself.</p><p>I considered myself a forward looking and positive person, but I kept thinking to myself “What if I had never played that game of volleyball.” “What if I just had landed better?” It was very hard for me to accept the fact that I was injured.</p><p>The whole foundation of my identity was crumbling. I felt lost again. I felt like I lacked direction again. The way I thought my life was going to be was challenged, and I needed to think of alternatives. I realized I was more insecure in myself than I had thought before. I began doubting myself in ways I hadn’t for a long time. “Will people still want to hang out with me if it’s not during a sport?” “Will girls still find me attractive?” “What am I going to do with all my time?” “What do I even like doing?”</p><p>During my recovery, I felt as if I had a choice. I could be miserable until I got better, change nothing, and leave myself susceptible to this mental torment all over again if I get hurt. Or I could find a way to be happy and like who I was WITHOUT my athletic ability.</p><p>The next few months weren’t initially easy. But I approached the situation with a positive and very proactive attitude. I knew it was going to take work, and I knew I had to take charge of it. Sitting alone in my room wasn’t going to help me feel better. I had to go out, I had to try new things, and I needed to approach new hobbies with an open mind.</p><p>I started growing relationships with the people around me more, I took more advantage of social opportunities, and I took a more proactive approach to my mental health. By the time that my injury healed and I was back to my normal athletic self, I was a whole different person.</p><p>Instead of creating an identity that was built on such a fragile foundation as athletic ability, I instead built a stronger foundation that couldn’t be as easily shaken. Not to say that after my injury I had completely figured myself out, there were many more trials and tribulations that sparked a similar change in how I viewed myself. But this injury in particular created largest and the most impactful change in my view of myself, and I came out of it so much stronger.</p><p>I know how hard this was for me and that’s why I created <a href="https://www.whatnow.co/">https://www.whatnow.co</a> — a place where people who are injured can talk to other people about what they’re going through. Getting injured is weird and scary. I never felt like I talked to anyone who ‘understood’ what I was going through. This platform tries to connect people who have been injured before to the people who are scared about their situations. If you have time, check it out, and consider volunteering your time to offer support to someone who is going through something similar.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dc061e8d8cf9" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why web developers should still use Wordpress]]></title>
            <link>https://medium.com/@lukebarwikowski/why-web-developers-should-still-use-wordpress-95736ed82b5?source=rss-395d59f878ff------2</link>
            <guid isPermaLink="false">https://medium.com/p/95736ed82b5</guid>
            <category><![CDATA[bootstrap-4]]></category>
            <category><![CDATA[bootstrap]]></category>
            <category><![CDATA[full-stack-development]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[wordpress]]></category>
            <dc:creator><![CDATA[Luke Barwikowski]]></dc:creator>
            <pubDate>Thu, 25 Jul 2019 16:16:21 GMT</pubDate>
            <atom:updated>2019-07-25T16:17:30.254Z</atom:updated>
            <content:encoded><![CDATA[<p>As a full-stack web developer, I had a large aversion to installing and using Wordpress for any project I was working on. I knew how to code a website on my own, why would I ever need something to do it for me? I would always say to myself, ‘Wordpress is for people who don’t know how to code’, and I knew that Wordpress has a reputation for being bloated, slow, and frustrating. However my opinion has changed after my most recent project — <a href="http://www.whatnow.co/">www.whatnow.co</a> — a website that focuses on finding injured athletes mental health support from peers.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Ip6_KesWTWCUxlbx5KE72A.png" /><figcaption>WhatNow.co — My most recent project where I instead opted to use Wordpress as the website’s frontend.</figcaption></figure><p>Whenever I made a website for a project I was working on I would instead choose to write the entire thing myself. From scratch — every time. There is a certain level of satisfaction that comes with this. It feels really good to know that you can do it all by yourself. And the work you put into making a landing page, or a website feels more earned when you do it the hard way.</p><p>However the core of the issue here, is we as web developers underestimate how much work there is to do to make a website.</p><p>I most often would create the websites I needed using Bootstrap, but I normally would find that I would need heavy customization from whatever Bootstrap gave me, and I would find myself spending wasting too much time writing CSS to fix such small things like padding and margins. If I had more than one page to my website, I could still use NGINX but I would normally find myself also using running another server running either Flask or Express on my machine and then using NGINX as a reverse proxy and have that utilizing SSL. And all of that adds so much dev-ops work.</p><p>So then I would have the first iteration of my website live. It would be on the internet, and I would be tempted to call it a day. But you can’t stop there. I would notice that elements of my page are loading slowly. So I’d need to run all my images through an optimizer and re-upload them. A pain in the ass when you’re doing everything from scratch. I’d realize I didn’t like a few elements of my website after launch, and I’d want to change or iterate on how it looked — or even consider a full redesign. Changing the design of a live site like that is even more of a pain in the ass. Attempting to rapidly iterate designs can be even harder.</p><p>But you haven’t even gotten half the work yet. You need services to tell when your site goes down, there’s SEO to do, you need to start content marketing, attracting users, and so much more. If you wrote your entire website from scratch, doing all of this becomes so much more impossible. And not worth your time.</p><p>Wordpress IS bloated, and it CAN BE slow. But it makes rapidly iterating a production website, SEO, and content marketing so much easier. It makes the end game of your finished website all around better.</p><p>Wordpress seems also slow at first, but there are also a ton of optimizations you can do to make it faster. By minimizing plugins on your website, and utilizing a lot of caching strategies, you can bring down your page loads down to comparable speeds of other websites without doing anything too crazy.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_TkyEs-MEzxajEOVaDY2HA.png" /><figcaption>This is my page speed score after optimizing my Wordpress site to Cache &amp; Compress better.</figcaption></figure><p>PageSpeed insights by Google (<a href="https://developers.google.com/speed/pagespeed/insights/">https://developers.google.com/speed/pagespeed/insights/</a>) can give you a great look into what you can do to perform your site’s performance. It was a god-send from me and I took my page loads from 10 seconds down to ~1 second. And there’s a lot of potential to get it down way lower. For example, I’m still hosting all of my assets on my own server. If I was utilizing a CDN to distribute my CSS, Javascript, and images my page loads would be significantly better. But I’m still getting these speeds on a $5 Digital Ocean server that’s running 3 other Wordpress Websites.</p><p>I could write a whole other article on Wordpress Optimization. There’s a lot of cool tricks you can do. But my hope is that I can at least convince you to consider using Wordpress as an alternative to building an entire website on your own. If you don’t use Wordpress you are missing out on the SEO and marketing benefits that come with it and you could be wasting your own time.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=95736ed82b5" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>