<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Janet Malone</title>
    <description>The latest articles on DEV Community by Janet Malone (@janetmalone64).</description>
    <link>https://dev.to/janetmalone64</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4203195%2Ff35a1616-452e-4d0f-b776-e8fe9414ae7e.png</url>
      <title>DEV Community: Janet Malone</title>
      <link>https://dev.to/janetmalone64</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/janetmalone64"/>
    <language>en</language>
    <item>
      <title>The setup screen I deleted from my custom Wordle tool</title>
      <dc:creator>Janet Malone</dc:creator>
      <pubDate>Tue, 17 Nov 2026 05:23:15 +0000</pubDate>
      <link>https://dev.to/janetmalone64/the-setup-screen-i-deleted-from-my-custom-wordle-tool-24oe</link>
      <guid>https://dev.to/janetmalone64/the-setup-screen-i-deleted-from-my-custom-wordle-tool-24oe</guid>
      <description>&lt;p&gt;When I started building a make-your-own-Wordle tool, the obvious plan was a builder page. Pick the word length, set the number of guesses, add a title, hit Create puzzle, then copy your link. That is how most generators work, and it is what I had sketched in a notebook.&lt;/p&gt;

&lt;p&gt;I threw it out after watching one person use the prototype. They filled in the form, hit Create, got a link, opened it, and then typed the same word into the puzzle again to check that it worked. Nobody trusts the output until they have played it. The form was a step between them and the only thing they wanted to see.&lt;/p&gt;

&lt;p&gt;So the version I shipped has no setup screen. You land on a five-letter grid with an empty first row and type your answer straight into it. That row is the secret word. The share link updates as you type, and a play button appears next to it so you can test your own puzzle in the same tab. There is nothing to configure, because for most people the answer is the only setting that matters.&lt;/p&gt;

&lt;p&gt;Two things this forced me to think about.&lt;/p&gt;

&lt;p&gt;The guess dictionary and the answer are not the same list. Guesses need to be real words, otherwise people just cycle through letters and it stops being a puzzle. But answers should be allowed to be anything five letters long: a nickname, a pet, a street name. So the answer is validated on length only, and guesses go through a word list. It feels loose until you remember the whole point is making one for your friend Sam.&lt;/p&gt;

&lt;p&gt;The URL is the storage. The word lives in the link fragment, so there is no database and no account, and the answer never goes to a server. The tradeoff is real: you cannot edit a puzzle after you share it. If you mistype the answer, you send a second link. I decided that was fine, because it is one short string, not a document.&lt;/p&gt;

&lt;p&gt;The use I did not anticipate was a teacher building a warm-up where the answer was a vocabulary word from the previous lesson, then dropping a single link into the class chat. No signup, no email collection, no student accounts. That mattered more than any option I could have put on the form.&lt;/p&gt;

&lt;p&gt;If you want to see the version with no setup screen, it is here: &lt;a href="https://customwordle.gamecodeshq.com" rel="noopener noreferrer"&gt;make your own Wordle&lt;/a&gt;. Type the answer, copy the link, done.&lt;/p&gt;

</description>
      <category>make</category>
      <category>your</category>
      <category>webdev</category>
      <category>tools</category>
    </item>
    <item>
      <title>The calorie tracker UX decision: log first, organize later</title>
      <dc:creator>Janet Malone</dc:creator>
      <pubDate>Wed, 04 Nov 2026 01:23:19 +0000</pubDate>
      <link>https://dev.to/janetmalone64/the-calorie-tracker-ux-decision-log-first-organize-later-4opn</link>
      <guid>https://dev.to/janetmalone64/the-calorie-tracker-ux-decision-log-first-organize-later-4opn</guid>
      <description>&lt;p&gt;When I started building a calorie tracker, the obvious approach was a food database. Search for an item, choose a serving size, pick a meal, save it to a daily log. I sketched that flow, then tried it with my own lunch from a small cafe. The bowl was not in the database. I got a dozen near matches, guessed a serving size, picked 'lunch', and spent more time operating the app than eating. That was the moment I changed the core interaction.&lt;/p&gt;

&lt;p&gt;The decision was to make calorie logging a single quick-add field. You type what you ate and an approximate calorie number, then press enter. Macros are optional. Meal categories do not exist. The app does not ask you to search a database before you can record anything. If you want more detail, you can put it in the label: 'chicken rice bowl 650'. The entry lands in today's list and you move on.&lt;/p&gt;

&lt;p&gt;This is less precise than a full food database, and that is intentional. Most people I know who track calories casually are not trying to produce an audit-grade ledger. They want a rough daily number that takes seconds to update. One friend logs 'toast and coffee 320' before she opens her email. That habit would not survive a five-step search flow.&lt;/p&gt;

&lt;p&gt;The tradeoff is that quick-add depends on your own estimate. If you have no idea what a portion contains, a database can help. We kept the app narrow instead: a free browser tool, no signup, and a fast entry path for people who already have a rough sense of their food.&lt;/p&gt;

&lt;p&gt;There is one gotcha worth knowing. Since there is no account, the log is stored in the browser's localStorage. That keeps setup at zero, but clearing site data or switching browsers means the history is gone. I put a short note about that on the page instead of pretending local-only storage is the same as an account.&lt;/p&gt;

&lt;p&gt;I put the current version at &lt;a href="https://calorietracker.theblissfulmind.store" rel="noopener noreferrer"&gt;calorietracker.theblissfulmind.store&lt;/a&gt; if you want to see the quick-add flow. The lesson for me was not that databases are bad. It was that the first screen should solve the smallest useful version of the job.&lt;/p&gt;

</description>
      <category>best</category>
      <category>calorie</category>
      <category>webdev</category>
      <category>tools</category>
    </item>
    <item>
      <title>The UX decision that made our mouse tester simpler: no Start button</title>
      <dc:creator>Janet Malone</dc:creator>
      <pubDate>Sat, 31 Oct 2026 05:23:21 +0000</pubDate>
      <link>https://dev.to/janetmalone64/the-ux-decision-that-made-our-mouse-tester-simpler-no-start-button-249j</link>
      <guid>https://dev.to/janetmalone64/the-ux-decision-that-made-our-mouse-tester-simpler-no-start-button-249j</guid>
      <description>&lt;p&gt;When I built the first version of our mouse tester, the obvious approach was a start screen: pick a test, click Start, then interact with a test area. That felt safe. It also made the tool feel like a form.&lt;/p&gt;

&lt;p&gt;The problem showed up in the first few user sessions. People opened the page, moved the mouse, and nothing happened. They clicked the big button, but then weren't sure what counted as "inside" the test. Some clicked the Start button again. One person told me they clicked the Start button, then clicked the test area, and only later realized the first click was just part of the UI. Others assumed the tool was broken because the first click they cared about was the one that launched the test.&lt;/p&gt;

&lt;p&gt;So we removed the start step entirely. The page listens as soon as it loads. Moving the mouse updates the cursor position. Pressing any button updates the button state. Scrolling updates the wheel readout. There is no "test mode" to enter or exit. If you opened the page, the test is already running.&lt;/p&gt;

&lt;p&gt;That decision forced a few implementation details. We attach listeners at the window level instead of to a specific test box, because users naturally move outside the box while testing. We also listen to mousedown and mouseup rather than click for button state. Side buttons are the gotcha: many browsers fire press/release events for button 3 and 4 but don't give you a normal click event. If the whole test is built around click handlers, those buttons look dead.&lt;/p&gt;

&lt;p&gt;The live page also made the double-click check more honest. Instead of asking users to "start double-click test", we show the timing between consecutive presses as they happen. Same for scroll: we show raw wheel deltas, because different browsers report lines versus pixels, and hiding that behind a single smooth number would be misleading.&lt;/p&gt;

&lt;p&gt;The result is less UI, not more. No setup, no tabs for each test, no reset flow. The page is just a readout of what your mouse is doing right now. If something feels wrong, you click and watch the state change. A common reason people land there is a right-click or side button that stopped responding after a cable swap.&lt;/p&gt;

&lt;p&gt;You can try the approach here: &lt;a href="https://mousetester.novadailyhub.com" rel="noopener noreferrer"&gt;mousetester.novadailyhub.com&lt;/a&gt;. Open it, move the mouse, and you're already testing.&lt;/p&gt;

</description>
      <category>mouse</category>
      <category>tester</category>
      <category>webdev</category>
      <category>tools</category>
    </item>
    <item>
      <title>Keeping a home security camera helper out of your request logs</title>
      <dc:creator>Janet Malone</dc:creator>
      <pubDate>Tue, 27 Oct 2026 11:23:23 +0000</pubDate>
      <link>https://dev.to/janetmalone64/keeping-a-home-security-camera-helper-out-of-your-request-logs-1nj1</link>
      <guid>https://dev.to/janetmalone64/keeping-a-home-security-camera-helper-out-of-your-request-logs-1nj1</guid>
      <description>&lt;p&gt;When I started building a small browser tool for narrowing down home security cameras, the default architecture was a form that posted answers to a backend and returned recommendations. I didn't want that. The questions get personal fast: do you rent, is there a power outlet near the door, do you need night vision in a shared hallway, are you home during the day. Those answers can describe a home's layout and occupancy. Putting them in a request log felt like a bad trade for a free tool.&lt;/p&gt;

&lt;p&gt;So the recommendation logic runs entirely in the browser. No signup, no email gate, no server-side session. I keep in-progress answers in sessionStorage rather than localStorage, which means they disappear when the tab closes. That's deliberate: I don't need a persistent profile of someone's security setup sitting on their device for months.&lt;/p&gt;

&lt;p&gt;The first gotcha was the back button. People would answer a few questions, click back to change one, and the draft state would reappear in odd places because I was mutating a single object and re-rendering everything. I ended up treating each step as a small state transition, writing the draft once per change, and restoring from the stored object on load. It sounds minor, but it fixed the "why did my answer to the power-outlet question vanish?" bug that showed up in my own testing.&lt;/p&gt;

&lt;p&gt;The performance detail was the recommendation data. I had a fairly large JSON file mapping camera types to tradeoffs: battery vs wired, local storage vs subscription, motion detection range, and so on. Shipping all of it up front made the first load slower than it needed to be, especially on a phone. I split it into small chunks and fetch only the branch the current question needs. Static chunks, no user answers included. I also prefetch the likely next chunk on focus, so the next step usually feels instant. The gotcha there was the first click after a cold load could briefly wait on a chunk; I added a small inline placeholder instead of a spinner so the layout doesn't jump.&lt;/p&gt;

&lt;p&gt;The result is a tool that does one job and forgets you when the tab closes. That felt like the right default for this category. If you want to try the browser-only version, it's at &lt;a href="https://homesecuritycamera.buyerguidehq.com" rel="noopener noreferrer"&gt;Home Security Camera Buyer Guide&lt;/a&gt;.&lt;/p&gt;

</description>
      <category>home</category>
      <category>security</category>
      <category>webdev</category>
      <category>tools</category>
    </item>
  </channel>
</rss>
