<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/">
  <channel>
    <title>Technology on D. Sébastien Taggart</title>
    <link>https://www.sebastientaggart.com/categories/technology/</link>
    <description>Recent content in Technology on D. Sébastien Taggart</description>
    <generator>Hugo</generator>
    <language>en</language>
    <lastBuildDate>Sun, 04 Oct 2026 22:13:05 -0400</lastBuildDate>
    <atom:link href="https://www.sebastientaggart.com/categories/technology/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Building the Ultimate Content Editor Experience in a CMS</title>
      <link>https://www.sebastientaggart.com/writing/building-the-ultimate-content-editor-experience-in-a-cms/</link>
      <pubDate>Tue, 02 Jul 2024 21:15:54 -0400</pubDate>
      <guid>https://www.sebastientaggart.com/writing/building-the-ultimate-content-editor-experience-in-a-cms/</guid>
      <description>Principles and feature buckets that keep a rich text editor trustworthy for authors, aligned with the live site, and usable under Section 508.</description>
      <content:encoded><![CDATA[<p>Most CMS projects treat the editor as a checkbox. Pick CKEditor or TinyMCE, wire a media library if you are lucky, and call it done. Then the help desk tickets start: &ldquo;Why does my table look different on the page?&rdquo; &ldquo;I pasted from Word and the whole section broke.&rdquo; &ldquo;Screen reader users say the accordion is unreadable.&rdquo; I have been on both sides of that (the person filing the ticket and the person explaining why WYSIWYG is a lie). The editor is where content quality and accessibility are won or lost before anything ships.</p>
<p>Here is how I think about building an editor experience that authors will actually use without breaking the site, and that auditors will not torch in a 508 review.</p>
<h2 id="principles-that-actually-constrain-design">Principles that actually constrain design</h2>
<p><strong>Simplicity</strong> sounds obvious until you add every toolbar button a power user asked for in a requirements workshop. My bias: default to a short toolbar and progressive disclosure. If a control does not apply to most pages, it belongs in a secondary panel or a structured field, not next to Bold. Fewer choices means fewer wrong choices.</p>
<p><strong>Impossible to break</strong> is not literal (authors are creative), but the goal is valid output. That usually means constraining the schema: allowed tags, allowed classes, sanitization on save, and patterns that map cleanly to your theme. If someone can paste arbitrary HTML from a random website and have it stick, you do not have an editor; you have a footgun with a blinking cursor.</p>
<p><strong>Perfect previews</strong> are the hardest item on the list. &ldquo;What you see is what you get&rdquo; is only honest if the iframe (or live preview route) loads the same CSS, typography, spacing tokens, and component markup as production. If preview runs against a stripped admin theme, you are training authors to distrust the tool. I treat preview parity as a build pipeline problem, not a nice-to-have.</p>
<p><strong>Familiarity</strong> is cheaper than novelty. Bold looks like bold, lists look like lists, link dialogs look like the link dialogs people know from Google Docs or Word. Custom iconography for standard actions is how you get &ldquo;where is underline?&rdquo; in every training session.</p>
<p><strong>508 compliance</strong> is not a bolt-on color contrast pass at the end. It is keyboard operability for every control in the chrome, visible focus, text alternatives for images at insert time, table headers that survive round-trips, and collapsible regions that expose state to assistive tech. If your accordion widget relies on click-only divs with no role or name, you have built technical debt that WCAG will find.</p>
<h2 id="image-management-and-embedding">Image management (and embedding)</h2>
<p>Authors need crop, resize, focal point, and alt text in the flow where they insert the image, not in a separate &ldquo;fix accessibility later&rdquo; queue. Drag-and-drop is fine if the same fields appear for uploads and library picks. Lazy loading and <code>srcset</code> should be automatic; the author should not hand-edit attributes.</p>
<p>I also like forcing a decision on decorative images (empty alt with a real checkbox) so people do not default to filename-as-alt-text, which is a classic 508 fail in the wild.</p>
<h2 id="file-management-and-embedding">File management (and embedding)</h2>
<p>PDFs and Office files are where CMSs quietly become file shares. If you allow arbitrary uploads from the body field, you get duplicate assets, stale versions, and broken links after migrations. A dedicated media or document entity with a stable URL, virus scanning if you are in a regulated environment, and a picker that inserts a shortcode or structured link beats &ldquo;attach whatever.&rdquo;</p>
<p>For federal-ish work, think about whether the file should open in a new tab, whether you need a file type and size label next to the link, and whether the title text is meaningful (not <code>FINAL_v7_really.pdf</code>).</p>
<h2 id="tables-sortable-interactive-compliant">Tables (sortable, interactive, compliant)</h2>
<p>Tables are the #1 source of &ldquo;looks fine in the editor&rdquo; grief. Authors merge cells because Excel trained them to; then responsive CSS and screen readers suffer.</p>
<p>My hierarchy: (1) simple data tables with explicit header rows and no layout tables in the body, (2) optional sorting or filtering only if you have a component that preserves <code>&lt;th scope&gt;</code> relationships and does not rely on hover alone, (3) if you need pivot-level behavior, that is probably not inline rich text; it is a block, a view, or an embed.</p>
<p>Interactive tables also need a plan for mobile (horizontal scroll with a visible cue, or reflow) and for keyboard users (sort controls must be focusable buttons, not mystery clicks on <code>&lt;th&gt;</code>).</p>
<h2 id="accordions-and-collapsible-content">Accordions and collapsible content</h2>
<p>Collapsible sections are great for long policy pages and terrible when implemented as anonymous divs with a chevron image. Use a pattern with <code>button</code> triggers, <code>aria-expanded</code>, and an associated region id. Keyboard support should mirror disclosure widgets in the ARIA authoring practices guide.</p>
<p>From an editor UX angle, treat accordion as a structured block (Gutenberg-style, Layout Builder section, whatever your stack supports) instead of free-form HTML. That keeps markup consistent and previews honest.</p>
<h2 id="formatting-and-cleanup">Formatting and cleanup</h2>
<p>Bold, italic, underline, links, headings, lists, blockquote: ship the boring set first. &ldquo;Clear formatting&rdquo; and &ldquo;paste as plain text&rdquo; (or paste filtering) save more production bugs than another alignment option. Paste from Word is a product feature; test it with real government templates and tables.</p>
<p>If you support subscript or strikethrough for legal or scientific content, expose them; if you do not need them, hide them so authors do not build layouts out of character-level styling.</p>
<h2 id="pre-styled-body-content-callouts-boxes-div-based-patterns">Pre-styled body content (callouts, boxes, div-based patterns)</h2>
<p>Callout boxes (&ldquo;blue half-width note,&rdquo; warning panels, pull quotes) should not be &ldquo;paste this div from the wiki.&rdquo; They should be either (a) a small set of allowed wrapper classes your CSS documents, inserted through a dropdown the editor understands, or (b) custom components with their own fields. Half-width layouts need a grid or breakpoint story on small screens (usually full width below a breakpoint), or mobile users eat the cost.</p>
<p>Whatever you do, avoid unbounded class names in the body. Whitelist classes, map them to components, and validate on save. That is how you keep the experience &ldquo;impossible to break&rdquo; without turning authors into front-end developers.</p>
<p>The editor worth shipping is not the one with the most buttons. It is the one where authors recognize the interface, the preview matches what visitors see, and the output is boring enough for automated checks and screen readers to handle. I am still looking for a stack that nails all of that without custom glue; if you have one, I would like to hear about it.</p>
]]></content:encoded>
    </item>
    <item>
      <title>Running a Self-Hosted Bookmarking Service with Shaarli</title>
      <link>https://www.sebastientaggart.com/writing/running-a-self-hosted-bookmarking-service-with-shaarli/</link>
      <pubDate>Sat, 29 Aug 2020 18:28:27 +0000</pubDate>
      <guid>https://www.sebastientaggart.com/writing/running-a-self-hosted-bookmarking-service-with-shaarli/</guid>
      <description>&lt;p&gt;In a world awash with digital content, finding a reliable way to manage your bookmarks without sacrificing privacy can feel like searching for a needle in a haystack. I wanted a solution that offered:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;A seamless user experience&lt;/li&gt;
&lt;li&gt;Robust privacy through self-hosting&lt;/li&gt;
&lt;li&gt;Simple setup via &lt;code&gt;docker&lt;/code&gt; and &lt;code&gt;docker-compose&lt;/code&gt;&lt;/li&gt;
&lt;li&gt;Minimal resource consumption&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;While there’s no shortage of alternatives, like &lt;a href=&#34;https://github.com/wallabag/wallabag&#34;&gt;Wallabag&lt;/a&gt;, I found it to be slower and more resource-heavy than my chosen solution: Shaarli. Other options, such as &lt;a href=&#34;https://github.com/cdevroe/unmark&#34;&gt;Unmark&lt;/a&gt;, &lt;a href=&#34;https://github.com/MarceauKa/shaark&#34;&gt;Shaark&lt;/a&gt;, and &lt;a href=&#34;https://github.com/Kovah/LinkAce&#34;&gt;LinkAce&lt;/a&gt;, often attempt to offer too much, leading to unnecessary complexity and resource demands.&lt;/p&gt;</description>
      <content:encoded><![CDATA[<p>In a world awash with digital content, finding a reliable way to manage your bookmarks without sacrificing privacy can feel like searching for a needle in a haystack. I wanted a solution that offered:</p>
<ul>
<li>A seamless user experience</li>
<li>Robust privacy through self-hosting</li>
<li>Simple setup via <code>docker</code> and <code>docker-compose</code></li>
<li>Minimal resource consumption</li>
</ul>
<p>While there’s no shortage of alternatives, like <a href="https://github.com/wallabag/wallabag">Wallabag</a>, I found it to be slower and more resource-heavy than my chosen solution: Shaarli. Other options, such as <a href="https://github.com/cdevroe/unmark">Unmark</a>, <a href="https://github.com/MarceauKa/shaark">Shaark</a>, and <a href="https://github.com/Kovah/LinkAce">LinkAce</a>, often attempt to offer too much, leading to unnecessary complexity and resource demands.</p>
<h2 id="features-and-interface">Features and Interface</h2>
<p>Upon logging in, the interface presents a bit of visual clutter, yet it remains remarkably intuitive:</p>
<p><img alt="Shaarli Interface" loading="lazy" src="/writing/running-a-self-hosted-bookmarking-service-with-shaarli/assets/index-0e5ee54e.png"></p>
<p>The <strong>search</strong> functionality stands out—it&rsquo;s quick, highlights results, and operates seamlessly. No bells and whistles; it just works.</p>
<p>The feature set is well-curated, though I find the <strong>&ldquo;Picture wall&rdquo;</strong> somewhat superfluous. It pulls in logos or generic stock images from bookmarked pages, which, to be candid, doesn’t add much value for me.</p>
<p><img alt="Picture Wall" loading="lazy" src="/writing/running-a-self-hosted-bookmarking-service-with-shaarli/assets/index-ff7cf9c6.png"></p>
<p>The <strong>Tag Cloud</strong> is standard fare, but if you’re a fan, you’ll appreciate its instant search results. Typing reveals results quickly, which is a boon for efficiency.</p>
<p><img alt="Tag Cloud" loading="lazy" src="/writing/running-a-self-hosted-bookmarking-service-with-shaarli/assets/index-9ab903db.png"></p>
<p>One limitation is the lack of support for multi-word tags (think &ldquo;Professional Development&rdquo; instead of splitting it into &ldquo;Professional&rdquo; and &ldquo;Development&rdquo;). However, the team is aware of this issue and is working on it—progress is always welcome!</p>
<p><strong>Importing Links</strong> from the &ldquo;Tools&rdquo; menu is a breeze, handling my ~200 bookmarks effortlessly. Reports from other users indicate it scales well, accommodating thousands of links without a hitch.</p>
<h2 id="browser-plugins">Browser Plugins</h2>
<p>For Firefox enthusiasts, there exists a straightforward <a href="https://github.com/ikipatang/shaarli-web-extension">web extension</a> that allows you to add the current page to your Shaarli instance via a simple popup:</p>
<p><img alt="Firefox Extension" loading="lazy" src="/writing/running-a-self-hosted-bookmarking-service-with-shaarli/assets/index-3fa4acfc.png"></p>
<p>That said, the user experience here could use some refinement. The popup can sometimes hide behind the main window—an annoying hiccup. Scrolling is required, which begs the question: why not simplify the interface? An additional click is needed to save your bookmark, unlike the smooth single-click experience in native browsers. Despite these quirks, it does serve its purpose.</p>
<p>For Chromium users, there’s a similar <a href="https://chrome.google.com/webstore/detail/add-to-shaarli/jhfblapoehcfajokolimghdfmeeakbee">extension</a> available, alongside an <a href="https://play.google.com/store/apps/details?id=com.dimtion.shaarlier">Android utility</a>. However, I found the latter a bit cumbersome for regular use.</p>
<p>It’s important to note that neither extension currently allows for bookmark syncing.</p>
<h2 id="performance">Performance</h2>
<p>Shaarli excels in performance. Even when running in Docker on a five-year-old machine, the interface feels responsive, as if it&rsquo;s been pre-fetched. Actions like viewing bookmark details, searching, and adding tags occur with instant responsiveness.</p>
<h2 id="configuration">Configuration</h2>
<p>Setting up Shaarli locally is straightforward with <code>docker-compose</code>:</p>
<div class="highlight"><pre tabindex="0" class="chroma"><code class="language-yaml" data-lang="yaml"><span class="line"><span class="cl"><span class="nt">version</span><span class="p">:</span><span class="w"> </span><span class="s1">&#39;3&#39;</span><span class="w">
</span></span></span><span class="line"><span class="cl"><span class="w">
</span></span></span><span class="line"><span class="cl"><span class="nt">services</span><span class="p">:</span><span class="w">
</span></span></span><span class="line"><span class="cl"><span class="w">  </span><span class="nt">shaarli</span><span class="p">:</span><span class="w">
</span></span></span><span class="line"><span class="cl"><span class="w">    </span><span class="nt">image</span><span class="p">:</span><span class="w"> </span><span class="l">shaarli/shaarli:master</span><span class="w">
</span></span></span><span class="line"><span class="cl"><span class="w">    </span><span class="nt">container_name</span><span class="p">:</span><span class="w"> </span><span class="l">shaarli</span><span class="w">
</span></span></span><span class="line"><span class="cl"><span class="w">    </span><span class="nt">restart</span><span class="p">:</span><span class="w"> </span><span class="l">unless-stopped</span><span class="w">
</span></span></span><span class="line"><span class="cl"><span class="w">    </span><span class="nt">ports</span><span class="p">:</span><span class="w">
</span></span></span><span class="line"><span class="cl"><span class="w">      </span>- <span class="m">8093</span><span class="p">:</span><span class="m">80</span><span class="w">
</span></span></span><span class="line"><span class="cl"><span class="w">    </span><span class="nt">volumes</span><span class="p">:</span><span class="w">
</span></span></span><span class="line"><span class="cl"><span class="w">      </span>- <span class="l">./cache:/var/www/shaarli/cache</span><span class="w">
</span></span></span><span class="line"><span class="cl"><span class="w">      </span>- <span class="l">./data:/var/www/shaarli/data</span><span class="w">
</span></span></span></code></pre></div><p>In my setup, port <code>8093</code> is managed through a Caddy reverse proxy:</p>
<div class="highlight"><pre tabindex="0" class="chroma"><code class="language-yaml" data-lang="yaml"><span class="line"><span class="cl"><span class="c"># Shaarli Bookmark Manager</span><span class="w">
</span></span></span><span class="line"><span class="cl"><span class="l">links.MYDOMAIN.com:80 {</span><span class="w">
</span></span></span><span class="line"><span class="cl"><span class="w">  </span><span class="l">proxy / localhost:8093 {</span><span class="w">
</span></span></span><span class="line"><span class="cl"><span class="w">    </span><span class="l">transparent</span><span class="w">
</span></span></span><span class="line"><span class="cl"><span class="w">    </span><span class="l">header_upstream X-Forwarded-Proto {scheme}</span><span class="w">
</span></span></span><span class="line"><span class="cl"><span class="w">  </span>}<span class="w">
</span></span></span><span class="line"><span class="cl"><span class="w">  </span><span class="c"># Wildcard TLS is handled upstream</span><span class="w">
</span></span></span><span class="line"><span class="cl"><span class="w">  </span><span class="l">tls off</span><span class="w">
</span></span></span><span class="line"><span class="cl">}<span class="w">
</span></span></span></code></pre></div><p>Shaarli integrates smoothly behind any reverse proxy, requiring no special configuration tweaks.</p>
<h2 id="conclusion">Conclusion</h2>
<p>If you seek a fast, straightforward alternative to less secure public bookmark managers, I highly recommend giving Shaarli a try. As we continue to navigate the complexities of digital information, having a reliable and private bookmarking solution at your fingertips is not just beneficial—it’s essential.</p>
]]></content:encoded>
    </item>
    <item>
      <title>Comparing YouTube Downloaders: Why I Chose Tubearchivist</title>
      <link>https://www.sebastientaggart.com/writing/youtube-downloaders-compared/</link>
      <pubDate>Mon, 25 Feb 2019 13:19:50 -0500</pubDate>
      <guid>https://www.sebastientaggart.com/writing/youtube-downloaders-compared/</guid>
      <description>&lt;p&gt;In the digital landscape, where algorithms often feel like labyrinthine mazes and ads spring up like weeds, the YouTube viewing experience can be downright frustrating. The relentless pursuit of revenue has led to a user interface cluttered with distractions, leaving many of us yearning for a simpler, more enjoyable way to consume content. Fortunately, a community of intrepid open-source developers has risen to the challenge. They’ve created tools designed to help us reclaim control over our viewing experience.&lt;/p&gt;</description>
      <content:encoded><![CDATA[<p>In the digital landscape, where algorithms often feel like labyrinthine mazes and ads spring up like weeds, the YouTube viewing experience can be downright frustrating. The relentless pursuit of revenue has led to a user interface cluttered with distractions, leaving many of us yearning for a simpler, more enjoyable way to consume content. Fortunately, a community of intrepid open-source developers has risen to the challenge. They’ve created tools designed to help us reclaim control over our viewing experience.</p>
<p>The goal here is straightforward: download YouTube videos to consume them ad-free, without the invasive suggestions or comments that often accompany them. We&rsquo;ll play these videos through the Jellyfin media player, ensuring that a separate app, YouTube or otherwise, is unnecessary.</p>
<p>While tools like <code>yt-dl</code> are certainly powerful, their user experience can be daunting for the average family member. Thus, my criteria for a more user-friendly solution are clear:</p>
<ul>
<li>A web-based interface that is simple to navigate.</li>
<li>Automatic downloading of new videos from subscribed channels.</li>
<li>Seamless cleanup of old or watched videos to manage storage effectively.</li>
<li>Compatibility with the jellyfin-youtube-metadata-plugin format, so downloaded videos display with thumbnails and appealing titles in Jellyfin.</li>
</ul>
<p>Imagine this: you open Jellyfin, navigate to &ldquo;YouTube,&rdquo; and browse through a neatly organized list of videos complete with thumbnails. A simple click, and you&rsquo;re off to enjoy your chosen content.</p>
<h3 id="introducing-tubearchivist">Introducing Tubearchivist</h3>
<p>To achieve this streamlined experience, I recommend using Tubearchivist. Here’s a brief overview of how to set it up:</p>
<ol>
<li><strong>Container Configuration</strong>: Begin by defining your Tubearchivist container. Below is a simplified Docker configuration to get you started:</li>
</ol>
<pre tabindex="0"><code>tubearchivist:
  container_name: tubearchivist
  restart: unless-stopped
  image: bbilly1/tubearchivist
  ports:
    - 8000:8000
  volumes:
    - /mnt/user/media/youtube:/youtube
    - cache:/cache
  environment:
    - ES_URL=http://archivist-es:9200     # Specify protocol and port
    - REDIS_HOST=archivist-redis          # No protocol needed
    - HOST_UID=1000
    - HOST_GID=1000
    - TA_HOST=10.0.50.10         # Set your hostname
    - TA_USERNAME=tubearchivist           # Initial TA credentials
    - TA_PASSWORD=verysecret              # Initial TA credentials
    - ELASTIC_PASSWORD=verysecret         # Password for Elasticsearch
    - TZ=America/New_York                 # Set your time zone
  depends_on:
    - archivist-es
    - archivist-redis
</code></pre><ol start="2">
<li><strong>Dependencies</strong>: Don’t forget to set up the necessary dependencies:</li>
</ol>
<pre tabindex="0"><code>archivist-redis:
  image: redislabs/rejson                 # For arm64, use bbilly1/rejson
  container_name: archivist-redis
  restart: unless-stopped
  expose:
    - &#34;6379&#34;
  volumes:
    - redis:/data
  depends_on:
    - archivist-es

archivist-es:
  image: bbilly1/tubearchivist-es         # Use the official version for amd64
  container_name: archivist-es
  restart: unless-stopped
  environment:
    - &#34;ELASTIC_PASSWORD=verysecret&#34;       # Matching Elasticsearch password
    - &#34;ES_JAVA_OPTS=-Xms512m -Xmx512m&#34;
    - &#34;xpack.security.enabled=true&#34;
    - &#34;discovery.type=single-node&#34;
    - &#34;path.repo=/usr/share/elasticsearch/data/snapshot&#34;
  ulimits:
    memlock:
      soft: -1
      hard: -1
  volumes:
    - es:/usr/share/elasticsearch/data    # Check for permission errors with bind mounts
  expose:
    - &#34;9200&#34;
</code></pre><ol start="3">
<li><strong>Volumes</strong>: Define your storage volumes to keep things organized:</li>
</ol>
<pre tabindex="0"><code>volumes:
  cache:
  redis:
  es:
</code></pre><h3 id="conclusion">Conclusion</h3>
<p>By implementing Tubearchivist, you not only enhance your YouTube experience but also take a significant step toward reclaiming your digital interactions. It’s about more than just watching videos—it&rsquo;s about creating a media consumption environment that respects your time and attention.</p>
<p>In a world where technology can often feel overwhelming, simplicity and clarity remain powerful allies. So let&rsquo;s take action—set up your Tubearchivist instance today and enjoy a more tailored, distraction-free viewing experience. You deserve it.</p>
]]></content:encoded>
    </item>
  </channel>
</rss>
