<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/">
    <channel>
        <title>ELVIRA news</title>
        <link>https://elvira.digital/news</link>
        <description>Release announcements and project news from ELVIRA.</description>
        <lastBuildDate>Mon, 16 Nov 2026 00:00:00 GMT</lastBuildDate>
        <docs>https://validator.w3.org/feed/docs/rss2.html</docs>
        <generator>https://github.com/jpmonette/feed</generator>
        <language>en</language>
        <copyright>© 2026 EvilFlowers Catalog</copyright>
        <item>
            <title><![CDATA[Where to follow ELVIRA releases]]></title>
            <link>https://elvira.digital/news/announcing-releases</link>
            <guid>https://elvira.digital/news/announcing-releases</guid>
            <pubDate>Mon, 16 Nov 2026 00:00:00 GMT</pubDate>
            <description><![CDATA[Every release of every public ELVIRA repository is now listed on one page, and notable releases get a post here.]]></description>
            <content:encoded><![CDATA[<p>ELVIRA is made of many repositories, and their releases were scattered across
GitHub. From now on there are two places to follow them on this site.</p>
<ul>
<li class=""><strong><a class="" href="https://elvira.digital/releases">Releases</a></strong> lists every release of every public repository in
the <a href="https://github.com/EvilFlowersCatalog" target="_blank" rel="noopener noreferrer" class="">EvilFlowersCatalog</a> organization,
newest first, with a link to its notes. It is rebuilt from GitHub every day.</li>
<li class=""><strong>News</strong>, this blog, has a post for releases that change something for
readers, librarians or the people who run ELVIRA. Posts explain what changed
and what to do about it; the full notes stay on GitHub.</li>
</ul>
<p>The first posts are <a class="" href="https://elvira.digital/news/catalog-0.15.0">EvilFlowersCatalog 0.15.0</a>, the
<a class="" href="https://elvira.digital/news/calibre-plugin-2.1">Calibre plugin 2.1</a>, and an introduction to the
<a class="" href="https://elvira.digital/news/books-digitalizer">books digitalizer</a>, which has no release yet.</p>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="how-a-release-gets-here">How a release gets here<a href="https://elvira.digital/news/announcing-releases#how-a-release-gets-here" class="hash-link" aria-label="Direct link to How a release gets here" title="Direct link to How a release gets here" translate="no">​</a></h2>
<figure class="figure_hSKo"><a href="https://elvira.digital/diagrams/activity-release-flow.svg" target="_blank" rel="noopener noreferrer" class="frame_JUsk panel_nJ0A"><img src="https://elvira.digital/diagrams/activity-release-flow.svg" alt="Activity diagram: a maintainer publishes a GitHub release with tag X.Y.Z; CI builds what the tag ships; the site rebuild fetches the releases of every public repository and lists them on /releases; a notable release also gets a post on /news." loading="lazy"></a><figcaption class="caption_ON5J"><a href="https://github.com/EvilFlowersCatalog/EvilFlowersCatalog.github.io/blob/main/diagrams/activity-release-flow.puml" target="_blank" rel="noopener noreferrer">PlantUML source</a></figcaption></figure>
<p>A release is a GitHub release with a version tag. Nothing else needs to happen
for it to appear on the releases page. The site picks it up at the next daily
rebuild, or sooner if the repository asks for a rebuild.</p>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="which-repositories-release-today">Which repositories release today<a href="https://elvira.digital/news/announcing-releases#which-repositories-release-today" class="hash-link" aria-label="Direct link to Which repositories release today" title="Direct link to Which repositories release today" translate="no">​</a></h2>
<p>Four repositories publish GitHub releases: the
<a href="https://github.com/EvilFlowersCatalog/EvilFlowersCatalog/releases" target="_blank" rel="noopener noreferrer" class="">catalog</a>,
the <a href="https://github.com/EvilFlowersCatalog/EvilFlowersViewer/releases" target="_blank" rel="noopener noreferrer" class="">PDF reader</a>,
the <a href="https://github.com/EvilFlowersCatalog/calibre-evilflowers/releases" target="_blank" rel="noopener noreferrer" class="">Calibre plugin</a>
and the <a href="https://github.com/EvilFlowersCatalog/evilflowers-postgresql/releases" target="_blank" rel="noopener noreferrer" class="">PostgreSQL image</a>.
The portal, the digitalizer and the workers do not tag releases yet. The
<a class="" href="https://elvira.digital/docs/releases">releases guide</a> describes how we intend to tag them.</p>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="subscribe">Subscribe<a href="https://elvira.digital/news/announcing-releases#subscribe" class="hash-link" aria-label="Direct link to Subscribe" title="Direct link to Subscribe" translate="no">​</a></h2>
<ul>
<li class="">This blog has an <a href="https://elvira.digital/news/rss.xml" target="_blank" rel="noopener noreferrer" class="">RSS feed</a> and an
<a href="https://elvira.digital/news/atom.xml" target="_blank" rel="noopener noreferrer" class="">Atom feed</a>.</li>
<li class="">On GitHub, <em>Watch → Custom → Releases</em> on a repository notifies you of each of
its releases.</li>
</ul>]]></content:encoded>
            <category>project</category>
        </item>
        <item>
            <title><![CDATA[Introducing the books digitalizer]]></title>
            <link>https://elvira.digital/news/books-digitalizer</link>
            <guid>https://elvira.digital/news/books-digitalizer</guid>
            <pubDate>Mon, 16 Nov 2026 00:00:00 GMT</pubDate>
            <description><![CDATA[A pipeline and a web studio that turn scans of printed books into searchable PDFs, PDF/A masters and EPUBs, and publish them to the catalog.]]></description>
            <content:encoded><![CDATA[<p><a href="https://github.com/EvilFlowersCatalog/evilflowers-books-digitalizer" target="_blank" rel="noopener noreferrer" class="">evilflowers-books-digitalizer</a>
turns scans of printed books into files a reader can search and an archive can
keep, and publishes them to the catalog. It has no tagged release yet; the
code and its
<a href="https://evilflowerscatalog.github.io/evilflowers-books-digitalizer/" target="_blank" rel="noopener noreferrer" class="">documentation</a>
are public.</p>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="why-a-studio-and-not-just-a-batch-job">Why a studio and not just a batch job<a href="https://elvira.digital/news/books-digitalizer#why-a-studio-and-not-just-a-batch-job" class="hash-link" aria-label="Direct link to Why a studio and not just a batch job" title="Direct link to Why a studio and not just a batch job" translate="no">​</a></h2>
<p>The project started as an unattended batch pipeline: scans in, PDFs out. On a
set of 880 scanned university books it gets most pages right and some badly
wrong. A stamp covers a paragraph, a formula is read as prose, a crop loses a
column. A batch job can only rerun everything. The studio lets a person fix one
region on one page, rerun that page in seconds, and approve the book.</p>
<figure class="figure_hSKo"><a href="https://elvira.digital/diagrams/activity-digitalizer.svg" target="_blank" rel="noopener noreferrer" class="frame_JUsk panel_nJ0A"><img src="https://elvira.digital/diagrams/activity-digitalizer.svg" alt="Activity diagram: TIFF scans are split and cleaned with ScanTailor, OCR'd with Tesseract and analysed for layout and formulas. A quality gate approves the book automatically or sends it to a review queue, where an operator fixes regions and re-runs pages. Then the PDF, PDF/A-2b master and EPUB are rendered and the book is published to the catalog." loading="lazy"></a><figcaption class="caption_ON5J"><a href="https://github.com/EvilFlowersCatalog/EvilFlowersCatalog.github.io/blob/main/diagrams/activity-digitalizer.puml" target="_blank" rel="noopener noreferrer">PlantUML source</a></figcaption></figure>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="what-it-does-with-each-book">What it does with each book<a href="https://elvira.digital/news/books-digitalizer#what-it-does-with-each-book" class="hash-link" aria-label="Direct link to What it does with each book" title="Direct link to What it does with each book" translate="no">​</a></h2>
<ol>
<li class="">Splits two-page spreads, then crops, deskews and cleans each page with
ScanTailor.</li>
<li class="">Runs Slovak OCR with Tesseract and finds layout regions: text, headings,
pictures, captions, formulas.</li>
<li class="">Scores the book: OCR confidence, regions the detectors disagree on, formulas
it could not read, missing pages. A book that scores well is approved; the
rest wait in a review queue.</li>
<li class="">Renders three files: a distribution PDF that opens fast in any viewer, a
PDF/A-2b master for preservation, and a reflowable EPUB.</li>
<li class="">Publishes the book to the catalog through its REST API. Publishing again
updates the same entry.</li>
</ol>
<p>The output folder of every book is also a valid Internet Archive item, so the
standard Internet Archive tools work on it directly.</p>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="the-review-screens">The review screens<a href="https://elvira.digital/news/books-digitalizer#the-review-screens" class="hash-link" aria-label="Direct link to The review screens" title="Direct link to The review screens" translate="no">​</a></h2>
<p>Only books the gate flagged need a person. The review queue lists them with the
reason, lowest score first.</p>
<p><span class="figure_hSKo"><a href="https://elvira.digital/assets/images/review-queue-566429fecfb72a5354f876e5039a898c.png" target="_blank" rel="noopener noreferrer" class="frame_JUsk"><img src="https://elvira.digital/assets/images/review-queue-566429fecfb72a5354f876e5039a898c.png" alt="Review queue: one flagged book with a quality score of 0.48 and the reason, 17 disputed regions and 54 of 55 formulas left as images" loading="lazy" decoding="async" width="1440" height="232"></a><span class="caption_ON5J">The keyboard works too: j and k move through the queue, a approves the selected book.</span></span></p>
<p>In the page studio, regions are drawn over the cleaned page, coloured by kind.
Where two detectors disagree, for example formula against text, the operator
decides with one click and can re-run OCR on that region.</p>
<p><span class="figure_hSKo"><a href="https://elvira.digital/assets/images/page-studio-86ebe2d2ddc5aad66431580346007fbd.png" target="_blank" rel="noopener noreferrer" class="frame_JUsk"><img src="https://elvira.digital/assets/images/page-studio-86ebe2d2ddc5aad66431580346007fbd.png" alt="Page studio: a page from a Slovak physics textbook with text, heading, picture and formula regions outlined, and a disputed formula region waiting for a decision" loading="lazy" decoding="async" width="1440" height="900"></a><span class="caption_ON5J">Blue marks text, purple headings, pink pictures, teal captions and orange formulas.</span></span></p>
<p>A swipe compares the original scan with the cleaned page, so a bad crop or a
lost column is easy to spot.</p>
<p><span class="figure_hSKo"><a href="https://elvira.digital/assets/images/compare-swipe-97d3eab3dc98e2f99faed9c9c5c6c742.png" target="_blank" rel="noopener noreferrer" class="frame_JUsk"><img src="https://elvira.digital/assets/images/compare-swipe-97d3eab3dc98e2f99faed9c9c5c6c742.png" alt="Swipe compare: the original scan on the left and the cleaned page with its regions on the right, divided by a slider" loading="lazy" decoding="async" width="1440" height="900"></a><span class="caption_ON5J">Drag the slider to move the line between the scan and the cleaned page.</span></span></p>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="status">Status<a href="https://elvira.digital/news/books-digitalizer#status" class="hash-link" aria-label="Direct link to Status" title="Direct link to Status" translate="no">​</a></h2>
<p>The digitalizer has not been released yet. When it has a first release, it will
be announced here and listed on the <a class="" href="https://elvira.digital/releases">releases page</a>. Until then, the
<a href="https://github.com/EvilFlowersCatalog/evilflowers-books-digitalizer" target="_blank" rel="noopener noreferrer" class="">repository</a>
has setup instructions for the studio and for the batch command line.</p>]]></content:encoded>
            <category>project</category>
            <category>digitalizer</category>
        </item>
        <item>
            <title><![CDATA[Sync a Calibre library with ELVIRA]]></title>
            <link>https://elvira.digital/news/calibre-plugin-2.1</link>
            <guid>https://elvira.digital/news/calibre-plugin-2.1</guid>
            <pubDate>Mon, 16 Nov 2026 00:00:00 GMT</pubDate>
            <description><![CDATA[calibre-evilflowers 2.1.0 keeps a Calibre library and an EvilFlowers catalog in sync, in both directions, and never deletes a book.]]></description>
            <content:encoded><![CDATA[<p><a href="https://github.com/EvilFlowersCatalog/calibre-evilflowers/releases/tag/2.1.0" target="_blank" rel="noopener noreferrer" class="">calibre-evilflowers 2.1.0</a>
is a <a href="https://calibre-ebook.com/" target="_blank" rel="noopener noreferrer" class="">Calibre</a> plugin that keeps a Calibre library
in sync with a catalog on an EvilFlowers server, such as ELVIRA. Books new in
Calibre are uploaded, books new in the catalog are downloaded, and a change on
one side is applied to the other. Point several computers at the same catalog
and they share one library.</p>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="before-anything-changes-you-see-the-plan">Before anything changes, you see the plan<a href="https://elvira.digital/news/calibre-plugin-2.1#before-anything-changes-you-see-the-plan" class="hash-link" aria-label="Direct link to Before anything changes, you see the plan" title="Direct link to Before anything changes, you see the plan" translate="no">​</a></h2>
<p>Clicking the toolbar button compares the library with the catalog and lists
every book with what the plugin will do: upload, download or skip. You can
change any row. When a book was edited on both sides, the plugin does not guess:
it shows the two versions field by field and you pick one.</p>
<figure class="figure_hSKo"><a href="https://elvira.digital/assets/images/sync-plan-conflict-light-8163dd2706644c831269599d5d79f288.png" target="_blank" rel="noopener noreferrer" class="frame_JUsk"><img src="https://elvira.digital/assets/images/sync-plan-conflict-light-8163dd2706644c831269599d5d79f288.png" alt="The sync window: seven books grouped as to upload, to download, needs a decision and in sync, with a side-by-side comparison of a book changed in both Calibre and the catalog" class="themedComponent_mlkZ themedComponent--light_NVdE" loading="lazy" decoding="async"></a><figcaption class="caption_ON5J"><p>Each row is coloured by what will happen to the book. The panel on the right
compares the two versions of a book that changed on both sides.</p></figcaption></figure>
<p>What travels: title, authors, language, tags (as categories), comments (as the
summary), publisher, publication date, identifiers such as ISBN and DOI, the
cover, and EPUB, PDF and MOBI files. Series, ratings and custom columns stay in
Calibre and a download never overwrites them.</p>
<p><strong>Nothing is ever deleted, on either side.</strong> A book removed in Calibre is not
downloaded again as new, and it is not removed from the catalog either. The
plugin only ever removes files it uploaded itself, when it replaces them with a
newer version.</p>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="browse-the-catalog-keep-your-shelf-read-your-highlights">Browse the catalog, keep your shelf, read your highlights<a href="https://elvira.digital/news/calibre-plugin-2.1#browse-the-catalog-keep-your-shelf-read-your-highlights" class="hash-link" aria-label="Direct link to Browse the catalog, keep your shelf, read your highlights" title="Direct link to Browse the catalog, keep your shelf, read your highlights" translate="no">​</a></h2>
<p>You do not have to sync a whole catalog. <strong>Browse catalog</strong> searches it, filters
by language and category, and downloads only the books you pick. Badges show
which books are already in your library or on your EvilFlowers shelf.</p>
<p><span class="figure_hSKo"><a href="https://elvira.digital/assets/images/browse-results-light-4c8aee8450a5d37a4e1c6447a968c3c4.png" target="_blank" rel="noopener noreferrer" class="frame_JUsk"><img src="https://elvira.digital/assets/images/browse-results-light-4c8aee8450a5d37a4e1c6447a968c3c4.png" alt="The catalog browser: search box, language and category filters, and a list of books with &quot;In your library&quot; and &quot;On your shelf&quot; badges" loading="lazy" decoding="async" width="900" height="650"></a></span></p>
<p><strong>Highlights for selected book</strong> shows the passages you highlighted while
reading in the ELVIRA reader, with page numbers. Copy them, or save them into a
Calibre column. Nothing is sent back to the catalog.</p>
<p><span class="figure_hSKo"><a href="https://elvira.digital/assets/images/highlights-light-f288ee4a2fea17b91084fae562ad6074.png" target="_blank" rel="noopener noreferrer" class="frame_JUsk"><img src="https://elvira.digital/assets/images/highlights-light-f288ee4a2fea17b91084fae562ad6074.png" alt="The highlights window: highlighted passages grouped by book, each with its page number" loading="lazy" decoding="async" width="900" height="650"></a></span></p>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="signing-in">Signing in<a href="https://elvira.digital/news/calibre-plugin-2.1#signing-in" class="hash-link" aria-label="Direct link to Signing in" title="Direct link to Signing in" translate="no">​</a></h2>
<p>You sign in with your EvilFlowers username and password once. The plugin uses
them to create an API key named <em>Calibre on &lt;computer&gt;</em> and keeps only
that key. You can see and revoke the key in your profile, and <strong>Sign out</strong>
revokes it for you.</p>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="install">Install<a href="https://elvira.digital/news/calibre-plugin-2.1#install" class="hash-link" aria-label="Direct link to Install" title="Direct link to Install" translate="no">​</a></h2>
<ol>
<li class="">Download <code>evilflowers-2.1.0.zip</code> from the
<a href="https://github.com/EvilFlowersCatalog/calibre-evilflowers/releases/tag/2.1.0" target="_blank" rel="noopener noreferrer" class="">release</a>.
You need Calibre 7 or newer.</li>
<li class="">In Calibre, open <strong>Preferences → Plugins → Load plugin from file</strong> and pick
the ZIP. Add the <strong>EvilFlowers</strong> button to a toolbar.</li>
<li class="">Open <strong>Settings…</strong> from the button's menu, enter your library's address (for
STU, <code>https://elvira.stuba.sk</code>), sign in and pick a catalog.</li>
</ol>
<p>The <a href="https://github.com/EvilFlowersCatalog/calibre-evilflowers#readme" target="_blank" rel="noopener noreferrer" class="">README</a>
covers the settings, how changes are detected, and the known limitations.</p>]]></content:encoded>
            <category>release</category>
            <category>calibre</category>
        </item>
        <item>
            <title><![CDATA[EvilFlowersCatalog 0.15.0]]></title>
            <link>https://elvira.digital/news/catalog-0.15.0</link>
            <guid>https://elvira.digital/news/catalog-0.15.0</guid>
            <pubDate>Mon, 16 Nov 2026 00:00:00 GMT</pubDate>
            <description><![CDATA[Search rebuilt inside PostgreSQL, ratings and reviews, lending statistics, an assistant that cites pages, and list endpoints that no longer slow down with page size.]]></description>
            <content:encoded><![CDATA[<p><a href="https://github.com/EvilFlowersCatalog/EvilFlowersCatalog/releases/tag/0.15.0" target="_blank" rel="noopener noreferrer" class="">EvilFlowersCatalog 0.15.0</a>
was released on 12 November 2026. Search now runs inside the catalog's own
PostgreSQL, and the catalog can search inside its publications and answer
questions about them with page citations. After a round of load testing, the
list endpoints also run far fewer database queries.</p>
<p>This post summarises the release for the people who use and run the catalog.
The <a href="https://github.com/EvilFlowersCatalog/EvilFlowersCatalog/releases/tag/0.15.0" target="_blank" rel="noopener noreferrer" class="">full release notes</a>
list every change with its issue.</p>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="for-readers">For readers<a href="https://elvira.digital/news/catalog-0.15.0#for-readers" class="hash-link" aria-label="Direct link to For readers" title="Direct link to For readers" translate="no">​</a></h2>
<ul>
<li class=""><strong>Better search.</strong> Full-text search with Slovak and Czech stemming, so a
query also finds other forms of the same word. Hits are highlighted, a
misspelled query gets a "did you mean", and a new endpoint suggests titles as
you type (<code>GET /api/v1/entries/suggest</code>).</li>
<li class=""><strong>Facets with exact counts</strong> for filtering results, in the API and in OPDS
1.2 and 2.0 search feeds, so e-reader apps get them too.</li>
<li class=""><strong>Ratings and reviews</strong> on publications, with moderation.</li>
<li class=""><strong>Reading history</strong> of what you opened.</li>
<li class=""><strong>Category and feed titles</strong> in English and Slovak.</li>
</ul>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="for-librarians">For librarians<a href="https://elvira.digital/news/catalog-0.15.0#for-librarians" class="hash-link" aria-label="Direct link to For librarians" title="Direct link to For librarians" translate="no">​</a></h2>
<ul>
<li class=""><strong>Lending statistics</strong> for the admin dashboard.</li>
<li class=""><strong>Branded notification e-mails</strong>, and e-mail addresses imported from LDAP.</li>
<li class=""><strong>Publishing from AI agents.</strong> The MCP server gained tools to create
publications and upload their files, and to manage users and API keys.</li>
<li class="">Readers can <strong>change their own password</strong>.</li>
<li class="">Fixes: one-week loans can be renewed, returning a loan that was never opened
sends its e-mail, and empty uploads are refused.</li>
</ul>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="search-by-meaning-extraction-and-the-assistant">Search by meaning, extraction and the assistant<a href="https://elvira.digital/news/catalog-0.15.0#search-by-meaning-extraction-and-the-assistant" class="hash-link" aria-label="Direct link to Search by meaning, extraction and the assistant" title="Direct link to Search by meaning, extraction and the assistant" translate="no">​</a></h2>
<p>These parts are in the release. Each works only once the service behind it is
configured.</p>
<ul>
<li class=""><strong>Ranking by meaning.</strong> With an embedding service configured, search combines
word ranking (BM25) with semantic ranking. Without one, it ranks on words only.</li>
<li class=""><strong>Extraction</strong> is off by default (<code>EVILFLOWERS_EXTRACTION_ENABLED</code>). The
catalog stores paragraphs, figures and their embeddings
produced by the <a class="" href="https://elvira.digital/docs/architecture#processing">extraction workers</a>. That enables
search inside publications (<code>GET /api/v1/entries/passages</code>), a publication's
figures, and similar publications.</li>
<li class=""><strong>The assistant</strong> answers questions from the library's passages, labels how
strongly each passage supports the answer, and links to the pages it used. In
the PDF reader it explains a selected passage or figure. It talks to a model
through the Ollama API, so you can host the model yourself. Without a model
endpoint configured, it cannot answer.</li>
<li class=""><strong>Titles lent with Readium LCP are never read by AI.</strong></li>
</ul>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="for-people-who-run-it">For people who run it<a href="https://elvira.digital/news/catalog-0.15.0#for-people-who-run-it" class="hash-link" aria-label="Direct link to For people who run it" title="Direct link to For people who run it" translate="no">​</a></h2>
<ul>
<li class=""><strong>List endpoints cost the same number of queries for any page size.</strong>
<code>GET /api/v1/entries?limit=30</code> used to run 226 SELECTs.</li>
<li class="">The signing key is loaded once per process instead of on every request, which
had cost about a third of the CPU.</li>
<li class="">Covers and thumbnails have versioned URLs, so browsers and proxies cache them
and a repeat request answers 304 without touching storage.</li>
<li class="">Downloads support HTTP range requests, so large files stream.</li>
<li class="">OpenTelemetry traces, metrics and JSON logs, and optional profiling.</li>
<li class="">Optional connection pooling, PgBouncer transaction pooling, and Celery
settings for autoscaled workers.</li>
<li class=""><code>manage.py backup</code> works with PostgreSQL 18 and fails loudly when the dump
fails.</li>
</ul>
<h2 class="anchor anchorTargetStickyNavbar_Vzrq" id="before-you-upgrade">Before you upgrade<a href="https://elvira.digital/news/catalog-0.15.0#before-you-upgrade" class="hash-link" aria-label="Direct link to Before you upgrade" title="Direct link to Before you upgrade" translate="no">​</a></h2>
<p>Read the <a href="https://github.com/EvilFlowersCatalog/EvilFlowersCatalog/releases/tag/0.15.0" target="_blank" rel="noopener noreferrer" class="">upgrade notes</a>
at the end of the release. In short:</p>
<ul>
<li class="">PostgreSQL needs the <code>vector</code> and <code>pg_trgm</code> extensions. <code>pg_search</code> is
optional and enables BM25 ranking. Our
<a href="https://github.com/EvilFlowersCatalog/evilflowers-postgresql" target="_blank" rel="noopener noreferrer" class="">PostgreSQL image</a>
has all three.</li>
<li class="">CORS is no longer open to every origin. Set <code>CORS_ALLOWED_ORIGINS</code>, or rely on
the default, which allows the origin of <code>EVILFLOWERS_PORTAL_URL</code>.</li>
<li class="">The MCP server accepts only Bearer API keys by default. Password clients need
<code>EVILFLOWERS_MCP_AUTHENTICATION_SCHEMAS=Bearer,Basic</code>.</li>
<li class="">The integration with the separate search service is removed. Search runs in
the catalog's database, and <code>evilflowers-search-service</code> is archived.</li>
<li class="">After migrating, run <code>manage.py embed_entries --missing</code> if you configured an
embedding service.</li>
</ul>]]></content:encoded>
            <category>release</category>
            <category>catalog</category>
        </item>
    </channel>
</rss>