3.2.4

by
Updated

In this minor update, we bring you the most advanced description generation yet. The generator is now context-sensitive, so you can expect the descriptions to be even more natural; a real time-saver!

Oh, we also fixed some bugs, and we’ve implemented some improvements in preparation for version 3.3.0.

Detailed log

> View code changes.
> View closed issues.

For everyone:

  • Added:
    • Documentation links have been added to the plugin details overview row.
    • Auto-generated descriptions no longer use elements of these types for generation:
      • address, bdo, br, button, canvas, code, fieldset, form, h1, h2, h3, h4, h5, h6, header, hr, input, label, link, meta, nav, noscript, option, pre, samp, script, select, style, svg, table, textarea, var, video
    • Auto-generated descriptions now add spaces around elements of these types (they’re considered as blocks) for generation:
      • article, aside, blockquote, dd, div, dl, dt, figcaption, figure, footer, li, main, ol, p, section, tfoot, ul
  • Improved:
    • Google can now respond to the noindex directive quicker as we’ve removed the canonical URL on these pages.
      • When noindex is set via The SEO Framework, we now remove the autogenerated canonical URL when it points to the current page.
      • Filtered and custom canonical URLs will still be rendered.
    • SEO attack mitigation is now also invoked on a paginated homepage-as-page.
  • Changed:
    • The “About” link on the plugin activation page now leads to our “about us” page.
    • The links on the plugin activation page are now prepended, instead of appended.
    • When trailing slashes are omitted via the permalink settings, the home page URL now gets one regardless if it’s a document root URL.
      • This fixes a bug in Google Search Console, where they couldn’t recognize the URL in the sitemap.
  • Fixed:
    • Titles:
      • Author archives no longer use protective title prefixes (like “Private: “, or “Protected: “) from a post with the same ID as the author.
    • Accessibility:
      • The primary term selector in the Classic Editor now correctly checks the selected primary term button when interacting via a keyboard.
        • This bug only affected the interface; internally, the primary term was correctly selected.
      • Read this aloud: “Less than BR greater than generated: automatically generated”; Yes, that’s stupid, and it will no longer be enunciated by screen readers.
        • Essentially, we’ve cleanly stripped <br> and other meta tags from the SEO Bar and Pixel Counter items’ ARIA-labels.
    • URLs:
      • When the home page has query parameters attached–like, for example, via WPML, it won’t have a query-breaking slash added any longer.
      • When the home page is a page, and when <!--nextpage--> is used, the relationship links are now correct on the first page.
      • When a custom canonical URL is provided, no link relationship URLs will be generated.
        • For developers: The link relationship filters still work.
      • When pagination occurs on pages with query parameters, the query parameters are forwarded.
        • Keep in mind that most plugins don’t support proper pagination with query parameters, like WPML. This is beyond the capabilities of WordPress.
          • For WPML, please don’t use the ?lang={langid} (parameter) setting. Use a the directory or (sub-)domain options instead, they’re superior in behavior.
          • With WPML, paginated pages’ canonical URL when using the query parameter setting won’t be indexed. We’ve forwarded this issue to their developers.
        • Besides, if you require query parameters for your front-end plugin pages, then you’ve already broken WordPress. I urge to look into the WordPress Rewrite API.
      • Polylang’s home canonical and Open Graph URLs are now correct.
        • We did this by whitelisting our files in their plugin; an auto-incompatible implementation.
        • We couldn’t fix the sitemap URLs without breaking your site, because Polylang messes with the site URL too.
          • The home URL in the sitemap is still and will always be incorrect.
          • Other URLs don’t always have a trailing slash, while they should have them.
        • No, we’re never going to blindly mix the languages into one sitemap. Here’s the reason.

For translators:

  • Added:
    • A few sentences have been added, visible on the plugin overview page.
  • Changed:
    • We’ve changed various instances of “tags”, which should’ve been “meta tags”.
    • “Two parts” should’ve been “three parts”; we’ve decided to make it “multiple parts” to future-proof it.
    • All instances of “home page” have been changed to “homepage”, to be in line with WordPress’ glossary.
    • Removed unnecessary capitalization.
    • Various other sentences have been improved.
  • Updated:
    • Translation POT file.

For developers:

  • File changes:
    • Added:
      • ../inc/functions/upgrade-suggestion.php
    • Removed:
      • ../inc/functions/tsfem-suggestion.php
    • Changed:
      • Added namespace The_SEO_Framework\Bootstrap to:
        • ../bootstrap/activation.php
        • ../bootstrap/deactivation.php
        • ../bootstrap/upgrade.php
        • ../bootstrap/activation.php
  • Method notes:
    • Added:
      • strip_tags_cs(), a context-sensitive version of strip_tags().
      • get_robots_meta()
      • use_title_protection()
      • use_title_pagination()
      • has_custom_canonical_url()
      • remove_pagination_from_url(), this used to be protected.
    • Changed:
      • s_excerpt() now uses the new strip_tags_cs() method (with modified arguments), instead of WordPress’ wp_strip_all_tags().
      • page() and paged() now point to the correct page if the page number doesn’t exist, and they now always return 1 on the admin screens to prevent
        excessive processing and unpredictable behavior of methods that rely on these.
      • get_singular_canonical_url() now no longer tries to add pagination, as WordPress does this for us.
      • add_url_pagination() now considers query arguments with pretty permalinks, and the third parameter is now optional (auto-determine).
      • numpages() now always returns 1 on the admin screens.
      • is_multipage() now always returns false on the admin screens.
    • Fixed:
      • get_paged_urls() now returns the correct singular URLs.
        • In extent, get_paged_url() also works as intended for singular pages.
        • This was found after noticing an inconsistency with the homepage-as-page versus real pages.