Resumo Diário (5h UTC)

Notícias e Eventos Resumos do fórum
834

Today on meta.discourse.org…

Key Stats

In the last 24 hours (2026-05-14 → 2026-05-15):

  • New posts: 53
  • New topics: 10

Top users (posts / likes in the last 24h):

User Posts Likes
EricGT 8 8
Moin 6 8
Lilly 6 7
ted 4 4
putty 3 3
RGJ 2 6
martin 2 3
gilles 2 2
ToddZ 1 2
noahl 1 2

Most active @team members (posts / likes in the last 24h):

Interesting Topics

#customization:theme-component

  • EricGT highlighted how nested replies can reintroduce “wrong reply button” confusion, arguing the plain Reply label isn’t specific enough and should better distinguish replying-to-topic vs replying-to-post (read more). The discussion explicitly ties back to nested replies UX concerns raised in the nested replies announcement thread (read more, read more), and Lilly quickly shipped improvements—first enabling customization of both reply button labels (read more), then adding the floating topic reply button into nested topic views (read more).

  • A new component aims to remove the “X votes left” line when topic-voting is configured with effectively “unlimited” votes (read more). NateDhaliwal suggested conditional visibility when a user is close to running out (read more), and awesomerobot noted a newly-added site setting—Topic voting enable vote limits—that can eliminate the need for absurdly high vote caps in the first place (read more). Related reference points in the ecosystem include the official topic-voting documentation topic (read more).

#announcements

  • The newly-improved “members can indicate they’re experiencing a reported issue” flow sparked a semantics-and-UX debate about whether the “Me too” counter should start at 1 (including the OP) or 0 (only “additional” reporters) (read more, read more). lindsey clarified the intent: the count represents “how many people are impacted,” hence starting at 1 to reflect the author’s impact (read more), while RGJ argued the label should change if the number includes the author (read more). The discussion also referenced how this looks “in the wild,” pointing to an API-related support thread where the button appeared to imply someone else was affected, but was just the OP (read more).

  • The Horizon theme now has high context topic cards enabled by default, expanding the information shown in topic lists and leaning into at-a-glance signals like solved, topic-voting, assign, tags, and excerpts (read more). This builds directly on the earlier “Horizon high context topic cards” introduction (read more) and references core/official feature docs for Solved (read more), Topic Voting (read more), and Assign (read more).

  • “Upcoming Changes” got a small but impactful UX report: when admins land on a filtered Upcoming Changes URL from a notification, the Preview link may be missing—despite being present when you browse the page normally (read more). martin explained that “Feedback…” and “Preview” are optional metadata per change item and offered to ping internally so more changes include them (read more).

#feature

  • A request proposes suppressing the “New features available!” notification on freshly created forums, arguing that on a brand-new instance everything is new and the alert adds noise rather than value (read more). The suggestion echoes the broader principle of reducing early-admin cognitive load (and references prior precedent for suppressing some “upcoming changes” messaging on new sites, via the linked GitHub PR from within the topic: read more).

  • Another branding/UX request asks for a separate mobile_logo_dark (or login_mobile_logo_dark) upload specifically for the login/signup splash screen, so admins can optimize the unauthenticated first impression without compromising the logged-in mobile header logo (read more).

bug

  • A composer interaction bug report shows that when the cursor sits just above a horizontal rule, pressing ↓ while the emoji picker is open can cause the horizontal rule to steal focus and close the picker (read more). ted also linked a proposed fix PR for Discourse core right from the report (read more).

  • The locale detector for ai content localization hit a mismatch: Norwegian is detected as no, but supported locales use nb_NO. nat confirmed that adding explicit mappings to the system prompt is the right direction and shared an upcoming improvement—pulling in the “supported languages” setting so other communities avoid similar issues (read more).

ux

  • A small but visible UI polish issue: user profile bios can be slightly clipped on the right edge (especially noticeable with italicized text), suggesting padding/overflow adjustments may be needed (read more). The report usefully references an older analysis of italics overflowing their container (quoted in the topic) to help narrow the likely CSS cause (read more).

  • The edit bookmark dialog UI got scrutiny: ted questioned whether a particular button is necessary at all, and noted that its icon and placement are misleading/confusing in its current form (read more).

  • A proposal suggests showing uneditable user fields in profile preferences as disabled inputs instead of hiding them, to better match user expectations and reduce confusion—while acknowledging moderation “staff notes” style use-cases might require configurability (read more).

  • Meta’s experimentation with “simple mode” for doc categories raised two UX questions: how to get the TOC back after clicking “timeline,” and why the toggle translations button is only visible while the timeline is open—especially when testing doc-categories, disco-toc, and content-localization together (read more). A related “how do I set the same sidebar categories for all users?” thread also saw continued troubleshooting around default vs user-customized sidebar state, relevant to admin predictability expectations (read more).

#contribute:site-feedback

  • “Simple mode” appearing in Meta’s Wiki doc categories sparked feedback about contributor recognition: Moin argued that showing authorship matters more in the community wiki than in official documentation, and linked several historical threads about docs/wiki discoverability and contribution pathways (read more; related: read more, read more).

#enterprise

  • A scaling story underscored that performance failures can come from information architecture rather than media volume—especially when category trees explode into “sub-sub-subcategories.” jpishgar described a large community buckling under category depth, and HAWK added a real-world customer migration anecdote where over-deep categorization became a dealbreaker (and reinforced “tags over endless nesting”) (read more).

#customization:plugin

  • The official Discourse Subscriptions plugin (subscriptions payments) got a high-signal edge-case bug report: in certain flows, the plugin can create a Stripe PaymentIntent without attaching a valid payment method (especially when an expired card exists), leading to silent subscription failures for end users (read more). Related plugin areas also saw activity elsewhere on Meta, including Discourse Workflow’s new visual editor PR announcement (read more) and an OpenID Connect corner case involving disallowed usernames ending with underscores (read more).

Activity by the @team Group

  • martin posted an announcement that high context topic cards are now enabled by default in the Horizon theme, pointing admins to the upcoming change toggle and recapping what the new cards surface (read more). He also followed up in the “Upcoming Changes” system thread to clarify that “Feedback…” and “Preview” links are optional per change item—and offered to prompt internal improvements to metadata coverage (read more).

  • awesomerobot replied to the “Hide Votes Left” component thread with timely product news: a new topic-voting site setting (Topic voting enable vote limits) can remove vote limits entirely without the “set it to 100,000” workaround—while noting the component might still be useful for mixed trust-level configurations (read more).

  • HAWK added an enterprise perspective to the scaling/IA discussion, describing a prospective customer with extreme geographic sub-categorization, Discourse’s preference for tags in such cases, and why the team declined to support an unsustainable structure (read more).

  • nat confirmed the fix direction for Norwegian locale handling in the locale detector, validating the “common language codes” mapping approach and committing to improve the generic system prompt by using the site’s supported languages setting (read more).

  • lindsey clarified the intent behind the new “Me too” counter behavior: it starts at 1 to represent the topic author as impacted, framing the counter as “people impacted” rather than “people who clicked” (read more).

If today’s Meta was any indication, tomorrow’s Discourse admins will have clearer Reply buttons, fewer “100,000 votes left” humblebrags, and emoji pickers that (hopefully) won’t get bullied by horizontal rules.