Linkerd
Linkerd keeps trading features for fewer operational surprises — 2.20 is tuning, not expansion.
A side-by-side editorial comparison of Refine and Supabase — release velocity, themes, recent moves, and the top alternatives to consider.
Refine's v5 majors are out; what's shipping now is the bug tail, mostly from contributors
Refine is a React framework for building internal tools and admin panels, and this window is entirely patch releases against the already-shipped v5 core and v8 MUI packages. The fixes cluster in three places: URL state parsing under syncWithLocation, access-control metadata not reaching button hooks, and CLI binary resolution inside monorepo workspaces.
Supabase is closing its most-exploited default: tables stop being public unless you say so.
The last month is dominated by tightening defaults rather than adding surface. New tables in the public schema will no longer be auto-exposed to the Data and GraphQL APIs, pg_graphql stops being enabled by default, and the OAuth token endpoint is being corrected to return 200 per OAuth 2.1. Alongside that, custom OAuth/OIDC providers landed for Auth and an RLS Tester entered preview.
Refine is a React framework for building internal tools and admin panels, and this window is entirely patch releases against the already-shipped v5 core and v8 MUI packages. The fixes cluster in three places: URL state parsing under syncWithLocation, access-control metadata not reaching button hooks, and CLI binary resolution inside monorepo workspaces.
The pattern is a framework absorbing the edge cases that only appear once a major is in real projects. Deeply nested conditional filters were being mangled because a parser depth limit was too shallow; useDeleteButton was dropping the meta prop that custom access rules depend on; refine dev broke under workspace hoisting. Notably, most of these carry different contributor names, several first-time — the maintenance load is being spread across the community rather than driven by a single team push.
Filters, access control, and router adapters have each produced multiple fixes in a single window, so expect continued patch releases in those areas; nothing in these entries points to a new major or a change in framework direction.
The last month is dominated by tightening defaults rather than adding surface. New tables in the public schema will no longer be auto-exposed to the Data and GraphQL APIs, pg_graphql stops being enabled by default, and the OAuth token endpoint is being corrected to return 200 per OAuth 2.1. Alongside that, custom OAuth/OIDC providers landed for Auth and an RLS Tester entered preview.
Supabase is paying down the security cost of its own convenience. Auto-exposing every public-schema table made the product fast to start with and easy to misconfigure; the fix is explicit Postgres grants, staged over six months from new projects in April to all existing projects on October 30. The RLS Tester is the same theme from the other side — the auto-exposure default only worked if row-level security was correct, and until now there was no way to check.
Expect the remaining rollout dates to hold and more of the developer experience to be rebuilt around explicit grants, with the RLS Tester graduating from preview as it becomes the tool people need to verify the new model. The October cutover for existing projects is the moment where this stops being an announcement and starts breaking things.
Other DevOps products tracked by Sparkpulse, ranked by recent ship velocity. Tap any card for the full editorial trajectory or compare directly with Refine.
Linkerd keeps trading features for fewer operational surprises — 2.20 is tuning, not expansion.
PlanetScale is giving Postgres users budgets, guardrails, and a read-only door for agents.
Vercel is building the durable-execution layer — and just learned its model catalog isn't its own.
Auth0 is rebuilding itself as the identity layer between B2B apps, their customers, and AI agents.
Sanity is spending this month on the surfaces around the editor, not the editor itself.
Prometheus has a new LTS baseline, and it tightened how credentials survive a redirect.
Other DevOps products tracked by Sparkpulse, ranked by recent ship velocity. Tap any card for the full editorial trajectory or compare directly with Supabase.
Expo shut down its own coding agent and kept the MCP server — it wants to be the substrate.
Clerk is building auth for machines: API keys, a CLI agents can drive, per-verification pricing.
Airbyte 2.0 made the pipeline run both ways: data activation turns the warehouse into a source.
Mixpanel is pushing analytics out of its own tab — into Postgres, Slack, and now Glean's assistant.
Groq is climbing off raw inference: hosted tools and MCP connectors now sit above the chips.
Windsurf is being rebuilt as Devin's front end — the IDE now ships the agent, the CLI, and the reviewer.
Latest ship moves from both products, interleaved chronologically. ⚡ = editorial spark.
They serve adjacent needs but don't currently overlap on shipped themes. Refine and Supabase are shipping at a similar cadence (velocity 0.0 vs 0.0, both within Sparkpulse's "active" band). See the at-a-glance table above for a side-by-side breakdown of velocity, recent sparks, and editorial themes.
Sparkpulse doesn't pick a winner — we score release velocity, not feature parity. Refine and Supabase are shipping at a similar cadence (velocity 0.0 vs 0.0, both within Sparkpulse's "active" band). For your specific use case, the alternatives sections above list other DevOps products to evaluate alongside.
Top Refine alternatives in DevOps are ranked by recent ship velocity. Browse the "Refine alternatives" section above for the current picks, or visit /alternatives/refine for the full list with editorial commentary on each.
Top Supabase alternatives in DevOps are ranked by recent ship velocity. Browse the "Supabase alternatives" section above for the current picks, or visit /alternatives/supabase for the full list with editorial commentary on each.