chore(deps): update mix dependencies #702

Open
renovate wants to merge 1 commit from renovate/mix-dependencies into main
Collaborator

This PR contains the following updates:

Package Type Update Change
ash_admin (source) prod minor 1.4.0 → 1.5.0
hammer (source) prod minor 7.5.1 → 7.6.0
phoenix (source) prod patch 1.8.16 → 1.8.17

Release Notes

ash-project/ash_admin (ash_admin)

v1.5.0

Compare Source

Features:
  • show what a create or update action returns besides the record by Fredrik Teschke

  • choose the fields shown for each relationship on the show page by Etienne Lacoursiere

  • hide the table search with the filters by Etienne Lacoursiere

  • searchable table columns via table_searchable_columns by Etienne Lacoursiere

  • add host_app_path option to link back to the host app by Etienne Lacoursiere

Bug Fixes:
  • only let the actor plug grant tenant controls by Paul Schoenfelder

  • only let the actor plug grant actor and authorization controls by Paul Schoenfelder

  • send no metadata message when an action returns none by Fredrik Teschke

  • register Cinder's colocated hooks in the admin bundle by azmaveth

  • spark formatter entry and Tailwind scan of the theme by Etienne Lacoursiere

  • give phx-change forms an id by Fredrik Teschke

ExHammer/hammer (hammer)

v7.6.0

Compare Source

  • Fix LeakyBucket (ETS and Atomic backends) and TokenBucket (Atomic backend) returning a flat {:deny, 1000}; they now return the real time in milliseconds until the hit would be allowed, as the ETS TokenBucket has since 7.5.0. The leaky buckets count in whole seconds, so their value runs to the second boundary at which the bucket has leaked enough. Behaviour change: the value is no longer always 1000. It can be much larger (e.g. up to 144000 for a bucket 144 over capacity at leak_rate: 1) or smaller (down to 1 when the next second is about to start). Code that converts it to a retry-after header should round up and floor at 1 second. A rate of 0 still answers {:deny, 1000}. (#​221)
  • Fix LeakyBucket (ETS and Atomic backends) allowing a hit whenever the level was below capacity, whatever its cost, so a bucket could be filled to capacity - 1 + cost. A hit is now allowed only if the level after it stays within capacity, as the moduledoc always said and as the Mnesia backend does. Behaviour change: hits with cost > 1 that used to overfill the bucket are now denied; hits with cost: 1 behave as before. A cost larger than capacity is never allowed. (#​222)
  • use Hammer, backend: :atomic, algorithm: :sliding_window now raises an ArgumentError saying that sliding window is not available for the Atomic backend and pointing to the ETS backend. It already failed at compile time, but with an unhelpful "could not load module" error. The error for an unknown Atomic algorithm now says "algorithm" rather than "backend" and no longer lists :sliding_window. (#​222)
  • Fix Atomic FixWindowPerKey hanging on a key forever. Resetting an expired window took a lock in the key's atomic; a process killed while holding it (for example a request process shut down mid-call) left the lock behind, and every later hit or inc on that key spun on it without returning. An expired window is now replaced by a fresh atomic, so there is no lock. A lock left behind by 7.5.1 or earlier is treated as an expired window. (#​222)
  • Fix TokenBucket ETS backend letting far more than refill_rate through. The refill clock advanced by a whole number of milliseconds, rounded down, so part of each elapsed stretch was credited again on the next hit. At refill_rate: 600 a caller hitting in a tight loop got about 1000 hits per second, and above 1000 tokens per second, where a token is worth less than a millisecond, the clock did not move at all and the limiter was effectively off (over a million hits in half a second at refill_rate: 1500, capacity: 10). Present since 7.5.0. The clock now advances by the exact time credited. (#​222)
  • Fix Atomic TokenBucket refilling in whole seconds, which capped sustained throughput at capacity tokens per second when refill_rate > capacity and delivered tokens in a lump at each second boundary. It now refills by the millisecond and keeps fractions of a token, so its {:deny, ms} value is exact to the millisecond (e.g. 19 at refill_rate: 55). Buckets held in memory across a hot code upgrade start over full. (#​222)
  • Fix LeakyBucket (ETS and Atomic backends) raising the level, and TokenBucket ETS backend taking tokens away, when the stored clock was ahead of the caller's. That happens when a concurrent hit reads the time later or the wall clock steps back; the negative elapsed time was applied as a negative leak or refill. It now counts as no time elapsed, and a TokenBucket (both backends) that denies in that state includes the time until the stored clock is reached in its {:deny, ms} value, since nothing refills before then. (#​222)
phoenixframework/phoenix (phoenix)

v1.8.17

Compare Source

Security fixes
  • CVE-2026-129344: Sending a channel message that cannot be serialized can lead to a denial of service for all clients on a shared topic
  • CVE-2026-128361: Unsanitized presence key crashes Presence tracker
Bug fixes
  • [Phoenix.PubSub] an Endpoint broadcast now works even when the endpoint is not started (#​6846)
  • [phoenix.js] Send the fallback health check before a buffered join (#​6770)
Enhancements
  • [phoenix.js] Allow enabling reconnections while a tab is hidden with reconnectWhenHidden (#​6883)
  • [Channels] send fastlane messages as shared binaries (#​6884)
  • Store verified route warnings relative to cwd (#​6889)
  • Support binary messages in LongPoll transport

Configuration

📅 Schedule: (UTC)

  • Branch creation
    • Every minute (* * * * *)
  • Automerge
    • At any time (no schedule defined)

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.

👻 Immortal: This PR will be recreated if closed unmerged. Get config help if that's undesired.


  • If you want to rebase/retry this PR, check this box

This PR has been generated by Mend Renovate.

This PR contains the following updates: | Package | Type | Update | Change | |---|---|---|---| | [ash_admin](https://hex.pm/packages/ash_admin) ([source](https://github.com/ash-project/ash_admin)) | prod | minor | `1.4.0` → `1.5.0` | | [hammer](https://hex.pm/packages/hammer) ([source](https://github.com/ExHammer/hammer)) | prod | minor | `7.5.1` → `7.6.0` | | [phoenix](https://hex.pm/packages/phoenix) ([source](https://github.com/phoenixframework/phoenix)) | prod | patch | `1.8.16` → `1.8.17` | --- ### Release Notes <details> <summary>ash-project/ash_admin (ash_admin)</summary> ### [`v1.5.0`](https://github.com/ash-project/ash_admin/blob/HEAD/CHANGELOG.md#v150-2026-11-20) [Compare Source](https://github.com/ash-project/ash_admin/compare/v1.4.0...v1.5.0) ##### Features: - show what a create or update action returns besides the record by Fredrik Teschke - choose the fields shown for each relationship on the show page by Etienne Lacoursiere - hide the table search with the filters by Etienne Lacoursiere - searchable table columns via table\_searchable\_columns by Etienne Lacoursiere - add host\_app\_path option to link back to the host app by Etienne Lacoursiere ##### Bug Fixes: - only let the actor plug grant tenant controls by Paul Schoenfelder - only let the actor plug grant actor and authorization controls by Paul Schoenfelder - send no metadata message when an action returns none by Fredrik Teschke - register Cinder's colocated hooks in the admin bundle by azmaveth - spark formatter entry and Tailwind scan of the theme by Etienne Lacoursiere - give phx-change forms an id by Fredrik Teschke </details> <details> <summary>ExHammer/hammer (hammer)</summary> ### [`v7.6.0`](https://github.com/ExHammer/hammer/blob/HEAD/CHANGELOG.md#760---2026-11-20) [Compare Source](https://github.com/ExHammer/hammer/compare/v7.5.1...v7.6.0) - Fix `LeakyBucket` (ETS and Atomic backends) and `TokenBucket` (Atomic backend) returning a flat `{:deny, 1000}`; they now return the real time in milliseconds until the hit would be allowed, as the ETS `TokenBucket` has since 7.5.0. The leaky buckets count in whole seconds, so their value runs to the second boundary at which the bucket has leaked enough. **Behaviour change:** the value is no longer always 1000. It can be much larger (e.g. up to 144000 for a bucket 144 over capacity at `leak_rate: 1`) or smaller (down to 1 when the next second is about to start). Code that converts it to a `retry-after` header should round up and floor at 1 second. A rate of 0 still answers `{:deny, 1000}`. ([#&#8203;221](https://github.com/ExHammer/hammer/issues/221)) - Fix `LeakyBucket` (ETS and Atomic backends) allowing a hit whenever the level was below capacity, whatever its cost, so a bucket could be filled to `capacity - 1 + cost`. A hit is now allowed only if the level after it stays within capacity, as the moduledoc always said and as the Mnesia backend does. **Behaviour change:** hits with `cost > 1` that used to overfill the bucket are now denied; hits with `cost: 1` behave as before. A `cost` larger than `capacity` is never allowed. ([#&#8203;222](https://github.com/ExHammer/hammer/issues/222)) - `use Hammer, backend: :atomic, algorithm: :sliding_window` now raises an `ArgumentError` saying that sliding window is not available for the Atomic backend and pointing to the ETS backend. It already failed at compile time, but with an unhelpful "could not load module" error. The error for an unknown Atomic algorithm now says "algorithm" rather than "backend" and no longer lists `:sliding_window`. ([#&#8203;222](https://github.com/ExHammer/hammer/issues/222)) - Fix Atomic `FixWindowPerKey` hanging on a key forever. Resetting an expired window took a lock in the key's atomic; a process killed while holding it (for example a request process shut down mid-call) left the lock behind, and every later `hit` or `inc` on that key spun on it without returning. An expired window is now replaced by a fresh atomic, so there is no lock. A lock left behind by 7.5.1 or earlier is treated as an expired window. ([#&#8203;222](https://github.com/ExHammer/hammer/issues/222)) - Fix `TokenBucket` ETS backend letting far more than `refill_rate` through. The refill clock advanced by a whole number of milliseconds, rounded down, so part of each elapsed stretch was credited again on the next hit. At `refill_rate: 600` a caller hitting in a tight loop got about 1000 hits per second, and above 1000 tokens per second, where a token is worth less than a millisecond, the clock did not move at all and the limiter was effectively off (over a million hits in half a second at `refill_rate: 1500`, `capacity: 10`). Present since 7.5.0. The clock now advances by the exact time credited. ([#&#8203;222](https://github.com/ExHammer/hammer/issues/222)) - Fix Atomic `TokenBucket` refilling in whole seconds, which capped sustained throughput at `capacity` tokens per second when `refill_rate > capacity` and delivered tokens in a lump at each second boundary. It now refills by the millisecond and keeps fractions of a token, so its `{:deny, ms}` value is exact to the millisecond (e.g. 19 at `refill_rate: 55`). Buckets held in memory across a hot code upgrade start over full. ([#&#8203;222](https://github.com/ExHammer/hammer/issues/222)) - Fix `LeakyBucket` (ETS and Atomic backends) raising the level, and `TokenBucket` ETS backend taking tokens away, when the stored clock was ahead of the caller's. That happens when a concurrent hit reads the time later or the wall clock steps back; the negative elapsed time was applied as a negative leak or refill. It now counts as no time elapsed, and a `TokenBucket` (both backends) that denies in that state includes the time until the stored clock is reached in its `{:deny, ms}` value, since nothing refills before then. ([#&#8203;222](https://github.com/ExHammer/hammer/issues/222)) </details> <details> <summary>phoenixframework/phoenix (phoenix)</summary> ### [`v1.8.17`](https://github.com/phoenixframework/phoenix/releases/tag/v1.8.17) [Compare Source](https://github.com/phoenixframework/phoenix/compare/v1.8.16...v1.8.17) ##### Security fixes - [CVE-2026-129344](https://github.com/phoenixframework/phoenix/security/advisories/GHSA-5vmc-qgf8-rw36): Sending a channel message that cannot be serialized can lead to a denial of service for all clients on a shared topic - [CVE-2026-128361](https://github.com/phoenixframework/phoenix/security/advisories/GHSA-pvfx-rx4x-phrr): Unsanitized presence key crashes Presence tracker ##### Bug fixes - \[Phoenix.PubSub] an Endpoint broadcast now works even when the endpoint is not started ([#&#8203;6846](https://github.com/phoenixframework/phoenix/issues/6846)) - \[phoenix.js] Send the fallback health check before a buffered join ([#&#8203;6770](https://github.com/phoenixframework/phoenix/issues/6770)) ##### Enhancements - \[phoenix.js] Allow enabling reconnections while a tab is hidden with `reconnectWhenHidden` ([#&#8203;6883](https://github.com/phoenixframework/phoenix/pull/6883)) - \[Channels] send fastlane messages as shared binaries ([#&#8203;6884](https://github.com/phoenixframework/phoenix/pull/6884)) - Store verified route warnings relative to cwd ([#&#8203;6889](https://github.com/phoenixframework/phoenix/pull/6889)) - Support binary messages in LongPoll transport </details> --- ### Configuration 📅 **Schedule**: (UTC) - Branch creation - Every minute (`* * * * *`) - Automerge - At any time (no schedule defined) 🚦 **Automerge**: Disabled by config. Please merge this manually once you are satisfied. ♻ **Rebasing**: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox. 👻 **Immortal**: This PR will be recreated if closed unmerged. Get [config help](https://github.com/renovatebot/renovate/discussions) if that's undesired. --- - [ ] <!-- rebase-check -->If you want to rebase/retry this PR, check this box --- This PR has been generated by [Mend Renovate](https://github.com/renovatebot/renovate). <!--renovate-debug:eyJjcmVhdGVkSW5WZXIiOiI0My4xNjUuMiIsInVwZGF0ZWRJblZlciI6IjQzLjE2NS4yIiwidGFyZ2V0QnJhbmNoIjoibWFpbiIsImxhYmVscyI6WyJhdXRvbWF0ZWQiLCJkZXBlbmRlbmNpZXMiLCJyZW5vdmF0ZSJdfQ==-->
chore(deps): update mix dependencies
All checks were successful
continuous-integration/drone/push Build is passing
continuous-integration/drone/pr Build is passing
bfc4980720
All checks were successful
continuous-integration/drone/push Build is passing
continuous-integration/drone/pr Build is passing
Required
Details
This pull request can be merged automatically.
You are not authorized to merge this pull request.
View command line instructions

Checkout

From your project repository, check out a new branch and test the changes.
git fetch -u origin 'renovate/mix-dependencies:renovate/mix-dependencies'
git switch 'renovate/mix-dependencies'
Sign in to join this conversation.
No description provided.