• Home
  • Features
  • Pricing
  • Docs
  • Announcements
  • Sign In

supabase / supabase-swift / 44374170227
96%

Build:
DEFAULT BRANCH: main
Ran 18 Nov 2026 07:00PM UTC
Jobs 1
Files 224
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst

18 Nov 2026 06:51PM UTC coverage: 96.2% (+0.007%) from 96.193%
44374170227

push

github

web-flow
fix(storage): keep a minimum wait between resumable upload retries (#1866)

* fix(helpers): use full jitter for the retry backoff

The draft auto_retry specs (supabase/sdk PR 190) draw the wait before the
n-th retry from 0...min(maxDelay, baseDelay * 2^(n-1)). RetryPolicy drew it
from cap/2...cap (equal jitter). Full jitter spreads clients that failed
together over the whole window, and matches the Realtime reconnect backoff.

backoffDelay(retry:random:) takes the jitter source so tests can pin it.

* refactor(helpers): share one full-jitter backoff between HTTP retries and Realtime

RetryPolicy.backoffDelay repeated the full-jitter math of Realtime's
BackoffPolicy. Move BackoffPolicy into Helpers, which Realtime already
depends on, and have RetryPolicy delegate to it. Move its tests to
HelpersTests and drop the randomized RetryPolicy backoff test that the
pinned-jitter test now covers.

* docs(helpers): fix backoffDelay doc comment for swift-format lint

* test(helpers): cover BackoffPolicy's default random source

* test(helpers): tighten backoff tests after code review

- The interceptor test now runs ten retries and checks that the wait window grows,
  so a wrong retry number fails it.
- The default random test draws 200 delays and checks they spread over the window.
- RetryPolicyTests keeps one delegation check instead of repeating BackoffPolicy's tests.
- The RetryPolicy doc no longer says it drives Realtime's reconnect delays.

* fix(storage): keep a minimum wait between resumable upload retries

The resumable upload loop has a time budget (retryTimeout) but no attempt
limit. With full jitter (0...cap) in the shared backoff, waits can be near
zero, so the loop could send a burst of retries to a failing server. Each
retry now waits a random duration in cap/2...cap, so the waits grow as the
docs say and the budget bounds the attempt count.

* refactor(storage): draw the resumable retry floor in one backoff call

3 of 3 new or added lines in 1 file covered. (100.0%)

3 existing lines in 1 file now uncovered.

16555 of 17209 relevant lines covered (96.2%)

640003.8 hits per line

Coverage Regressions

Lines Coverage ∆ File
3
97.52
-2.48% Sources/Helpers/JSONValue/JSONValue.swift
Jobs
ID Job ID Ran Files Coverage
1 44374170227.1 18 Nov 2026 07:00PM UTC 224
96.2
GitHub Action Run
Source Files on build 44374170227
  • Tree
  • List 224
  • Changed 3
  • Source Changed 1
  • Coverage Changed 3
Coverage ∆ File Lines Relevant Covered Missed Hits/Line
  • Back to Repo
  • Github Actions Build #44374170227
  • 908d3a5b on github
  • Prev Build on main (#44363301106)
  • Next Build on main (#44376694894)
STATUS · Troubleshooting · Open an Issue · Sales · Support · CAREERS · ENTERPRISE · START FREE TRIAL · SCHEDULE DEMO
ANNOUNCEMENTS · TWITTER · TOS & SLA · Supported CI Services · What's a CI service? · Automated Testing

© 2026 Coveralls, Inc