Skip to content

chore: refresh stale test snapshots left over from #363 and #367#369

Merged
collinsezedike merged 1 commit into
mainfrom
chore/refresh-stale-test-snapshots
Jul 12, 2026
Merged

chore: refresh stale test snapshots left over from #363 and #367#369
collinsezedike merged 1 commit into
mainfrom
chore/refresh-stale-test-snapshots

Conversation

@collinsezedike

Copy link
Copy Markdown
Collaborator

Summary

  • main had 7 committed Soroban test snapshots that no longer matched what cargo test actually produces, in two batches:
    • 5 in blend-adapter (accrue_is_idempotent_at_a_stable_rate, accrue_reflects_yield_from_a_rate_increase, deposit_supplies_to_pool_and_tracks_total, withdraw_after_accrue_pays_out_appreciated_value, withdraw_returns_usdc_and_reduces_total), left stale from fix: correct BlendAdapter.accrue() rate divisor #367: changing the mock pool's SCALAR constant to match the real RATE_SCALAR fix changed the recorded auth/footprint for every test sharing that setup(), not just the one new test added in that PR.
    • 2 in defindex-adapter (get_pool_returns_the_configured_defindex_vault, get_protocol_returns_defindex), left stale since test: add unit tests for Blend and DeFindex adapter contracts #363: rewriting setup() to add mint/set_admin calls for the new tests in that PR changed the fingerprint of the two pre-existing tests that share the same setup, which weren't re-committed at the time.
  • Root cause: on both PRs, only the snapshot files for genuinely new tests were staged; the git checkout -- <modified test_snapshots> cleanup step used to discard unrelated non-deterministic noise from earlier local runs also silently discarded these legitimately-changed files without checking each one individually.
  • Verified the fix properly this time: ran the full cargo test (matching CI's exact invocation, working-directory: packages/contracts, run: cargo test) twice independently, diffed every touched file against main, and confirmed the same 7 files show real, stable, non-trivial content drift both times — everything else is line-ending/generator-artifact noise from Windows core.autocrlf, discarded.
  • These are diagnostic snapshot captures only (soroban-sdk testutils writes them after each run); nothing asserts against them, so this was never a CI-failing bug, just accumulating drift between what's committed and what the current code actually produces.

Test plan

  • cargo test in packages/contracts: 43 tests pass, snapshots now match exactly
  • Reran full suite twice independently and confirmed no further drift on these 7 files
  • cargo clippy --all-targets -- -D warnings clean

@vercel

vercel Bot commented Jul 11, 2026

Copy link
Copy Markdown

The latest updates on your projects. Learn more about Vercel for GitHub.

Project Deployment Actions Updated (UTC)
meridian Ready Ready Preview, Comment Jul 11, 2026 10:56pm

@collinsezedike collinsezedike merged commit f1dc40f into main Jul 12, 2026
8 checks passed
@collinsezedike collinsezedike deleted the chore/refresh-stale-test-snapshots branch July 12, 2026 09:21
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant