Closed fxamacker closed 6 months ago
Attention: Patch coverage is 62.92135%
with 66 lines
in your changes are missing coverage. Please review.
Project coverage is 70.38%. Comparing base (
2fbf860
) to head (35fdb7e
). Report is 8 commits behind head on feature/array-map-inlining.
Files | Patch % | Lines |
---|---|---|
typeinfo.go | 61.48% | 43 Missing and 14 partials :warning: |
array_debug.go | 30.76% | 6 Missing and 3 partials :warning: |
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
I found some improvements to deduplication while migration + validation was running last night.
Work on this PR is paused this morning to meet about and begin work on urgent request to help with Cadence 1.0 migration.
This PR has about 1/2 day of effort left to improve its deduplication for further reduction of memory and persistent storage.
Yesterday, this passed atree inlining & deduplication migration with validation enabled (for all payloads for all accounts) using mainnet Jan 19, 2024 checkpoint as input.
Closes #358 Updates epic #292
This PR deduplicates Cadence dictionary type resulting in reduced memory and persistent storage.
More specifically, this encodes inlined atree slab extra data section as two-element array:
TODO (est. 1.5 - 2 days effort total): [x] - initial deduplication of dict type info [x] - build, run, and pass migration with validation enabled for all accounts and all payloads from mainnet snapshot [x] - improve deduplication (1/2 day effort to wrap up + rerun test using mainnet snapshot)
UPDATE (Feb 29, 2024): :stop_sign: Work on this PR is paused to work on urgent request to create tooling for Cadence 1.0 migration.
NOTE: This PR is expected to produce small improvements to atree inlining & deduplication results :bar_chart:.
main
branchFiles changed
in the Github PR explorer