apollo_central_sync,apollo_starknet_client: collapse three feeder calls into one per block

[dafnamatsry]

No description provided.

[cursor]

PR Summary

High Risk
Changes the core central-sync download and storage path; incorrect handling could break block ingestion or marker consistency despite the intentional atomic commit fix.

Overview
Central sync now fetches each block with one feeder get_state_update call (includeBlock=true, includeSignature=true) instead of separate block, signature, and state-update requests. The central source exposes a unified CentralStateUpdate stream via stream_new_blocks; stream_state_updates and the old parallel block/signature download path are removed.

The sync loop merges the former block and state-diff pipelines into a single StateUpdateAvailable event handled by store_state_update. Persistence uses one read-write transaction for header, signature, body, and state diff (plus related class-manager updates), so header_marker and state_marker advance together and avoid a recoverable failure leaving mismatched markers.

apollo_starknet_client adds BlockStateUpdate and StarknetReader::block_state_update_and_signature; StateUpdateStream schedules that call and maps the response into API types. Tests and mocks are updated accordingly.

^{Reviewed by Cursor Bugbot for commit 45aea34. Bugbot is set up for automated code reviews on this repo. Configure here.}

[reviewable-StarkWare]

This change is 

[dafnamatsry]

• apollo_central_sync,apollo_starknet_client: collapse three feeder calls into one per block #14186 👈 (View in Graphite)
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, have a team admin enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 45aea34. Configure here.}

[cursor]

Stale marker read causes incorrect backward-compatibility classification

Medium Severity

The compiler_backward_compatibility_marker is read at line 435 (before the storage transaction), but is updated at line 495 (inside the transaction). In the old code, the header stream always ran ahead of the state diff stream, so the marker was already updated when store_state_diff read it. Now that both are combined, for a block N with old starknet version where the marker equals N, the check compiler_backward_compatibility_marker <= block_number is incorrectly TRUE, causing the block to be treated as backwards-compatible when it isn't. The stale comment at line 426–428 ("this stream is behind the header stream") confirms this ordering assumption was not updated.

Additional Locations (1)

• crates/apollo_central_sync/src/lib.rs#L487-L497

 

^{Reviewed by Cursor Bugbot for commit 45aea34. Configure here.}