Fig 1.
Proposed system push process: Source code is encrypted and stored on IPFS.
The encryption key is split into three shares via SSS: One is retained by the owner, one is sent to the middleman for fast retrieval, and one is registered on the blockchain asynchronously.
Fig 2.
Proposed system pull process: After an on-chain permission check, the system first queries the authoritative blockchain for a key share.
If unavailable due to latency, it optimistically falls back to the middleman, ensuring a responsive user experience.
Fig 3.
Performance comparison: proposed system vs. centralized git.
User-perceived push time is faster than git push for common repository sizes, while pull performance is highly competitive, validating our asynchronous, optimistic design.
Table 1.
User-perceived performance: proposed system vs. centralized git (in seconds).
Fig 4.
Breakdown of total system workload during a push operation: The Pure Blockchain Latency represents a large, constant overhead, confirming it as the primary system bottleneck that our asynchronous design successfully abstracts away from the user’s workflow.
Fig 5.
System latency comparison: Local vs. public testnet environments.
This chart quantifies the performance overhead of public networks. Minimal local latency (lighter bars) reflects efficient core logic, while higher public latency (darker bars) is attributable to real-world network and consensus delays.