KarpelesLab / libwallet / 24435361820

25%

Build Type

push

github

Committed by MagicalTux

Commit Message

Network.DoRPC: default 30 s timeout (bound every untimeouted RPC)

Root cause of the iOS CI hang: every `n.DoRPC(...)` call went through
`DoCtx(context.Background(), ...)` — no deadline. The moment any
upstream endpoint (Ethereum mainnet public RPC, Solana, modchain,
etc.) stopped responding, the goroutine blocked forever. The balance
poller I just added multiplied the problem because it fires every
60 s in the background.

Fix applied at the lowest level so no call site is missed:

- DoRPC and DoRPCNamed now wrap their call in a 30 s
  context.WithTimeout. A single RPC can no longer wedge a goroutine.
  Callers that genuinely need longer deadlines (there are none today)
  can still go through DoRPCCtx / DoRPCNamedCtx with their own ctx.
- The balance poller's poll() already wraps currentAssets in a 15 s
  timeout (tighter than the global default — a poll that's taking
  more than 15 s is almost certainly the user's network being down
  and should be abandoned).
- nativeBalance / bitcoinBalance / NativeAsset / SolanaTokenBalances
  threaded through ctx so both the poller and the API endpoint
  propagate deadlines through to the RPC.

CI:
- Drop iOS job timeout from 60 to 20 min. Happy path is ~8 min;
  20 min is ~2.5× that. With the RPC timeout cap in place, a stuck
  network can't consume the whole window anymore.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Coverage Stats

0 of 25 new or added lines in 4 files covered. (0.0%)

2 existing lines in 1 file now uncovered.

2109 of 9676 relevant lines covered (21.8%)

2.62 hits per line