fix: clean-build failures on ip_local_network#71
fix: clean-build failures on ip_local_network#71MoonBoi9001 wants to merge 3 commits intoip_local_networkfrom
Conversation
Both contract-deploy run.sh files trip shellcheck on a fresh lint pass. Suppress SC1091 above the source lines (the paths only exist at container runtime), rewrite the SC2010 ls|grep patterns as find -name -quit, and rename the unused for-loop counter to `_` for SC2034. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The network subgraph still references two pre-Horizon contracts that don't exist on hardhat. Earlier local-network builds wrote zero-address placeholders to keep the manifest valid; 66cf07d removed them. Restore the placeholders so subgraph-deploy succeeds again on a fresh stack. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
graph-contracts-issuance and graph-contracts-tap run in parallel and sign from the same Ethereum account; when their transactions race for the same nonce the loser fails with NonceTooLowError, leaving no pending TXs. Replace the activation-goals exit-1 with sleep-and-retry. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
MoonBoi9001
changed the title
MoonBoi9001
changed the title
|
Fine to merge this. FYI in the process of replacing this branch with a rebased clean version Some clean-up pending on |
RembrandtK
left a comment
RembrandtK
left a comment
There was a problem hiding this comment.
Fine to merge this.
FYI in the process of replacing this branch with a rebased clean version gip-88 and also progressively merging more of it into main (the divergence has been significantly reduced already). The fixes here has two aspects, one which has now been merged to main (where the zero address issue was introduced in 642978b, fixed in 5c2eded), and the nonce issue, which is not in new gip-88 branch that does not have split deployment containers.
Some clean-up pending on gip-88.
2 participants
TL;DR
A clean
ip_local_networkbuild currently fails in the contract-deploy phase, blocking the network subgraph and seven downstream services. This PR re-adds the legacy address placeholders and replaces the issuance nonce-raceexit 1with a sleep-and-retry. Touches only the two*.run.shfiles plus shellcheck cleanup.Motivation
The network subgraph manifest references two pre-Horizon contracts (
LegacyServiceRegistry,LegacyDisputeManager) that hardhat does not deploy. Earlier local-network builds wrote zero-address placeholders for them; commit66cf07d(the gateway pin bump) removed that step. Without the placeholders,subgraph-deployrejects the manifest and the seven services that depend on the network subgraph never come up.Separately,
graph-contracts-issuanceandgraph-contracts-taprun in parallel, sign from the same Ethereum account, and race for the same nonce; the loser'sexit 1takes the issuance container down on roughly half of cold starts. Together these mean anyone trying the clean-build path you documented gets a stack that looks healthy at the docker level but is silently missing the subgraph layer.Summary
*.run.shfiles (SC1091 / SC2010 / SC2034)