D3597 — A later minimal.toml layer that sets one network sub-table keeps the network keys of an earlier layer · Issue #3600 · gominimal/minimal · GitHub
You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
D3597 — A later minimal.toml layer that sets one network sub-table keeps the network keys of an earlier layer #3600
### D3597 — A later minimal.toml layer that sets one network sub-table keeps the network keys of an earlier layer
kind: task
size: M
est: 120 code, 110 tests
executor: agent
origin: bug
epic: 1501
layers:
- expand:
- crates/mfile/src/expand.rs: merge_value merges `network` by sub-table and key across layers, and the report records an origin for each `network.<sub>` key
test: cargo test -p mfile network_merges_by_sub_table
- cli:
- crates/minimal/src/cmd/boxes/create.rs: entry_egress reads the per-sub-table merge and names the layer that set `network.egress`, not the last layer that declares `network`
- crates/minimal/src/cmd/box_spec.rs: source comments and setting_table read the per-sub-table origins, so `network.bep` and `network.egress` each name their own layer
- crates/minimal/src/tests.rs: adds the issue reproduce case: the spec of `dev` holds both `bep` and `egress`, and entry_egress sees the `[defaults.network.egress]` hosts
test: cargo test -p minimal box_spec_keeps_earlier_network_egress
- docs:
- docs/specs/29-spec-box-local-first/29-spec-box-local-first.md: BOX-048 states that `network` merges by sub-table, as `vars` merges by name
- docs/specs/18-spec-box-networking/18-spec-box-networking.md: replaces the text that says the merge replaces the `network` table whole
test: cargo test -p mfile network_merges_by_sub_table
proof: With `[defaults.session.network.bep]` and `[defaults.network.egress]` set, `min box spec dev` shows both `bep` and `egress` under `[network]`, each with its source layer.
That /implement is queued, not refused. 4 follow-up coders are in flight, and 3 is the most that run at once, so no coder started. 8 queued commands are ahead of it. The runtime starts queued follow-up coders oldest first as room frees, looking every ten minutes, and refuses one still waiting after 14 days.
Nothing to type again: the command gets a rocket when its coder starts, or a refusal that names the reason. /implement stop takes it off the queue.
changed nothing: Revise round 1: no change. The failed checks are the macOS 'unit' job and its aggregator 'ci-macos-success'. On macOS, 'just test' runs only -p minvmd -p boxes -p bep. cargo tree shows that none of th
4ed74864
2
pushed
5a631e94
3
pushed
baa11fae
4
pushed
a76bc01f
5
pushed
1eec8258
test
exit
tail
cargo test -p mfile network_merges_by_sub_table
0
running 3 tests test expand::tests::network_merges_by_sub_table_mode_none_with_own_egress_is_refused ... ok test expand::tests::network_merges_by_sub_table_mode_none_clears_inherited_egress_and_ingress ... ok test expand::tests::network_merges_by_sub_table ... ok test result: ok. 3 passed; 0 faile
cargo test -p minimal box_spec_keeps_earlier_network_egress
0
gress_beside_entry_egress ... ok test result: ok. 5 passed; 0 failed; 0 ignored; 0 measured; 1063 filtered out; finished in 0.01s running 0 tests test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 43 filtered out; finished in 0.00s running 0 tests test result: ok. 0 passed; 0 failed;
cargo test -p mfile network_merges_by_sub_table
0
running 3 tests test expand::tests::network_merges_by_sub_table_mode_none_with_own_egress_is_refused ... ok test expand::tests::network_merges_by_sub_table_mode_none_clears_inherited_egress_and_ingress ... ok test expand::tests::network_merges_by_sub_table ... ok test result: ok. 3 passed; 0 faile
Verify min task run agent-verify: exit 100 — FAIL [ 13.076s] (1473/1473) minimal::otel_cli exec_run_attach_and_task_run_share_the_commands_trace thread 'exec_run_attach_and_task_run_share_the_commands_trace' (7079) panicked at crates/minimal/tests/otel_cli.rs:805:13: attach: no daemon attach span under cmd d868acebccd5d319: [
That /implement is held, not dropped. The container platform had no instance to give the coder: its pool of prepared containers was full, which happens most just after a deploy. It is asked again every 10 minutes for up to 30.
Nothing to type again: the command gets a rocket when its coder starts, or a refusal that names the reason.
What the platform said
could not start the coder: Error: there is no container instance that can be provided to this durable object
Activity
norrietaylor commentedon Nov 18, 2026
/implement
minimal-foundry commentedon Nov 18, 2026
That
/implementis queued, not refused. 4 follow-up coders are in flight, and 3 is the most that run at once, so no coder started. 8 queued commands are ahead of it. The runtime starts queued follow-up coders oldest first as room frees, looking every ten minutes, and refuses one still waiting after 14 days.Nothing to type again: the command gets a rocket when its coder starts, or a refusal that names the reason.
/implement stoptakes it off the queue.minimal-foundry commentedon Nov 19, 2026
Pull request #3619 is open as a draft for D3597.
branch
agent/3600-a-later-minimal-toml-layer-that-sets-one-network· epic #15014ed748645a631e94baa11faea76bc01f1eec8258cargo test -p mfile network_merges_by_sub_tablecargo test -p minimal box_spec_keeps_earlier_network_egresscargo test -p mfile network_merges_by_sub_tableVerify
min task run agent-verify: exit 100 — FAIL [ 13.076s] (1473/1473) minimal::otel_cli exec_run_attach_and_task_run_share_the_commands_trace thread 'exec_run_attach_and_task_run_share_the_commands_trace' (7079) panicked at crates/minimal/tests/otel_cli.rs:805:13: attach: no daemon attach span under cmd d868acebccd5d319: [Outside the task's
layers:, so the PR is a draft:crates/minimal/src/task.rsUpdated 2026-11-21T05:01:19.585Z.
minimal-foundry commentedon Nov 19, 2026
That
/implementis held, not dropped. The container platform had no instance to give the coder: its pool of prepared containers was full, which happens most just after a deploy. It is asked again every 10 minutes for up to 30.Nothing to type again: the command gets a rocket when its coder starts, or a refusal that names the reason.
What the platform said
could not start the coder: Error: there is no container instance that can be provided to this durable object