test: contract tests improving coverage (toward #79)#81
Merged
markovejnovic merged 1 commit intoJul 12, 2026
Conversation
…uting Toward harmont-dev#79 (>=95% coverage). Each addition targets a real Contract path — a wire-format conversion, a gRPC status classification, or a refusal — not a getter or a pass-through. Triage per AGENTS.md: grpc/codec.ex 54% -> 100% from_grpc/to_grpc wire contracts + the rpc_error status-classification table (every domain reason -> its promised gRPC status). cfg/network.ex 86% -> 100% non-string network.uplink -> ArgumentError. cfg/grpc.ex 89% -> 100% plaintext-default cred(nil) via load/0. cluster/routing 85% -> 100% register_self already-registered + all/0 node mapping (new test file; reuses the named singleton to avoid a start_supervised race). cfg/budget.ex 80% -> 90% non-number / bad unit string / wrong type -> specific error tuples (table-driven). redist/file.ex 87% -> 93% {:install_failed, _} on an unwritable dest. redist/targz.ex 88% -> 94% {:extract_failed, _} on a corrupt archive with a valid checksum. Deliberately declined (mechanical / privileged), with rationale recorded: - redist download_error Req pass-throughs (already declined in moduledocs); - cfg/budget.get/0 getter and the dead required/3 :error branches (a default is always present, so they are unreachable); - cfg/toml.ex /sys/fs/cgroup /etc/subuid /etc/subgid (need Linux root/cgroupfs); - node/fire_vmm/{meter,daemon,relay,guest_agent} (integration-only: cgroup, Firecracker, sockets, built musl agent — covered in CI's integration job). All +20 tests pass; mix format --check-formatted and mix credo --strict clean.
markovejnovic
approved these changes
Jul 12, 2026
Comment on lines
+53
to
+57
| test "with no cred configured, load returns nil (plaintext)" do | ||
| # The plaintext default: an absent cred must yield nil, never a defaulted | ||
| # credential that would silently enable TLS. | ||
| assert Grpc.load().cred == nil | ||
| end |
Comment on lines
+83
to
+101
| # `from_grpc/1` -- the decode boundary. The contract: an inbound CreateVmRequest | ||
| # maps to a domain Spec with the right field placement, or is refused with a | ||
| # specific reason the server then classifies to a gRPC status. | ||
| describe "from_grpc/1 CreateVmRequest" do | ||
| test "a well-formed request decodes to a Spec preserving every field" do | ||
| assert {:ok, | ||
| %Spec{ | ||
| img_id: "img", | ||
| type: :micro, | ||
| arch: :aarch64, | ||
| boot_args: "console=ttyS0" | ||
| }} = | ||
| Codec.from_grpc(%CreateVmRequest{ | ||
| img_id: "img", | ||
| instance_type: :INSTANCE_TYPE_MICRO, | ||
| arch: :ARCHITECTURE_AARCH64, | ||
| boot_args: "console=ttyS0" | ||
| }) | ||
| end |
Contributor
There was a problem hiding this comment.
I generally don't see that much utility in these tests frankly. I will be honest I mostly vibe-coded these tests, so that's how we ended up with codec_test.exs, but in the future I'd like to see these tests replaced with e2e grpc tests.
i'm comfortable with your PR, considering the existing precedent =)
Contributor
Author
There was a problem hiding this comment.
Agreed, more following through on the issue's ask. I'l look into creating an e2e ask and see if I can help with that.
Comment on lines
+94
to
+97
| test "install/3 surfaces a corrupt archive as {:extract_failed, _}", %{ | ||
| server: server, | ||
| dest: dest | ||
| } do |
Comment on lines
+17
to
+32
| # Under `mix test --no-start` the app tree (which provisions Routing) is | ||
| # absent; start a test-scoped registry so these contracts are testable | ||
| # without the full supervision tree. Routing is a named singleton, so if | ||
| # another test (e.g. HyperTest) already started it, reuse that one rather | ||
| # than racing on start_supervised!/1. | ||
| if Process.whereis(Routing) do | ||
| :ok | ||
| else | ||
| _ = Application.ensure_all_started(:horde) | ||
|
|
||
| case start_supervised(Routing) do | ||
| {:ok, _pid} -> :ok | ||
| # Lost the start race to a parallel test: reuse the existing registry. | ||
| {:error, {:already_started, _pid}} -> :ok | ||
| end | ||
| end |
Contributor
There was a problem hiding this comment.
my gut feeling is that this is going to cause some flaky tests in the future, but OK, good enough for now
Codecov Report✅ All modified and coverable lines are covered by tests. 📢 Thoughts on this report? Let us know! |
Contributor
|
thank you for your PR! |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Toward #79 (>=95% coverage).
What
Each addition targets a Contract path — a wire-format conversion, a gRPC status classification, or a refusal — not a getter or a pass-through. Triage per
AGENTS.md(Contract / Mechanical / Privileged):hyper/grpc/codec.exfrom_grpc/to_grpcwire contracts + therpc_errorstatus-classification table (every domain reason → its promised gRPC status)hyper/cfg/network.exnetwork.uplink→ArgumentErrorhyper/cfg/grpc.excred(nil)viaload/0hyper/cluster/routing.exregister_selfalready-registered +all/0node mapping (new test file; reuses the named singleton to avoid astart_supervisedrace)hyper/cfg/budget.exredist/file.ex{:install_failed, _}on an unwritable destredist/targz.ex{:extract_failed, _}on a corrupt archive with a valid checksum+20 tests, table-driven where the assertion shape repeats.
mix format --check-formattedandmix credo --strictare clean on all of them.Deliberately declined
Per
AGENTS.md's "Answering coverage reports" — these gaps are not worth a unit test, with rationale:redistdownload_errorReq pass-throughs (already declined in those modules' own moduledocs);cfg/budget.get/0getter;cfg/budget'srequired/3:errorbranches (a default is always present, so they are unreachable).cfg/toml.ex(needs root to write/etc/hyper/config.toml; the absent-file-> %{}default is covered),sys/linux/cgroup/v2.ex(cgroupfs),sys/linux/subid.ex(/etc/subuid+/etc/subgid; theranges/1parsing is fixture-tested), andnode/fire_vmm/{meter,daemon,relay,guest_agent}(need live cgroup / Firecracker / sockets / the built musl agent — covered in CI'sintegrationjob).A patch may legitimately merge below 100%; these uncovered mechanical lines cost less than tests that pin implementation detail.
Notes
RoutingTeststarts the registry withstart_supervised(non-bang) and reuses an already-started one, sinceHyper.Cluster.Routingis a named singleton andHyperTestalso starts it — a bang call raced underasync.GLM-5.2 did this work.
Closes toward #79.