Skip to content

Latest commit

 

History

History
152 lines (125 loc) · 9.08 KB

clients.md

File metadata and controls

152 lines (125 loc) · 9.08 KB

The clients a release was proved against

Read this first if you pin. This emulator's whole claim is that real clients drive it. That claim is only ever true of the versions those clients had when the suites ran, and this page is where they are written down — generated from the workflow that installs them and from the fixtures that pin the providers, never typed.

It exists because of one measured morning. A consumer running feint as a required CI gate, with no credentials, resolves its Terraform providers fresh on every pull request. Scaleway published provider 2.81.0 on 2026-08-17; it moved private NICs from instance/v1 · private_nics onto instance/v2alpha1 · private-network-interfaces, their ~> 2.68 constraint picked it up the next morning, and their lane went red. They pinned back to ~> 2.80.0 and wrote down what it cost: the lane stopped emulating the provider our clusters actually run. Nothing we published would have let them pin deliberately instead (#325).

How to read it

A version in the first table is what the conformance workflow installs and drives on every pull request. It is not a floor and not a recommendation: it is the only version any claim in a release note is true of. A newer client is not known to work until the pin moves and the suite passes on it.

The second table is the one that decides whether a Terraform lane will meet the same emulator you did, and its third column is doing real work:

  • exact — the constraint names one version. That version is what answered.
  • constraint — terraform init -upgrade resolves it on the runner, on every run. It is a floor. The version that answered is not recorded anywhere in this repository, and this page will not pretend otherwise.
  • not pinned — no constraint at all. Whatever the registry served that day.

The last column is derived from the conformance workflow and from the suites that apply the examples — tools/conformance/stacks.sh for the qualification stacks, tools/conformance/quickstart.sh for the quick start ones — so a fixture nothing applies cannot pass for a proof by sitting in the directory.

Two things this table said the day it landed, which nothing had said before, and which are now checked rather than read. A stack CI applies must carry a provider constraint — an unconstrained one is resolved from the whole registry on every run, so the apply proves the emulator answered something and nothing anybody can replay. And a stack CI does not apply must be declared, with its reason, in stacksRunByHand (internal/cli/docs_stacks.go); feint docs --check exits 2 on a stack that is neither applied nor declared, and on a declaration for a stack that no longer exists or that CI has started applying. The list below the table is that declaration, printed from the same source the refusal reads.

Both families are in that population, and deliberately: examples/quickstart holds the example the most people copy first, and an example nobody runs is a README that rots (#593).

The versions

The clients

Client Version Pinned in Drives, in CI
scw 2.56.3 .github/workflows/conformance.yml (SCW_VERSION) Scaleway
Terraform 1.13.3 .github/workflows/conformance.yml (TERRAFORM_VERSION) Exoscale, Outscale, Scaleway
OpenTofu 1.12.5 .github/workflows/conformance.yml (TOFU_VERSION) Exoscale, Outscale, Scaleway
octl 0.0.32 .github/workflows/conformance.yml (OCTL_VERSION) Outscale
exo 1.95.6 .github/workflows/conformance.yml (EXO_VERSION) Exoscale

The Terraform providers

The client above is the engine; what answers this emulator is the provider. Each row is one required_providers entry, read where it is written.

Fixture or stack Provider Constraint What that is worth Applied in CI
tools/conformance/faults outscale/outscale ~> 1.7 constraint: resolved fresh on each run, so the version that answered is not knowable here yes
tools/conformance/outscale/terraform outscale/outscale ~> 1.7 constraint: resolved fresh on each run, so the version that answered is not knowable here yes
tools/conformance/outscale/terraform-doorway outscale/outscale ~> 1.7 constraint: resolved fresh on each run, so the version that answered is not knowable here yes
tools/conformance/scaleway/terraform scaleway/scaleway 2.83.0 exact: the version that answered yes
examples/stacks/exoscale exoscale/exoscale 0.72.0 exact: the version that answered yes
examples/stacks/exoscale hashicorp/aws 6.72.0 exact: the version that answered yes
examples/stacks/outscale outscale/outscale 1.8.1 exact: the version that answered yes
examples/stacks/outscale/modules/net outscale/outscale ~> 1.7 constraint: resolved fresh on each run, so the version that answered is not knowable here yes
examples/stacks/scaleway scaleway/scaleway 2.83.0 exact: the version that answered yes
examples/quickstart/outscale outscale/outscale ~> 1.7 constraint: resolved fresh on each run, so the version that answered is not knowable here yes
examples/quickstart/scaleway scaleway/scaleway 2.83.0 exact: the version that answered yes

Read the third column narrowly, because it is the one a consumer pins against. An exact constraint names the version that answered. A constraint is resolved by terraform init -upgrade on the runner, every run, so the version that answered is not knowable from this repository — it is a floor, not a proof. Not pinned is not pinned: nothing here says which version ran, and no number is invented to fill the cell.

Every stack under examples/stacks/ is applied on every pull request.

Which client may drive which pack

The table above says in which version a client was proved. This one says whether it may drive a pack at all, and it exists because of a sentence that was false for two days on the front line of this repository (#592).

README.md:41 read "Run your Terraform against Scaleway, Outscale or Exoscale". docs/confidence.md read "A Terraform run against Exoscale — no, and refused rather than half-served". And feint up had been refusing iac.engine: terraform for that pack since #525 landed on 2026-08-26, because the published provider splits a single apply between this emulator and a paying account. Every doc gate was green throughout, and none of them was wrong: they check values — a coverage count, a route total, an image tag, a client version — and each of those is something an artefact also holds. "Run your Terraform against three clouds" is a claim about capability, and nothing owned it.

So the matrix below owns them, and every sentence in this repository that claims a client for a provider is generated from it. Its proof column is what makes it more than a table: a row names the instrument that establishes it, and a row whose instrument does not carry it fails feint docs --check at exit code 2 — in both directions, because a check that only refuses what is written down stops measuring the day something stops being written down.

Provider Client Mode Support Proof Reason
Exoscale exo control plane supported conformance workflow —
Exoscale OpenTofu control plane supported conformance workflow —
Exoscale Terraform control plane supported conformance workflow —
Outscale octl control plane supported conformance workflow —
Outscale OpenTofu control plane supported conformance workflow —
Outscale Terraform control plane supported conformance workflow —
Scaleway OpenTofu control plane supported conformance workflow —
Scaleway scw control plane supported conformance workflow —
Scaleway Terraform control plane supported conformance workflow —

Every sentence in this repository that claims a client for a provider is generated from this table, and every row is resolved against the thing its proof column names: conformance workflow means .github/workflows/conformance.yml runs a suite driving that client against that pack on every pull request, and up.go VetoEngine means the pack itself refuses the engine before feint up starts a process. A row whose proof does not carry it fails feint docs --check at exit code 2, and so does a page that claims a pair this table does not.

Where else this appears

The same block is spliced into the body of every GitHub Release, so the page and the release say the same thing by construction. feint docs --check fails with exit 2 the moment either drifts from the files above, which is how a generated table stays a measurement rather than becoming a comment.