Skip to content
SHC Docs

The flagship bar is payment, not license family, owner ruling 2026-08-07

Status: OWNER RULING: LANDED IN THE CONTRACT. This ruling rewrites requirement 4 of docs/decisions/2026-08-06-flagship-tier.md and moves what requirement 4 used to test onto a separate, non-flagship axis. No code changes; the flagship-contract fitness check never measured license posture and still does not (it cannot; see §5).

Date: 2026-08-07. Measured against shc origin/main @ 5c83f061b.


“if you are forced to pay to use the app it’s not a flagship.”

Ten words that replace a license-family test with a user-facing test.

Requirement 4 as shipped on 2026-08-06 read, in its last sentence:

SSPL/BSL-style no-SaaS licenses disqualify.

That is the wrong axis, and it was already producing wrong answers in the candidate table it shipped with: one stated outright, one waiting.

The worked case: huly. apps/huly is the projects candidate and the only candidate in the whole table with a real user’s full OIDC login proven live (2026-08-05 10:59:15). Huly itself is EPL-2.0, verified in-repo against the platform repo’s LICENSE, recorded at docs/guide/apps/custom-builds.md §“The Huly soft fork”, and nobody is forced to pay to use it. But the package bundles two components under no-SaaS terms (apps/huly/vars.yaml:147-148):

Bundled pinLicense familyVerified how
elasticsearch:7.14.2Elastic License 2.0, dual with SSPL (ELv2 applies from 7.11)license family recorded at docs/guide/apps/licensing-postures.md; the pin read from apps/huly/vars.yaml:148
docker.redpanda.com/redpandadata/redpanda:v24.3.6BSL/BUSL 1.1family recorded in the same page; the Additional Use Grant on this exact tag is unverified in-repo; the page’s own rule is “never generalize BSL: read the grant on the exact version pinned”

Read literally, the old requirement 4 disqualified huly, the best candidate in the table, on a technicality about two sidecars. The table never actually applied it there (the projects row said only “License verify”), so this was a wrong answer waiting rather than a wrong answer stated. Under the owner’s criterion it does not arise at all: a self-hoster installing huly pays nobody.

The wrong answer that WAS stated: keygen. The license row read “requirement 4 disqualifies no-SaaS licenses, so this needs an explicit owner ruling before any SSO work is even worth doing”, a candidate blocked outright, and packaging work deferred, purely on the retired axis. That block is withdrawn; the tier doc’s §7 license row now carries the corrected verdict, and what remains owed there is an SSO plug, not a ruling.

The two things being conflated. “May I use this without paying?” and “may SHC sell this as a managed service?” are different questions with different answers, and only the first one is what the flagship label claims. A flagship is SHC’s pick for an operator installing an app on their own estate (§1 of the tier doc). What SSPL/BSL/ELv2/RSALv2 restrict is offering the software to third parties as a service, a constraint on SHC as a provider, not on the operator, and therefore not a property of the app’s fitness for the label.

3. Half one: requirement 4 is now the payment bar

Section titled “3. Half one: requirement 4 is now the payment bar”

The rewritten requirement lives in docs/decisions/2026-08-06-flagship-tier.md §1.4 and is authoritative there. Its shape:

  • The disqualifier is payment by the user: to use the app at all, to reach its core functionality, or to reach its SSO.
  • The test is what the user must pay. Not the license family on the code. Not what SHC would owe a vendor as a managed provider.
  • A redistribution grant is still a packaging precondition (SHC ships images, soft-forks and custom builds), but it is a precondition, not the disqualifier.

This is not a loosening. The apps the bar correctly disqualifies were already segregated into apps/enterprise/ before this ruling, and both are disqualified by the new wording as squarely as by the old requirement 1:

AppWhy the user must payIn-repo citation
apps/enterprise/highlightSSO is gated behind a license key the deployment must carryapps/enterprise/highlight/compose.yaml:57 (LICENSE_KEY), apps/enterprise/highlight/app.yaml:18
apps/enterprise/metabase”OSS ships password, Google sign-in and LDAP; SAML/JWT/OIDC SSO are paid”apps/enterprise/metabase/app.yaml:75

The bucket definition and requirement 4 now state the same test. The apps/enterprise/ bucket was defined when the buckets landed as the apps where “SSO/core needs a paid license with no free workaround” (commit bcc4ca425). That is the payment bar, app-side, written a day before the ruling that generalizes it. The same commit records the contrapositive: sonarqube was kept at apps/ root “because it now has a working free SAML path AND an auth.disable gated path”, free to use, so not enterprise. The two definitions are compatible and no bucket membership moves.

4. Half two: no-SaaS becomes a managed-cloud eligibility axis

Section titled “4. Half two: no-SaaS becomes a managed-cloud eligibility axis”

The concern requirement 4 used to carry is real; it was filed under the wrong heading. It is now its own axis, recorded in docs/guide/apps/licensing-postures.md §“Managed-cloud eligibility: a separate axis, and not a flagship one”.

It is explicitly NOT a flagship criterion. A component under SSPL, BSL, ELv2, RSALv2 or FCL never costs an app the flagship label. What it does is constrain a product SHC has not shipped.

And it is a recorded future constraint, not a live one. SHC’s managed cloud is Phase 2 roadmap: docs/internal/product/business-plan.md:855, “§14.3 Phase 2 (Months 6–12): Add Managed Cloud”, with the Managed tiers launching progressively from month 7. Nothing in the catalogue is offered as a managed service today, so no clause in this family is currently binding on SHC. The axis exists so that the day the managed cloud does ship, the eligibility question is already answered per component instead of being rediscovered under delivery pressure.

  • The fitness check. flagship-contract measures the mechanical half only (SSO surface, backup.yaml, tests/tests.bats, league exclusivity, the -classic ban). License posture was already on the tier doc’s deliberately-not-measured list (“a grep cannot referee them, so the check does not pretend to”), and a payment bar is no more greppable than a license family. This ruling adds no check and changes no check.
  • Requirement 1. The SSO ladder’s “the claimed tier must be reachable WITHOUT a paid license” clause stays exactly as written. It is now the SSO-specific case of the general bar rather than a separate idea, and the two are worded to agree.
  • apps/enterprise/ membership. Both members fail the new bar too (§3).
  • Zero apps are labeled. This ruling changes the contract, not the roster.

Verified in-repo at 5c83f061b: huly’s EPL-2.0 identification (docs/guide/apps/custom-builds.md); huly’s two bundled pins and that both are tag-pinned rather than digest-pinned (apps/huly/vars.yaml:147-148); elasticsearch and redpanda license families (licensing-postures.md); highlight’s LICENSE_KEY wiring (compose.yaml:57); metabase’s paid-SSO statement (app.yaml:75); the apps/enterprise/ bucket definition (commit bcc4ca425); the managed-cloud Phase 2 placement (business-plan.md:855).

unverified: do not assert these without a source: the Redpanda BSL Additional Use Grant text for the exact pinned tag v24.3.6; whether Keygen CE self-hosting is free of charge (the repo records it as “fair-source (self-hostable CE)” with the restriction “targets reselling Keygen itself as a service” (docs/internal/ee-licensing-design.md:286-292), which speaks to the no-compete axis, not to price); the exact Fair Core License terms, which docs/decisions/2026-08-06-flagship-tier.md has carried as verify since it landed.