C.32.ACE:11 - SoTA-Echoing
These rows document how source practice contributes to eval-program fields, result-use boundaries, and refresh conditions in C.32.ACE.
| Source to inspect | Why this source is load-bearing here | Transfer into ACE | Concrete ACE mutation | Blocked overread |
|---|---|---|---|---|
FPF source presentation ТриПрототипаТриОшибки (2022-03-26) | Separates variant, prototype, candidate, stake, solution, error, eval as variant comparison, and testing as error checking; also requires fair comparison and indicator selection. | Make eval a typed architecture evaluation over declared candidates and criteria. | test is admitted only as one evalOperation when expectation failure or hard-constraint checking is current; parity frame and result form are mandatory. | A test, check, or pass-fail result is not the whole eval program, not the criterion, and not the decision. |
Ford, Parsons, Kua, and Sadalage, Building Evolutionary Architectures, 2nd ed. (https://www.oreilly.com/library/view/building-evolutionary-architectures/9781492097532/) | Current practitioner source for incremental architecture governance and feedback under source-side fitness-function terminology. | Restore the source term to FPF eval programs over ACS rows, Q-Bundle slots, candidate structures, and parity frames. | ACE record names evaluated rows, purpose, scope, eval operation, trigger mode, result form, run context, receiving use, and refresh or retire condition. | Fitness-function wording is not imported as the FPF object name or as a new architecture characteristic kind. |
Software Architecture Metrics (https://www.oreilly.com/library/view/software-architecture-metrics/9781098112226/) | Current practitioner source for metric categories and governance practice after quality goals are named. | Carry metric-cadence distinctions as eval-program fields, not as criteria rows. | ACE distinguishes scope, trigger mode, result form, run context, method refs, and refresh or retire condition. | A metric, dashboard, rank, or score is not a project criterion, selected architecture, or architecture decision. |
Ford, Richards, Sadalage, and Dehghani, Software Architecture: The Hard Parts (https://www.oreilly.com/library/view/software-architecture-the/9781492086888/) | Mature practitioner source for objective definitions, trade-off analysis, and least-worst choices under competing characteristics. | Require declared criteria rows and protected counter-characteristics before synthesis, comparison, or selection uses eval results. | ACE rows carry proxy risk, protected counter-characteristics, and receiving use before result-driven action. | A better reading or rank does not authorize comparison, selection, choice, G.5 selected-set result declaration, actual publication, or decision by itself. |
Goodhart and proxy-risk line, plus current FPF E.13 | Optimized proxies can detach from the declared architecture concern. | Keep proxy repair in E.13 while ACE records the risk before result use. | ACE requires proxy risk and protected counter-characteristics; proxy drift requires E.13. | An eval result cannot replace the declared architecture concern. |
Current FPF C.16, C.25, E.23, A.19.CPM, A.19.SelectorMechanism, G.5, E.17, E.24.PUB, and C.11 | Existing patterns for the next questions for measurement, Q-Bundles, repeated improvement, comparison, selection, selected-set result declaration, publication, and local choice. | Keep ACE as the eval-program framing and typed-result dispatch boundary. | Use C.16 for measurement validity and readings, C.25 for composite quality, E.23 for improvement feedback, A.19.CPM for comparison, A.19.SelectorMechanism for set-returning selection, G.5 for selected-set result declaration, E.17 for a source-backed publication face and return to source, E.24.PUB for an actual publication occurrence and availability, and C.11 for local choice. Accept each actual typed result only under the definition and test for that result. | ACE does not validate measurement, define every eval result, define Q-Bundles, compare, select, declare a selected-set result under G.5, publish it to an audience, choose, or decide. |
Source-currentness boundary. Use each source row only for the ACE eval-program field, result-use boundary, or refresh condition named in that row. Recheck the row when a named book edition, source presentation, FPF pattern for the next question, metric practice, or evolutionary-architecture practice changes the transferred move. If the project wants criteria-row admission, measurement validity, Q-Bundle structure, explicit comparison, selection, selected-set result declaration, actual publication, local choice, evidence, assurance, or decision use, leave ACE and open the pattern for the next question.