{# T12 - Software assurance record, CSA-structured (D1 section 1's T12 row; A-P3). Every
   value comes from fixtures_report.json through proofpack.render.t12's context; no
   arithmetic here. The intended-use slot is the manufacturer's (placeholder until they
   write it); the sign-off cells are blank: ProofPack signs nothing. #}
{% extends "base.html" %}
{% block pages %}
<section class="page" id="page-1">
<header class="page-header">{{ header_engine }}</header>
<h1>{{ template_name }}</h1>
<p class="cover-note disclaimer">Attachment set prepared for the manufacturer's own submission or record. Not a submission. No regulator has endorsed this tool.</p>
{{ cover_stamps() }}
<p>{{ csa_sentence }} <span class="unverified">{{ csa_unverified }}</span></p>
<p class="supports">{{ supports_sentence }}</p>
<p class="print-note">{{ print_sentence }}</p>
<h2 id="t12-s1">1. Intended use</h2>
{{ margin_note(anchor.scope) }}
<p>The manufacturer states here what ProofPack is used for in their process and which of their records its output enters. ProofPack does not draft this text.</p>
{{ placeholder(slot) }}
<h2 id="t12-s2">2. Failure modes the engine inspects</h2>
<p>Each code below is defined in <code>proofpack/errors.py</code>. The second column is the condition the code inspects; the third is the engine action, or that no engine path raises the code in this version. The last column is for the manufacturer's own risk rating.</p>
<div class="table-wrap">
<table class="failure-modes">
<caption>Engine codes, from the engine's own code table</caption>
<thead><tr><th scope="col">Code</th><th scope="col">Condition inspected</th><th scope="col">Engine action</th><th scope="col">Manufacturer's risk rating</th></tr></thead>
<tbody>
{% for f in failure_modes %}
<tr data-code="{{ f.code }}"><td class="mono">{{ f.code }}</td><td>{{ f.inspects }}</td><td>{{ f.does }}</td><td class="signoff"></td></tr>
{% endfor %}
</tbody>
</table>
</div>
{{ page_footer() }}
</section>
<section class="page" id="page-2">
<header class="page-header">{{ header_engine }}</header>
<h2 id="t12-s3">3. Assurance evidence: the fixture register (fixtures_report.json)</h2>
{{ margin_note(anchor.evidence) }}
<p>Every row of the report is printed, whatever its status. A row reads "matched" only when every compared value lies within the row's tolerance; "not matched", "no oracle recorded", "not built in this version" and "compared by the test suite only" are printed as they stand.</p>
<table class="environment">
<caption>Where the report was generated</caption>
<tbody>
{% for label, value in environment %}
<tr><th scope="row">{{ label }}</th><td class="mono">{{ value }}</td></tr>
{% endfor %}
</tbody>
</table>
<table class="tolerances">
<caption>Tolerance classes (D1 sections 3.2 and 9)</caption>
<tbody>
{% for label, rule in tolerance_rules %}
<tr><th scope="row">{{ label }}</th><td>{{ rule }}</td></tr>
{% endfor %}
</tbody>
</table>
<table class="counts">
<caption>Report rows by status</caption>
<tbody>
<tr><th scope="row">Rows</th><td class="num" data-count="rows">{{ summary.rows }}</td></tr>
<tr><th scope="row">matched</th><td class="num" data-count="matched">{{ summary.matched }}</td></tr>
<tr><th scope="row">not matched</th><td class="num" data-count="not_matched">{{ summary.not_matched }}</td></tr>
<tr><th scope="row">no oracle recorded</th><td class="num" data-count="no_oracle_recorded">{{ summary.no_oracle_recorded }}</td></tr>
<tr><th scope="row">no independent oracle</th><td class="num" data-count="no_independent_oracle">{{ summary.no_independent_oracle }}</td></tr>
<tr><th scope="row">not built in this version</th><td class="num" data-count="not_built">{{ summary.not_built }}</td></tr>
<tr><th scope="row">compared by the test suite only</th><td class="num" data-count="suite_only">{{ summary.suite_only }}</td></tr>
<tr><th scope="row">Exit code of proofpack fixtures</th><td class="num" data-count="exit_code">{{ report.exit_code }}</td></tr>
</tbody>
</table>
<div class="table-wrap">
<table class="evidence">
<caption>One row per comparison or register entry; the largest deviation is the largest absolute difference between engine and oracle values in the row, and "—" where a value was not compared (the note names it) or the row has no comparison</caption>
<thead><tr><th scope="col">Row</th><th scope="col">What is compared</th><th scope="col">Oracle</th><th scope="col">Tolerance</th><th scope="col" class="num">Values</th><th scope="col" class="num">Largest deviation</th><th scope="col">Status</th><th scope="col">Note</th></tr></thead>
<tbody>
{% for r in evidence %}
<tr data-row="{{ r.id }}"><td class="mono">{{ r.id }}</td><td>{{ r.what }}</td><td class="mono">{{ r.oracle.file }}{% if r.oracle.entry %} ({{ r.oracle.entry }}){% endif %}{% if r.oracle.marking %} <span class="unverified">{{ r.oracle.marking }}</span>{% endif %}</td><td>{{ r.tolerance }}</td><td class="num">{{ r.n }}</td><td class="num mono">{{ r.deviation }}</td><td class="fixture-status" data-status="{{ r.status }}">{{ r.status_text }}</td><td>{% if r.reason_unverified %}<span class="unverified">{{ r.reason }}</span>{% else %}{{ r.reason }}{% endif %}{% if r.suite_tests %} ({{ r.suite_tests }}){% endif %}</td></tr>
{% endfor %}
</tbody>
</table>
</div>
{{ page_footer() }}
</section>
<section class="page" id="page-3">
<header class="page-header">{{ header_engine }}</header>
<h2 id="t12-s4">4. Installation and operational checks (performed and signed by the manufacturer)</h2>
<p>ProofPack prints the results below; the person, the date and the signature are the manufacturer's. ProofPack signs nothing.</p>
<h3 id="t12-s4-1">4.1 proofpack doctor --offline, as run by proofpack fixtures</h3>
<table class="doctor">
<caption>Environment checks recorded in the report ({{ doctor_flagged }} flagged by doctor); the last column is what doctor recorded, which the manufacturer reads</caption>
<thead><tr><th scope="col">Check</th><th scope="col">Flagged by doctor</th><th scope="col">Essential</th><th scope="col">Recorded</th></tr></thead>
<tbody>
{% for d in doctor %}
<tr data-check="{{ d.name }}"><td>{{ d.name }}</td><td>{{ d.flagged }}</td><td>{{ d.essential }}</td><td class="mono">{{ d.info }}</td></tr>
{% endfor %}
</tbody>
</table>
<p>The licence check is not printed in this table: T12 is written only under a licence that is usable (not refused, not expired past grace); its mark, if any, is on the cover.</p>
<table class="checklist">
<caption>Checklist: each step, what it inspects, and blank fields for the manufacturer's record</caption>
<thead><tr><th scope="col">Step</th><th scope="col">Action</th><th scope="col">What the step inspects</th><th scope="col">Record reference</th><th scope="col">Performed by</th><th scope="col">Date</th><th scope="col">Signature</th></tr></thead>
<tbody>
<tr data-step="doctor"><td>1</td><td><code>proofpack doctor --offline</code></td><td>Python, numpy, pyyaml and jsonschema present; the packaged schemas and guidance map load; write access; the licence file; the platform against the reference platform</td><td class="signoff"></td><td class="signoff"></td><td class="signoff"></td><td class="signoff"></td></tr>
<tr data-step="fixtures"><td>2</td><td><code>proofpack fixtures --offline</code></td><td>The fixture register rows above: {{ summary.matched }} matched and {{ summary.not_matched }} not matched of {{ summary.rows }} rows; exit code {{ report.exit_code }}</td><td class="signoff"></td><td class="signoff"></td><td class="signoff"></td><td class="signoff"></td></tr>
<tr data-step="mapping"><td>3</td><td><code>proofpack map --input &lt;your table&gt;</code>, then a person compares the mapping table with the manufacturer's data dictionary</td><td>That each column's role (label, score, prediction, case, site, attributes) is the one the data dictionary gives; the mapping.json SHA-256 printed in T8 is the file that person confirmed</td><td class="signoff"></td><td class="signoff"></td><td class="signoff"></td><td class="signoff"></td></tr>
<tr data-step="declarations"><td>4</td><td>The RA lead reads T8 section 3 (declarations echo) beside the criteria.yaml the manufacturer's team authored</td><td>Positive class, score orientation, thresholds, intended-use prevalence, subgroups, and each criterion's author, date and justification, as echoed</td><td class="signoff"></td><td class="signoff"></td><td class="signoff"></td><td class="signoff"></td></tr>
</tbody>
</table>
<h2 id="t12-s5">5. Determinism and tolerances</h2>
{{ margin_note(anchor.determinism) }}
<p class="tolerance">{{ tolerance_policy }}</p>
<p class="f17">{{ f17_sentence }}</p>
<p class="platform">{{ platform_sentence }}</p>
<p class="supports">{{ supports_sentence }}</p>
<h2 id="t12-s6">6. Guidance references</h2>
<table class="guidance">
<caption>Every anchor used on this page, labelled from the guidance map; a draft is a draft in the label and in the structured data</caption>
<thead><tr><th scope="col">Internal id</th><th scope="col">Document, version or date, status</th><th scope="col">Draft</th></tr></thead>
<tbody>
{% for ref in guidance_refs %}
<tr id="{{ ref.id }}"{% if ref.draft %} class="draft"{% endif %}><td class="mono">{{ ref.id }}</td><td>{{ ref.label }}</td><td>{% if ref.draft %}yes{% else %}no{% endif %}</td></tr>
{% endfor %}
</tbody>
</table>
{{ page_footer() }}
</section>
{% endblock %}