>>> [!note] Migrated issue <!-- Drupal.org comment --> <!-- Migrated from issue #3619204. --> Reported by: [mably](https://www.drupal.org/user/3375160) Related to !81 >>> <p>Ninety method names lead with a noun or an adjective rather than a verb. <span class="drupalorg-gitlab-issue-link drupalorg-gitlab-link-wrapper"><a href="https://git.drupalcode.org/project/pdv/-/work_items/3618947" class="drupalorg-gitlab-link">https://git.drupalcode.org/project/pdv/-/work_items/3618947</a></span> brought the published interfaces in line with core and stopped there by design, so the internals were never swept: <em>consumerLabel()</em>, <em>recordsTable()</em>, <em>tenantId()</em>, <em>aad()</em>, <em>pending()</em>, <em>total()</em>.</p> <p>Measured against all 23,174 method declarations in core, before this change: 817 of pdv's 993 declarations (82.3%) lead with a token core uses the same way, 68 (6.8%) lead with a domain verb core has no word for, and 108 (10.9%) lead with a noun or adjective. Accessor density is already right: <em>get</em> prefixes 26.9% of pdv's methods and 26.9% of core's, so this is about the remaining tenth, not about a missing accessor habit.</p> <h3>What the names become</h3> <p>The return type decides most of them: a value getter takes <em>get</em>, a render array takes <em>build</em>, an object factory takes <em>create</em>, an int count takes <em>count</em>, a bool takes <em>is</em> or <em>has</em>. Every verb used here is attested in core: build 724 times, create 1814, is 649, has 262, process 254, count 60, exists 47.</p> <p>Eight needed the code read rather than a rule. <em>pending()</em> and <em>total()</em> both return an int and become <em>countPendingForTenant()</em> and <em>countSubjectsForTenant()</em>, because a bare <em>getTotal()</em> tells a reader nothing and the docblock already calls total that tenant's rotation denominator. <em>siteFeaturesDisabled()</em> and <em>vaultFeaturesDisabled()</em> become <em>isDisabledSiteWide()</em> and <em>isDisabledForCurrentUser()</em>: the second is not about the vault at all, it ORs the site switch with the per-user opt-out, and the old pair hid that difference. <em>blankPassphraseValue()</em> takes an element and returns an element, so it is a #process callback and takes core's <em>processX</em> shape. Three more spell out a scalar argument at the call site, applying the same rule to itself: <em>suiteFor($tag)</em> becomes <em>getSuiteForTag()</em>.</p> <h3>Ten names are deliberately kept</h3> <p>Two procedural hooks (<em>pdv_install</em>, <em>pdv_demo_page_attachments</em>); three {@inheritdoc} overrides, whose names belong to the parent; three hook-derived methods on Hooks classes, which follow the hook name as <span class="drupalorg-gitlab-issue-link drupalorg-gitlab-link-wrapper"><a href="https://git.drupalcode.org/project/pdv/-/work_items/3618947" class="drupalorg-gitlab-link">https://git.drupalcode.org/project/pdv/-/work_items/3618947</a></span> established; and three route callbacks, <em>raw()</em>, <em>kindAccess()</em> and <em>kindLabels()</em>. Core names a route callback after the resource rather than verb-first, which is why its own <em>overview()</em> appears six times, and why pdv's four <em>overview()</em> and its <em>items()</em> never came up as findings.</p> <h3>Why now rather than at 1.0</h3> <p>A rename crossing a module boundary breaks a named-argument caller fatally, and yoyaku consumes this module's @api. Pre-1.0 there are no update hooks and no stored value is involved, so the whole cost is textual today and is a downstream breakage later. Eighteen of the ninety are public, so this is not purely internal churn.</p> <p>AI-Generated: Yes (Claude Code was used to run the census against core, choose the names, and write the renames on the merge request. I reviewed and ran the work myself before posting it.)</p>