docs: add design-rationale commentary to core support classes

Enriches class-level docblocks and inline comments across the shared
utility classes with the "why" behind design decisions that aren't
obvious from reading the code alone:

- Cache       two-tier rationale, atomic-write semantics, failure modes
- Curl        single-use-per-instance rationale, default option choices
- Log         wrapper rationale, redaction expectations for callers
- Database    auto-migration philosophy, schema-versioning approach
- ServerResource  flat-array rationale, interfaces[0]-only limit called
              out for future maintainers, unit-conversion map
- ConfigureService  why a sibling of ModuleFunctions, catalogue caching
              policy, cp-in-constructor reasoning

Pure documentation — no code changes, all files remain lint-clean and
Pint-formatted.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
Prophet731
2026-04-17 21:08:37 -04:00
parent ad85439dfb
commit 8a88862364
6 changed files with 197 additions and 12 deletions

View File

@@ -3,11 +3,47 @@
namespace WHMCS\Module\Server\VirtFusionDirect;
/**
* Two-tier cache: uses Redis when the ext-redis extension is available, with an atomic
* filesystem fallback stored in the system temp directory.
* Two-tier cache: Redis when ext-redis is available, atomic filesystem fallback otherwise.
*
* WHY TWO TIERS
* -------------
* The module is deployed to every kind of WHMCS install — shared hosting, dedicated
* VPS, bare-metal. Requiring Redis would exclude the long tail of smaller operators
* who never installed the extension. But operators who DO have Redis get a huge
* performance win for cross-request caching (PowerDNS zone lists, OS template
* listings, traffic stats), so we opportunistically use it when present.
*
* The fallback is filesystem-based, using the OS temp directory. Writes are atomic
* via the classic tmp-file + rename pattern so a process crash mid-write can never
* corrupt an existing cache entry for another concurrent reader.
*
* EXPIRY SEMANTICS
* ----------------
* Redis: native SETEX — the key auto-expires on the server side.
* Filesystem: we store a JSON envelope {expires, data} and check expiry on read,
* deleting stale entries lazily. This means a cache with lots of expired entries
* will slowly accumulate files until accessed — acceptable for the module's scale
* (tens-to-hundreds of keys per install) but worth noting if someone ports this
* to a higher-volume context.
*
* NAMESPACE
* ---------
* Every key is prefixed with "vfd:" to avoid collisions with anything else that
* shares the Redis instance. Nested keys add their own sub-prefix (e.g.
* "pdns:zones:<hash>" for PowerDNS zone lists) for semantic clarity in the logs.
*
* FAILURE MODES
* -------------
* Redis unreachable: we set $redisAvailable = false on first failure, which
* permanently disables Redis for the rest of this PHP process (subsequent calls
* skip straight to the file cache). Prevents paying reconnect overhead on every
* miss when Redis is down.
* File cache write fails: silently skipped. Cache is best-effort; a failed SET
* just means the next GET will re-fetch from the authoritative source.
*/
class Cache
{
/** Module-global key prefix — keeps us out of Redis key collisions on shared installs. */
const PREFIX = 'vfd:';
/** @var \Redis|null */
@@ -150,12 +186,18 @@ class Cache
}
}
// File cache fallback with atomic write (race condition safe)
// File cache fallback with atomic write.
// Writing to a temp file + rename ensures that readers either see the
// complete previous entry or the complete new entry — never a truncated
// or partially-written file. getmypid() suffix lets concurrent PHP
// processes write to the same key without stomping each other's temp files.
$path = self::filePath($key);
$tmp = $path . '.' . getmypid() . '.tmp';
$entry = json_encode(['expires' => time() + $ttl, 'data' => $value]);
if (@file_put_contents($tmp, $entry, LOCK_EX) !== false) {
// rename() is atomic on POSIX when source and target are on the same
// filesystem (which they always are here — both in sys_get_temp_dir).
@rename($tmp, $path);
}
}