Bulk Data

The *Many methods take arrays and return the created records or ids; the *Stream methods take any iterable, hold nothing in memory, and return a count. Use streams for ingest jobs where the input is larger than you want a PHP array to be. Bulk methods take the collection lock once and, on the log and queue, buffer many records per fsync, which makes their bulk writes several times faster per record than one-at-a-time calls. DocStore bulk writes still sync each record file; they save lock churn, not fsyncs.

DocStore bulk APIs:

$records = $docs->putMany([
    ['slug' => 'home', 'kind' => 'page'],
    ['slug' => 'about', 'kind' => 'page'],
]);

$count = $docs->putStream($largeIterable);

$docs->exportJsonl(__DIR__ . '/pages.jsonl');
$docs->importJsonl(__DIR__ . '/pages.jsonl');

JSONL rows are one object per line, either {"id": "...", "data": {...}} or a bare data object (an id is generated on import). Rows may carry explicit ids, so export and import round-trip a collection exactly.

SegmentedLog bulk APIs:

$log->appendMany([
    ['type' => 'page.saved'],
    ['type' => 'page.rendered'],
]);

$log->appendStream($largeIterable);

Queue bulk APIs:

$queue->enqueueMany([
    ['task' => 'render', 'pageId' => $home->id()],
    ['task' => 'notify', 'pageId' => $home->id()],
]);

$jobs = $queue->claimMany(100);
$queue->completeMany(array_map(fn ($job) => $job->id(), $jobs));

Optional partitions place a log's segments under a per-day or per-month directory (events/partitions/2026-07-05/), so old periods can be archived or deleted as whole directories:

$daily = new Storh\SegmentedLog($root, 'events', partition: 'daily');
$monthly = new Storh\SegmentedLog($root, 'events', partition: 'monthly');

Retention deletes log records older than a cutoff and compacts the reclaimed space; purgeDone drops completed queue jobs from memory:

$deleted = $log->retain()->olderThanDays(90)->compact();
$queue->purgeDone(olderThanSeconds: 86400);