/lab
Small tools from the work I do every day.
Built for poking at. Everything runs in your browser or against this site's own endpoints, and nothing you do here is stored.
01 · GET /api/me
See exactly what your browser sends.
Every debugging session starts with “what did the server actually receive?” Send a real GET /api/me and read it back.
// press send. the server replies with exactly what it received from you.
02 · schedule
Read any cron line or Laravel schedule.
Paste a cron expression like 0 9 * * 1-5 or a Laravel chain like ->weekdays()->dailyAt('09:00'). Get it in plain English, the next runs in your time zone, and a map of the week.
At 09:00, on Monday through Friday.
cron 0 9 * * 1-5tz UTC
- next 8 runs
- 1in 28h
- 2in 2d
- 3in 3d
- 4in 4d
- 5in 5d
- 6in 8d
- 7in 9d
- 8in 10d
the next 7 days, hour by hour5 runs
03 · queue:work
Watch a queue fail, retry and recover.
--tries, --backoff and the number of workers decide whether a bad minute becomes an incident. Drag the sliders and see why.
❯ php artisan queue:work --tries=3 --backoff=3 × 3
pending 0
workers 3
- worker 1idle
- worker 2idle
- worker 3idle
backoff 0
- processed
- 0
- retried
- 0
- failed_jobs
- 0
- throughput
- 0.0/s
select * from failed_jobs order by id desc limit 4;
(0 rows)
Try it: raise the failure rate and drop --tries to 1, and watch failed_jobs fill up. Then add workers and a backoff, and see the queue drain while retries recover. Simulated in your browser.
04 · N+1
Watch one innocent loop send a query per row.
The most common Laravel performance bug hides in a Blade loop. Flip between Post::all() and Post::with('author')->get() and watch the query log.
PostController.php · index.blade.php
// controller $posts = Post::all(); {{-- view --}} @foreach ($posts as $post) {{ $post->author->name }} @endforeach
// each ->author fires its own query: 50 posts, 50 extra queries
0 / 51 queries · 0ms waiting on the database
- lazy
- 51 queries · 204ms
- eager
- 2 queries · 8ms
Catch it before production: add Model::preventLazyLoading(! app()->isProduction()); to AppServiceProvider::boot() and every lazy load throws in development. Query counts are exact; the timing is simulated at the round trip you pick.