Django Chat Django Tasks - Jake Howard
6 snips
Apr 15, 2026 Jake Howard, senior systems engineer at Torchbox and Django/Wagtail contributor, talks about creating a generic background task API and how it was upstreamed into Django. He covers cross-platform challenges, task timeouts, JSON-only arguments, and why production backends stay separate. He also discusses security triage and the rising noise from AI-generated reports.
AI Snips
Chapters
Books
Transcript
Episode notes
Django Tasks Originated In Wagtail's Publish Problems
- Jake explains Django Tasks grew from Wagtail's need to move publish-time work (indexing, CDN cache) out of request/response to background processing.
- Wagtail initially considered its own solution, but Tom Usher pushed for a generic API so multiple backends could be swapped without forcing Celery or RQ on users.
A Shim API Lets Packages Be Backend Agnostic
- Jake realized a shim API above concrete task systems lets package maintainers write background calls without choosing a single queue implementation.
- That abstraction makes it trivial to switch between Celery, RQ, database backends at deployment time, keeping package APIs stable.
Killing Long Running Python Tasks Became A Side Quest
- Jake dove into the complex problem of reliably killing long-running Python tasks across platforms and even gave a PyCon UK talk about it.
- Cross-platform timeouts are tricky because Linux tricks may not work on Windows, prompting deeper investigation and platform-specific approaches.
