v0.1.27

What's Changed

• Pin flake8<4 by @dchiquito in #561
• Always add schemakey by @dchiquito in #562

Full Changelog: v0.1.26...v0.1.27

v0.1.26

What's Changed

• Expose port for RabbitMQ management console by @mvandenburgh in #554
• Don't validate Assets with no sha256 by @dchiquito in #558
• Add rudimentary admin-only dashboard by @dchiquito in #557

Full Changelog: v0.1.25...v0.1.26

v0.1.25

What's Changed

• Bump dandischema to 0.4.3 by @dchiquito in #556

Full Changelog: v0.1.24...v0.1.25

v0.1.24

What's Changed

• Checksum queue by @dchiquito in #551
  Create a separate Celery queue for checksum calculations to avoid clogging the global task queue during uploads.

Full Changelog: v0.1.23...v0.1.24

v0.1.23

• limit admin inline tables to 10 entries
• bump dandischema to 0.4.2

v0.1.22

Optimize Celery tasks

• Asset metadata validation is very fast, so it is now done in the request thread to reduce the number of tasks dispatched.
• Version metadata validation is now done once a minute (configurable) rather than a new validation task being dispatched on every dandiset modification.
  During large uploads, each file upload would trigger it's own version metadata validation, which created an excessive number of tasks that were simply revalidating for no reason.
  This new scheme effectively debounces version metadata validation.
• Draft metadata manifests are now written out whenever metadata is changed, rather than nightly.

v0.1.21

Separate celery beat process into a separate dyno in Heroku Procfile.
This allows us to scale the worker without inadvertently triggering duplicate beat events.

v0.1.20

Bump dandischema to 0.4.1, schemaVersion to 0.6.0

v0.1.19

Add CELERY_WORKER_CONCURRENCY environment variable setting

v0.1.18

• Fix admin console blankable field bug
• Use GitHub Actions for CI
• Add manual Asset garbage collection script
• Add manual versioned object deletion script
• Add dandiset import script
• Write draft manifest files nightly