Upcoming release 0.8 - czcorpus/kontext GitHub Wiki

Notes for the upcoming release 0.8

The 0.8 release is planned to follow soon after 0.7 but the transition is expected to be much easier than from 0.6 to 0.7. API changes will be negligible at worst and changes in configuration will be auto-upgradeable.

Asynchronous calculations

(concordance, wordlist and collocations auxiliary data)

  • new alternative backend based on the Celery project
    • it will finally detach the calculation from KonText's process tree
    • the previous solution based on the multiprocessing module will be still available
    • it will be still limited to sorted concordances (just as in bonito-open) because Manatee does not offer any sort of progressive result fetching once "shuffle" is enabled

Periodic maintenance tasks

  • it will be possible to run regular maintenance scripts/tasks within Celery Beat
    • among other things, this will allow a common administration of asynchronous and periodic tasks (e.g. via Celery Flower web interface)

Configuration cleanup

  • some values will be moved to more convenient locations
  • unused values will be removed
  • a RelaxNG schema will be available to be able to validate general KonText configuration

Subcorpus "mixer"

This is an experimental module allowing user to mix a subcorpus with defined ratios of specified content type (e.g. 30% journalism, 50% fiction, 20% poetry).

Other updates

  • the release will be primarily tested with new WSGI environment Gunicorn along with a reverse proxy (Nginx, Apache).
    • it will still run within mod_wsgi (and possibly other compatible servers)