This post is by Martin Keegan, project lead on OpenSpending.
Changes in v0.10:
Data loading has been separated from the main web application. Web-based and command-line tools for data wranglers to load/reload datasets
have been separated from the main end-user facing web application.
They now reside in separate code repositories; there has been signficant
reorganisation of the resulting source trees
More tests. Test coverage and organisation has been improved, and considerably more
of the tests pass
Model overhaul. The integration between python and MongoDB has been effectively replaced
Removed dependency on celery. Long-running import tasks used to use a third-party subsystem called
celery, which proved an administration and reliability hassle. It has
been replaced by our own code.
Command line interface tidied up
Data-wrangler workflow improvements. Drop dataset is now supported, as are CLI and WUI tools for tagging CKAN packages for use with OpenSpending