Closed dcloud closed 9 years ago
When generating a CSV in a celery task, the file pointer location is on the task queue machine, which is a problem.
Switch search views to new-style view classes as part of this.
Progress in 3e99ec2...9d37ffe for exporting to CSV from search. Waiting for user research to determine if JSON API is needed/desirabled.
Also, Haystack's new-style view classes are broken in 2.4.0, so no dice there.
1b817de...19dd056 improve the CSV export using postgreSQL intarray extension to query and order Datasets by list of ids. Better than iterating over searchqueryresult and getting each object individually.
Now there is a CSVExportMixin for objects in 4f143c208bb9f40960eec420b7587622339ce8a1. Seems like we're good here.
Probably best to think of in terms of user downloads (over API). So this should probably include:
Having search-based CSV creation could be transactionally expensive, but perhaps we could optimize PostgreSQL for generating CSV data efficiently.