History

v1.0 (TBD)

Update to python-bmds v1.0, including BMDS 3.3
Add full javascript frontend

v0.6 (2018-09-16)

Update to BMDS v0.11.0
Use black to standardize code formatting
Update libraries to latest versions; including Django 2.1 LTS
Add CI {tests, black, flake8}

v0.5.1 (2018-04-19)

Update to BMDS v0.10.0

v0.5 (2017-02-27)

Refactor application
Updated deployment procedure
- use ini file instead of environment variables
- add instructions for installing as Windows services
Update to BMDS v0.9.1
Update to Bootstrap v4

v0.4.1 (2017-11-07)

Update to BMDS v0.9.0 (added BMDS version 2.7)
Backwards-compatible updates to dataset API to include new optional dose/response units
Updated deployment procedure

v0.4.0 (2017-06-06)

Vacuum sqlite3 database after job deletion to propagate data removal
Update to BMDS v0.8.1
Use asyncio for drunner tasks (unfortunately, more instable for linux/mac users)

v0.3.0 (2017-03-17)

In addition to catching global errors in BMD execution, try to catch errors the individual dataset level. Therefore, if a batch has a thousand datasets and only one fails, note this, but return the successful results
Log failed dataset with traceback
Update docs on deployment procedure and simple R usage
Update to BMDS v0.7.0

v0.2.4 (2017-03-03)

Updated to BMDS v0.5.4 (don't run models w/ parameters > dose groups)

v0.2.3 (2017-03-02)

Updated to BMDS v0.5.3 (more resilient to bad dataset errors)

v0.2.2 (2017-02-15)

Updated to BMDS v0.5.2 (can run datasets with 3 dose groups w/o failure)

v0.2.1 (2016-12-23)

Updated to BMDS v0.5.0
Minor updates to output JSON. See details in the output documentation
- For each output, added dataset_index integer.
- For each model, added name, model_index, and model_version

v0.2.0 (2016-12-14)

For continuous summary datasets, rename responses array to means
Added quickstart for using R
Added the ability to enter continuous individual datasets
Added new option on requests to allow immediate - if True and enabled on the server, a user can request an immediate response to a job, and therefore polling for results is no-longer required. False by default for server performance.

v0.1.0 (2016-12-05)

Initial release