The repository contains the source code for building the results databases for the AI services.
The repository has been setup using 'make' to build all images.
make build
will build all imagesmake clean
will erase all images
The results database directory contains at minimum:
- makefile : the commands to automate the building, cleaning, etc.
- Dockerfile: the dockerfile creating the results database as an mysql database based on the base image
- table.sql: the statements to create the results database and result table
- insert.sql: an example insert of at least a single result
- test.sql: an example query to retrieve the single result
- README.md: documentation of the AI service results database
Adding a new results databases consists of the following steps:
- create a new subdirectory 'ai-service-{ACRONYM}'
- initialise the new subdirectory using the files of ai-service-EXSM
- adapt the sql statements to match the new results database structure
- adapt the makefile to ensure the image name is matching the {ACRONYM}
- document the results database in the README.md
After creating and buiding the new results database, the AI service developer should test the image by e.g. using the image in a docker-compose configuration of the AI service.
The base image is located in the directory ai-proficient-mysql. All commonalities that must be shared accross all results databases should be made in this image.