mike1813
commented
1 year ago Replaced incorrect package assignments. Also added a PackageInterdependencies.csv table which I found useful when trying to sort out what they should have been. Now commited on branch 6a.
Entities are assigned to 'packages' which are used as labels for identifiable sub-model within the overall domain model.
The 'package' field is then used as a sort index to provide a 'canonical' sequence within each CSV table. Changes that affect one specific package, although distributed across CSV files, will be localised within each CSV file. This gives Git at least some chance to handle merges between branches where different submodels have been changed in each branch. Without this it would be almost impossible for two sets of logically sensible changes not to create merge conflicts.
Unfortunately, in the rush to produce a stable version prior to migrating the repository to GitLab, some package assignments were made automatically using a SQL query, and while most assignments are sensible, some threat assignments became scrambled.
This should be fixed as a matter of urgency on branch 6a, and then merged into development branches, before any other development of any branches. If we don't do this, we'll just produce a lot of merge conflicts.