After adding layer 2 and 3 back to the UML model, do this work.
Instead of copying all the files and then running the conversion do both steps for each file.
Update the service that uploads files to target/mbt to immediately update the model.
Use a queue to make it go faster.
Also make sure only step libraries and interfaces are sent and not every .java file.
To save time writing the files generated from the model, do a POST and check if the file has changed, if it hasn't, sent a blank file back. The idea is that only the feature file is sent and only java files that are changed are transmitted back and written to the disk.
Another change to make things efficient, and this probably needs its own work item is to select UML classes/interactions by tags when generating the code. Without this, all the adoc/feature files have to be parsed again, just to skip the unwanted features/scenarios etc.
After adding layer 2 and 3 back to the UML model, do this work. Instead of copying all the files and then running the conversion do both steps for each file. Update the service that uploads files to target/mbt to immediately update the model. Use a queue to make it go faster. Also make sure only step libraries and interfaces are sent and not every .java file. To save time writing the files generated from the model, do a POST and check if the file has changed, if it hasn't, sent a blank file back. The idea is that only the feature file is sent and only java files that are changed are transmitted back and written to the disk. Another change to make things efficient, and this probably needs its own work item is to select UML classes/interactions by tags when generating the code. Without this, all the adoc/feature files have to be parsed again, just to skip the unwanted features/scenarios etc.