Hi, yesterday I fixed a bug in the post-processing of large requests (Ones that are split up) … if some items return “INVALID_ADDRES” or “UNKNOWN_OBJECT” the code failed. Now it checks the return code. I tried for about 2 hours to produce a test-scenario to reproduce this problem, but finally gave up. I fixed the bug by debugging a real S7 interaction and it was pretty obvious what was the reason, but it was not so easy to reproduce the situation in a unit test without manually modelling the interaction …
For the new generated drivers I made the code generator generate all classes so I can parse and serialize them to JSON and XML using Jackson. This made it really easy to write unit tests … now writing such tests for the old ones it pretty nerve-wrecking. Does it make sense to annotate the existing ones so we can write simpler tests? Or should we concentrate on the new world and rather work on building the replacements faster? Chris
