updates to prebuilder

[xtuml]

updates to prebuilder

[cortlandstarrett]

With the command line nature, some testing could be added pretty easily. The approach to manual testing I was doing was comparing the output of BridgePoint prebuilder with pyxtuml prebuilder. First, I wanted to understand the differences. But if we achieve getting them to match, we can then test them against each other going forward. It would be easy to supply some 'golden SQL' from models providing substantial coverage.
If I begin using this utility more, I will help add some tests.

[john-tornblom]

Unfortunately, there is a lack of unit tests in pyxtuml for actions that interact with e.g. instances, state machines and ports. In particular, the modules sourcegen.py, prebuild.py, and oal.py desperately needs attention from a test engineer (see links for coverage metrics).

Some of these things where testing manually on a few test models available at the time, but as pointed out above, there are missing pieces. Going forward, I believe it is essential to have a test model/set of test models that systematically covers the input space of the prebuilder.

[leviathan747]

Can this issue now be closed? PR serviced here: #13

[cortlandstarrett]

It could. However, I have one more quick check I want to make. If it is easy to link up S_IRDTs instead of core types in set arithmetic, that would be an improvement.

I will confirm that set arithmetic now parses cleanly. My test included all of the OAL in mc/model/mcooa, escher and mcshare.

[leviathan747]

I recommend you raise a separate issue for the IRDT issue. The two problems you called out originally in this issue have been resolved.

[cortlandstarrett]

Closing issue as fixed. I will raise another on IRDTs if I think it is a good idea to make that change.