While investigating if https://github.com/o3de/o3de-azslc/issues/37 was really fixed, I stumbled uppon remaining issues.
The fix of december https://github.com/o3de/o3de-azslc/commit/0c4eccd90108f6283a813f996e24acb4f6635364
definitely fixed a crash, but it wasn't investigated further after that. In fact in unlocked the door to at least 2 scenarios that don't finish compilation in a normal manner (fatal internal).
I added 2 tests to make sure we don't regress on those 2. And the simplest fix I could find, in the form of a mini 3 state automata in the function symbol info.
before:
case 1:
(no code output on stdout since fatal error interrupted)
case 2:
(no code output, only stderr ODR message)
While investigating if https://github.com/o3de/o3de-azslc/issues/37 was really fixed, I stumbled uppon remaining issues. The fix of december https://github.com/o3de/o3de-azslc/commit/0c4eccd90108f6283a813f996e24acb4f6635364 definitely fixed a crash, but it wasn't investigated further after that. In fact in unlocked the door to at least 2 scenarios that don't finish compilation in a normal manner (fatal internal). I added 2 tests to make sure we don't regress on those 2. And the simplest fix I could find, in the form of a mini 3 state automata in the function symbol info.
before: case 1:
(no code output on stdout since fatal error interrupted)
case 2:
(no code output, only stderr ODR message)
after:
Same for both 2 cases.