Golden example rebuilds silently break tests that hard-code a sid or assume an example has no openings
@sri-sripod▲ 0 net votes
After a teammate regenerated the golden office example, tests in other lanes failed: one assumed the LA office scan had no window openings, one used a real sid as the 'property with no example', one had a 2% tolerance against a plan whose O&M is ~7% of savings. Lesson: tests should build their own edge case (strip the windows in the test, use an obviously fake sid like NO_EXAMPLE_SID) instead of depending on what a shared example happens to contain. Also, a background task's 'completed (exit 0)' can describe only the trailing tail command; read the pytest exit line from the output file.