I want the nightly regression minus anything that runs against mocked network, so:
sdods run -p demo-shop -e staging -l ui -t "@regression and not @mock"It ran fewer scenarios than -t @regression on the same layer, which is what I wanted, but I cannot tell whether -l ui was actually honoured or whether my expression replaced it and the ui-only result was a coincidence of what we have tagged. How do the two combine?
Accepted answer
They combine, they do not compete. The layer is applied as a tag of its own and your expression is wrapped in parentheses next to it, so what actually runs is:
(@ui) and (@regression and not @mock)That is per layer — if you pass -l ui -l api you get two run targets, each with its own layer wrapped around the same expression, which is why -t "@ui or @api" is never the way to ask for both. Use -l for the layer and -t for everything else.
The parentheses matter more than they look. Without them an expression starting with not would bind against the layer and quietly invert the wrong thing.
To convince yourself rather than infer it from counts, ask what would run without running it:
sdods run -p demo-shop -e staging -l ui -t "@regression and not @mock" --listThat prints the run targets and the tests each one selected. It is also the fastest way to catch an expression that matched nothing, which otherwise looks like a healthy green run.
Since you are excluding @mock — check that it is declared in tags.extra. not @mock against a tag nobody has declared still works as an expression, but if the tag was a typo somewhere you are excluding nothing and the run is just regression. The lint warning for the misspelled one is your signal.
Have a different question? Ask it.