1) share fix-all test process
2) define common/consistent unit test patterns (using embedded strings? what about driving analysis from loose files to assist with editing test cases?)
3) pull together common test cases (with a special emphasis on negative patterns)
4) provide a 'reference repo' and automation that applies submitted analyzers in a repo branch and provides some baselining/diffing capabilities
5) provide performance testing automation
6) Roslyn diagnostics analyzers (mentioned previously): analyzers must be clean for these checks to be the best they can be
Ideas include:
1) share fix-all test process 2) define common/consistent unit test patterns (using embedded strings? what about driving analysis from loose files to assist with editing test cases?) 3) pull together common test cases (with a special emphasis on negative patterns) 4) provide a 'reference repo' and automation that applies submitted analyzers in a repo branch and provides some baselining/diffing capabilities 5) provide performance testing automation 6) Roslyn diagnostics analyzers (mentioned previously): analyzers must be clean for these checks to be the best they can be