I'd like to be able to set
*epxected-failures* for cmucl to the set of known failures. This is to make it easy for me to track when old failures get fixed or new failures happen.
It seems less than ideal to have this set in the ansi-test code. It would be nice to be able to set this in some other way. I guess possibly complicated because the ansi-tests are well-contained. As I understand it, to run the tests you basically need to run doit.lsp to set up everything and run the tests. I suppose I could set expected-failures there, but that seems not right either.
Could these things be factored out a bit to have a setup.lsp file that does what doit.lsp does (mostly). Then I can have my own script that loads setup.lsp, sets my expected failures, runs the tests, and examines the list of unexpected failures and passes. Then I don't have to touch ansi-tests itself when failures are fixed and new tests fail.
Any other ideas?