Make 3c -verify an error for now.
#519
Conversation
PR #488 made `3c -verify` cover only the compiler diagnostics, but none of our regression tests actually use that functionality. Instead, one regression test (macro_function_call) used `3c -verify` to try to test the absence of 3C warnings, and we were unaware that the test wasn't testing what it was supposed to. I think it's best to make `3c -verify` an error for now so we don't make that mistake again. I'm deleting the addVerifyAdjuster code because I think we're unlikely to ever want to use it. If we do, it should be easy enough to bring back by reverting this commit.
|
I'd prefer you leave the unused code with a "TODO: #425", until we give up on it entirely and close that issue. I'm still hopeful that we can try again and get it to work, since it's such a great tool. Your error message should prevent it from causing problems, and the issue can be amended to include a reminder to delete TODOs on close. |
Sure; done. I don't feel strongly about this. I was assuming that if we did get the diagnostic verifier to work for 3C diagnostics (as opposed to ordinary compiler diagnostics), we would need different code, but I may be wrong; I haven't researched it. |
PR #488 made
3c -verifycover only the compiler diagnostics, but none of our regression tests actually use that functionality. Instead, one regression test (macro_function_call) used3c -verifyto try to test the absence of 3C warnings, and we were unaware that the test wasn't testing what it was supposed to. I think it's best to make3c -verifyan error for now so we don't make that mistake again.