2008/12/29 Talin <span dir="ltr"><<a href="mailto:viridia@gmail.com">viridia@gmail.com</a>></span><br><div class="gmail_quote"><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Do the existing tests use a wildcard rule to gather all test sources? We would need to insure that the wildcard rule for the unit tests and the large tests are mutually exclusive.<div></div></blockquote><div><br>By "large tests", I assume you mean the .ll tests? Those are one-file-per-test. Also, if you use Chris' suggestion on directories, it will separate the unit tests from the large tests.<br>
</div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><div>Also, will the unit tests be one executable or many? They will probably compile/run faster if there is a smaller number of executables.</div>
</blockquote><div><br>Yes, I agree, they'll link faster if we have fewer tests. I would not want a single massive test for all of LLVM -- how about 1 test binary per major directory in llvm/lib, unless it's really necessary to split them up, e.g. target-dependent unittests, but each .cpp file would ideally have its own _test.cpp file, so it's easier to keep the tests for different modules separate.<br>
</div></div>