Hi Simon, > > $ ./gnulib-tool --test --with-tests parse-duration > > Using '--test --with-tests' or '--create-testdir --with-tests' is quite > common. Is there any reason why --with-tests isn't the default?
It's for consistency with --import. Many packages like to have --import without tests, and having --with-tests be enabled for --test but not for --import would/may seem odd. Maybe it's also historical: when the core of gnulib-tool was written (in August 2004), gnulib did not contain any unit tests, but the option --create-testdir exists already since 2002. Support for unit test modules was only added on 2005-08-25. > It seems rare to want to test a module without also wanting to include its > self-tests. Naively, it also seems surprising that '--test' and > '-create-testdir' does not bring in self-*tests*. I agree. > What do you think of making --with-tests a no-op, and add a > --without-tests to disable it? This would save us all some typing... You mean to make this change only in combination with --test, --create-testdir, and --create-megatestdir? Not with --import. I'm undecided. What do the others think? > I haven't looked into the gnulib-tool changes at all so I don't know how > difficult it would be to implement. It'd be pretty trivial to implement. Bruno
