On Wed, 23 Oct 2002, William A. Rowe, Jr. wrote: > At 02:44 AM 10/23/2002, Aaron Bannert wrote: > >On Tue, Oct 22, 2002 at 09:03:02PM -0500, William A. Rowe, Jr. wrote: > >> >Hmm... I haven't looked at the new test stuff, but I wonder how it's > >> >going to deal with the testshm* tests, since those tests depend on > >> >fork() and exec() to work. > >> > >> As well as they did before... > >> > >> ... that is, not at all. > > > >How do you mean that they don't work? Anonymous shared memory is not > >currently supported on systems that lack sufficient process inheritance > >mechanisms (ie. fork() and exec()). > > I mean, the tests couldn't work (couldn't compile for that matter.) > So we simply walked around those tests.
That is because the tests weren't written properly. APR provides features macros for everything that isn't completely portable. If the tests used those macros, every test would have always been buildable on every platform. This is one of the goals of the new test suite. There will be few (if any) programs that don't compile and run everywhere. My goal is for all tests to run everywhere, although some of the tests will return NotImplemented on some platforms. BTW, once I get the tests all converted, there is another project that needs to be undertaken. Our test suite tests about half of functionality we provide. And, even for the stuff that is tested, it isn't tested well. I will continue to port the tests, and then I will begin to fill out the suite, but some help would be appreciated. I have already added some thoughts to test/README about how to write good tests, but more thoughts in that area would be great. Ryan _______________________________________________________________________________ Ryan Bloom [EMAIL PROTECTED] 550 Jean St Oakland CA 94610 -------------------------------------------------------------------------------
