>>>>> "Paul" == Paul Jenner <[EMAIL PROTECTED]> writes:
Paul> However I wonder if cross reference against daily mauve regressions Paul> would help identify and fix these more quickly? Yes. The GCC community has a number of auto-builders that send email (using the ChangeLog) to anybody who breaks the build (including regression tests). Something like that may be helpful. I've got a bunch of test code running atm, though I'm not sending the output anywhere yet. Specifically right now my tester: * builds gcj * runs its test suite including mauve and jacks * builds gcjx * builds classpath with both gcj and gcjx * builds jamvm Next step is to run mauve against the just-built jamvm. I'd also like to set it up to build and test some applications (hopefully ones that are relatively simple to set up -- for larger things I would like to see us more actively involved with Gump). Anyway, the question is, what to do with the info? The reason I turned off my nightly tests the last time was that only I could see the results, and I didn't have time to track down regressions. Should we set up a new list? Try to implement some gcc-like solution? Send them to an existing list? Tom _______________________________________________ Classpath mailing list Classpath@gnu.org http://lists.gnu.org/mailman/listinfo/classpath
