On Wed, 11 Feb 2009 at 04:07PM -0800, mabshoff wrote: > On Feb 11, 3:54 pm, Dan Drake <dr...@kaist.edu> wrote: > > I'm one of the people who did an upgrade, and have now run "make > > test", "make ptest", and "make ptestlong" with no persistent > > failures. > > Thanks, but why run all three? "make ptestlong" is all you need since > it does everything and more than the other two. You can always run > ptestlong a couple times if you want to and there is even undocumented > support in ptest to run each doctest n times in a row instead of only > once. What is missing to make that useful is reports that tell you > that $FOO failed x out of y times instead of telling you that it > failed. :)
I run all three because different things happen with them. I've noticed that I tend to get fewer Lisp crashes when I run ptest, which seems strange -- you'd think things would be more reliable when the system isn't as heavily loaded, but that's not how it works. I also seem to get random tests timing out during one flavor of test that don't time out during other flavors. I know that there isn't a big difference -- and that there *shouldn't* be -- but the results of running the different tests aren't identical, and it makes me feel better to run the different flavors of 'make test'. So I do. :) I would love to see support for running each test n times and reporting the number of failures, if any. Dan -- --- Dan Drake <dr...@kaist.edu> ----- KAIST Department of Mathematical Sciences ------- http://mathsci.kaist.ac.kr/~drake
signature.asc
Description: Digital signature
