On Tue, Oct 6, 2009 at 10:08 PM, Andrew Deason <[email protected]>wrote:
> On Tue, 06 Oct 2009 16:50:35 -0700 > Russ Allbery <[email protected]> wrote: > > > I've been playing with the idea of proposing we use C TAP Harness, > > which has the nice property of using the same protocol as Perl's test > > suite. > > > > > > http://www.eyrie.org/~eagle/software/c-tap-harness/<http://www.eyrie.org/%7Eeagle/software/c-tap-harness/> > > > > The hard part is to find a way to do testing on both UNIX and > > Windows, and I'm not sure C TAP Harness will get us there. > > So far I've been using Perl and its testing classes, so +1 for TAP at > least. :) > > The 'framework' is more than that, though. In particular, how we > configure what we're testing and how we set up the infrastructure are > areas I think we could most stand to improve. Running the tests in the > simple 'make check'-style process as part of a build or something is one > thing, but I'd like to be able to easily point it at a particular > existing cell, server, volume, etc and say "go test that". > > I've got a vague notion of some system of dependencies where it uses > existing pieces of a cell, and creates that which doesn't already > exist... but it's, uh, vague. > > src/tests was built with that idea in mind, but never got there. -- Derrick
