Theo Van Dinter writes: > On Thu, May 25, 2006 at 06:08:15PM +0100, Justin Mason wrote: > > fwiw, a long term fix would be option (b) -- tests should not be > > using the user's "real" userstate dir. > > Yeah, but that's not actually the issue. > > The warn() (now info()) is in the M::SA::get_and_create_userstate_dir() > function which always gets called (assuming sed_path() is called, which > is almost always,) and isn't actually related to using the config. > It's just there to create the dir and return the path to the caller. > > It looks like the tests already skip the user's userstate dir via the -p > option.
ah ok then ;) --j.
