RE: LSB woody test results
At 2003/9/16 17:46-0700 Wichmann, Mats D writes:
> > > happily use pax if available to fulfill that need. This is how other
> > > distributions have been dealing with it and IMHO is rather stupid
> > > but oh well. A dependency on pax has recently been added to the lsb
> > > package to make this easier.
> > I figured that this would be the case. How do you feel about adding
> > some of the other dependencies? I notice, for example, that
> > Tcl/Tk and expect constitute some of the failures.
> Sigh. expect is used to automate the process of feeding
> the root password to tests that need it, but has nothing
> at all to do with LSB conformance... but it seems most
> other distros do include it in their "keystone" package,
> so that installing "lsb" gives you not just LSB conformance,
> but also enough to run the test suite, if that makes sense.
Expect isn't needed to run the lsb-runtime-test suites, just to build
them. I'm currently trying to work out how to make the
lsb-runtime-test build process a lot friendlier (ie, non-root,
non-hard coded paths etc), but haven't managed to work out a clean
solution yet. fakeroot gets close, though the inability to su to other
users is rather problematic.
That being said vsc-lite (commands and utilities) which may be
introduced for 2.0 will be using expect at runtime, though its bundled
with the test suite.
IBM OzLabs Linux Development Group