On Thu, 2011-07-07 at 17:14 -0700, Edward Pilatowicz wrote:
> so the pkg history fix looks fine to me. i do have some comments on the
> test change though...
>
> - your deleting two functions, but you're not deleting some variables
> only used by those functions (pyver_test_manf_1).
Oops, I'll fix that.
> - you added a comment saying the tests will need to be brought back if
> we ship another python interpreter. is this really true? or do they
> only need to be brought back if we ship another python interpreter and
> parts of the packaging system use it?
Nope, this is explicitly for testing the pkgdepend generate path where
we have to use a different depthlimitedmf(xy).py in order to find
imports for a python module that runs using a version of python other
than the one that pkg(5) is using.
Rather than calling depthlimitedmf.py classes/methods inline, we execs
the alternate python runtime with the alternate depthlimitedmf(xy)
module. If Python 2.4 isn't installed on systems, we don't run that
code.
> - given that we're deleting a bunch of code that we might just need to
> resurrect soon, would it be possible to update these tests to run with
> the 2.6 interpreter and just leave them in place so when the next
> version comes along we just update them again? (instead of undeleting
> them.)
We could, though that's more work than I think is warranted here - we'd
need to port the existing 2.6 version depthlimitedmf.py so that it could
run standalone. I can do that if people really think it's worthwhile?
cheers,
tim
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
