Smart 0.52

[Gustavo Niemeyer]

Hey guys!

> congratulations to the new release! I have a couple of patches lying around
> (and applied to the SUSE package), which I would like to get upstream:
>
>   * smart-yast-media.patch
>     handle yast repos that a spread over multiple dirs correctly
>
>   * smart-yast2_compressed-packages.patch
>     support compress yast2 packages file, introduced in openSUSE 10.3

As we discussed on friday, I have started to look over all of your
current patches.  I'm being a bit slower than I'd like because I'm
implementing tests for the merged changes, and also because I'm doing a
few cleanups when possible.

I'm providing below a breakdown of what I did today.

> (I noticed that I had to remove /var/lib/smart/cache manually, after adding
> smart-yast2_compressed-packages.patch, because smart didn't seem to notice
> that its cache was outdated with the changed YaST2Channel object.

As we talked, this is just a matter of bumping the __stateversion__ for the
loader.

So, here we go.  I've looked at 5 patches so far.  I'll continue
on the others ASAP.


smart-yast2-media.patch
-----------------------

This was applied with a minor fixup to an edge case,
and tests.


smart-broken-repo-without-summary-or-description-workaround.patch
-----------------------------------------------------------------

This has been applied with changes in the implementation, and
I've also added tests.


smart-yast2_compressed-packages.patch
-------------------------------------

This has been applied.  I've copied over the doctest for the yast2
channel and modified it to work with compressed data.  There's one
detail that I also changed, which I'd like to ask you guys about.
I've modified the default value of "compressed" to be false, so that
old configurations won't break. Without that, even the original
doctest itself was failing, because it has compression enabled, and
the automatic detection wasn't working with the channel metadata.
I guess that even with compression being False by default, it should
work ok if the automatic detection is possible in most cases.  Do you
think that's ok?

Besides applying the patch, I've also cleaned up the channel
implementation quite a bit.  These changes may conflict with the
other patches, but it won't be hard to integrate them even then,
as the implementation is pretty much the same.


smart-channelsync.patch
-----------------------

Applied.


smart-ksmarttray.patch
----------------------

This seems to change the path to smart-update so that one
doesn't have to rely on $PATH.  Being system-specific, I
guess it makes sense to continue being a SuSE patch, right?

-- 
Gustavo Niemeyer
http://niemeyer.net