Smart needs some testing

[Anders F Björklund]

Christian Metzen wrote:

>> If I'm reading this correctly, it worked OK in Smart 1.2/1.3
>> but does not with Smart 1.4 ? If so, that is indeed trouble.
>> 
> 
> Yes you are right, same problem here. We already talked about that some
> weeks ago. Can´ t find any usable solution. Hope you keep up the good work
> and find any fix for this.

The way to get this fixed is to set up a more "real" test scenario.
None of the current tests are enough to test the dependency solver.

This could possibly include "smartstrap" too, to do entire distros.
But so far there hasn't been any offers to look into this properly.

I don't think this will happen for Smart 1.4.x, but then again there
is no reason why Smart 1.4 should do any worse than previous either ?

So what could be done is to look into whether parts of the changes
to transaction should be reversed, or if there are "simple fixes" ?

> We are already discussing a smart alternative for the next major release,
> but i better want back my smart pm. :-)

There's some other major changes that need to happen for Smart 2000,
but there is no use in developing those if nobody is using it anyway.

--anders