ratings and scores use 0 for default value

[Jeff Mitchell]

Dan Meltzer wrote:
> We're going to want to change tables down the road probalby anyways...
> so we'll have to come up with a good way of doing updates (probably
> something like quassel does, I imagine.  I cannot think of any
> possible justification for making this change between a release
> candiate and the final release.  The idea of unrated ==0 is used
> throughout the amarok code, and it would be relativaly easy for
> regressions to appear.

Yes, agreed, the major Amarok codebase using null=0 shouldn't be changed
now.  Caleb also said he may try to come up with a better schema we
could use for i.e. 2.1.

>> Probably, the fix is to either modify the schema or the query (so that
>> the query uses values of 0 as acceptable for all values > X).
> 
> ... Though modifying the query would probably be less
> regression-likely... though it has the (probably minor) side effect of
> people who explicity rate their songs as 0 still getting them...

Even in 1.4, 0 has *never* meant bad -- 1 (which was = to 1/2 star) was
the lowest rating you could give a song, which meant, essentially
"Unbearably Awful".  The star was tri-state -- click once for 1/2, twice
for 1, thrice on the first star to entirely remove the rating.

This might stem from a confusion (as a result of our UI or workflow)
where the user sees no perceptible difference between unrated and 0
stars -- or doesn't understand that 0 *is* unrated.  But from a codebase
perspective, zero has always meant unrated.

That's why I'm in favor of trying to fix this before release...

--Jeff