Code review comment for lp://qastaging/~sil2100/compiz-core/scale_all-0.9.7

> I'm going to say Needs fixing, just because there is no fix upstream yet :)
>
> Also, I strongly recommend investigating the other scale regressions and
> starting by simply reverting the code that caused the regressions

This will inherently "cause" regressions by the fact that you just reverted important bugfixes. It does not take into account the waste of developer time as they are chasing up old code that they probably don't even remember writing.

If you want to fix things safely, analyse how the code works, analyse what its meant to do, write unit tests to ensure that it matches the required behaviour and implement your fixes. I don't revert things simply because I don't understand them - I learn the code by making it better and writing tests to ensure that it does what I think it does.

Lucasz - can you propose this fix to lp:compiz and then resubmit it here? Thanks for taking the time to look into this.