José Matos wrote: > I took some time to look this problem. > > The logic seems right, there are one place where there is a off by one, > because in python range(a,b), assuming a and b integers goes from a to b-1. > This logic may seems strange but it pays because. > > range(a,b) + range(b,c) == range(a,c)
I see. I fixed this. > Since you are only having problems in the revert stage put this in and we > will fix the remain problems in svn. Conversion also doesn't fully work yet. It puts in the param also in cases it shouldn't. See FIXMEs. > Yes this is asymmetric with relation to updates, updates should always > work, revert can be fixed at any time, even if in this case means before > release. I've done so. Maybe this is easier for you to fix the remeining issues. I'll leave the bug report open as a reminder. Jürgen
