(Moved the conversation out of #692 into this thread to keep the issue focused).

Holger Just wrote:

We have spend way more time than initially planned to bring the patches to the current unstable branch.

I think we have to reduce that time for future patches, let's talk at the next team meeting what went wrong and how we can improve our process and communication. The merges and rebase attempts were a serious PITA and I wouldn't want to do something like this again (and don't wish that anyone else too).

We don't need to wait for that, it sounds like the exact same problem as with acts_as_journalized:

• a branch was forked
• the branch was long lived, which meant a divergence from the source branch
• the branch had changes that impacts many parts of the system, thus merge problems

To make things even worse, this design was a long lived branch of another long lived branch (#263). The only way around this is to make smaller more frequent changes.

I'm not saying this was bad or wrong, only that because of how the design was handled we now have to do more work to get it done.

Eric Davis