Richard Heck wrote: > I'll have to think about this. The current policy of committing first to > trunk makes a lot of sense, as far as keeping branch stable. There are > plenty of cases where a bug gets fixed in trunk first, and then we wait > to commit to branch until we see how that goes. Of course, the downside > to this is that there may be less testing of trunk than of branch. E.g., > I use BRANCH_2_0_X for actual work, not trunk.
This is all true IMHO. This brings us back to your idea of an experimental extension of the stable branch (a thing easily possible with git AFAIU). In the past, I have simply applied such patch candidates locally and used this for daily work over some weeks before applying to branch. With such an extension, many more people could (in theory) join in real-work testing. We could even keep such an extension over a minor release cycle (i.e., only merge it in if it is really mature). Jürgen