Richard Heck wrote:

> We always have people who continue with the previous series, for one
> reason or another. I don't see any reason to deprive them of bug-fixes
> we already have.

Well, we need to avoid that a fix to a minor problem creates a major 
regression. I am a big fan of backporting fixes to the stable branch where 
it is sensible, but to each backport there is a risk attached that we can't 
quantify. IMO the most important property of the stable branch should be 
that users can _always_ upgrade from version x.y.n to x.y.n+1 without even 
thinking whether this might cause a problem.

Therefore I do not tend to think in terms of depriving users from bug fixes, 
but the other way round: Unless I have strong evidence that a bug fix is 
harmless I would not backport it. This might look awkward at first sight. 
Maybe it is only understandable if you already were the one who created a 
major problem by backporting a minor bugfix (Unfortunately I know that 
situation far too well).


Georg

Reply via email to