On Fri, Jan 21, 2011 at 02:59:37AM -0800, JamieEchlin wrote:
> Stefan Sperling-7 wrote:
> >
> > There is nothing that tells you this information directly.
> > The changes brought in by a merge are treated just like any other
> > local changes. As far as Subversion is concerned, you could have made
> > any merged changes manually.
> >
> > Defining a good branching/merging strategy, making sure everyone knows it,
> > and consistently keeping to it may help with avoiding such questions
> > in the first place.
> >
> > Stefan
> >
>
> OK, thanks.
>
> With respect, your second paragraph is a bit of a cop out. We do have a well
> defined branching and merging strategy, the problem is making sure everyone
> knows it and follows it, even in the heat of trying to get out a production
> fix on a Sunday.
>
> The development mentality in the corporate environment is not the same as in
> open source land, where people agonise over their commits for a long time
> because they know they're going to be peer-reviewed by many people.
>
> Some of the users here are not really programmers, they're rocket
> scientists, and they only write code as a side-effect of their main role.
> They expect things to just work, and they're too fantastically intelligent
> to learn how to respect the merging rules defined. So they keep telling us
> anyway.
:)
> I exaggerate but I hope you see my point... we do spell these things out but
> we need the big stick too.
Good point. Documenting the strategy is only part of the problem in real
life. So it would help if Subversion had a better way of letting people
know that the strategy is being followed.
It would be even nicer if Subversion had a way of guiding users to
perform merges in accordance with a pre-defined strategy.
So if the user tried to perform e.g. a cherry-picking merge to a branch
for which cherry-picking isn't defined in the strategy, svn would error
out and let the user know that this type of merge isn't allowed for the
merge target.
Neither would be trivial to design and implement, but both could
be possible.
But of course, educating users is a social problem, not a technical one.
No technical solution will ever solve this completely.
If rocket scientists were perfect, there wouldn't be a problem.
Stefan
Received on 2011-01-21 12:47:42 CET