> > "Bruce A. Mah" <[EMAIL PROTECTED]> writes:
> > > Differences of opinion on naming aside...the branch isn't supposed to
> > > last long at all.  The point is to provide a slightly polished snapshot
> > > to the wider developer community.  We can't do the QA/releng work on
> > > HEAD without calling for a code freeze (which we early on decided that
> > > we would *not* do).
> > 
> > Then you don't need a branch, you just need a simple tag, and you can
> > slide it forward if something needs fixing, and remove it after rolling
> > and shipping the snapshot. 
> No, in this case that doesn't help.  What we want is to grab a stable
> moment, then to allow development to continue.  However, we may then want
> to tweak that stable moment without impinging on development, which
> requires a branch.

Not necessarily.  What we've done at work is *NOT* create a branch
unless absolutely necessary.  The only time a branch is requires is *if*
a file changes out from underneath the developer *AND* it that files
needs modifying but *must not* contain that same change.

We play it by ear, since in almost all cases, the change can be made to
any necessary file(s) and the file(s) updated by hand.  Otherwise, often
the change that happens are necessary to merge into the build, so we do
an update.

Only in very rare cases do we run into a problem where we have to create
a branch.  In that case, the developer responsible for the release
creates a branch from his checked out tree (there's no law against
creating a branch from sources that are older than the HEAD), and then
makes any necessary changes.

It's *not* that hard to do.  Otherwise, once the release is made using
the files, a point-tag is laid down and we've saved the hassle of the

> The QA/releng work requires us to modify the stuff being released
> following the branchpoint.

See above.


To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message

Reply via email to