[cdt-dev] logistices for the EFS work

[Chris Recoskie <recoskie@xxxxxxxxxx>]

Hello everyone,

Greg and I are at the point where we want to start making changes to the
code as we implement EFS support in CDT.  This is a large task which will
take a while to complete, and will have direct implications on APIs people
are using.

As such, we want to get feedback from everyone as to how to handle putting
the changes in.  We can:

a)  Make incremental checkins as we go on HEAD.  Pros:  everyone can adapt
to API changes incrementally.  People can collaborate more easily on the
new changes.  Cons:  EFS support will trickle in, and will not necessarily
provide a complete or functional solution until we are done.  Things are
apt to be broken frequently.

b) Work for months until we think we're done, then do one gigantic checkin
once things are "working."  Pros:  nothing changes until stuff works.
Cons:  It's very hard for everyone to react to the new APIs and
functionality.  People may modify the same code we're trying to modify,
which will making staying in sync difficult.

c)  Work on a branch in the CDT, then merge it back to HEAD when we're
done.  Pros:  Nothing changes until stuff works.  Cons:  CVS merge tools
suck.  Also, no one is forced to look at the changes because most people
will work on HEAD and not our branch.  Staying in sync is still a problem.

We are leaning towards doing a).  Does anyone have any concerns with this
before we start doing anything?

===========================

Chris Recoskie
Team Lead, IBM CDT Team
IBM Toronto
http://www.eclipse.org/cdt