Re: [Self-platform-dev] Re: [teg] Self-Platform Updates (Tentative Version Control UI)

[Federico Heinz]

On 15/02/2008, Nagarjuna G. wrote:
> On Fri, Feb 15, 2008 at 12:13 AM, Federico Heinz <address@hidden>
> wrote:
> > Well, actually, every change *is* a branch, only in most cases is appears
> > to be the only branch.
> "every change is a branch" to my understanding is not correct and will
> mislead.  Your sense of 'actully' and 'in most cases' is not helping me at
> all.  Please make it clear, what you meant.

What I mean is that all branches are equipotential. In other words, there is no
"trunk".

> My understanding is as follows:
> A branch gets created when a change is made not to the recent object, but to
> an older one.

What is created there is not a branch, but a fork, which is not the same.

Say, you have the following history:

0 - 1 - 2 - 3

When the users committed the changes that created versions 1, 2, and three,
they created a branch (albeit the only one), of the predecessor version. Then,
along comes another user who again drives version 1:

        2 - 3
      /
0 - 1
      \
        4

What this user has created is *another* branch of version 1, which configures a
fork. From the system's point of view, 2 is no different from 4. both will have
to stand on their merits.

> This is our understanding.  the storage algorithm
> that is implemented in the backend  tracks each objects' history.   We are
> calculating the branches by taking the history of each object.

If I understand you right, this algorithm implements the behaviour described
above, so there should be no worries.

        Fede

signature.asc
Description: PGP signature