[p4] Video of Guido on Google internal review process, Mondrian and Perforce usage

[Chuck Karish]

Keep these requirements in mind:

- The developer must be able to build and test in the source tree
  that holds the modified files.

- Each developer's changes must be separate from changes made
  by other developers or  by the same developer in another
  workspace.

Perforce branches aren't sufficiently lightweight that it's practical
to make a branch of a large codebase to back up each client
workspace.  This might be manageable with sparse per-client
branches but it would be complicated.

The feature we're looking for here is promotions.  Before review
a set of changes would be visible only in a user-specific code line.
After review the changelist would be promoted to global visibility
on the shared code line.

  Chuck

On 12/13/06, David Weintraub <qazwart at gmail.com> wrote:
> Very interesting, but dependent upon NFS. I am wondering if the
> process might be better if they used a REVIEW branch.
>
> 1). Developer does an integration of the files they want from the
> "MAIN" branch to the REVIEW branch.
> 2). Developer checks out files on REVIEW branch and makes the changes.
> 3). Reviewer examines the files on the REVIEW branch. Reviewer could
> use their own client and not depend upon the Developer's client.
> 4). Reviewer makes suggestions which the Developer fixes. Suggestions
> could be done via checking out and in the file without making any
> changes and putting the comments in the changelist description.
> 5). When developer and reviewer agree to everything, Reviewer
> integrates the changed file from the REVIEW branch to the MAIN branch.
>
> The advantages I see is that both the Developer and Reviewer are both
> documented in Perforce, and that the developer and reviewer don't have
> to share a Perforce client or an NFS directory. You can also determine
> if a file is in review, or if it has finished being reviewed, when the
> review started, and when it was completed.
>
> Similar "g4" wrappers could also be used in this process.
>
> --
> David Weintraub
> qazwart at gmail.com
> _______________________________________________
> perforce-user mailing list  -  perforce-user at perforce.com
> http://maillist.perforce.com/mailman/listinfo/perforce-user
>


-- 
Chuck Karish   karish at well.com   (415) 317-0182