Yes Pramod I must understood your suggestion. I prefer option 1, with the
exception of cases where multiple commits are required to preserve
attribution.

Utlimately I think the choice depends on what function you want a branch to
serve. If you want the branch to tell the story of how the code was
developed then option 2 or 3 is superior. If you want the branch to
represent purely functional changes that can be easily ported, then option
1 is superior. I prefer the latter view of a branch.

Tim

On Mon, Nov 2, 2015 at 11:11 AM, Pramod Immaneni <[email protected]>
wrote:

> Tim are you saying at the end of the process you should only see one commit
> for a JIRA/change. If that is the case then you may be referring to 1.
>
> On Mon, Nov 2, 2015 at 11:09 AM, Timothy Farkas <[email protected]>
> wrote:
>
> > +1 for 3. After review I think everything should still be squashed into 1
> > commit, with some exceptions like preserving attribution (for example
> when
> > you rename a file or when multiple authors work on a feature).
> >
> > On Mon, Nov 2, 2015 at 11:00 AM, Pramod Immaneni <[email protected]
> >
> > wrote:
> >
> > > I wanted to find out how folks feel about squashing commits for reviews
> > on
> > > pull request. This is not for the initial pull request but only for
> > > subsequent commits to address the reviews. Here are some options.
> > >
> > > 1. Squash everything to a single commit. This is the process we are
> > > following today. Advantage is there is one commit per JIRA and self
> > > contained. Easy to cherry-pick if needed.
> > > 2. Preserve the individual commits for pull request reviews. Advantage
> is
> > > you preserve the review history in the pull request, the thoughts and
> > > discussions that went behind the changes and you see the incremental
> > > changes in separate commits. Disadvantage is you will have to work with
> > > multiple commits if you are trying to something with the change like
> > > re-apply it elsewhere.
> > > 3. A hybrid of 1. and 2. where you don't let the commits grow large. No
> > > standard set limit but the contributor and committer work keep it to a
> > > reasonable amount squashing smaller commits.
> > > 4. Anything you would like to propose.
> > >
> > > I would like to add +1 for option 3.
> > >
> > > Thanks
> > >
> >
>

Reply via email to