Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [egit-dev] GitMoveDeleteHook spec

On Fri, Mar 25, 2011 at 02:14, Robin Rosenberg
<robin.rosenberg@xxxxxxxxxx> wrote:
> I wonder what your ideas are regarding what the GitMoveDeleteHook. Here
> is what I think.
>
> 1 Moves inside a repo
>
> 1.1. A Move of something inside a repo in EGit will change the index to
> reflect that a path has been moved.
> 1.2. A Move of something inside a repo in EGit should not change the
> SHA-1 of corresponding dircache
>    entries, i.e. if you move a file or project with unstaged changes,
> the unstaged changed shall be
>    unstaged also after the move.
> 1.3  A Delete of something inside a repo should delete those resources
> from the index.
> 1.4. A Move/Delete on conflicting entries shall affect stage 0 only.

+1 to all of section 1.

> 2 Moves of resources containing repo
> 2.1 The User should be able to move a project or folder containg a repo
> (or get an error until that works)
> 2.2. Special care is needed for moving repos in and out of other repos.
> Immediate reaction, NO, but that
> is not really reasonable since we may want to do that anyway.
> 2.2 a A move of a repo to within another: Does the use want to create a
> git subproject, a subtree merge,
> or just move the project and repo or just move the content of the repo.
> Undoable?

Yikes. Lets at least prevent them from doing this until we do
something correct here. What correct is, I don't know. I would say git
submodule, but *shrug*.

> 3. Moves of resourcers between repo
> 3.1 Handle staged/unstaged as in 1.1

Yes, +1.

> 3.2 Mixed cases (shudder)

By mixed you mean partially staged stuff in the source repository?
Yikes. Don't let the user do that.

> The Commit Dialog of today does not really respect the state of the
> index. That should be fixed, so it
> works more like Git Gui.

+1.

-- 
Shawn.


Back to the top