When staging by hunk and a file is fully staged, put cursor on first hunk of next file vs next file name

[gibfahn]

Versions

• vim-fugitive a93ceff61f070a095c3c712c389247a80897e3b0 (latest master)
• Using neovim v0.6.1 (latest neovim)

Background

When staging, I'm either staging by file or by hunk. When staging by file, I'm giving each file a quick look then staging the whole thing. When staging by hunk, I'm carefully reading each diff.

When you stage a hunk, fugitive automatically puts the cursor on the hunk marker (@@ -1,9 +1,9 @@) for the next hunk, which is great as it lets you tap s repeatedly to go through the open hunks.

However, when you stage the last hunk in a file, fugitive puts the cursor on the next file marker (e.g. M README.markdown) instead of the first hunk marker in that file. This means you have to constantly be checking

Worked Example

1. Modify a couple of files by running this shell snipped in this repo:

printf '%s\n%s\n%s\n' firstR "$(cat README.markdown)" lastR >README.markdown 
printf '%s\n%s\n%s\n' firstC "$(cat CONTRIBUTING.markdown)" lastC >CONTRIBUTING.markdown

2. Open fugitive

Screenshot:

3. Go down to Unstaged (2) and hit = to expand the hunks

Screenshot:

4. Move down two lines to @@ -1,3 +1,4 @@, and hit s to stage the first hunk

Screenshot:

5. The cursor has automatically moved to the next hunk (which is excellent). Hit s again to stage that one too.

Screenshot:

6. Note that the cursor is now on M README.markdown, so hitting s a third time would stage all changes in that file, but we just want to stage hunk-wise. Therefore we have to move one line down to stage only the next (firstR) hunk rather than everything in this file.

Screenshot:

[tpope]

I'm not convinced this should be changed. A possible alternative workflow would be to start with everything closed, press i to cycle through changes (which expands diffs automatically), and press s on the hunks you want to stage.

[gibfahn]

A possible alternative workflow would be to start with everything closed, press i to cycle through changes (which expands diffs automatically), and press s on the hunks you want to stage.

The expanding diffs automatically is nice, but this doesn't really make things any easier. Previously to stage the 4 hunks in the above example (cursor starting in the first hunk) it would be ssjs, and with i it's ssis.

Using i and s isn't so bad when you're just picking out a few diffs to stage out of a bunch of unconnected changes (when you're mostly not staging things). It's much more painful though when you're staging ~everything, and just using this as a point to do some code review on your changes before submitting.

As I mentioned above though, it's not so much the extra keystrokes as having to constantly remember to check whether the cursor is on a file or a hunk, which distracts from the focus on the code itself.

It looks like this mapping works as a workaround:

" Stage, go up a line, then skip to the next hunk.
:nmap S sk]c