random/sparse-checkout

Sparse-Checkout

This code tries to simulate the way sparse-checkout works in recent Git versions.

The Problem

The problem is that the behaviour of sparse-checkout changed somewhere in the line. In most recent versions, sparse-checkout init, on a clone, will hide every single file; older versions, though, will refuse to init the sparse-checkout if that would leave the current directory empty.

Solution

The idea is to push update-index --skip-worktree to tag a file as "being ignored" (Git will always assume the file checksum is still the same as in the index, no matter what you do with the file). Then, when a user wants to re-enable the directory/file, use update-index --no-skip-worktree.

Although that sounds simple, there an issue that Git deals only with files and not directories, and trying to disable/enable the file in the index requires to pass the files; for example, on the re-enable, how to find the names of the files we just deleted?

That requires a combination of use with ls-files --cached [pattern]. This way we can get all the files that need to be pushed back/ignored, reading directly from the Git information.