Fixes checkout not possible with (untracked) files under gitignore

[osklyar]

This is a follow up for #429. The original implementation was based on the assumption that Worktree.Status is used to check differences between the index and the worktree. However, the Worktree.Checkout uses Worktree.diffStagingWithWorktree (via Worktree.cointainsUnstagedChanges), which is also used by the Status. This has prevented a checkout of a branch due to "unstaged changes" even when the status shows clean.

In this PR the application of ignores is moved from Status directly into diffStagingWithWorktree, so the Status logic remains unchanged, but now the same logic applies to Checkout. With this change Checkout proceeds as there are no unstaged changes any longer (listed in ignores).

[mcuadros]

can you provide a test?

[osklyar]

The Status test is still covered by the original one. A test for Checkout is added below

[codecov]

Codecov Report

Merging #444 into master will decrease coverage by 0.58%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #444      +/-   ##
==========================================
- Coverage   78.01%   77.43%   -0.59%     
==========================================
  Files         127      127              
  Lines        9171     9173       +2     
==========================================
- Hits         7155     7103      -52     
- Misses       1236     1305      +69     
+ Partials      780      765      -15

Impacted Files	Coverage Δ
worktree_status.go	71.36% <100%> (+0.26%)	⬆️
plumbing/transport/ssh/common.go	0% <0%> (-50.91%)	⬇️
plumbing/transport/ssh/auth_method.go	33.33% <0%> (-24.77%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 65bf694...81dbc6a. Read the comment docs.

[mcuadros]

You should provide a test where the checkout has ignored files.

[osklyar]

This is exactly what the test does: (1) checkout on an empty repo -- no errors, (2) file file added and checkout produces an error, (3) .gitignore added and checkout still produces an error (file is no more a problem, but .gitignore itself is, (4) .gitignore staged and now checkout produces no error because file is excluded from unstaged by .gitignore. Well, this was the intention. Do you see a mistake in this reasoning?