I can't do my usual "add the tests first, with EXPECTED/ACTUAL sections that
document the bug" method here, because the tests would hang without the bug
being fixed.
We need two different tests here: one where a cherry-picked commit simply
becomes empty "by itself", because the change is already present on the
destination branch (this was only a problem with git versions older than 2.45),
and the other where the cherry-pick stops with conflicts, and the user resolves
them such that no changes are left, and then continues the cherry-pick. This
would still fail even with git 2.45 or later. The fix is the same for both cases
though.
The tests show that the selection behavior after skipping an empty cherry-picked
commit is not ideal, but since that's only a minor cosmetic problem we don't
bother fixing it here.
Our logic to decide if a file needs to be checked out from the previous commit
or deleted because it didn't exist in the previous commit didn't work for
submodules. We were using `git cat-file -e` to ask whether the file existed, but
this returns an error for submodules, so we were always deleting those instead
of reverting them back to their previous state.
Switch to using `git ls-tree -- file` instead, which works for both files and
submodules.
It works for submodules too.
Also, pass file name and file content explicitly; the existing tests don't care
about these, but when writing tests that do, it makes them easier to understand.
Replace merge-tool with merge options menu that allows resolving all
conflicts for selected files as ours, theirs, or union, while still
providing access to the merge tool.
It seems useful to have the flexibility to remap "enter" in confirmations to
"y", but keep "enter" for menus and suggestions (even though we sometimes use
menus as confirmations, but it's still good to give users the choice).
This one doesn't make a difference in practice because we don't remap the key in
tests, but if we would, then this would no longer work correctly. It's just more
correct this way.
So far, confirmations and prompts were handled by the same view, context, and
controller, with a bunch of conditional code based on whether the view is
editable. This was more or less ok so far, since it does save a little bit of
code duplication; however, now we need separate views, because we don't have
dynamic keybindings, but we want to map "confirm" to different keys in
confirmations (the "universal.confirm" user config) and prompts (hard-coded to
enter, because it doesn't make sense to customize it there).
It also allows us to get rid of the conditional code, which is a nice benefit;
and the code duplication is actually not *that* bad.
To fix the problem described in the previous commit, iterate backwards over the
stashes that we want to delete. This allows us to use their Index field.
The test shows that we are currently auto-forwarding branches even if they are
checked out by another worktree; this is quite bad, because when you switch to
that other worktree you'll see that the files that are touched by the fetched
commits are all modified (which we don't test here).
This was needed in an earlier version of the test, when we asserted the file
content in a more complicated way. It should have been removed in caca62b89e.
This assertion didn't test anything useful, given that the filtered view already
has two lines. This should have been adapted in 9f0b4d0000 when we added
section headers while filtering.
Also, fix two other commands that stage all files under the hood:
- when continuing a rebase after resolving conflicts, we auto-stage all files,
but in this case we never want to include untracked files, regardless of the
filter
- likewise, pressing ctrl-f to find a base commit for fixup stages all files for
convenience, but again, this should only stage files that are already tracked
These should have been added when we started rendering this information in
e5b09f34e0; apparently I was too lazy back then. Adding them now to guard
against breaking it in the next commit.
I'm adding these to the CRUD tests, it doesn't seem worth adding separate tests
just for these assertions.
There's no reason not to allow these.
Technically we could enable a few more, but I chose not to because some might be
surprising or confusing in filtering mode. For example, creating a fixup commit
would work (shift-F), but the newly created commit might not show up if it
doesn't match the filter. Similarly, pressing `f` to fixup a commit into its
parent would work, but that parent commit might not be visible, so users might
expect to be fixing up into the next visible commit.
When shift-selecting a range of commits across a file rename in
filtering-by-path mode, the diff currently shows an added file rather than a
renamed file. Add a test that demonstrates this, we'll fix this in the next
commit.
When filtering for a file path we use the --follow option for "git log", so it
will follow renames of the file, which is great. However, if you then selected
one of the commits before a rename, you didn't see a diff, because we would pass
the original filter path to the "git show" call.
To fix this, call git log with the --name-status option when filtering by path,
so that each commit reports which file paths are touched in this commit;
remember these in the commit object, so that we can pass them to the "git show"
call later.
Be careful not to store too many such paths unnecessarily. When filtering by
folder rather than file, all these paths will necessarily be inside the original
filter path, so detect this and don't store them in this case.
There is some unfortunate code duplication between loading commits and loading
reflog commits, which I am too lazy to clean up right now.
The test shows that selecting a commit before the rename shows an empty diff,
and selecting the rename itself shows an added file rather than a rename.
This broke with the introduction of the age in the stashes list in bc330b8ff3
(which was included in 0.41, so 12 minor versions ago).
This makes it work again when filtering by file, but it still doesn't work when
filtering by folder (and never has). We'll fix that next.
The test shows that it doesn't work, the list is empty both when filtering by
file and by folder.
I could have added test cases for these to the unit tests in
stash_loader_test.go instead, but I don't find tests that need to mock git's
output very valuable, so I opted for an integration test even though it takes
much longer to run.
Similar to what was done in 457cdce61d, and for the same reason.
However, instead of waiting and fixing them one by one as we see them fail, I
decided to go about it more systematically. To do that, I added calls to
`time.Sleep(1 * time.Second)` in all the Shell.Commit* helper functions; this
ensures that all the commits we make get different committer time stamps, making
all these tests fail. With this I'm pretty confident that we're good now.
When entering filtering we would only call FocusLine, which takes care of
highlighting the selected line in the commits list, but not of re-rendering the
main view. HandleFocus does that.
When exiting filtering, the HandleFocus call was missing entirely.
The tests needed to be reworked a little bit to make this testable.
Now that -committerdate is the default sort order, we could get different
results for the sort order of the branches list depending on whether the commits
on both branches have the same committer time stamp (likely in an integration
test, since git time stamps have second resolution), in which case git will fall
back to alphabetical order, or not (rare, but possible), in which case master
will have the newer commit and will come first. Make this stable by forcing the
sort order to alphabetical.
We might have more tests with this problem, we'll just have to fix them one by
one as we see them fail.
At the same time, we change the defaults for both of them to "date" (they were
"recency" and "alphabetical", respectively, before). This is the reason we need
to touch so many integration tests. For some of them I decided to adapt the test
assertions to the changed sort order; for others, I added a SetupConfig step to
set the order back to "recency" so that I don't have to change what the test
does (e.g. how many SelectNextItem() calls are needed to get to a certain
branch).
While it's true that the behavior is a little different from the staging panel,
where the staged lines are actually removed from the view and in many cases the
selection stays more or less in the same place, it is still very useful to move
to the next stageable thing in the custom patch building view too.
