I'm just going to the end of the soft-wrapped line, mirroring the behavior with hard wraps. It's maybe not perfectly technically correct but that behavior would require us to consider word boundaries outside of the current line, which doesn't seem worth the complexity.
If the range *touches* the fold, we unfold. This was needed to fix the behavior for unfolding at the current selection position. Previously, there was some kind of translation issue that was allowing us to accidentally work the way we wanted without this.
Co-Authored-By: Antonio Scandurra <me@as-cii.com>
This removes the need for a lock around in `WrapMap` and also removes
`WrapMap::notifications` because gpui already has a standard way of
notifying when a model updates.
We want to move a bunch of methods on FoldMap to fold_map::Snapshot. This captures a buffer snapshot, and we'll need a bunch of methods that currently are on Buffer to also be implemented on buffer::Snapshot. This Content reference struct will be a good zero-cost place to store that logic.
Rather than computing a diff after processing a batch of FSEvents, we instead detect renames as we're inserting entries. We store an entry_id on the File object that is owned by each buffer, and use this to detect when the path of the File has changed.
We now also manage all File-related state and event emission for Buffers in the LocalWorktree, since the logic will need to be totally different in the remote case.
Co-Authored-By: Max Brunsfeld <maxbrunsfeld@gmail.com>
Co-Authored-By: Antonio Scandurra <me@as-cii.com>
This commit does too much. The first goal was to change our approach to saving new buffers so that we don't need to construct a File for an entry that doesn't exist. Rather than doing that, we call `Worktree::save_buffer_as` with the buffer handle, the path, and the contents. This then saves the buffer and returns a handle to a `File` that references an entry that actually exists. I needed to do this so that we can store an entry id on `File`.
In the process, I noticed intermittent test failures on `test_rescan_simple`, so I made some changes required to fix those related to our reuse of existing ids. Our previous approach of removing a path when inserting a new entry was broken, because of the recursive nature of `remove_path`. Instead, I simply recycle the id of an existing worktree entry with the same path if one is present, then allow it to be replaced.
Remove the `Start` and `End` variants, and always
use the structure that was previously called `Middle`.
This makes Anchors simpler to serialize and deserialize.
When running this command with multiple cursors, if one of them was at a
later position in the buffer but lying on a shallower node, it could
happen that its start could move prior to cursors that were before it
but lying on a deeper node.
This could cause the selection merging algorithm to mistakenly keep some
selections even if they overlapped. With this commit we now sort
selections prior to merging them in `Editor::select_larger_syntax_node`.
Antonio Scandurra