Setting options directly in (h.)Buf.Settings without calling SetOption() or
SetOptionNative() is generally not the best idea, since it may not
trigger the needed side effects.
In particular, after https://github.com/zyedidia/micro/pull/3343,
directly setting `diffgutter` and `ruler` causes them not being tracked as
locally overridden per buffer, so if we run the `reload` command,
it unexpectedly replaces them with the default ones.
In `ReloadSettings()` the `filetype` can change upon globbed path given by
the `settings.json` or by identifying a different `filetype` based on the
file name given or pattern present inside the file.
To prevent further recursion caused by checking the `filetype` again, its
processing stops here and isn't considered in `DoSetOptionNative()`
once again where the callbacks are usually triggered.
Like in NewBuffer(), we need to update glob-based local settings
before updating the filetype, since the filetype itself may be among those
glob-based local settings.
* `UpdatePathGlobLocals()`
* to apply the settings provided within e.g. "/etc/*": {}
* `UpdateFileTypeLocals()`
* to apply the settings provided within e.g. "ft:shell": {}
We don't need to call `InitLocalSettings()` twice any longer.
Fix regression introduced while implementing nano-like page up/down in
commit b2dbcb3e: if the view is already at the end of the buffer and
the last line is even above the bottom, i.e. there are some empty
lines displayed below the last line (e.g. if we have scrolled past the
last line via the mouse wheel), pressing PageDown not just moves the
cursor to the last line but also unexpectedly adjusts the view so that
the last line is exactly at the bottom.
Fix regression introduced in commit fdacb28962 ("CopyLine, CutLine,
DeleteLine: respect selection"): when CopyLine, CutLine or DeleteLine is
done in the last line of the buffer and this line is not empty, this
line gets selected but does not get copied/cut/deleted (and worse, it
remains selected).
Commit 9fdea82542 ("Fix various issues with
`SpawnMultiCursor{Up,Down}`") changed SpawnMultiCursorUp/Down actions to
honor softwrap, i.e. spawn cursor in the next visual line within a
wrapped line, which may not be the next logical line in a buffer. That
was done for "consistency" with cursor movement actions CursorUp/Down
etc. But it seems there are no actual use cases for that, whereas at
least some users prefer spawning multicursor in the next logical line
regardless of the softwrap setting. So restore the old behavior.
Fixes#3499
In cursor's Goto(), which is currently only used by undo and redo, we
restore remembered LastVisualX and LastWrappedVisualX values. But if
the window had been resized in the meantime, the LastWrappedVisualX
may not be valid anymore. So it may cause the cursor moving to
unexpected locations.
So for simplicity just reset these values on undo or redo, instead of
using remembered ones.
