This commit refactors the "nerdtree#ui_glue#UpDir()" function to fix
a bug in the behavior of the 'u' mapping. To reproduce the bug:
1. Open a bookmarked directory with 'o'
2. Press 'u' twice, observing normal behavior
3. Open the same bookmark again with 'o'
4. Press 'u' again, observe the new root is closed
The reason for this bug is that the parent node already existed and
was closed by the second 'u' press in step two above.
This commit fixes this bug by being careful to always open the new
root and transplant the child node properly. Also, the internals of
the function in question were refactored to make use of valuable
NERDTree functions that ultimately help to reduce the amount of code
needed for this operation.
The definitions for certain bookmark mappings reference callbacks
that don't actually support operation on bookmarks. Furthermore,
the mappings that are defined are not documented anywhere. Removing
them seems reasonable since they serve no real purpose.
If a file does not exist for the current buffer, this function
should fail with a clear warning message.
Here, I improved the messages that this function prints so that it
fails gracefully when no path can be determined in the calling
context.
The ":NERDTreeFind" command calls the "reveal()" method on the
NERDTree root node. The "reveal()" method would, in turn, call the
node's "open()" method. Since the "open()" method would only
initialize the child nodes of the root (i.e., read them from disk)
when the list of child nodes was empty, new paths would not be
included in the list.
This commit will result in the refreshing of the child node list
whenever "reveal()" is called on a directory node (unless it is the
first time the node is being opened... the most efficient option).
The result is that ":NERDTreeFind" will discover newly created paths
that exist on disk but are not cached in the NERDTree.
A stray debugging message is also removed.
Fixes issue #779.
Function-local variables, instead of script-local variables, should
be used here. In addition, "empty()" is a better choice for testing
for the absence of an argument. Finally, the use of "else" is
removed.
The docstring is also updated to include the new argument.
I contend that we should use confirm() whenever possible. It makes
the code cleaner and uses a builtin feature rather than a custom
one. Doing it the "Vim way" is always preferable in my mind.
The "o" mapping, which toggles directory nodes open/closed, allowed
the user to close the tree root. This was not consistent with the
"x" mapping which stops the user from doing this. This applies to
a double-click on the root node as well.
It should be noted that, if the root node is somehow closed, "o" and
double click can still re-open the tree, even with this change. In
other words, I was careful.
Previously, pressing "x" on the tree root would result in
unpredictable behavior. The user would either an receive an error
message or the parent of the tree root (which is not visible) would
be closed. This commit repairs this problem.
In addition, some code duplication was removed.
This bug is subtle! Opening a directory node in a new tab (with the
"t" or "T" mappings) would previously fail and require a refresh
because it called the directory node's "activate()" method.
In reviewing that method (i.e., "activate()"), I discovered that the
directory node's NERDTree is rendered before the method returns,
which overwrites the content of the tree in the new tab or window.
To clarify, when "t" or "T" is used on a directory node, a new
directory node and tree must be created to be rendered in a new tab.
So, calling "self.getNerdtree().render()" at the bottom of
"activate()" will render the NERDTree instance from which "t" or "T"
was invoked, not the new NERDTree that is being displayed. This
overwrites the new NERDTree text, and, thus, a refresh is required.
Since a call to "render()" is almost always necessary at the bottom
of "activate()" to keep everything in sync for other mappings, we
avoid this problem entirely by using the "open()" method directly
(works for files and directories) in the callbacks.
Fixes#549.
Directory tree navigation is broken because of directory signs which is
shown in UTF-8.
Cygwin is a Windows application, so it uses ASCII codepages and so
directory signs must be in ASCII, but if to modify
"nerdtree#runningWindows" function there are many other functions break
that convert paths, e.g.
So, the quick and reliable solution is to add a
separate function "nerdtree#runningCygwin" and use it in a specific
place.
As reported in issue #67, the function driving the `p` mapping was
not updated to work as expected when the cursor is positioned on a
cascade. This problem is addressed here.
Fixes#67.
Issues #597, #642, and #650 all report problems with the NERDTree
handler function for middle mouse clicks. In all cases, the problems
arose from the use of a function that didn't exist and from the use
of a bad argument in the call to the "g:NERDTreeAddKeyMap" function.
The fix for the first problem is obvious, but the solution to the
second bug merits explanation.
Previously, middle click events in the NERDTree window were
triggered with the "<MiddleRelease>" Vim key code. Since
"<MiddleMouse>" is always triggered before "<MiddleRelease>", The
error in #642 was bound to occur (because of the default behavior
for middle mouse clicks). Thus, the problem was easily solved by
using "<MiddleMouse>" instead of "<MiddleRelease>" in the mapping.
As an enhancement, I added the trigger of a "<LeftMouse>" event as
the first command in the handler function. This will cause the
middle click to reposition the cursor below the pointer before
continuing with its normal behavior. The benefits of this are clear.
This mapping has no defined behavior for bookmarks. Unless an issue
is raised to address this, it will be left just so for now.
Fixes#597, fixes#642, and fixes#650.
I altered the behavior of the ":OpenBookmark" command to match that of
the "NERDTree-o" mapping. This is acceptable for the following reasons:
1. It was broken, so no one was using it.
2. The name matches its behavior.
If a bookmark is to be opened in an explorer window, we should have a
command with a matching name for that behavior (":ExploreBookmark", for
example). This can be added later if there is enough demand for the
feature. Otherwise, this is a perfectly valid change.
Sorting the list of user bookmarks requires care to ensure that Vim's
builtin sort function is called correctly. Previously, this function was
called incorrectly. This is why the sorting of bookmarks never worked.
The offending functions have been removed here and replaced with
"s:Bookmark.CompareBookmarksByName". To understand the necessity for
this change, read ":h sort()" for the requirements of the function
reference argument (esp., note that it must return -1, 0, or 1).
In addition to fixing this problem, the new comparison function will
inspect the "g:NERDTreeBookmarksSort" setting to determine whether
case-sensitivity is preferred in the sort. The documentation has been
modified to accurately reflect this adjustment. The change is also made
in such a way as not to break any existing configurations.
Fixes#361 ("My bookmarks aren't sorted").
So :bprev was failing to jump back to the right buffer. I dont fully
understand this yet, but I'm fairly sure this is because there is a
magic <directory> buffer that we delete when creating a wintree :-/
Anyway, we are explicitly storing the prev buffer again - and updating
it when reusing treewins. It's worky, but there may be a superior way...
Some will have a "NERDTree" buffer variable and others will not. In the
ones that do, getbufvar will return a dictionary. getbufvar will return
an empty string from the other buffers. When looping through the buffers,
let will throw an error if nt already exists and is a different type than
what is trying to be put into it. This easily can be illustrated by these
two statements:
:let x = ""
:let x = {}
E706: Variable type mismatch for: x
This commit gets rid of the variable before setting it so that the
mismatch cannot occur.
This is the counterpart to a PR I just submitted to undotree
(https://github.com/mbbill/undotree/pull/61).
I noticed that my statusline doesn't update properly when using NERDTree to move
between revisions of a file with `go` or `gi`
(https://github.com/wincent/wincent/issues/16). I established that this was
because it was using `'eventignore'` to suppress all autocmds, which in turn
prevents the statusline from updating.
Commenting out the `set eventignore=all` line makes the failure to update go
away, at the cost of firing more autocmds.
I considered adding an option for opting out of this behavior (eg. `let
g:NERDTreeEventignore=0` or something), or rearchitecting my statusline to use
an approach like vim-airline does based on CursorMoved autocmds (see
https://github.com/vim-airline/vim-airline/issues/82; see also
30f078daf5/plugin/airline.vim (L36-L50)
for current implementation), but then realized that a simpler fix is to have
NERDTree just disable only the autocmds that it uses instead of disabling all of
them.
This is probably not enough to unbreak every bit of code in the world that
depends on those autocmds, but it does at least unbreak my use case, because it
allows my `WinLeave` autocmd to run and update the statusline.
Just disable the warning for now. The bookmark system will be replaced
with the project system soon - so no sense putting much effort into
this.
Closes#168
