真的很酷，但是我对映射的建议是使用^W^J而不是J(因为所有的H J K L已经有意义了)，另外我还会引入新的缓冲区，因为当你想要交换的时候，你可能不想继续编辑你已经在的缓冲区。是:

function! MarkSwapAway()
    " marked window number
    let g:markedOldWinNum = winnr()
    let g:markedOldBufNum = bufnr("%")
endfunction
function! DoWindowToss()
    let newWinNum = winnr()
    let newBufNum = bufnr("%")
    " Switch focus to marked window
    exe g:markedOldWinNum . "wincmd w"
    " Load current buffer on marked window
    exe 'hide buf' newBufNum
    " Switch focus to current window
    exe newWinNum . "wincmd w"
    " Load marked buffer on current window
    exe 'hide buf' g:markedOldBufNum
    " …and come back to the new one
    exe g:markedOldWinNum . "wincmd w"
endfunction
nnoremap <C-w><C-h> :call MarkSwapAway()<CR> <C-w>h :call DoWindowToss()<CR>
nnoremap <C-w><C-j> :call MarkSwapAway()<CR> <C-w>j :call DoWindowToss()<CR>
nnoremap <C-w><C-k> :call MarkSwapAway()<CR> <C-w>k :call DoWindowToss()<CR>
nnoremap <C-w><C-l> :call MarkSwapAway()<CR> <C-w>l :call DoWindowToss()<CR>

我从sgriffin的解决方案中有一个稍微增强的版本，您可以在不使用两个命令的情况下交换窗口，但是使用直观的HJKL命令。

事情是这样的:

function! MarkWindowSwap()
    " marked window number
    let g:markedWinNum = winnr()
    let g:markedBufNum = bufnr("%")
endfunction

function! DoWindowSwap()
    let curWinNum = winnr()
    let curBufNum = bufnr("%")
    " Switch focus to marked window
    exe g:markedWinNum . "wincmd w"

    " Load current buffer on marked window
    exe 'hide buf' curBufNum

    " Switch focus to current window
    exe curWinNum . "wincmd w"

    " Load marked buffer on current window
    exe 'hide buf' g:markedBufNum
endfunction

nnoremap H :call MarkWindowSwap()<CR> <C-w>h :call DoWindowSwap()<CR>
nnoremap J :call MarkWindowSwap()<CR> <C-w>j :call DoWindowSwap()<CR>
nnoremap K :call MarkWindowSwap()<CR> <C-w>k :call DoWindowSwap()<CR>
nnoremap L :call MarkWindowSwap()<CR> <C-w>l :call DoWindowSwap()<CR>

试着在正常节点中使用大写HJKL来移动你的窗口，这真的很酷:)

真的很酷，但是我对映射的建议是使用^W^J而不是J(因为所有的H J K L已经有意义了)，另外我还会引入新的缓冲区，因为当你想要交换的时候，你可能不想继续编辑你已经在的缓冲区。是:

function! MarkSwapAway()
    " marked window number
    let g:markedOldWinNum = winnr()
    let g:markedOldBufNum = bufnr("%")
endfunction
function! DoWindowToss()
    let newWinNum = winnr()
    let newBufNum = bufnr("%")
    " Switch focus to marked window
    exe g:markedOldWinNum . "wincmd w"
    " Load current buffer on marked window
    exe 'hide buf' newBufNum
    " Switch focus to current window
    exe newWinNum . "wincmd w"
    " Load marked buffer on current window
    exe 'hide buf' g:markedOldBufNum
    " …and come back to the new one
    exe g:markedOldWinNum . "wincmd w"
endfunction
nnoremap <C-w><C-h> :call MarkSwapAway()<CR> <C-w>h :call DoWindowToss()<CR>
nnoremap <C-w><C-j> :call MarkSwapAway()<CR> <C-w>j :call DoWindowToss()<CR>
nnoremap <C-w><C-k> :call MarkSwapAway()<CR> <C-w>k :call DoWindowToss()<CR>
nnoremap <C-w><C-l> :call MarkSwapAway()<CR> <C-w>l :call DoWindowToss()<CR>

上面所有的答案都很好，不幸的是，这些解决方案与QuickFix或LocationList窗口结合使用不能很好地工作(我在尝试让Ale错误消息缓冲区与此一起工作时遇到了这个问题)。

解决方案

因此，我添加了额外的一行代码，在进行交换之前关闭所有这些窗口。

exe ' if windo &buftype = =“quickfix | | &buftype = =“locationlist | lclose | endif’

整个代码看起来像;

" Making swapping windows easy
function! SwapWindowBuffers()
    exe ':windo if &buftype == "quickfix" || &buftype == "locationlist" | lclose | endif'
    if !exists("g:markedWinNum")
        " set window marked for swap
        let g:markedWinNum = winnr()
        :echo "window marked for swap"
    else
        " mark destination
        let curNum = winnr()
        let curBuf = bufnr( "%" )
        if g:markedWinNum == curNum
            :echo "window unmarked for swap"
        else
            exe g:markedWinNum . "wincmd w"
            " switch to source and shuffle dest->source
            let markedBuf = bufnr( "%" )
            " hide and open so that we aren't prompted and keep history
            exe 'hide buf' curBuf
            " switch to dest and shuffle source->dest
            exe curNum . "wincmd w"
            " hide and open so that we aren't prompted and keep history
            exe 'hide buf' markedBuf
            :echo "windows swapped"
        endif
        " unset window marked for swap
        unlet g:markedWinNum
    endif
endfunction

nmap <silent> <leader>mw :call SwapWindowBuffers()<CR>

交换功能的功劳归布兰登·奥瑟

为什么需要它

如果不先删除所有QuickFix (QF)和LocationList(LL)窗口，swap函数就不能正常工作的原因是，如果QF/LL缓冲区的父缓冲区get被隐藏(并且在窗口中不显示)，耦合到它的QF/LL窗口将被删除。这本身不是一个问题，但是当窗口隐藏时，所有的窗口号码都被重新分配，交换被打乱，因为第一个标记的窗口的保存号码(可能)不存在了。

从这个角度来看:

第一个窗口标记

____________________
| one              | -> winnr = 1    marked first    g:markedWinNum=1
|                  | -> bufnr = 1
|__________________|
| two (QF window   | -> winnr = 2
| coupled to one   |
|__________________|
| three            | -> winnr = 3
|                  | -> bufnr = 2
|__________________|

第二个窗口标记

____________________
| one              | -> winnr = 1                    g:markedWinNum=1
|                  | -> bufnr = 1
|__________________|
| two (QF window   | -> winnr = 2
| coupled to one)  |
|__________________|
| three            | -> winnr = 3    marked second    curNum=3
|                  | -> bufnr = 2                     curBuf=2
|__________________|

第一个缓冲开关，窗口1被窗口3的缓冲区填充。因此，QF窗口将被删除，因为它不再有父窗口。这将重新排列窗口编号。请注意，curNum(第二个选择窗口的编号)指向一个不再存在的窗口。

____________________
| three            | -> winnr = 1                    g:markedWinNum=1
|                  | -> bufnr = 2
|__________________|
| three            | -> winnr = 2                     curNum=3
|                  | -> bufnr = 2                     curBuf=2
|__________________|

所以当切换第二个缓冲区时，它会尝试选择curNum窗口，这个窗口已经不存在了。因此，它创建了缓冲区并切换缓冲区，导致一个不需要的窗口仍然打开。

____________________
| three            | -> winnr = 1                    g:markedWinNum=1
|                  | -> bufnr = 2
|__________________|
| three            | -> winnr = 2
|                  | -> bufnr = 2
|__________________|
| one              | -> winnr = 3                     curNum=3
|                  | -> bufnr = 1                     curBuf=2
|__________________|

看看:h ctrl-w_ctrl-x和/或:h ctrl-w_ctrl-r。这些命令允许您在当前布局中交换或旋转窗口。

编辑:实际上，在这种情况下，这将不起作用，因为它只会交换当前列或行。您可以转到每个窗口并选择目标缓冲区，但这相当繁琐。

从这个开始:

____________________
| one       | two  |
|           |      |
|           |______|
|           | three|
|           |      |
|___________|______|

将“three”设置为活动窗口，然后发出命令ctrl+w j，这将移动当前窗口以填充屏幕底部，留给你:

____________________
| one       | two  |
|           |      |
|___________|______|
| three            |
|                  |
|__________________|

现在将'one'或'two'作为活动窗口，然后发出命令ctrl+w r。这将'旋转'当前行中的窗口，留下你:

____________________
| two       | one  |
|           |      |
|___________|______|
| three            |
|                  |
|__________________|

现在将“two”设置为活动窗口，并发出ctrl+w h命令，这将移动当前窗口以填充屏幕左侧，留下:

____________________
| two       | one  |
|           |      |
|           |______|
|           | three|
|           |      |
|___________|______|

如你所见，这个动作有点乱。它有3个窗口，有点像“瓷砖游戏”谜题。如果你有4个或更多的窗口，我不建议你尝试这个方法——你最好关闭它们，然后在需要的位置重新打开它们。

我做了一个视频演示如何在Vim中使用分割窗口。