as requested by the exception thrown:
fugitive: A third-party plugin or vimrc is calling
fugitive#buffer().commit() which has been removed. Replace it with
matchstr(FugitiveParse()[0], '^\x\+')
This allows for advanced customization where the shortening of the
branch name is up to the caller, unlike before.
This change is observable from the outside, and as such can be
dangerous. But AFAIK it does not change the default behavior.
This removes the hardcoded minwidth limits for both the `hunks` and
`branch` parts. It replaces these with safe accesses to the `minwidth`
setting as done by `airline#parts#define_minwidth`.
This is an experimental feature that allows to display the statusline in
the tabline. It might still be a bit rough, but seems to work so far.
Remaining problem:
- Mode changes are not immediately detected, only after moving the
cursor
fixes#1388closes#1867
for some reasons, calling async functions might cause an error, if a
SourcePre command has been defined. So explicitly call the functions
using `:noa` modifier to prevent triggering the SourcePre autocommand.
fixes#1742
this caused that the branch extension was not correctly working anymore
because of a refactoring of fugitive tpope/vim-fugitive@5d11ff7
Solution: Move the existence check for the fugitive plugin into a
separate function and call it from there in all places that check the
fugitive plugin. Do the same for lawrencium and vcscommand check.
fixes#605#1739
Instead of requiring each version control plugin to modify airline to
show the current branch, provide a customization function we can check
instead.
Following the example of airline_theme_patch_func, you define the
variable like so:
let g:airline#extensions#branch#custom_head = 'david#svn#get_branch'
Custom functions should cache their value. They may need an autocmd to
invalidate their cache:
" Use a buffer-unique group name to prevent clearing autocmds for other
" buffers.
exec 'augroup svndavid-'. bufnr("%")
au!
autocmd BufWinLeave <buffer> unlet! b:svndavid_branch
augroup END
This change lets me integrate with vc.vim (I couldn't get VCSCommand
working for svn) or write my own thing for perforce.
Additionally, always load whole file and check for existence.
Instead of determining up front whether various scm plugins are
installed, check for them on use so they can be added after this script
is sourced.
This also mitigates the problem of checking for existence of autoload
functions (which are not loaded by exist()). Since we're checking
root-level functions, they're likely to be loaded once we're using any
part of the plugin.
previously, the branch extension used name[0:6], however that would
break with multibyte characters, since this is a byte index and not a
character index.
fixes#1686
The initial reason in #237 is not valid anymore, as vim-gitgutter
functions as expected for editing files that are not part of a repo,
whether they are:
- in a different repo
- in a parent repo (cwd being a submodule)
- outside of the repo
Furthermore, removing this check allows to show relevant info for
specific fugitive file names that are fugitive://..../sha1//...
which are hard to parse manually, especially in complicated situations
such as submodules.
rename g:airline#init#async variable to g:airline#init#vim_async
because that is what it is for: showing whether vim supports async. Is
not set vor nvim, because nvim always supports jobs.
When using vim-gitgutter and fugitive:
The hunks extension and the branch extension work as expected when the
file is first loaded; both parts are added to the statusline.
Once the cursor is moved and stopped for &updatetime ms:
1. The branch extension clears b:airline_head on CursorHold
2. Somehow (?) airline#statusline gets called on CursorHold
3. The hunks extension returns '' when b:airline_head is empty, causing
the hunks to be removed from the statusline.
It doesn't make sense to clear airline_head just because the cursor
moved, and the commit message adding the line doesn't say why:
13297cee03
Commit 174b7e1962 relies on airline_head
being set.
Debug detail:
Executing CursorHold Auto commands for "*"
autocommand unlet! b:airline_head
[...]
continuing in CursorHold Auto commands for "*"
calling function airline#statusline(1)
[...]
line 1: return exists('*airline#extensions#branch#head') && empty(get(b:, 'airline_head', ''))
function airline#extensions#hunks#get_hunks[13]..<SNR>32_get_hunks[14]..<SNR>32_get_hunks_gitgutter[1]..<SNR>32_is_branch_empty returning #1
function airline#extensions#hunks#get_hunks[13]..<SNR>32_get_hunks[14]..<SNR>32_get_hunks_gitgutter returning ''
function airline#extensions#hunks#get_hunks[13]..<SNR>32_get_hunks returning ''
function airline#extensions#hunks#get_hunks returning ''
:au CursorHold
--- Auto-Commands ---
gitgutter CursorHold
* call gitgutter#process_buffer(bufnr(''), 1)
CursorHold
* unlet! b:airline_head
airline_whitespace CursorHold
* call <sid>ws_refresh()
------------------------
Sample vimrc:
set nocompatible
if empty(glob('~/.vim/autoload/plug.vim'))
silent !curl -fLo ~/.vim/autoload/plug.vim --create-dirs
\ https://raw.githubusercontent.com/junegunn/vim-plug/master/plug.vim
autocmd VimEnter * PlugInstall --sync | q | doautocmd WinEnter
endif
autocmd VimEnter *
\ if len(filter(values(g:plugs), '!isdirectory(v:val.dir)'))
\| PlugInstall --sync | q
\| endif
call plug#begin('~/.vim/bundle/')
Plug 'vim-airline/vim-airline'
Plug 'tpope/vim-fugitive'
Plug 'airblade/vim-gitgutter'
call plug#end()
set laststatus=2
set updatetime=250
let g:airline_theme = 'dark'
Currently, vim-airline will check untracked status for files e.g. living
in .git. So when editing .git/config it will show as being untracked.
While technically, this is correct I prefer not to see this for those
files. So skip the check for those files.
If a file is edited inside a git repository, which lies within a git
repository, the branch extensions shows 'gitmaster | hgdefault'
To make it more obvious, that we are looking into both repositories
here, use 'git:master | hg:default'
closes#1482
Currently vim-airline assumes, that the git_dir is part of the path for
the file being edited. This has changed, since git supports worktrees.
So take care of b:git_dir (which is set by fugitive) being a path
differently from the absolute path of the file being edited (however, it
should include the substring worktree in that case).
This commit makes branch.vim use neovim's async jobs instead of a system()
function. This way we avoid the v:shell_error overwrite bug and allow live
updates of the untracked status.
* The head string is now calculated iff it has changed.
* The not exists symbol for current file appears as soon as its status is known.
* Fixes various problems with asynchronous status checking, such as:
* The not exists symbol keeps appearing and disapearing. This happened when
file was marked as not existing, the untracked cache was invalidated, and
the cache update is started, but in the meantime, the head string
calculation used the current (empty) value of the cache.
* The not exists symbol never appears, because cache keeps getting invalidated
before b:airline_head is emptied and updated.
closes#1306
* Introduce a config variable that holds the vcs-dependent parts of code.
* Removes `get_*_untracked` duplication by merging their logic together.
* Removes custom checks for 'git' or 'hg'. Functions now rely on provided config
argument.
* Use loops instead of manually specifying each handled VCS.
closes#1303
This commit fixes a bug, where untracked files in Git repos did not get the not
exists symbol displayed. The fix works by replacing the previous check for
whether currently edited file is a directory or not with a check based on
`findfile`. More detailed explanation follows.
VCS extension to vim-airline checks whether currently edited file is untracked.
The previous `s:get_git_untracked` implementation, which was used for the
aforementioned purpose for Git repos, was buggy, because it did not return the
untracked symbol in most situations, i.e. the edited file is untracked, but it
was treated as if it wasn't.
The root cause was the second clause in boolean expression checking the output
of `git status`:
if output[0:1] is# '??' && output[3:-2] is? a:file
It was added to make sure we are not checking a directory, but at this point in
the program `a:file` is an absolute path, while output of `git status` is a
relative path from the root of git repo. So the `is?` expression failed in most
situations.
