Groupnames like 'airline_c1_to_airline_x_inactive' still have to be
processed, so do not skip them.
Also for the tabline, '_inactive' groups are not used, so skip them
Replace by a function that retuns the to be executed highlighting string
Should in theory be a bit faster, since the same function does not have
to be called 5 times per highlighting group.
It probably is not much better, but here are some random numbers:
Profiling:
Previously:
count total (s) self (s)
199 0.022973 0.009909 let cmd = printf('hi %s %s %s %s %s %s %s %s', a:group, s:Get(colors, 0, 'guifg=')…
New:
count total (s) self (s)
79 0.010166 0.000862 let cmd = printf('hi %s%s', a:group, s:GetHiCmd(colors))
Make sure, to also check, if the separator groups have already been
defined. Once they are defined initially, they are usually available in
the palette group. So do not redefine them once they have been created
initially.
references #1779
Previously, we only checked for if the name starts with 'airline_c'
But there could as well be highlighting groups starting being called
'airline_b_to_airline_c5', but if buffer 5 is no longer visible in the
current viewport, then we do not need to recreate those groups.
currently, if the matchstr() evaluates to '', it would skip the required
section, because bnr would be evaluated to zero and since there is no
buffer zero, skip that highlighting group.
Make sure, to only skip the highlighting group if the buffer number is
actually > 0
In a long editing session, there could happen to accumulate several
highlighting groups for buffers that might no longer be visible.
Therefore, only re-create the highlighting group for buffers that are
actually displayed in the current tabpage. If not, skip them.
references #1779
previously, it could happen that the same highlighting group was defined
several times, because it was available in several modes within
g:airline#theme[mode]. So the second one would always win.
Therefore, loop through all modes in reverse order and define the group
and remember what group has already been defined. If we happen to have
to re-define the same group, skip it. Since we are traversing the list
in reverse order, this should make sure the last definition wins.
This has the benefit of being more performant and hopefully helps with
e.g. #1779 and similar issues.
At the moment you can never be sure whether you look at the results
that ale just produced after your last changes or you are looking at
outdated information.
Instead of showing the last results while ale executes checks on the
current buffer it can now show a specific indicator.
This commit introduces the new config variable
`g:airline#extensions#ale#checking_symbol` which defaults to "...". In
case you don't want to see this, you may set it to an empty string.
This will make sure that the current window will be highlighted as
expected, even if there are no other windows and all highlighting groups
will be correctly re-created as 'inactive'
references #1807
Previously the ale refresh was triggered after user interactions only.
When linting takes some time and the user is not actively working the
ale information was not updated while the things ale directly controls
were updated. This change makes showing linting results a loot smoother.
for older Vims, the wordcount extension falls back to `g<c-g>` and
parses the messaage that is generated.
When doing so, it expects the English locale, which might not be true.
Therefore, try to regex parse the result without using the english words
(assuming the output is still in the same order as in English).
Compatibility:
- Don't use a script-local function to update the format strings
- Protect against `:normal!' moving the cursor on on the wordcount check
Bugfix:
- Let to_string() try to return something for all values
- Now returns correctly when passed both 0 and '0'
Upkeep:
- Simplify check again no valid key from winwidth()
- Old wordcount check: use matchstr() - more expressive and fewer steps
- Improve documentation style/clarity/detail
v:t_list is an internal variable describing the type of a list.
Unfortunately, this is not available in older vims. So let's fall back
to using `type([])`
see #1795
This prevents a warning message, when the function is tried to add
several times. Also while at it, add a '!' so that no error is thrown if
the file is sourced a second times.
Checking for wordcount changes now uses a b:changedtick comparison in
the statusline funcref. The autocommand strategy that used to do this is
removed, simplifying the code.
- Formatter is now only called when the wordcount changes
- ...#apply() now only compares against the filetype list when necessary
- Old format() function is no longer called for (unsupported) counting
of visual words
- Misc code quality improvements
Various improvements:
- Seperate out visual mode detection.
- Use TextChanged rather than CursorMoved where supported.
- Let users specify the filetypes for which wordcounting is enabled
with a list rather than a pattern.
- Move the filetype check to when airline is refreshed, as opposed to
on every update; autocommands are not created if wordcounting is
disabled
Currently the formatter, and not the wordcount plugin, is responsible
for providing the wordcount as well as formatting it. The default
formatter allows visual mode word counting, although this is not
documented.
The new interface - a transform() function, allows the main wordcount
plugin to internalise this logic. Providing the wordcount simplifies
formatter implementations:
- All formatters can display the visual wordcount.
- Formatters do not have to worry about compatibility with different
vim versions.
The old format() function can now be deprecated, although the wordcount
plugin retains compatibility with formatters using it. The default
formatter will also be used as a fallback if no suitable function is
found.
The default formatter is rewritten to use the new interface.
Add 'arduino' filetype to list of C-like languages. This is needed in
order to avoid mix-indent-file warnings when using hard tabs for
indentation in Arduino source files (*.ino, *.pde) containing top level
block comments with leading space before a '*'.
Similar to: 460ed02864
Original fix: 8fde76dd63
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
closes#1730
- Make sure to always call the term extension on TermOpen autocommands
- consistently use the airline_term highlighting group
- hard code the airline_term highlighting group, because by the time the
function airline#themes#patch() is called the highlighting group
airline_c would no be defined, so we cannot dynamically get those
attributes
- In the documentation, mention how the g:airline_mode_map can be set
including the terminal section
The default formatter uses a hand-build regexp for shortening the path
in the tabline. However, since it uses the \w regex atom, this won't
match e.g. cyrillic letters.
To fix this, use the builtin pathshorten() function which does handle
this case correctly. For a test, use e.g. 'D/Учёба/t.c'
closes#1737
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
Startify unfortunately does not set the buftype option, so the
`[noperm]` would be shown, since the buffer is not really a file.
(see mhinz/vim-startify#324)
So allow to blacklist startify from this setting. To do so, the logic
from the tabline extension was refactored out into a separate function
that returns true, if the buffer matches
`g:airline#extensions#tabline#ignore_bufadd_pat` and that function is
then reused for the readonly() function as well.
Recently, NERDTree added the &readonly setting to its buffer.
Unfortunately, this caused airline to render the '[noperm]' string in
it.
Fix this by only making the readonly check for buffers that actually
represent files (e.g. the buftype option is empty).
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.
275ec4fe63 broke this so that numbering
started from 0 instead of 1 with
g:airline#extensions#tabline#buffer_idx_mode = 1
Specifically, I overlooked that |index| actually tracked the index of
the current buffer in the buffer list *starting from 1*, whereas I
assumed it tracked the true index (starting from 0).
