Previously if a code block was part of a list and o or O was used a new
list marker would be inserted. Also moved the is_codeblock check
function to utils file for use elsewhere.
- Create function wrapper around lvimgrep for input checking, pattern
quoting, and error handling.
- Add Vader tests for VimwikiSearch.
- Change syntax loading from try/catch to explicit file check (to
prevent Vader test bug).
- Update doc/vimwiki.txt for changes.
- Change test script to run Vader tests separately
Previously a new list marker would get added when working within a code
block that was part of a list. This change excludes code blocks from
list marker creation.
This change modifies the behavior of VimwikiReturn to not insert a new
list marker if not within a list. Mostly useful when mapping <CR> to
VimwikiReturn 3 5 or similar. The behavior of 'o' was also modified to
properly insert a marker in a list with hard line wraps.
The functions stopped at the end/start block marker (respectively)
instead of returning the following line. This caused issues when the
function was subsequently called in markdown mode since the start and
end block markers are the same.
This fixes#407
For markdown syntax the extension was added to the url multiple times if
the option g:vimwiki_markdown_link_ext was set which is now fixed.
Normalizing links no longer adds the extension to the description field.
Fixes#240. Also fixes a related issue where markdown links where not
correctly highlighted in todo lists. Adds concealment for markdown links
in todo lists.
The changes made with #686 did not ensure the ftplugin mappings were
specific to the buffer containing a Vimwiki file. This resulted in
undesired keymappings with no defined behavior since the <Plug>
definition was buffer specific.
All key mappings have the option of being disabled but the <Plug>
definitions are still available for user customization. Key sequences
are no longer overwritten if a mapping already exists. Fixes#671,
fixes#425.
Vim used to throw the E706 error if you tried to change a variable's
type. This error can be triggered by vimwiki if, during the user
settings loop in `read_global_settings_from_user()`, the variable type
for `users_value` changes.
The vim error was removed in v7.4.1546 as part of
f6f32c38bf.
Changing this option to allow empty strings places the diary files in the same directory as the main wiki files. This is more backwards compatible with previous releases.
Names can be used for interwiki links in the format wn.name:link.
Additionally, :VimwikiUISelect and the menu created in GVim utilize the
wiki name. Fixes#612, closes#477.
* Temporary fix for omnicomplete of vimwiki links - #456.
This fixes the omnicomplete of wiki links under Windows which were
not working since paths on Windows use '\' instead of '/'. This is
a temporary fix until path refactoring is done.
* Update changelog with description of fix for #456
Update CSS Styles
Add Splash Image
Compress PNG
Remove header
Change headings sizes
Add background colors to code blocks and use HEX
Update fonts and put back old screenshots
Change font-size in code blocks
This adds the vimwiki_local `html_filename_parameterization` option (0|1)
which performs the same sanitization of filenames as does the
vimwiki_markdown gem.
For example, if your file is called "My File.md" it will be written out
as "my-file.html".
If the html_filename_parameterization options is enabled, we also _do
not delete_ non-matching html files when the VimwikiAllToHtml command is
run.
Closes#129
Adds new configuration variable, "conceal_onechar_markers", defaulting
to on (preserving default behaviour)
Adds if-statement around relevant parts of code (as suggested in the
issue), which uses the new configuration variable.
Fix#315 - Don't conceal one-character markers
The option 'links_space_char' replaces spaces with the specified
character when creating a link from a visual selection. This modifies
the behavior to not replace spaces in the link description but to still
replace spaces in the filename.
The optimized table formatting algorithm was still O(n) because
calculating aligns required traversal to the separator line on the top
of a table. This heuristic algorithm calculates aligns in O(1) in the
best case by looking at cells in two rows above.
Say, if two above rows are
| uau | uauyaya | ya |
| ua | uaua | uaaua |
| <cursor> | | |
then the aligns can be figured out without need to find the separator
line. If aligns cannot be figured out after this fast check then the
algorithm falls back to O(n) with searching for the separator line.
This formatting is needed in the optimized table formatting algorithm
for shrinking the current table line to try the best to avoid
re-formatting the whole table. However, this breaks cursor motion and
may also affect other table related behaviors. The fix is to move this
formatting to get_aligned_rows(). Now table rows are not affected by any
means, because the current row is only replaced temporarily for the
formatting purpose only and gets reverted as soon as the check is
finished.
Fixes#656.
This reverts commit 23d273d547, which
fixes#655.
Description of fixed problem:
1. When looping through rows, if a separator was found, all subsequent
rows would not get an assigned alignment.
2. This breaks `s:get_aligned_rows` because it calls `s:fmt_sep` and
`s:fmt_row`, both of which expect `aligns` to have a value for every
row number.
