I just installed sweet vim plugin that provides context colouring for javascript. I've fiddled with the colours so that they are just right, and am pretty pleased. However, there are times when I don't want the context colouring:
in insert mode, since the colouring doesn't work well while I'm editing
when I'm search, since the colouring seems to override the search hl
I resolved the first problem with:
autocmd InsertEnter *.js :JSContextColorToggle
autocmd InsertLeave *.js :JSContextColorToggle
However, the second problem is trickier. At first I thought I could just map /, my search key, so that it would first toggle the context colouring and then perform the search. I haven't been able to figure out how to write that mapping, though. How would I store "the original meaning of /" for use in my map?
Thanks,
p.s. check out this sweet context colouring (with solarized).
Hi I'm the author of the plugin. I fixed the conflict with hlsearch so it should just work now (try pulling latest version from git, I've not updated vim.org yet..)
As for the insert mode behaviour, there is a difference in behaviour between vim 7.3 and 7.4. 7.4 has the 'TextChanged' and 'TextChangedI' events, which fire when text is changed in normal , and insert mode respectively. This triggers the highlighting to update. However the TextChangedI event only fires on leaving insert mode. So if this is the behaviour you want, you can get it by upgrading to 7.4. In 7.3 I had to hook into the cursormoved event, which checks the 'b:changedtick' variable, which vim updates whenever changes occur... I'm not sure if this can occur during insert mode, but I think it might, which would explain odd behaviour if you're using 7.3.
I'm still trying to figure out what the best behaviour should be in insert mode...its tricky because the code syntax my be invalid during editing, and when the code cannot be parsed the plugin cannot work (and you will see regular syntax highlighting appear.. this may be a good thing as one value of syntax highlighting is visual syntax checking!). Another option would be to assume it is the same level as at the point where editing started, and offset the following text by however many characters are added/deleted during editing. Yet another option would be to have syntax highlighting in the area being edited (current line?) .
/ will enter command-mode.
if you only want to toggle highlighting with /, I think you need map <expr>.
something like
nnoremap <expr> / YourFunction()
in YourFunction(), you first do turn off the syntax hi, then return a /.
However you have to think about when to restore the hi. you can create another command mode mapping, map <cr> to first turn on the js hi, then return <cr>.
or just create an autocmd, when entering Normal mode, turn on the highlighting.
I didn't test above idea, hope it helps.
Related
I am using GVim 7.4 and I would like to do some really simple mapping to use CtrlP fuzzy matching of tags when a key combination is used.
I tried 2 approaches and they all seem to fail when vim calls Control + [x] combination. While I do understand that there are restrictions when it comes to mapping Ctrl+[x] codes, I haven't found any information on why ctrl mapping wouldn't work.
noremap \t :CtrlPTag<CR><C-\>w
This one enters CtrlP tag mode but then it doesn't enter word from under the cursor.
noremap \t <C-p><C-\>w
Here we don't even get to CtrlP window (I even omit going into tag mode here for simplicity).
As far as I understand (I'm no CtrlP user), the plugin is triggered via some command / key combination, and then presents interactive selection and filtering. It even has different "source" modes.
Now, this is a pretty heavy integration into Vim, probably using scratch buffers and its own input loop. That's why you cannot simply append keys to the mapping and get them interpreted by the plugin "as typed".
Typically, these plugins offer mode selection and so on via (optional) command arguments. Check the plugin's help, and if you cannot get the plugin into the state you need, best contact the plugin's author and ask for such enhancement.
I use pandoc markdown in text files and want to automate links that refer to internal textnodes. For example I have a link like [\%110lund] going to the word "und" in line 110. To automate the jumping process I defined a keybinding:
nnoremap <Leader>l vi[y/<ctrl+r>0<CR>
Unfortunately <ctrl+r> is written as the query string instead of performed to copy the visual selection.
So my question is how do I have to notate <ctrl+r>0 at this location so that it is actually performed instead of written out
Use c+r instead of ctrl+r.
In order to avoid confusion, I am striking out the incorrect edit that someone else made, rather than reverting it. In the context of a vim mapping (such as the :nnoremap of this question) the following should be typed literally. For example, <c-r> really means 5 characters.
Use <c-r> instead of <ctrl+r>.
See :help keycodes for more options.
Is there any way to stop vim from automatically updating folds on the fly? I really love vim's folding, and I prefer having it in syntax mode so that folds are created as I type. But for instance when I code C++ and I write a bracket { it automatically closes all subsequent folds, and when I then close the bracket again with a }, vim automatically expands all subsequent folds, meaning that I have to refold everything.
Another related problem, if I have the same document open in a different buffer, say I have run ":split", then writing an open bracket { will nest all folds in the buffer under the fold in which I opened the bracket, and closing it will un-nest the folds but also close all of them. If I use either "." or "->" to access a member function/variable, it resets all folds in the buffer to be whatever the current foldlevel is, regardless of which folds I have opened/closed myself.
This is somewhat frustrating when I have the same document open in two buffers so I can read the contents of one function when writing another, as I constantly have to switch buffers and reopen my folds.
In my .vimrc I have
set foldmethod=syntax
and that is about it. For autocompletion I use clang-complete and supertab with:
let g:SuperTabDefaultCompletionType = "<c-x><c-u><c-p>"
I think that is everything which migh affect this.
Edit:
Added some pictures to help illustrate the problem
Both problems can be solved with the following two autocmds:
autocmd InsertLeave,WinEnter * setlocal foldmethod=syntax
autocmd InsertEnter,WinLeave * setlocal foldmethod=manual
This sets the buffer local 'foldmethod' to manual when insert mode is entered or its window (a buffer display) is left, and sets it to syntax when insert mode is left or its window is entered.
This works because setting 'foldmethod' to manual will keep the folds automatically created by syntax as if you set them yourself (manually), and manual folds are not updated based on the syntax of the file.
I've found two bugs with this solution:
When switching windows while in insert mode, the autocmd will set the 'foldmethod' to syntax for the new window, even though it's in insert mode and should be set to manual.
This isn't really a problem for me because I use Vim like a civilized person and operate in normal mode by default.
When
a new buffer is created (e.g. by reading a file)
and 'foldlevel' is 0
and a particular syntax is used (I'm able to duplicate the issue with a C file)
and the o or O command is used to enter insert mode for the first time in that buffer (doing i<esc>o does not duplicate the bug),
then all folds below the cursor will be opened.
I accidentally discovered this when testing the above solution, and now looking back I'm surprised I found it; it's almost not even worth mentioning. I don't intend on trying to write a test file that has the exact syntax necessary to duplicate the bug, so this may go unnoticed for another eon.
I actually discovered this question several months ago and used the solution Ben linked to for a while, before eventually being annoyed enough with the multiple window for one buffer issue (your second problem) to fix it.
So thanks to Ben for his solution, and you for asking this question!
I made a bit of a modification to user4830797's answer which helps deal with situations where the file you're editing doesn't use foldmethod=syntax (for example, a .vimrc file which might use foldmethod=marker):
autocmd InsertLeave,WinEnter * let &l:foldmethod=g:oldfoldmethod
autocmd InsertEnter,WinLeave * let g:oldfoldmethod=&l:foldmethod | setlocal foldmethod=manual
I think you need to check :h 'foldlevel. You should also perhaps use :mkview, probably in autocmd which restores manually open and closed folds.
Other then that you should perhaps set folding method differently on different file types (for instance on C you could set it to manual or marker)
If you temporarily set the foldmethod to manual, then Vim will keep all the folds currently defined by syntax, and keep them in the exact open/closed state you have them in now. This can be done automatically with an InsertEnter autocmd and restored on InsertLeave to protect your fold states in a single window. Unfortunately I have not yet spent time trying to get it working in split windows, but using window-local variables it is easy enough to even account for the user switching windows or something without leaving insert mode. See http://vim.wikia.com/wiki/Keep_folds_closed_while_inserting_text for details and discussion.
Sometimes when I'm working on a project I want to play around with some data. Often times the data is on one line and is huge (>25k characters). I understand I could set nowrap and have this line just run off the screen, but I tend to like set wrap for other reasons. So, as a workaround I want to hide these long lines in a marker fold (e.g. {{{ long line }}}). This works fine but I run into a problem with synmaxcol for some reason. If the folded line exceeds synmaxcol then when I open the file, the syntax highlighting runs over. For example:
However, as soon as I open the fold the syntax corrects itself:
Having to open the fold every time is annoying though. As you can see in this example the line is not actually all that long -- it just exceeds synmaxcol. Since synmaxcol is exceeded at a "string" element, the rest of the file is highlighted as a string (so nothing but a singular double quote will stop it).
Why is this happening and how can I fix it? I've tried this with different syntax files and filetypes and it still occurs. I've also tried it with no plugins, a minimal vimrc (containing only syn on) and a modeline to set fdm=marker:synmaxcol=60 and it still happens.
You can manually enter :syntax sync fromstart to force Vim to rescan the syntax from the beginning of the opened file.
I would suggest defining a hotkey for convenience:
noremap <F5> <Esc>:syntax sync fromstart<CR>
inoremap <F5> <C-o>:syntax sync fromstart<CR>
Now you can press F5 to clean up most syntax highlighting problems.
Also, have a look at Vim's fixing syntax highlighting - wiki page
Moreover reading :help :syn-sync-first might shed some more light on the issue.
UPDATE:
I was able to reproduce this behavior on my machine (I'm running Vim 7.3.429).
However, when I wrapped the fold markers {{{ and }}} in block comments, vim correctly rendered the syntax. You can create appropriately wrapped fold-markers using the zf command. See Vim tips: Folding fun.
Normally Vim picks the correct blockcomment string based on the currently active syntax. However, my Vim is pretty vanilla and didn't recognize Ruby syntax. I could specify autocmd FileType ruby set commentstring==begin%s=end in my .vimrc file to set the proper block comment. See :fold-create-marker for more details.
Another solution is to set synmaxcol=0, which will effectively set it to infinity. This causes Vim to check the syntax of the entire line, no matter how long it is. However, I'm not sure what kind of performance penalty you'll have to pay for that.
I've set this on my .vimrc:
let g:clang_snippets=1
let g:clang_snippets_engine='clang_complete'
let g:clang_conceal_snippets=1
set conceallevel=2 concealcursor=inv
I don't know how conceal is expected to work, maybe the clang_complete's docs should have a tip for a specific setting to hide the snippets adorns.
How do I hide it? I'm using MacVim built with +conceal, but it's not working. This is my messy .vimrc by now.
NOTE:
I'm sticking with g:clang_snippets_engine='clang_complete' because it seems to be more smart than the snipMate parameter completion, switching to NORMAL mode is a wiser choice to navigate between parameters since I can use SuperTab completion for params in INSERT mode while being able to navigate through them with the same tab at NORMAL mode. snipMate engine was acting weird to me sometimes too, sometimes it switched to a parameter after a completion, sometimes not.
Also, I'm missing a final tab to go after the last parameter, right after the function call (snipMate does that), so I can just insert ; and hit Enter.
Disclaimer: This question is related with the issue at https://github.com/Rip-Rip/clang_complete/issues/176.
EDIT:
My problem was with this line at my .vimrc:
au BufNewFile,BufRead *.cpp set syntax=cpp11
I'm using C++11 Syntax Support and #xaizek has discovered and pointed it out as the problem in the comments bellow in the accepted response, it seems the root cause is the use of the syntax clear command in it.
According to :help 'concealcursor':
Sets the modes in which text in the cursor line can also be concealed.
When the current mode is listed then concealing happens just like in
other lines.
n Normal mode
v Visual mode
i Insert mode
c Command line editing, for 'incsearch'
So with concealcursor=iv you asked Vim to hide concealed text in insert and visual modes, but show it in normal mode. So just do:
:set concealcursor=inv