tsukkee / unite-tag Goto Github PK

    autocmd BufEnter *
                \   if empty(&buftype)
                \|      nnoremap <buffer> <C-]> :<C-u>UniteWithCursorWord -immediately tag<CR>
                \|  endif

here is my config for unite-tag,and my os is ubuntu ,gvim 7.4
then i open vim with a new file named Test.java type into

class Test{
  String
}

and do not save it,my cursor in on String then i type c-]
the vim show me the resuly in the unite buffer,if i select one and type Enter,the file will be opened in the current buffer,and cover my new file without save it!
i think it is better to confirm whether the file in current buffer has been saved,if not open the selection in a split buffer.nerdtree also has this feature ,maybe you can have a try.

It would be nice to add support for alternative tags list specification.

Use case:

• use default tags option for navigation between project and its dependencies by tag under cursor
• use tags generated only from project files for unite candidates

{g:unite_data_directory}/tag ではなく、 {g:unite_data_directory}tag というパスになっているようなのですが、これは意図したものでしょうか？

In FuzzyFinder I use :FufBufferTag to show tags in the currently open file only, is that possible with unite-tag?

I can't find it in help.
Suppose I have multi tag file, can I define which one should be used?

Thanks

Hi,

I'm having trouble getting my tags to show up in unite.tag - the list of tags is empty. They show up automatically in tagbar, are they supposed to show up automatically in unite.vim as well, or do I need to generate them manually before it will work?

The same way it is done for fname and relative_fname. Right now kind maximum width is hardcoded to 10, which is a bit much if only shortened kind is shown.

Please add the similar g:unite_source_tag_max_kind_length option.

Hi

I have a few tags file > 4MB. Consequently, it took a while to cache for the first time using. Any workaround to trigger caching tags automatically after vim is loading? Thanks~

Hi @Shougo, thank you very much for adding kind support (following our discussion here).

I have a few questions:

i) In the mentioned issue you said:

If you want to filter by kind, you must use "converter_abbr_word".

Generally I want to filter by candidate but maybe I want to filter by kind. So: is it possible to switch the converter within the unite buffer? i.e pressing something like <C-s> and switch from the default converter to abbr_word?

ii) Consider the following foo.py file:

def foo(x):
    return x

y = 3

When I run :Unite tag/include I get the following unite buffer:

Instead of showing k:f and k:v is possible to directly display function and variable (i think that a) it is better to show the full kind instead of the abbreviation and b) there is no need to show k: since it is clear that this third column of the unite buffer displays the tag kind).

iii) When I run :Unite tag/include in my vimrc file I get:

Can the syntax highlighting and the spacing between columns be improved?

Hi,

There is any way that I can auto-cache the tags file as soon as vim is opened ?

Thanks in advance

PS: Thanks for this very useful plugin

Hello.

I have this tag/include profile:

call unite#custom#profile('tagbar', 'context', {
    \ 'quit': 0,
    \ 'buffer_name': 'Tagbar',
    \ 'start_insert': 0,
    \ 'vertical': 1,
    \ 'direction': 'topleft',
    \ 'toggle': 1,
\})

I toggle tag/include split with this mapping:

nmap <silent> <F1> :Unite -profile-name=tagbar tag/include<CR>

Steps to reproduce the problem:

1. Open a file with some definitions that tag/include can understand.
2. Hit <F1>. A split with tag/include is now open and focused.
3. Select any definition, hit <CR>. Cursor jumps to the selected definition in the file.
4. Hit <F1> again to close the tag/include split.

Expected results:
Cursor stays on the definition.

Actual results:
Cursor jumps away from the definition and positioned where it was when tag/include split was opened.

Please fix.

My tags file includes class names in it. It would be great to have unite-tag display the class name (if it exists) instead of the filename.

Is that something that's possible?

let g:unite_source_tag_max_fname_length = 64

This seems to have no effect regarding file name length.

Say VimL file has script-local function test() defined as

function! s:test()
...
endfunction

Then tag/include shows both test and s:test.

tags soruce for unite.vim - please fix the small mistake.

My project is wpa-supplicant. about 3w tags or more.
my map is nnoremap <leader>u :<C-u>Unite tag -start-insert file_rec/async:!<CR>
When type \u. and **scan in unite.
The tag population is to slow.
In 5 Mins, It calcs only 30000+ tags....

After executing :Unite -start-insert tag:%, the buffer is empty.

But :Unite -start-insert tag works well.

ctrlp.vim also works well for buffer tags.

This problem occurs in my php files, all plugins are the latest versions.

It doesn't work for javascript files.

I installed tagbar and jsctags, which works fine for generating tags for javascript files.

But I can't get any tags for javascript files when use unite-tag, I tried :Unite tag/include, but just an empty window is opened without any tag.

unite-tag.txt in section customize suggest to replace C-] for browsing sources. I like this better than the old behavior, but the suggestion introduces an annoying bug with the vim jumplist. When pressing C-o you don't jump to last visited position where you used C-] but to the previous buffer where C-] was used.

My suggestion is to use m' before the colon to record the position in the jumplist where C-] has been used.

Hi,

I use

call unite#filters#matcher_default#use(['matcher_fuzzy'])

to set default matcher, but unite-tag does not use the default matcher. This is easily worked around with

call unite#custom#source('tag', 'matchers', 'matcher_fuzzy')

I would expect that matcher_default should be used?

I have the same tag repeated multiple times, and this plugin only shows one of occurences.

load_one_ballot hardware/abstract_scanner.py    /^    def load_one_ballot(self, ballot):$/;"    m   class:AbstractScanner
load_one_ballot hardware/mock_scanner.py    /^    def load_one_ballot(self, ballot):$/;"    m   class:MockEpson
load_one_ballot hardware/network_scanner.py /^    def load_one_ballot(self, ballot):$/;"    m   class:NetworkScanner

When I run :Unite tag and filter for load_one_ballot I only get the first occurence. Would it be possible to show all of the occurences ? Thanks

When using :Unite tag, the path/filename part is basically useless because most of it is truncated and the rest of the space os occupied by the users home folder. Basically this is what all my matches look like:

@/Users/..spfront.py

That /Users/ part is useless information, I know what folder/project I'm in. So I would like an option to have the same behaviour as when using tag/include which uses fnamemodify(path, ':t') instead of filename here.

Perhaps it would be good to have two options, one for the relative path (fnamemodify(path, ':.')) and one for just the filename (fnamemodify(path, ':t')).

Basically something like this:

Current behaviour:

g:unite_source_tag_show_fname_only = 1

g:unite_source_tag_show_relative_fname = 1

Is it possible to sort tags with first priority for tags from the current file?

When I open the Unite window with the tag/include source, if the source has not been used yet or neocomplete has not been invoked in the current buffer, the tag list is empty. Forcing a unite_redraw does not cause the results to show up but they will show if I close and reopen the Unite window.

こういう要望がunite.vim本体にあったので、そろそろキャッシュをfile_rec/asyncのようにファイルに書き出す機能があっても良いのではないかと思いました。

Shougo/unite.vim#293

vital.vimにキャッシュ機能が実装されているので、unite.vimのvitalを使えばキャッシュ機能をそのまま利用できます。

私がやってもいいですが、なかなか暇が…。

The last one is over three years old.

Best

Matthias