Why MatchTag is consume resources on filetype=vim (maybe others too)?

[FelikZ]

I just notices that when Im moving across my custom colorsheme source, there is a huge lags during movements via jk. So I decided to profile what is currently happening and found that MatchTag is trying to match something on current filetype... Checked twice :set ft? so its really equals to vim.

Here is some statistic from profiles:

FUNCTION  <SNR>85_Highlight_Matching_Pair()
Called 85 times
Total time:   0.014041
Self time:   0.014041

389 FUNCTIONS SORTED ON TOTAL TIME
390 count  total (s)   self (s)  function
391    85   0.014041             <SNR>85_Highlight_Matching_Pair()
392    88   0.012402             airline#check_mode()
393    88   0.007818             airline#extensions#whitespace#check()
394   440   0.004540             airline#util#wrap()
395    88   0.004066             airline#statusline()
396   176   0.003881             airline#util#append()
397    88   0.003260             airline#parts#ffenc()
398    85   0.001647             <SNR>97_sync_active_winnr()
399    88   0.000984             airline#parts#iminsert()
400    88   0.000622             airline#parts#paste()
401    88   0.000597             airline#parts#mode()
402    88   0.000511             airline#parts#readonly()
403    88   0.000448             airline#parts#filetype()
404     1   0.000146   0.000129  unite#handlers#_on_cursor_hold()
405     1   0.000051             <SNR>137_garbage_collect()
406     1   0.000017             <SNR>144_restore_statusline()

As you can see, airline in pair with MatchTag crap a little performance. How I can prevent MatchTag to load (e.g. use Highlight_Matching_Pair function)?

[gregsexton]

You want to completely disable MatchTag? Just remove it from your vim directory.

If you want to conditionally disable, you can remove the auto group matchhtmlparen: augroup! matchhtmlparen.

Highlight_Matching_Pair is only called when you move your cursor.

If I'm reading your output correctly, 0.014041 seconds is ~14ms. Divided by 85 is ~0.16 ms per keypress. I doubt that you can notice this lag. I suspect your problem lies elsewhere.