Re: col() in lines with apostrophe characters

[Ingo Karkat]

On 15-Mar-2010 18:57, ZyX wrote:

Ответ на сообщение <<Re: col() in lines with apostrophe characters>>,
присланное в 20:15:39 14 марта 2010, Воскресенье,
отправитель Ingo Karkat:

      let len = strlen(substitute(str, ".", "x", "g"))

Never use this. It is three orders of magnitude slower than
     let len=len(split(str, '\zs'))

You can test this by launching
     vim -u NONE -c 'source test.vim'
My results are located in <<strprofile>> file.

Note that both methods assume that one symbol is symbol+combining diacritics (if
any).

-- regards, ingo

--
You received this message from the "vim_dev" maillist.
Do not top-post! Type your reply below the text you are replying to.
For more information, visit http://www.vim.org/maillist.php

FUNCTION  Test1()
Called 1 time
Total time:   0.840012
 Self time:   0.840012

count  total (s)   self (s)
    1              0.840012     return len(substitute(a:str, '.', '.', 'g'))

FUNCTION  Test2()
Called 1 time
Total time:   0.080001
 Self time:   0.080001

count  total (s)   self (s)
    1              0.080001     return len(split(a:str, '\zs'))

FUNCTIONS SORTED ON TOTAL TIME
count  total (s)   self (s)  function
    1   0.840012             Test1()
    1   0.080001             Test2()

FUNCTIONS SORTED ON SELF TIME
count  total (s)   self (s)  function
    1              0.840012  Test1()
    1              0.080001  Test2()


FUNCTION  Test1()
Called 1 time
Total time:   2.970045
 Self time:   2.970045

count  total (s)   self (s)
30001              0.060002 for i in s:range
30000              2.770042     let s:count += len(substitute(a:str, '.', '.', 
'g'))
30000              0.080000 endfor
    1              0.000000 call garbagecollect()

FUNCTION  Test2()
Called 1 time
Total time:   3.790057
 Self time:   3.790057

count  total (s)   self (s)
30001              0.070002 for i in s:range
30000              3.490053     let s:count += len(split(a:str, '\zs'))
30000              0.140002 endfor
    1              0.000000 call garbagecollect()

FUNCTIONS SORTED ON TOTAL TIME
count  total (s)   self (s)  function
    1   3.790057             Test2()
    1   2.970045             Test1()

FUNCTIONS SORTED ON SELF TIME
count  total (s)   self (s)  function
    1              3.790057  Test2()
    1              2.970045  Test1()

Thank you for this valuable insight. I had taken my substitution straight from Vim's :help strlen(). When re-producing your profiling on my (Linux x86, Vim 7.2.368) system, the difference is one order of magnitude (0.84 vs 0.08), still substantial. Your test processes a very long string once. When I changed this to processing a short string many times over (test2.vim, attached; an arguably more real-life scenario), the substitution eventually is faster than the split. I intended to ask Bram to update the documentation with your improved algorithm. With my inconclusive results, I'm not so sure any more.

test2.vim
Description: Binary data