Re: [PATCH] add 'string-distance' to calculate Levenshtein distance

[chen bin]

Hi, Eli,
I attached the new patch.

A software project usually contains nomal directory and file names.
Linux kernel source, for example.

My package `counsel-etags` will use byte compare version.

Regards,
Chen

On Thu, Apr 19, 2018 at 6:05 PM, Eli Zaretskii <address@hidden> wrote:
>> From: chen bin <address@hidden>
>> Date: Tue, 17 Apr 2018 22:31:20 +1000
>> Cc: address@hidden
>>
>> As you suggested, I re-write the code using 'FETCH_STRING_CHAR_ADVANCE'.
>
> Thanks.
>
>> I also implemented the byte comparing version. It's 4 times as fast. And I do
>> need use it to compare file path in my package 'counsel-etags'.
>
> File names are just strings for this purpose, and they can potentially
> include any non-zero characters.  So I don't see why they are special.
>
>> The fille path couldn't contain any funny characters (emoji). so
>> it'sperfectly fine
>> to use byte comparing version.
>
> File names can very well include emoji and other "funny" characters,
> Emacs supports that on all modern systems (including even MS-Windows).
>
>> diff --git a/etc/NEWS b/etc/NEWS
>> index 5aa92e2991..3cce2c48c7 100644
>> --- a/etc/NEWS
>> +++ b/etc/NEWS
>> @@ -490,6 +490,8 @@ x-lost-selection-hooks, x-sent-selection-hooks
>>  +++
>>  ** New function assoc-delete-all.
>>
>> +** New function string-distance to calculate Levenshtein distance between 
>> two strings.
>
> This long line should be filled using the fill-column setting we use
> in NEWS.  Even better, make the header a short summary, like
>
>   ** New function 'string-distance'
>
> and then describe its functionality in a separate sentence that starts
> immediately below that header.
>
>> +DEFUN ("string-distance", Fstring_distance, Sstring_distance, 2, 3, 0,
>> +       doc: /* Return Levenshtein distance between STRING1 and STRING2.
>> +If BYTECOMPARE is nil, we compare character of strings.
>> +If BYTECOMPARE is t, we compare byte of strings.
>
> Please lose the "we" part, it's inappropriate in documentation,
> because it describes what Emacs does.
>
>> +Comparing by byte is faster and non-ascii characters has weighted distance.
>
> I would delete this sentence, it is IMO confusing more than anything
> else.  (And I still think the bytewise comparison is not needed.)
>
>> +  bool use_bytecompare = !NILP(bytecompare);
>                                 ^^
> Space between these 2 characters.
>
>> +  else
>> +    {
>> +      int c1, c2;
>> +      ptrdiff_t i1, i1_byte, i2, i2_byte;
>> +      i2 = i2_byte = 0;
>> +      for (x = 1; x <= len2; x++)
>
> Please move the initialization of i2 and i2_byte into the for-loop
> initializer (suing the comma operator).
>
>> +          i1 = i1_byte = 0;
>> +          for (y = 1, lastdiag = x - 1; y <= len1; y++)
>
> Likewise here with i1 and i1_byte.
>
> Thanks.



-- 
help me, help you.

0001-add-api-string-distance.patch
Description: Text Data