[ruby-dev:34758] Re: [ruby-cvs:23717] Ruby:r16477 (trunk): * regparse.c (PINC): use optimized enclen() instead of

　ささだです．

　この変更から，開発用マシン（x86 linux on vmware）で

('A'..'Z').each{|w|
  eval("p(/#{w}/ =~ '#{w}')")
}


が 全部 nil を返すようになってしまいました（文字全般がヒットしな
い？）．しかし，別の amd64 マシンでは問題なく動きます．

　例えば，変数の初期化などかな，と思うのですが，何か心当たりはありま
せんか？

[e-mail] wrote:
> matz	2008-05-19 17:25:03 +0900 (Mon, 19 May 2008)
> 
>   New Revision: 16477
> 
>   Modified files:
>     trunk/ChangeLog
>     trunk/common.mk
>     trunk/encoding.c
>     trunk/include/ruby/encoding.h
>     trunk/re.c
>     trunk/regenc.h
>     trunk/regexec.c
>     trunk/regparse.c
>     trunk/string.c
> 
>   Log:
>     * regparse.c (PINC): use optimized enclen() instead of
>       ONIGENC_MBC_ENC_LEN().
>     
>     * regparse.c (PFETCH): ditto.
>     
>     * regparse.c (PFETCH): small optimization.
>     
>     * regexec.c (slow_search): single byte encoding optimization.
>     
>     * regenc.h (enclen): avoid calling function when encoding's
>       min_len == max_len.
>     
>     * re.c (rb_reg_regsub): rb_enc_ascget() optimization for single
>       byte encoding.
>     
>     * re.c (rb_reg_search): avoid allocating new re_registers if we
>       already have MatchData.
>     
>     * re.c (match_init_copy): avoid unnecessary onig_region_free()
>       before onig_region_copy. 
>     
>     * encoding.c (rb_enc_get_index): remove implicit enc_capable check
>       each time.
>     
>     * encoding.c (rb_enc_set_index): ditto.
>     
>     * encoding.c (enc_compatible_p): small refactoring.
>     
>     * include/ruby/encoding.h (rb_enc_dummy_p): inline
>       rb_enc_dummy_p() and export related code.

-- 
// SASADA Koichi at atdot dot net