Re: wcwidth replacement problems

[Alexander V. Lukyanov]

On Tue, Aug 26, 2008 at 09:32:32AM +0200, Bruno Haible wrote:
> Alexander V. Lukyanov wrote:
> > Let's measure it.
> > 
> > $ time ./wcwidth-solaris 
> > wcwidth(0x2022)=2
> > 
> > real    0m2.205s
> > user    0m2.200s
> > sys     0m0.000s
> > 
> > $ time ./wcwidth-rpl 
> > wcwidth(0x2022)=1
> > 
> > real    0m55.477s
> > user    0m55.350s
> > sys     0m0.000s
> > 
> > $ time ./wcwidth-mk 
> > wcwidth(0x2022)=1
> > 
> > real    0m1.944s
> > user    0m1.940s
> > sys     0m0.010s
> 
> This is not a fair comparison: wcwidth-mk works only in UTF-8 locales,
> whereas wcwidth() from the system and from gnulib return the right result
> in all locales. The test whether the locale encoding is UTF-8 is precisely
> what takes up most time in the gnulib replacement.

Ok. Would you accept this patch, which caches charset properties?
BTW, it also fixes a bug - cjk case could not ever be executed before.

$ time ./wcwidth-rpl1
wcwidth(0x2022)=1

real    0m4.160s
user    0m4.150s
sys     0m0.010s

Worse than Solaris native wcwidth (nl_langinfo+strncmp costs), but still 13
times faster than before.

-- 
   Alexander.

diff
Description: Text document