Re: Another patch, for discussion tho

[Bruce Korb]

On 04/16/12 09:27, Jim Meyering wrote:
> Bruce Korb wrote:
> > Attached are several files, one the patch, another a derived file based on
> > a file that is part of the patch, and the last is the tool used to derive
> > the header from the mapping file.
>
> I suppose the motivation is performance?
> If so, have you measured the improvement?

Actually, no, it is not for performance.  I found it wearisome to write:

   if (isalnum(x) || (x == '_') || (x == '.......

all over the place.  I saw what Zack W. was doing in CPP and decided there
had to be a way to specify the classifications, spit out a table and
not fret over it any more.  (I did a first incantation a decade ago.)

So, if there is a performance improvement, it would be amusing to know what
it might be.  I haven't measured tho.  If ctype.h completely satisfies your
need, I'm sure there is little to no improvement (other than for
strspn and strpbrk -- there is no reverse spanning or breaking).