Still improving! (Was: Re: [PATCH] Avoid strtok for thread safety)

[Martin Sebor]

Paolo Carlini wrote:
...
> Nice, I didn't consider that... However, in the meanwhile, I tried to post
> (without success, due to a problem with my smtp server :( a different 
> version
> (tested x86-linux), which basically avoid half the finds thanks to the a
> priori knowldege of the names of the various categories. Something like 
> this:
> 
>      {
>        char* __new;
>        const char* __end;
>        const char* __beg = __s + strlen(_S_categories[0]) + 1;
>        for (size_t __i = 0;
>             __i < _S_categories_size + _S_extra_categories_size - 1; ++__i)
>          {
>            __end = strpbrk(__beg, ";");
>            __new = new char[__end - __beg + 1];
>            memcpy(__new, __beg, __end - __beg);
>            __new[__end - __beg] = '\0';
>            _M_names[__i] = __new;
>            __beg = __end + strlen(_S_categories[__i + 1]) + 2;
>          }
>        __new = new char[__s + __len - __beg];
>        memcpy(__new, __beg, __s + __len - __beg);
>        _M_names[_S_categories_size + _S_extra_categories_size - 1] = __new;
>      }

Sorry for butting in, but the code above doesn't seem exception
safe :) (I.e., it will leak if any of the new expressions except
the first one throws).

Regards
Martin