no-reply@allegro.cc (archlinka) — Wed, 07 Apr 2021 12:51:53 +0000

Hi, I was just looking to Allegro source code and foundsome strange situation. Or it's also possible that I just do not understand the code:

#SelectExpand
  1/* Function: al_ustr_find_set
*/
  3int al_ustr_find_set(const ALLEGRO_USTR *us, int start_pos,
 const ALLEGRO_USTR *accept)
  5{
 int rc;
 int32_t c, d;
 int pos;
 int set_pos;
 10
 /* Fast path for ASCII characters. */
 if (all_ascii(accept)) {
    rc = _al_binchr(us, start_pos, accept);
    return (rc == _AL_BSTR_ERR) ? -1 : rc;
 }
 16
 /* Non-ASCII. */
 pos = 0;
 while ((c = al_ustr_get(us, pos)) != -1) {
    if (c == -2) {
       /* Invalid byte sequence. */
       pos++;
       continue;
    }
 25
    set_pos = 0;
    while ((d = al_ustr_get_next(accept, &set_pos)) != -1) {
       if (c == d)
          return pos;
    }
 31
    pos += al_utf8_width(c);
 }
 34
 return -1;
 36}

It should find first occurrence of code point starting at pos start_pos. The thing I do not understand is that the argument start_pos is used only in one place in the function and that's the ASCII only branch. But not in the non ASCII. Is it bug or just there is something I do not see?

no-reply@allegro.cc (amarillion) — Wed, 07 Apr 2021 16:27:02 +0000

I think you're right, this looks like a bug.

no-reply@allegro.cc (Peter Hull) — Wed, 07 Apr 2021 18:04:29 +0000

I agree, if you've got time please could you create an issue on the github?
https://github.com/liballeg/allegro5/issues

Thanks.
(p.s. looks like it would be easily fixed, too!)

Possible bug in allegro utf8 function