Re: [AD] New mode selector, or get_gfx_mode_list(), broken [djgpp, possibly others]

["Henrik Stokseth" <hstokset@xxxxxxxxxx>]

Peter Wang <tjaden@xxxxxxxxxx> wrote:

> I just don't like it returning through a global.  It should be like
> the bitmap interface.
>
>    GFX_MODE_LIST *list;
>
>    list = get_gfx_mode_list(GFX_FOO);
>    if (list) {
>       /* work with list */
>       destroy_gfx_mode_list(list);
>    }
>
> If it is really necessary to know if the call failed or not
> supported, an optional parameter:
>
>    list = get_gfx_mode_list(GFX_FOO, &error);

the failed vs. supported will be useful on for example the VBE/AF driver,
where fetch_mode_list will be supported and  but the lowlevel FreeBe/AF
driver cannot be loaded from disk.

if you think it matters i could rewrite it to return a pointer to a a
GFX_MODE_LIST struct like this:

---

typedef struct GFX_MODE {
   int width, height, bpp;
}

typedef struct GFX_MODE_LIST {
   int error; /* 0 = ok, negative for errors */
   int modes; /* number of gfx modes */
   int malloced; /* malloc'ed or hardcoded? */
   GFX_MODE *mode; /* pointer to the actual mode list array */
} GFX_MODE_LIST;

GFX_MODE_LIST *get_gfx_mode_list(int card);
int destroy_gfx_mode_list(GFX_MODE_LIST *mode_list);

and remove the global variables.

---

> I also think an array is the right choice here, rather than a linked
> list, but I'm not sure why.

i am sure why. it's simpler and more error-safe to use an array than a
linked list. tons of pointers only has it with giving the programmer
debugging headaches. the modeselector already seems to have a bug somewhere
and that's my #1 priority. perhaps the proposed rewrite will help though?
:o)

Sincerely Henrik Stokseth.
-----------------------------------------------------------------------
E-mail: hstokset@xxxxxxxxxx  Homepage: http://hstokset.n3.net