Re: [gnome-love] struct craziness

[Jose Marino <marinoj nso edu>]

Hello,

I believe this is the offending code at the top of function 
'option_parsing_results_init':


<code>

static OptionParsingResults *
option_parsing_results_init (OptionParsingResults *results, int *argc, 
char **argv)
 {
   int i;

   results = g_new0 (OptionParsingResults, 1);
   results->screen_number = -1;
   results->default_working_dir = NULL;
   results->zoom = "1.0";

</code>

You're allocating a new 'OptionParsingResults' and storing the new 
pointer in 'results' which basically overwrites the pointer that was 
passed to the function.

Thus, in main:

<code>
  results = &parsing_results;

  g_assert (initialization_complete);

  results = option_parsing_results_init (results, &argc, argv);
</code>

After the call to 'option_parsing_results_init' your pointer 'results' 
points to a different place.

Regards,
Jose


Christian Kirbach wrote:
> Hi,
>
> I am hacking on gnome-terminal to switch from popt to GOption command line
> option parsing. I have a problem adressing a struct by means of a pointer
> to that struct. Consider
>
> =============================================
> typedef struct
> {
>  gboolean option_fullscreen;
> /* stuff */
> }
>   OptionParsingResults;
>
> OptionParsingResults parsing_results;
>
> int main (int argc, char **argv)
> {
>
>   OptionParsingResults *results = &parsing_results;
> printf("%o %o\n", *results, parsing_results);
> }
> =============================================
>
> In the main function I create a pointer to the struct in the main scope.
> The printf shows that both dereference the same memory spot (equal
> numbers).
> At some point in main() the option parsing takes places. The
> parser writes its results directly into the 'struct parsing_results' like
>
>  &parsing_results.option_fullscreen = foo
>
> However when trying to access the results via the pointer 'results'
> I read different values in contrast to reading the outcome using 
> parsing_results,
> i.e.
>
>  parsing_results.option_fullscreen != results->option_fullscreen
>
> I have no explanation for this. Can someone shed some light on this, 
> please?
>
> I've attached the current state of my patch for gnome-terminal. You can 
> try this patch and run e.g.
>  gnome-terminal --zoom=5
>
> I've added some printf to show the difference.
>
>
> Regards
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> gnome-love mailing list
> gnome-love gnome org
> http://mail.gnome.org/mailman/listinfo/gnome-love