Bug#407205: Patch available, please test

[Attilio Fiandrotti <attilio.fiandrotti@gmail.com>]

Frans Pop wrote:
> Thanks Attilio.
>
> The patch seems to work fine. I have seen no regressions while testing it
> and the storage of selections in the debconf database seems good.

great!

> On Friday 19 January 2007 15:53, Attilio Fiandrotti wrote:
>
> > Moreover, the SELECT single handler no longer lets the user to unselect
> > all options toghether: one, at least, will be always selected.
>
>
> What exactly do you mean by this? Are you talking about multiselect
> questions? I tried with the "extra locales" question in localechooser
> (after modifying the script a little) and I could unset all selections
> without problem.

No, i'm talking about SELECT only: now there is always a row selected, 
so when the user presses <Go> you always end with "xxx" as value for a 
given question.
Previously the user was allowed to deselected all rows, hence you could 
end with "" (empty string) in the database, and i believe this latter 
behaviour is not correct.
Of course, for MULTISELECT options the user can still uncheck all 
checkboxes.

> About the patch itself. I cannot really judge the code, but I also don't
> see any obvious errors.
>
> There are a few places where the indentation is not correct:
> +    tindex = malloc(sizeof(int) * count);
> +       indices = q_get_indices(q);
>      ^^^
> [...]
> +        if((result != NULL) && bool_var==1)
> +        {
> +        printf ("Option %d active\n", i);
>         ^^^
> +            copy = g_strdup(result);
> [...]
> +       /* by default the first row gets selected if no default option is specified */
>      ^^^
> +    gtk_tree_model_get_iter_first (model,&iter);
> +    path = gtk_tree_model_get_path (model, &iter);

Alright, tomorrow i'll do some makeup to the patch before committing :)

Attilio