Re: Unexplainable crashes with GtkTreeModelSort

[Tim Müller <zen18864 zen co uk>]

On Monday 19 January 2004 02:13, Dimitar Haralanov wrote:

>       I think that I might have found where the problem stems from...
> According to the documentation, gtk_tree_path_next advances the path
> pointer to the next path. However, it never checks whether it is valid
> or even exists. Therefore, when I use gtk_tree_model_get_iter
> with the invalid path, I get a crash.
>
>       Considering that this is not the behaviour of gtk_tree_path_prev, I
> consider this to be a major bug in GTK.
>
>       If this is not a bug but an intended feature, please, someone tell me
> how to guard against using invalid paths...

Check the return value of gtk_tree_model_get_iter(). If it is TRUE, the path 
was valid, if it is FALSE, the path doesn't exist.

Or use iters from the start and then gtk_tree_model_iter_next() (with checking 
the return value).

gtk_tree_path_next() and _prev() cannot check whether a tree path is valid, 
because they are not bound to a specific model. gtk_tree_path_prev() also 
doesn't check whether a row really exists (you can construct an arbitrary new 
tree path and then do _prev() and still get an invalid tree path), it just 
returns FALSE if it's at the first node at the current depth already.

Cheers
-Tim