Re: [Nautilus-list] Patch for Sawfish-workaround

[Darin Adler <darin bentspoon com>]

On 2/22/02 8:31 AM, "Gaute Lindkvist" <lindkvis stud ntnu no> wrote:

In the future, please submit patches made with "diff -u" -- "unified diff"
format.

A few comments:

I'd prefer that this code go into its own function rather than just adding
the code into an existing function.

>     gint format;

Don't use gint in eel or nautilus code.

>     /* If the define is set, we activate the workaround */
>     if (WORKAROUND_FOR_SAWFISH_TRANSIENT_BUG) {

This should be an #ifdef or an #if, not an if statement.

>                 foundatom = (Atom) ((long *) data)[0];

This should be guint32, not long, right?

>                     gtk_window_set_transient_for (GTK_WINDOW (dialog), NULL);

It seems a bad hack to be doing this, rather than not marking the window
transient in the first place.

Also, is this the right place for this check? Are all the dialogs in
question using this function? This is the most serious issue for me. I don't
want to put this workaround here and miss some dialogs that don't happen to
use this function.

An alternative approach would be to make an eel version of
gtk_window_set_transient_for that does nothing if the parent window is the
desktop, and then change Nautilus to use that.

But perhaps we can just require a newer Sawfish and avoid this altogether?

> I added a define, that can turn on or off the workaround. Perhaps
> this should be a compile-time option, but I have very little
> experience with Makefiles.

Making it a compile-time option doesn't make sense to me.

    -- Darin