Wrapped in gaffa

Programming, Art, Music, Philosophy, Politics

Posts Tagged ‘Software

Trash on portable devices

leave a comment »

Experienced users know that Ubuntu (Gnome) creates a trash folder on removable devices so they use Shift+Delete to permanently delete data. Rookie users don’t care that there is a folder and empties the trash can.

There is a bug report in Ubuntu that says Shouldn’t put .Trash-$USER on removable devices and there is a bugreport in Nautilus that says Ask to empty when unmounting media with items in trash. I think both titles are a sign that this is not a bug, but a matter of intuitive behaviour.

If we chose to have a concept of trash cans that is what we should stick with. Inconsistent behaviour is not user friendly. That said it might be convenient to have an option to always permanently delete files or automatically empty the trash can before unmount on certain devices.

I came across an actual bug when I was looking at the default behaviour. It seems like the trash applet is not updating when I delete something on a portable device. That should be be fixed.

Trash is rediscovered when the device is remounted, which is more than I expected when I read the bug reports.

Inconvenience #1
I can not empty individual trash folders. I have to empty all my trash.
Solution: Is obvious isn’t it?

Inconvenience #2
The folder is left on the device even when empty. If many different users use the same device. It would end up looking like this:
This happens on a FAT formatted device which lack a concept of users.
I could also imagine that some devices can’t handle directories prefixed with “.” and that is another good reason to delete the trash folder on portable devices when it is empty.
Solution: Trash folder on portable devices should be deleted when empty.

Feature #1
Option to empty trash before unmount with Yes/No dialog.

I don’t like permanent deletion as an option, when you can always delete with Shift+Delete or just enable the option in Nautilus and use the Right-click menu.

What is your thoughts?


Written by gaffa

2007-01-22 at 00:48

Posted in Gnome

Tagged with , , ,

GtkNotebook Ramblings

The idea of tabs can be thought of as a stack of bricks. If you count the bricks from the bottom to the top and pull a brick out. The index of the new brick will be the same as the old. Top of the stack excluded. Notebook tabs should be the same and almost is in GtkNotebook.

Think of tabs in a notebook as a double linked-list namely GList.
typedef struct {
gpointer data;
GList *next;
GList *prev;
} GList;

Where data is a pointer to a GtkNotebookPage. When a page is removed you just update the next/prev pointers in the tabs around it – and we have a nice stack of bricks.
struct _GtkNotebookPage
GtkWidget *child;
GtkWidget *tab_label;
GtkWidget *menu_label;
GtkWidget *last_focus_child; /* Last descendant of the page that had focus */
guint default_menu : 1; /* If true, we create the menu label ourself */
guint default_tab : 1; /* If true, we create the tab label ourself */
guint expand : 1;
guint fill : 1;
guint pack : 1;
guint reorderable : 1;
guint detachable : 1;
GtkRequisition requisition;
GtkAllocation allocation;
gulong mnemonic_activate_signal;
gulong notify_visible_handler;

The programmer using the API knows the tabs by index and calls the function gtk_notebook_remove_page which translate the index to a list element and passes it on to gtk_container_remove which emits a signal “remove”. The remove signal is caught by gtk_notebook_remove which calls gtk_notebook_real_remove

gtk_notebook_real_remove is called with a GList* list (a double linked-list of GtkNotebookPages) and a GtkNotebook* notebook.

This snippet find the new page in the GList. It gets the previous page unless the current page is the first tab.
next_list = gtk_notebook_search_page (notebook, list, STEP_PREV, TRUE);
if (!next_list)
next_list = gtk_notebook_search_page (notebook, list, STEP_NEXT, TRUE);

If we delete tab1 this would happen.

tab deleted old
Why not select Tab 1? Wouldn’t that be the natural thing to do if tabs are left aligned?

tab deleted new

It can be done by simply switching the order of STEP_PREV and STEP_NEXT
next_list = gtk_notebook_search_page (notebook, list, STEP_NEXT, TRUE);
if (!next_list)
next_list = gtk_notebook_search_page (notebook, list, STEP_PREV, TRUE);

Additionally #131920 causes the switch_page signal to report wrong index numbers as the signal is generated before the deleted page element is unlinked from the GList.

This can be fixed by simply calling g_list_remove_link as soon as we are done with the next/prev pointers of the list element and before signals are being sent. The element can still be used as usual until it is freed later.

I’m restriced in which file formats I can upload so here is how the patch looks like (against 2.10.6):
--- gtk+-2.10.6/gtk/gtknotebook.c 2006-09-10 08:33:15.000000000 +0200
+++ gtknotebook.c 2007-01-20 22:16:17.000000000 +0100
@@ -4148,9 +4148,11 @@
priv = GTK_NOTEBOOK_GET_PRIVATE (notebook);
destroying = GTK_OBJECT_FLAGS (notebook) & GTK_IN_DESTRUCTION;
- next_list = gtk_notebook_search_page (notebook, list, STEP_PREV, TRUE);
+ next_list = gtk_notebook_search_page (notebook, list, STEP_NEXT, TRUE);
if (!next_list)
- next_list = gtk_notebook_search_page (notebook, list, STEP_NEXT, TRUE);
+ next_list = gtk_notebook_search_page (notebook, list, STEP_PREV, TRUE);
+ notebook->children = g_list_remove_link (notebook->children, list);
if (notebook->cur_page == list->data)
@@ -4187,7 +4189,6 @@
if (!page->default_menu)
g_object_unref (page->menu_label);
- notebook->children = g_list_remove_link (notebook->children, list);
g_list_free (list);
if (page->last_focus_child)

Seems like my patch has been comitted in Gtk+ 2.13.3:

2008-06-08 Björn Lindqvist

Bug 131920 – gtkNotebook sends incorrect switch_page value

* gtk/gtknotebook.c (gtk_notebook_real_remove): Make switch-page
signal send correct page index when page index 0 is active and
removed. (#131920, Samuel Fogh)

Written by gaffa

2007-01-21 at 19:24

Posted in Gtk

Tagged with , , , ,