Consider this one. I will write a \t as '[tab]', a newline as a '\n' and a space as a dot, '.', to illustrate this bug in a text-only field like this bugzilla's.

This is how Anjuta does auto-indentation right now:

static void\n
my_calling_funcion_wrong (void);\n
[tab]MyItem1 *item1;\n
[tab]MyItem2 *item2;\n
[tab]MyItem3 *item3;\n
\n
[tab]my_long_funcion (item1,\n
[tab][tab][tab][tab]..item2, \n
[tab][tab][tab][tab]..item3);\n
}\n

This is wrong, it should instead do this:

static void\n
my_calling_funcion_right (void);\n
[tab]MyItem1 *item1;\n
[tab]MyItem2 *item2;\n
[tab]MyItem3 *item3;\n
\n
[tab]my_long_funcion (item1, \n
[tab].................item2, \n
[tab].................item3);\n
}\n

The reason for that is compatibility with other people's editor settings for the tab size. For example with _wrong if the original author wrote it with 8-spaces per tab in mind, then another author who has his editor set to use 4-spaces per tab will often see the function as mis-indented. This wouldn't have happened if it was indented the way _right is indented.

I'm not asking to replace all tabs with spaces. That's for strange people who don't get the clue. 

Let's see another example:

static void\n
my_calling_funcion_right (void);\n
[tab]MyItem1 *item1;\n
[tab]MyItem2 *item2;\n
[tab]MyItem3 *item3;\n
\n
[tab]while (TRUE) {\n
[tab][tab]while (TRUE) {\n
[tab][tab][tab]my_long_funcion (item1, \n
[tab][tab][tab].................item2, \n
[tab][tab][tab].................item3); \n
[tab][tab]}\n
[tab]}\n
}\n

Note how easy it becomes to move indentation from more tabs to less tabs without having to worry about 8-space-sized tabs vs. 4-space-sized tabs.