summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorRan Benita <ran234@gmail.com>2012-11-04 14:07:30 (GMT)
committerRan Benita <ran234@gmail.com>2012-11-04 14:14:47 (GMT)
commitfb201645b2bae1f4a1f7a6dfa37e14a0de04556e (patch)
treed9c5cd02e3c1946f27d8b3a211a6442a17e2983e
parent998c957a3c7c861093f3f71f34989c258ed3073b (diff)
Add some explanations on consumed modifiers
This should hopefully clarify this somewhat subtle point to the uninitiated users. Signed-off-by: Ran Benita <ran234@gmail.com>
-rw-r--r--xkbcommon/xkbcommon.h45
1 files changed, 43 insertions, 2 deletions
diff --git a/xkbcommon/xkbcommon.h b/xkbcommon/xkbcommon.h
index 65c4dcc..f8b3d8b 100644
--- a/xkbcommon/xkbcommon.h
+++ b/xkbcommon/xkbcommon.h
@@ -1331,12 +1331,53 @@ xkb_state_mod_indices_are_active(struct xkb_state *state,
* active, so it is not consumed by this translation.
*
* It may be desireable for some application to not reuse consumed modifiers
- * for further processing, e.g. for hotkeys or keyboard shortcuts.
+ * for further processing, e.g. for hotkeys or keyboard shortcuts. To
+ * understand why, consider some requirements from a standard shortcut
+ * mechanism, and how they are implemented:
+ *
+ * 1. The shortcut's modifiers must match exactly to the state. For example,
+ * it is possible to bind separate actions to \<Alt\>\<Tab\> and to
+ * \<Alt\>\<Shift\>\<Tab\>. Further, if only \<Alt\>\<Tab\> is bound to
+ * an action, pressing \<Alt\>\<Shift\>\<Tab\> should not trigger the
+ * shortcut.
+ * Effectively, this means that the modifiers are compared using the
+ * equality operator (==).
+ * 2. Only relevant modifiers are considered for the matching. For example,
+ * Caps Lock and Num Lock should not generally affect the matching, e.g.
+ * when matching \<Alt\>\<Tab\> against the state, it does not matter
+ * whether Num Lock is active or not. These relevant, or significant,
+ * modifiers usually include Alt, Control, Shift, Super and similar.
+ * Effectively, this means that non-significant modifiers are masked out,
+ * before doing the comparison as described above.
+ * 3. The matching must be independent of the layout/keymap. For example,
+ * the \<Plus\> (+) symbol is found on the first level on some layouts,
+ * and requires holding Shift on others. If you simply bind the action
+ * to the \<Plus\> keysym, it would work for the unshifted kind, but
+ * not for the others, because the match against Shift would fail. If
+ * you bind the action to \<Shift\>\<Plus\>, only the shifted kind would
+ * work. So what is needed is to recognize that Shift is used up in the
+ * translation of the keysym itself, and therefore should not be included
+ * in the matching.
+ * Effectively, this means that consumed modifiers (Shift in this example)
+ * are masked out as well, before doing the comparison.
+ *
+ * To summarize, this is how the matching would be performed:
+ * @code
+ * (keysym == shortcut_keysym) &&
+ * ((state_modifiers & ~consumed_modifiers & significant_modifiers) == shortcut_modifiers)
+ * @endcode
+ *
+ * @c state_modifiers are the modifiers reported by
+ * xkb_state_mod_index_is_active() and similar functions.
+ * @c consumed_modifiers are the modifiers reported by
+ * xkb_state_mod_index_is_consumed().
+ * @c significant_modifiers are decided upon by the application/toolkit/user;
+ * it is up to them to decide whether these are configurable or hard-coded.
*
* @returns 1 if the modifier is consumed, 0 if it is not. If the modifier
* index is not valid in the keymap, returns -1.
*
- * @sa xkb_state_mod_mask_remove_consumend()
+ * @sa xkb_state_mod_mask_remove_consumed()
* @memberof xkb_state
*/
int