qmk_firmware/docs/tap_hold.md

29 KiB

Tap-Hold Configuration Options

While Tap-Hold options are fantastic, they are not without their issues. We have tried to configure them with reasonable defaults, but that may still cause issues for some people.

These options let you modify the behavior of the Tap-Hold keys.

Tapping Term

The crux of all of the following features is the tapping term setting. This determines what is a tap and what is a hold. The exact timing for this to feel natural can vary from keyboard to keyboard, from switch to switch, and from key to key.

?> DYNAMIC_TAPPING_TERM_ENABLE enables three special keys that can help you quickly find a comfortable tapping term for you. See "Dynamic Tapping Term" for more details.

You can set the global time for this by adding the following setting to your config.h:

#define TAPPING_TERM 200

This setting is defined in milliseconds and defaults to 200ms. This is a good average for the majority of people.

For more granular control of this feature, you can add the following to your config.h:

#define TAPPING_TERM_PER_KEY

You can then add the following function to your keymap:

uint16_t get_tapping_term(uint16_t keycode, keyrecord_t *record) {
    switch (keycode) {
        case SFT_T(KC_SPC):
            return TAPPING_TERM + 1250;
        case LT(1, KC_GRV):
            return 130;
        default:
            return TAPPING_TERM;
    }
}

Dynamic Tapping Term :id=dynamic-tapping-term

DYNAMIC_TAPPING_TERM_ENABLE is a feature you can enable in rules.mk that lets you use three special keys in your keymap to configure the tapping term on the fly.

Key Aliases Description
QK_DYNAMIC_TAPPING_TERM_PRINT DT_PRNT Types the current tapping term, in milliseconds
QK_DYNAMIC_TAPPING_TERM_UP DT_UP Increases the current tapping term by DYNAMIC_TAPPING_TERM_INCREMENTms (5ms by default)
QK_DYNAMIC_TAPPING_TERM_DOWN DT_DOWN Decreases the current tapping term by DYNAMIC_TAPPING_TERM_INCREMENTms (5ms by default)

Set the tapping term as usual with #define TAPPING_TERM <value> in config.h and add DYNAMIC_TAPPING_TERM_ENABLE = yes in rules.mk. Then, place the above three keys somewhere in your keymap and flash the new firmware onto your board.

Now, you can try using your dual-role keys, such as layer-taps and mod-taps, and use DT_DOWN and DT_UP to adjust the tapping term immediately. If you find that you frequently trigger the modifier of your mod-tap(s) by accident, for example, that's a sign that your tapping term may be too low so tap DT_UP a few times to increase the tapping term until that no longer happens. On the flip side, if you get superfluous characters when you actually intended to momentarily activate a layer, tap DT_DOWN to lower the tapping term. Do note that these keys affect the global tapping term, you cannot change the tapping term of a specific key on the fly.

Once you're satisfied with the current tapping term value, open config.h and replace whatever value you first wrote for the tapping term by the output of the DT_PRNT key.

It's important to update TAPPING_TERM with the new value because the adjustments made using DT_UP and DT_DOWN are not persistent.

The value by which the tapping term increases or decreases when you tap DT_UP and DT_DOWN can be configured in config.h with #define DYNAMIC_TAPPING_TERM_INCREMENT <new value>. Note that the tapping term is not modified when holding down the tap term keys so if you need to, for example, decrease the current tapping term by 50ms, you cannot just press down and hold DT_DOWN; you will have to tap it 10 times in a row with the default increment of 5ms.

If you need more flexibility, nothing prevents you from defining your own custom keys to dynamically change the tapping term.

enum custom_dynamic_tapping_term_keys = {
    DT_UP_50 = SAFE_RANGE,
    DT_DOWN_50,
    DT_UP_X2,
    DT_DOWN_X2,
}

bool process_record_user(uint16_t keycode, keyrecord_t *record) {
    switch (keycode) {
    case DT_UP_50:
        if (record->event.pressed) {
            g_tapping_term += 50;
        }
        break;
    case DT_DOWN_50:
        if (record->event.pressed) {
            g_tapping_term -= 50;
        }
        break;
    case DT_UP_X2:
        if (record->event.pressed) {
            g_tapping_term *= 2;
        }
        break;
    case DT_DOWN_X2:
        if (record->event.pressed) {
            g_tapping_term /= 2;
        }
        break;
    }
    return true;
};

In order for this feature to be effective if you use per-key tapping terms, you need to make a few changes to the syntax of the get_tapping_term function. All you need to do is replace every occurrence of TAPPING_TERM in the get_tapping_term function by lowercase g_tapping_term. If you don't do that, you will still see the value typed by DT_PRNT go up and down as you configure the tapping term on the fly but you won't feel those changes as they don't get applied. If you can go as low as 10ms and still easily trigger the tap function of a dual-role key, that's a sign that you forgot to make the necessary changes to your get_tapping_term function.

For instance, here's how the example get_tapping_term shown earlier should look after the transformation:

uint16_t get_tapping_term(uint16_t keycode, keyrecord_t *record) {
    switch (keycode) {
        case SFT_T(KC_SPC):
           return g_tapping_term + 1250;
        case LT(1, KC_GRV):
            return 130;
        default:
            return g_tapping_term;
    }
}

The reason is that TAPPING_TERM is a macro that expands to a constant integer and thus cannot be changed at runtime whereas g_tapping_term is a variable whose value can be changed at runtime. If you want, you can temporarily enable DYNAMIC_TAPPING_TERM_ENABLE to find a suitable tapping term value and then disable that feature and revert back to using the classic syntax for per-key tapping term settings. In case you need to access the tapping term from elsewhere in your code, you can use the GET_TAPPING_TERM(keycode, record) macro. This macro will expand to whatever is the appropriate access pattern given the current configuration.

Tap-Or-Hold Decision Modes

The code which decides between the tap and hold actions of dual-role keys supports three different modes, in increasing order of preference for the hold action:

  1. The default mode selects the hold action only if the dual-role key is held down longer than the tapping term. In this mode pressing other keys while the dual-role key is held down does not influence the tap-or-hold decision. In other words, this mode ignores interrupts.

  2. The “permissive hold” mode, in addition to the default behavior, immediately selects the hold action when another key is tapped (pressed and then released) while the dual-role key is held down, even if this happens earlier than the tapping term. If another key is just pressed, but then the dual-role key is released before that other key (and earlier than the tapping term), this mode will still select the tap action.

  3. The “hold on other key press” mode, in addition to the default behavior, immediately selects the hold action when another key is pressed while the dual-role key is held down, even if this happens earlier than the tapping term.

Note that until the tap-or-hold decision completes (which happens when either the dual-role key is released, or the tapping term has expired, or the extra condition for the selected decision mode is satisfied), key events are delayed and not transmitted to the host immediately. The default mode gives the most delay (if the dual-role key is held down, this mode always waits for the whole tapping term), and the other modes may give less delay when other keys are pressed, because the hold action may be selected earlier.

Comparison :id=comparison

To better illustrate the tap-or-hold decision modes, let us compare the expected output of each decision mode in a handful of tapping scenarios involving a mod-tap key (LSFT_T(KC_A)) and a regular key (KC_B) with the TAPPING_TERM set to 200ms.

By default, mod-taps behave like HOLD_ON_OTHER_KEY_PRESS, while layer-taps behave like "Ignore Interrupt" out of the box. If you want "Ignore Interrupt"-like behaviour for mod-taps, you must enable IGNORE_MOD_TAP_INTERRUPT, or return false in the get_hold_on_other_key_press function for all mod-taps.

Note: "kc held" in the "Physical key event" column means that the key wasn't physically released yet at this point in time.

Distinct taps (AABB) :id=distinct-taps

Time Physical key event Ignore Interrupt PERMISSIVE_HOLD HOLD_ON_OTHER_KEY_PRESS
0 LSFT_T(KC_A) down
199 LSFT_T(KC_A) up a a a
210 KC_B down ab ab ab
220 KC_B up ab ab ab
Time Physical key event Ignore Interrupt PERMISSIVE_HOLD HOLD_ON_OTHER_KEY_PRESS
0 LSFT_T(KC_A) down
200 LSFT_T(KC_A) held Shift Shift Shift
201 LSFT_T(KC_A) up Shift Shift Shift
205 KC_B down b b b
210 KC_B up b b b

Nested tap (ABBA) :id=nested-tap

Time Physical key event Ignore Interrupt PERMISSIVE_HOLD HOLD_ON_OTHER_KEY_PRESS
0 LSFT_T(KC_A) down
110 KC_B down B
120 KC_B up B B
199 LSFT_T(KC_A) up ab B B
Time Physical key event Ignore Interrupt PERMISSIVE_HOLD HOLD_ON_OTHER_KEY_PRESS
0 LSFT_T(KC_A) down
110 KC_B down B
120 KC_B up B B
200 LSFT_T(KC_A) held B B B
210 LSFT_T(KC_A) up B B B
Time Physical key event Ignore Interrupt PERMISSIVE_HOLD HOLD_ON_OTHER_KEY_PRESS
0 LSFT_T(KC_A) down
200 LSFT_T(KC_A) held Shift Shift Shift
205 KC_B down B B B
210 KC_B up B B B
220 LSFT_T(KC_A) up B B B

Rolling keys (ABAB) :id=rolling-keys

Time Physical key event Ignore Interrupt PERMISSIVE_HOLD HOLD_ON_OTHER_KEY_PRESS
0 LSFT_T(KC_A) down
110 KC_B down B
130 LSFT_T(KC_A) up ab ab B
140 KC_B up ab ab B
Time Physical key event Ignore Interrupt PERMISSIVE_HOLD HOLD_ON_OTHER_KEY_PRESS
0 LSFT_T(KC_A) down
110 KC_B down B
200 LSFT_T(KC_A) held B B B
205 LSFT_T(KC_A) up B B B
210 KC_B up B B B

Default Mode

Example sequence 1 (the L key is also mapped to KC_RGHT on layer 2):

             TAPPING_TERM
  +---------------|--------------------+
  | +-------------|-------+            |
  | | LT(2, KC_A) |       |            |
  | +-------------|-------+            |
  |               | +--------------+   |
  |               | | KC_L         |   |
  |               | +--------------+   |
  +---------------|--------------------+

The above sequence would send a KC_RGHT, since LT(2, KC_A) is held longer than the TAPPING_TERM.


Example sequence 2 (the L key is also mapped to KC_RGHT on layer 2):

                           TAPPING_TERM
  +-----------------------------|------+
  | +---------------+           |      |
  | | LT(2, KC_A)   |           |      |
  | +---------------+           |      |
  |            +--------------+ |      |
  |            | KC_L         | |      |
  |            +--------------+ |      |
  +-----------------------------|------+

The above sequence will not send KC_RGHT but KC_A KC_L instead, since LT(2, KC_A) is not held longer than the TAPPING_TERM.


Example sequence 3 (Mod Tap):

                         TAPPING_TERM
  +---------------------------|--------+
  | +-------------+           |        |
  | | SFT_T(KC_A) |           |        |
  | +-------------+           |        |
  |       +--------------+    |        |
  |       | KC_X         |    |        |
  |       +--------------+    |        |
  +---------------------------|--------+

Based on previous examples, you might have expected the output of the above sequence to be KC_A KC_X since SFT_T(KC_A) is NOT held longer than the TAPPING_TERM. However, the actual output would be capital X (SHIFT + x) due to reasons explained under Ignore Mod Tap Interrupt.

Permissive Hold

The “permissive hold” mode can be enabled for all dual-role keys by adding the corresponding option to config.h:

#define PERMISSIVE_HOLD

This makes tap and hold keys (like Layer Tap) work better for fast typists, or for high TAPPING_TERM settings.

If you press a dual-role key, tap another key (press and release) and then release the dual-role key, all within the tapping term, by default the dual-role key will perform its tap action. If the PERMISSIVE_HOLD option is enabled, the dual-role key will perform its hold action instead.

An example of a sequence that is affected by the “permissive hold” mode:

  • LT(2, KC_A) Down
  • KC_L Down (the L key is also mapped to KC_RGHT on layer 2)
  • KC_L Up
  • LT(2, KC_A) Up
                         TAPPING_TERM   
  +---------------------------|--------+
  | +----------------------+  |        |
  | | LT(2, KC_A)          |  |        |
  | +----------------------+  |        |
  |    +--------------+       |        |
  |    | KC_L         |       |        |
  |    +--------------+       |        |
  +---------------------------|--------+

Normally, if you do all this within the TAPPING_TERM (default: 200ms), this will be registered as al by the firmware and host system. With the PERMISSIVE_HOLD option enabled, the Layer Tap key is considered as a layer switch if another key is tapped, and the above sequence would be registered as KC_RGHT (the mapping of L on layer 2). We could describe this sequence as a “nested tap” (the modified key's key down and key up events are “nested” between the dual-role key's key down and key up events).

However, this slightly different sequence will not be affected by the “permissive hold” mode:

  • LT(2, KC_A) Down
  • KC_L Down (the L key is also mapped to KC_RGHT on layer 2)
  • LT(2, KC_A) Up
  • KC_L Up
                         TAPPING_TERM   
  +---------------------------|--------+
  | +-------------+           |        |
  | | LT(2, KC_A) |           |        |
  | +-------------+           |        |
  |       +--------------+    |        |
  |       | KC_L         |    |        |
  |       +--------------+    |        |
  +---------------------------|--------+

In the sequence above the dual-role key is released before the other key is released, and if that happens within the tapping term, the “permissive hold” mode will still choose the tap action for the dual-role key, and the sequence will be registered as al by the host. We could describe this as a “rolling press” (the two keys' key down and key up events behave as if you were rolling a ball across the two keys, first pressing each key down in sequence and then releasing them in the same order).

?> The PERMISSIVE_HOLD option is not noticeable if you also enable HOLD_ON_OTHER_KEY_PRESS because the latter option considers both the “nested tap” and “rolling press” sequences like shown above as a hold action, not the tap action. HOLD_ON_OTHER_KEY_PRESS makes the Tap-Or-Hold decision earlier in the chain of key events, thus taking a precedence over PERMISSIVE_HOLD. This remark also applies to default mod-taps.

For more granular control of this feature, you can add the following to your config.h:

#define PERMISSIVE_HOLD_PER_KEY

You can then add the following function to your keymap:

bool get_permissive_hold(uint16_t keycode, keyrecord_t *record) {
    switch (keycode) {
        case LT(1, KC_BSPC):
            // Immediately select the hold action when another key is tapped.
            return true;
        default:
            // Do not select the hold action when another key is tapped.
            return false;
    }
}

Hold On Other Key Press

The “hold on other key press” mode can be enabled for all dual-role keys by adding the corresponding option to config.h:

#define HOLD_ON_OTHER_KEY_PRESS

This mode makes tap and hold keys (like Layer Tap) work better for fast typists, or for high TAPPING_TERM settings. Compared to the “permissive hold” mode, this mode selects the hold action in more cases.

If you press a dual-role key, press another key, and then release the dual-role key, all within the tapping term, by default the dual-role key will perform its tap action. If the HOLD_ON_OTHER_KEY_PRESS option is enabled, the dual-role key will perform its hold action instead.

An example of a sequence that is affected by the “hold on other key press” mode, but not by the “permissive hold” mode:

  • LT(2, KC_A) Down
  • KC_L Down (the L key is also mapped to KC_RGHT on layer 2)
  • LT(2, KC_A) Up
  • KC_L Up
                         TAPPING_TERM
  +---------------------------|--------+
  | +-------------+           |        |
  | | LT(2, KC_A) |           |        |
  | +-------------+           |        |
  |       +--------------+    |        |
  |       | KC_L         |    |        |
  |       +--------------+    |        |
  +---------------------------|--------+

Normally, if you do all this within the TAPPING_TERM (default: 200ms), this will be registered as al by the firmware and host system. With the HOLD_ON_OTHER_KEY_PRESS option enabled, the Layer Tap key is considered as a layer switch if another key is pressed, and the above sequence would be registered as KC_RGHT (the mapping of L on layer 2).

?> The HOLD_ON_OTHER_KEY_PRESS option is essentially redundant with the default mod-tap behaviour. The only notable difference is that HOLD_ON_OTHER_KEY_PRESS reduces the delay before the key events are made visible to the host.

For more granular control of this feature, you can add the following to your config.h:

#define HOLD_ON_OTHER_KEY_PRESS_PER_KEY

You can then add the following function to your keymap:

bool get_hold_on_other_key_press(uint16_t keycode, keyrecord_t *record) {
    switch (keycode) {
        case LT(1, KC_BSPC):
            // Immediately select the hold action when another key is pressed.
            return true;
        default:
            // Do not select the hold action when another key is pressed.
            return false;
    }
}

Ignore Mod Tap Interrupt

To enable this setting, add this to your config.h:

#define IGNORE_MOD_TAP_INTERRUPT

?> This option affects only the Mod Tap keys; it does not affect other dual-role keys such as Layer Tap.

By default, the tap-or-hold decision for Mod Tap keys strongly prefers the hold action. If you press a Mod Tap key, then press another key while still holding the Mod Tap key down, the Mod Tap press will be handled as a modifier hold even if the Mod Tap key is then released within the tapping term, and irrespective of the order in which those keys are released. Using options such as PERMISSIVE_HOLD or HOLD_ON_OTHER_KEY_PRESS will not affect the functionality of Mod Tap keys in a major way (these options would still affect the delay until the common code for dual-role keys finishes its tap-or-hold decision, but then the special code for Mod Tap keys will override the result of that decision and choose the hold action if another key was pressed). In fact, by default, the tap-or-hold decision for Mod Tap keys is done in the same way as if the HOLD_ON_OTHER_KEY_PRESS option was enabled, but without the decreased delay provided by HOLD_ON_OTHER_KEY_PRESS.

If the IGNORE_MOD_TAP_INTERRUPT option is enabled, Mod Tap keys are no longer treated as a special case, and their behavior will match the behavior of other dual-role keys such as Layer Tap. Then the behavior of Mod Tap keys can be further tuned using other options such as PERMISSIVE_HOLD or HOLD_ON_OTHER_KEY_PRESS.

An example of a sequence that will be affected by the IGNORE_MOD_TAP_INTERRUPT option (assuming that options like PERMISSIVE_HOLD or HOLD_ON_OTHER_KEY_PRESS are not enabled):

  • SFT_T(KC_A) Down
  • KC_X Down
  • SFT_T(KC_A) Up
  • KC_X Up
                         TAPPING_TERM
  +---------------------------|--------+
  | +-------------+           |        |
  | | SFT_T(KC_A) |           |        |
  | +-------------+           |        |
  |       +--------------+    |        |
  |       | KC_X         |    |        |
  |       +--------------+    |        |
  +---------------------------|--------+

Normally, this would send a capital X (SHIFT+x), even if the sequence is performed faster than the TAPPING_TERM. However, if the IGNORE_MOD_TAP_INTERRUPT option is enabled, the SFT_T(KC_A) key must be held longer than the TAPPING_TERM to register the hold action. A quick tap will output ax in this case, while a hold will still output a capital X (SHIFT+x).

However, if the HOLD_ON_OTHER_KEY_PRESS option is enabled in addition to IGNORE_MOD_TAP_INTERRUPT, the above sequence will again send a capital X (SHIFT+x) even if performed faster than the TAPPING_TERM. The difference from the default configuration is that by default the host will receive the key events only after the SFT_T(KC_A) key is released, but with the HOLD_ON_OTHER_KEY_PRESS option, the host will start receiving key events when the KC_X key is pressed.

For more granular control of this feature, you can add the following to your config.h:

#define HOLD_ON_OTHER_KEY_PRESS_PER_KEY

?> This option affects all dual-role keys.

You can then add the following function to your keymap:

bool get_hold_on_other_key_press(uint16_t keycode, keyrecord_t *record) {
    switch (keycode) {
        case SFT_T(KC_SPC):
            // Do not force the mod-tap key press to be handled as a modifier
            // if any other key was pressed while the mod-tap key is held down.
            return false;
        default:
            // Force the dual-role key press to be handled as a modifier if any
            // other key was pressed while the mod-tap key is held down.
            return true;
    }
}

Note that you must return false in get_hold_on_other_key_press in order to apply IGNORE_MOD_TAP_INTERRUPT for a certain mod-tap key.

?> IGNORE_MOD_TAP_INTERRUPT[_PER_KEY] is being progressively phased out to align the (default) behavior and configuration of mod-taps with the rest of dual-role keys.

Quick Tap Term

When the user holds a key after tapping it, the tapping function is repeated by default, rather than activating the hold function. This allows keeping the ability to auto-repeat the tapping function of a dual-role key. QUICK_TAP_TERM enables fine tuning of that ability. If set to 0, it will remove the auto-repeat ability and activate the hold function instead.

QUICK_TAP_TERM is set to TAPPING_TERM by default, which is the maximum allowed value for QUICK_TAP_TERM. To override its value (in milliseconds) add the following to your config.h:

#define QUICK_TAP_TERM 120

Example:

  • SFT_T(KC_A) Down
  • SFT_T(KC_A) Up
  • SFT_T(KC_A) Down
  • (wait until tapping term expires...)

With default settings, a will be sent on the first release, then a will be sent on the second press allowing the computer to trigger its auto repeat function until the key is released.

With QUICK_TAP_TERM configured, the timing between SFT_T(KC_A) up and SFT_T(KC_A) down must be within QUICK_TAP_TERM to trigger auto repeat. Otherwise the second press will be sent as a Shift. If QUICK_TAP_TERM is set to 0, the second press will always be sent as a Shift, effectively disabling auto-repeat.

!> QUICK_TAP_TERM timing will also impact anything that uses tapping toggles (Such as the TT layer keycode, and the One Shot Tap Toggle).

For more granular control of this feature, you can add the following to your config.h:

#define QUICK_TAP_TERM_PER_KEY

You can then add the following function to your keymap:

uint16_t get_quick_tap_term(uint16_t keycode, keyrecord_t *record) {
    switch (keycode) {
        case SFT_T(KC_SPC):
            return QUICK_TAP_TERM - 20;
        default:
            return QUICK_TAP_TERM;
    }
}

?> If QUICK_TAP_TERM is set higher than TAPPING_TERM, it will default to TAPPING_TERM.

Retro Tapping

To enable retro tapping, add the following to your config.h:

#define RETRO_TAPPING

Holding and releasing a dual-function key without pressing another key will result in nothing happening. With retro tapping enabled, releasing the key without pressing another will send the original keycode even if it is outside the tapping term.

For instance, holding and releasing LT(2, KC_SPC) without hitting another key will result in nothing happening. With this enabled, it will send KC_SPC instead.

               TAPPING_TERM
  +-----------------|------------------+
  | +---------------|-------+          |
  | | LT(2, KC_SPC) |       |          |
  | +---------------|-------+          |
  |                 |                  |
  |                 |                  |
  |                 |                  |
  +-----------------|------------------+

For more granular control of this feature, you can add the following to your config.h:

#define RETRO_TAPPING_PER_KEY

You can then add the following function to your keymap:

bool get_retro_tapping(uint16_t keycode, keyrecord_t *record) {
    switch (keycode) {
        case LT(2, KC_SPC):
            return true;
        default:
            return false;
    }
}

Retro Shift

Auto Shift, has its own version of retro tapping called retro shift. It is extremely similar to retro tapping, but holding the key past AUTO_SHIFT_TIMEOUT results in the value it sends being shifted. Other configurations also affect it differently; see here for more information.

Why do we include the key record for the per key functions?

One thing that you may notice is that we include the key record for all of the "per key" functions, and may be wondering why we do that.

Well, it's simple really: customization. But specifically, it depends on how your keyboard is wired up. For instance, if each row is actually using a row in the keyboard's matrix, then it may be simpler to use if (record->event.row == 3) instead of checking a whole bunch of keycodes. Which is especially good for those people using the Tap Hold type keys on the home row. So you could fine-tune those to not interfere with your normal typing.

Why are there no *_kb or *_user functions?!

Unlike many of the other functions here, there isn't a need (or even reason) to have a quantum or keyboard-level function. Only user-level functions are useful here, so no need to mark them as such.