* shell.nix: fix gradle error
error was:
> \> Task :compileDebugJavaWithJavac FAILED
> error: Source option 7 is no longer supported. Use 8 or later.
> error: Target option 7 is no longer supported. Use 8 or later.
fixed by overriding the jdk version for gradle to openjdk17
* layouts/latn_bone: fix layout
- fix alignment with tabs mess
- make like actual bone layout, instead of trying to fit it into a 10 key
wide keyboard
- fixes missing üäö
- fix missing $
- moves q and ß where they belong
- remove all of the diacritic keys (they can be added through the
settings as extra keys)
- kept the number row extra keys integration into top row
- kept the idea of compressing , and . into the swipe actions of the
bottom row
- kept the number keys as key4 in the layer4 positions (instead of
moving them to a number row, which i also considered)
Include comprehensive ctrl modmap, enabling the use of ctrl modifier with Serbian Cyrillic keys. Remove clunky fn modmap previously used for just a few actions such as cut/copy/paste/etc.
The 'ctrl' modmap is different from the other modmaps as it also applies
the built-in Ctrl modifier to the resulting character, even if it was
first modified by the custom modmap.
For example, this will map Ctrl+в to Ctrl+V (not to v):
<ctrl a="в" b="v"/>
This is intended to add keyboard shortcuts in non-latin layouts.
A caveat is that the latin character appears on the keyboard while Ctrl
is activated.
Adds Serbian Latin QWERTY layout. Serbian language uses both Latin and Cyrillic as its official scripts, and this variant should cover anyone from Serbo-Croatian speaking group that uses Latin script.
Adds Serbian Cyrillic layout. Modmap of Fn key allows for desktop-like easier access to actions such as selectAll, copy, paste, etc. This is because, unlike desktop systems, on mobile ctrl does not combine with cyrillic keys to perform actions.
Allow combining circumflex \u0302 for Cyrillic vowels, enabling users to express themselves accurately.
ја сам са̂м = I am alone
ја сам сам = I am am
да = yes
да̂ = give
код = at
ко̂д = code
Similar to #437 for aigo.
This allows adding more compose sequences without modifying
en_US_UTF_8_Compose.pre.
This is done by grouping sequences files that should be merged together
into a directory. This also allows moving keysymdef.h into that
directory.
* Fix persian's half-space invisibility
* Add a 0.5 space before ظ in persian
* Change the place of چ in persian
* Change the width of backspace in persian
* Rename zwnj to halfspace in KeyValue.java and beng_provat layout
The arabic layout was used as the default on devices where all the
installed languages are not supported by the keyboard. This is not
intended.
This is probably caused by 'getCurrentInputMethodSubtype' returning the
first layout in the list of disabled subtypes in alphabetical or
language tag order.
Re-ordering the subtypes in method.xml had no effect.
Setting 'overridesImplicitlyEnabledSubtype' in method.xml has no
measured effect.
This adds the clipboard pane, which allows to save an arbitrary number of
clipboards and to paste them later. The key can be disabled in settings.
Checking the "Recently copied text" checkbox will cause the keyboard to keep a
temporary history of copied text. This history can only contain 3 elements
which expire after 5 minutes.
If this is unchecked, no history is collected.
History entries can be pinned into the persisted list of pins.
Internally "key repeat" is reword into "long press" when the same
mechanism was used for both features.
The constraint that 'timeoutWhat' must be set to '-1' when no message is
expected has been lifted to simplify the code.
This adds some errors:
- 'script' or 'numpad_script' is set an empty string.
- Multiple '<modmap>' elements.
- 'shift', 'width' and 'height' on every nodes that support them are
clamped to a valid value.
This replaces the switch cases in KeyModifier.java with JSON files, one
for each diacritic.
The number of states increases from 6727 to 7377. The apk size slightly
decreases (around 3kb).
Encoding errors in the compose data compiler due to:
- 'UTF-16' adds a BOM, use 'UTF-16-LE' instead
- 'str.encode' returns a byte array, use 'array' to have a 16-bit char
array.
Parse key names from keysymdef.h, which is distributed with Xorg. The
Greek, Cyrillic and Hebrew sequences referenced these keysyms.
This increases the number of sequences from 2043 to 2668.
Change the compose state machine definition to allow final states that
are wider than 16-bits.
This increases the number of sequences that can be used from
en_US_UTF_8_Compose.pre from 2013 to 2043 (of 3201).
