| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Currently layouts designed to be used with only one hand have the
English descriptions: “left handed” and “right handed”. This is
ambiguous, as it highlights *handedness* rather than one-hand typing.
Translations in some languages may thus convey only the former
understanding, which is incorrect.
Clarify that these layouts are for typing with only *one* hand.
Closes #483
Part-of: <https://gitlab.freedesktop.org/xkeyboard-config/xkeyboard-config/-/merge_requests/737>
|
|
Add the newest layout of the German Neo keyboard layout family. Noted is
a layout designed to make typing as comfortable and efficient as
possible in both German and English. It was developed with the help of a
layout optimizer, which is based on the metrics and concepts of
ArneBab's optimizer and adapts and extends them.
I switched from Neo 2 via Mine to Noted and I like to see this variant
to be available out of the box in Debian/Ubuntu.
Sources:
* https://neo-layout.org/Layouts/noted/
* https://dariogoetz.github.io/noted-layout/
|
|
|
|
Added `caps:return` to make the `Caps Lock` key an additional `Return` key.
|
|
According to our doc and practices, symbols should be defined only for
the first group. There are still a few layouts defining simultaneously
two groups. One of them is `nec_vndr/jp`, which has also a dedicated
rule. This commit fixes only this layout in preparation of a refactor
for `inet` rules.
- Split `nec_vndr/jp(pc98)` into two sections: `pc98` and `jp`, and
make the latter the default section.
- Fix the rules to ensure using this layout with the model `pc98` will
load *explicitly* 2 layouts, as this is the case for other models.
- Make the rules match `jp` layout, not `nec_vndr/jp`. No other rule
used the vendor path to match a layout.
The remaining layouts with 2 groups should also be fixed someday.
|
|
This option swaps right `Alt` with right `Win`. Symbols are already
there, so this commit just adds the corresponding rules via the registry.
Part-of: <https://gitlab.freedesktop.org/xkeyboard-config/xkeyboard-config/-/merge_requests/731>
|
|
Since c4f76b584fe10bc2037966a70efad9bc5b682a4a we merge rule sets
with and without explicit merge modes (i.e. *explicit* merge mode
are section starting with either `+` or `|`). This is wrong, as
these are not supposed to be matched at in the same rule set.
Part-of: <https://gitlab.freedesktop.org/xkeyboard-config/xkeyboard-config/-/merge_requests/729>
|
|
The previous update c4f76b584fe10bc2037966a70efad9bc5b682a4a removed
the initialization of a dictionary in the `merge` function, leading to
duplicates, namely the files *not* starting with a rule header.
While we introduced `defaultdict` previously to avoid initializing lists
manually, we still need to initialize the `dict` in order to keep it in
sync with `section_names`.
Part-of: <https://gitlab.freedesktop.org/xkeyboard-config/xkeyboard-config/-/merge_requests/729>
|
|
Diktor is an ergonomic alternative to the JCUKEN layout (see:
`ru(winkeys)`), adapting Dvorak design principles for Russian.
Although it is popular, there is no track of the original author.
The original implementation targeted Windows. The closest to an
“official” web page is: https://olegp.name/software/diktor-keyboard-layout/.
|
|
Layout compatibility rules are broken in configurations with 2 or more
layouts, e.g.:
- Works:
- config: `--layout fi --variant basic`
- symbols: `pc+fi(classic)+inet(evdev)`.
`fi(basic)` is correctly remapped to `fi(classic)`.
- Does not work:
- config `--layout fi,us --variant basic,`
- symbols: `pc+fi(basic)+us:2+inet(evdev)`.
`fi(basic)` is not remapped and leads to an include error.
- Does not work:
- config `--layout us,fi --variant ,basic`
- symbols: `pc+us:fi(basic):2+fi(classic):2+inet(evdev)`.
`fi(basic)` is not remapped and leads to an include error.
I initially thought merging the following rules sets to fix it:
1. `model layout[n] = symbols`
2. `model layout[n] variant[n] = symbols`
but it does not work: `*` wildcard does *not* match empty values.
There is another issue: using a variant will check and match *both*
rules sets. In the first rule set, compat rules are only for deleted
symbols files (e.g. `dev`). But when the symbols file exists, we end
up with an invalid include statement (see `fi(basic)` examples above).
This is because the first rules set has a catch-all rule that will
always match, while we want it to match only in the second rules set.
Unfortunately such rules cannot be expressed in rules files. So the
solution is to append generated compatibility sections in the
corresponding symbols files and remove the corresponding current rules.
Section name clashes are checked and will raise an error.
E.g.: currently we append the following to `symbols/de`:
partial xkb_symbols "lld" {
include "it(lldde)"
};
If we already had a section named lld, this would produce the following
error:
Cannot add compatibility section in symbols/de: "lld" already exists
Also:
- Move symbols handling in meson into the symbols subfolder.
- Sort compat layout rules.
- We ensure that rules with variant take priority over rules without one.
Part-of: <https://gitlab.freedesktop.org/xkeyboard-config/xkeyboard-config/-/merge_requests/728>
|
|
- Merge section with same header, independently of the whitespaces.
- Improve Python (typing)
Part-of: <https://gitlab.freedesktop.org/xkeyboard-config/xkeyboard-config/-/merge_requests/728>
|
|
|
|
Make `Caps Lock` an additional `Ctrl` and `Shift + Caps Lock` the
regular `Caps Lock`.
Part-of: <https://gitlab.freedesktop.org/xkeyboard-config/xkeyboard-config/-/merge_requests/673>
|
|
Use Caps Lock key as additional 3rd lever chooser, Ctrl + Caps Lock for
original Caps Lock behaviour.
The 2023 DIN standard for German keyboards recommends it as an option:
- https://de.wikipedia.org/wiki/E1_(Tastaturbelegung)#Feststelltaste/Umschaltsperre
- https://en.wikipedia.org/wiki/Caps_Lock#Abolition
Part-of: <https://gitlab.freedesktop.org/xkeyboard-config/xkeyboard-config/-/merge_requests/687>
|
|
Previously the compatibility rules where incorrect:
- Lines with a comment starting by “//” where parsed incorrectly,
because comments where in fact not supported.
- `variantsMapping.lst` was added to rules without variant.
This resulted in incorrect rules with MLVO = (model, layout), e.g. the
beautiful following rules:
```
! model layout = symbols
* mao(nz(mao)) = pc+//(Delete)
* ben(basic) = pc+in(ben)
```
Fixed the issue and ensure to raise an error if expectation on layout
variant is broken.
|
|
`Shift + Caps` locks the digits on the digits row, `Caps Lock` alone
behaves as usual (Azerty layouts).
The `Shift + Caps Lock` combination inverts the first two levels only.
Originally designed for layouts with digits on the second level of the
digits row (e.g. Azerty layouts).
This allows users to have direct access to digits while keeping
usual capitalization via `Caps Lock`.
Also add the `FOUR_LEVEL_LOCKABLE_LEVEL2` key type, a variant of
`FOUR_LEVEL_MIXED_KEYPAD` that uses `LevelFive` instead of `NumLock`,
to avoid interaction with the keypad. Originally created to lock the
digits on layouts with digits on the second level of the digits row
(e.g. Azerty layouts).
|
|
Fallback with package `strenum` is required for Python < 3.11.
Part-of: <https://gitlab.freedesktop.org/xkeyboard-config/xkeyboard-config/-/merge_requests/702>
|
|
|
|
This partially reverts commits:
- 91da938a16b4dc29df62fa3794d2d046daa2edbe
- f67ca9fbb974c57cdef9d2c9f422d618e1c4de4c
|
|
Ergo‑L is an ergonomic keyboard layout meant to optimize the typing
experience in French, English, and programming. Accents, diacritics
and fancy typographic symbols are typed using a custom dead key, but
it is fully implemented in the keymap itself as an `ISO_Level5_Latch.`
Below is a more detailed presentation of the layout (in French):
https://ergol.org/presentation/
This commit adds Ergo‑L `fr(ergol)`, along with an angle-mod variant
`fr(ergol_iso)` made for ISO keyboards.
|
|
|
|
|
|
This maps Scroll_Lock to Mod3, effectively enabling this virtual
modifier.
|
|
Use in-repository-only symlinks between symbols files that help us to
resolve option mismatches, e.g. where we need to map caps:foo to
capslock(foo).
This effectively reverts
symbols: symlink the special rules prefixes to our symbols files
|
|
The vast majority of options are a straightforward foo:bar = +foo(bar)
mapping. Instead of maintaining those mappings by hand let's generate
the rules file from the files we definitely maintain by hand: the
base.xml (and base.extras.xml) files.
This also makes it clearer which ones are exceptions and need to be
filled in through other means (or ignored because the option does not
affect any symbols).
The resulting rules file is identical to the previous hand-generated
one but it is alphabetically sorted and uses different whitespacing.
For merge.py to work we need to also fix the whitespaces in
rules/compat/0041-option_symbols.part, otherwise the duplicate header
detection won't work.
|
|
This makes it immediately readable what maps to what over e.g.
ml4_s.part.
|
|
|
|
This avoids having to special-case grab:debug and srvrkeys:none in the
rules files.
|
|
These used to have a special entry in the rules file - it's easier if we
just make a section for them and have the rules-file entry map 1:1.
For ctrl:grouptoggle_capscontrol see commit 6bba53fd
|
|
On branch ohu_sk
Changes to be committed:
modified: rules/base.extras.xml
modified: symbols/hu
|
|
|
|
See the previous commits, none of these machines need the latest
xkeyboard-config.
|
|
Best I can tell these are macbooks released from 2006 to 2012 or
something. Doesn't matter too much anyway since the only effect of these
rules was a different geometry file.
Since geometry is not supported at all in libxkbcommon we won't lose
much by dropping these special cases.
|
|
Released in 2008 we expect few of these devices still exist and those
that do probably don't need the latest xkeyboard-config release.
|
|
The Nokia RX-51 (N900) came out in 2009. 15 years later we can purge
this rule, the number of devices that still exist and need latest
xkeyboard-config is too small to worry about.
The Nokia RX-44 (N810) an SU-8W were released 2007.
|
|
The Azerty keyboards used in France and Belgium have the digits on the
shift level instead of on the base level of the top row. In Windows
the corresponding layouts have the curious characteristic that CapsLock
locks also the digits, meaning that some users have acquired the habit
of engaging CapsLock when they need to type some digits.
Fix this by adding a new option `caps:digits_row` that applies only to
the `$azerty` layouts group (currently `fr` and `be`). This option
changes the key type of the digits keys to `FOUR_LEVEL_SEMIALPHABETIC`.
There are two drawbacks with this solution:
- It is not possible to type the uppercase À, È, É and Ç. Without the
option, one gets those by the internal transformation rules, done
either by the X server or xkbcommon.
- It changes the behavior of the US variants from `fr`. But we do not
expect anyone to use at the same time e.g. `fr` and `fr(us-alt)` with
the new option `caps:number_row`. Instead we expect e.g. either one
uses `fr(us-alt)` alone or `fr,us`. In the former case the option is
not needed, while in the later case the option works correctly.
|
|
We have the rules in place for this, let's expose those.
Fixes #445
|
|
These were added as rules entries in ac8fa2656 but no-one noticed for 13
years that they had no corresponding base.xml entry. Let's remove them
because they look suspiciously like "added because we can" rather than
"added because there's a need for it". Google cannot find a single
reference to it that isn't this repo or a mirror of it.
Fixes #445
|
|
These have been commented out for over 12 years because they're
apparently broken, see 569dd72bd4b and
https://bugs.freedesktop.org/show_bug.cgi?id=41621
Let's remove those, we can reinstate those if there are users that
can make these rules work.
Fixes #445
|
|
The purpose of this option is to have Caps Lock behave as a Control
modifier, but still be able to distinguish it from the Control_L and
Control_R keys in things like games and emulators.
People tend to get confused when in some programs Caps Lock with this
option doesn't behave exactly like Control. ctrl:nocaps is probably the
option they really want.
Fixes #443
|
|
|
|
https://iso639-3.sil.org/sites/iso639-3/files/downloads/iso-639-3_Retirements.tab
ais Nataoran Amis S Split by part going to Amis [ami] and creating Sakizaya [szy] with the remaining part 2019-01-25
uun Kulon-Pazeh S Split into Kulon [uon] and Pazeh [pzh] 2022-01-20
See also the pycountry commit
https://github.com/pycountry/pycountry/commit/a2341a11df5b5fddb26519e9b9cbe1befdee74a9
Fixes #430
|
|
On branch master
Your branch is up to date with 'origin/master'.
Changes to be committed:
modified: rules/base.extras.xml
modified: symbols/hu
|
|
black is getting increasingly more difficult to use in the CI across
distribution versions (thanks to Python's messy packaging) so let's
replace it with ruff format which produces largely the same results.
|
|
use the builtin `install_symlink` target provided by meson instead of providing on external `ln` and `sh`.
|
|
Until commit 9350c7347a from two years ago, the `extended` variant
contained *fewer* symbols than the `basic` variant, and did not contain
any symbols that `basic` did not have. The extra symbols that it did
have over the `simple` variant were all present in the same locations
in the `basic` variant. So at that point the `extended` variant was
fully redundant. The mentioned commit 9350c7347a did not "perfect"
the `extended` layout but merely added some symbols in ill-considered
places. Such adjustments are stuff for the `custom` layout instead.
This addresses and closes issue #412.
Signed-off-by: Benno Schulenberg <bensberg@telfort.nl>
|
|
The layout is not international at all, and cannot be used to type
just German or French or Slovak either. It looks more like a first
draft of the `scn` layout.
The original contributor, David Paleino, agrees with its removal.
This addresses and closes issue #406.
Signed-off-by: Benno Schulenberg <bensberg@telfort.nl>
|
|
Zawgyi [1] is a font for the Burmese language where the characters
are encoded in a way that is incompatible with Unicode, so "Zawgyi"
is not part of the name of the language. Also, the addition "Tai"
for a specific version of the font is unneeded.
[1] https://en.wikipedia.org/wiki/Zawgyi_font
Furthermore, drop the `eng` marker for two layouts as no English
can be typed with them.
Signed-off-by: Benno Schulenberg <bensberg@telfort.nl>
|
|
In some circumstances the identifier of the variant is presented,
rather than the name. The identifier "two" is not descriptive
enough.
Thanks @benno for suggesting the name "si2".
|
|
|
