Prepare release 1.8

[wismill]

We have already quite a few new features and MR are stalled, let’s cut a release.

This is our current changelog for 1.8:

libxkbcommon 1.8 - 2024-11-29

API

Breaking changes

• Updated keysyms case mappings to cover full Unicode 16.0. This change
  provides a consistent behavior with respect to case mappings, and affects
  the following:

  • xkb_keysym_to_lower() and xkb_keysym_to_upper() give different output
    for keysyms not covered previously and handle title-cased keysyms.

    Example of title-cased keysym: 0x10001f2 (U+01F2 “Dz”):

    • xkb_keysym_to_lower(0x10001f2) == 0x10001f3 (U+01F3 “dz”)
    • xkb_keysym_to_upper(0x10001f2) == 0x10001f1 (U+01F1 “DZ”)

  • Implicit alphabetic key types are better detected, because they use the
    latest Unicode case mappings and now handle the title-cased keysyms the
    same way as upper-case ones.

  Note: As before, only simple case mappings (i.e. one-to-one) are supported.
  For example, the full upper case of U+01F0 “ǰ” is “J̌” (2 characters: U+004A
  and U+030C), which would require 2 keysyms, which is not supported by the
  current API.

New

• Implement the GroupLatch action, usually activated with the keysym ISO_Group_Latch. (#455.group-latch)

• symbols: Added support for multiple actions per levels:

  • 1 keysym == 1 action now holds for every keysym count.
  • When no action is specified, interpret statements are used to find an action corresponding for each keysym, as expected.
  • When both keysyms and actions are specified, they should have the exact same count for each level.
  • For now, at most one action of each following categories is allowed per level:
    • modifier actions: SetMods, LatchMods, LockMods;
    • group actions: SetGroup, LatchGroup, LockGroup.
      Some examples:
    • SetMods + SetGroup: ok
    • SetMods + SetMods: error
    • SetMods + LockMods: error
    • SetMods + LockGroup: ok

  (#486.multiple-actions-per-level)

• rules: Use XKB paths to resolve relative paths in include statements. (#501.rules-includes)

• • Updated keysyms using latest [xorgproto] (commit: d7ea44d5f04cc476dee83ef439a847172f7a6bd1):

    Additions:

    • XKB_KEY_XF86RefreshRateToggle
    • XKB_KEY_XF86Accessibility
    • XKB_KEY_XF86DoNotDisturb

    Relevant upstream merge request: xorgproto-91.

• Added deprecated keysym warnings:

  • name deprecation (typo, historical alias), reporting the reference name;
  • keysym deprecation (ambiguous meaning, all names deprecated).

  These warnings are activated by setting the log verbosity to at least 2.

  It is advised to fix these warnings, as the deprecated items may be removed in a
  future release.

• Added the following modifiers names definitions in xkbcommon-names.h:

  • XKB_MOD_NAME_MOD1
  • XKB_MOD_NAME_MOD2
  • XKB_MOD_NAME_MOD3
  • XKB_MOD_NAME_MOD4
  • XKB_MOD_NAME_MOD5
  • XKB_VMOD_NAME_ALT
  • XKB_VMOD_NAME_META
  • XKB_VMOD_NAME_NUM
  • XKB_VMOD_NAME_SUPER
  • XKB_VMOD_NAME_HYPER
  • XKB_VMOD_NAME_LEVEL3
  • XKB_VMOD_NAME_LEVEL5
  • XKB_VMOD_NAME_SCROLL

• Rules: Added support for special layouts indexes:

  • single: matches a single layout; layout[single] is the same as without
    explicit index: layout.
  • first: matches the first layout/variant, no matter how many layouts are in
    the RMLVO configuration. Acts as both layout and layout[1].
  • later: matches all but the first layout. This is an index range. Acts as
    layout[2] .. layout[4].
  • any: matches layout at any position. This is an index range.

  Also added:

  • the special index %i which correspond to the index of the matched layout.
  • the :all qualifier: it applies the qualified item to all layouts.

  See the documentation
  for further information.

Fixes

• Previously, setting explicit actions for a group in symbols files made the parser
  skip compatibility interpretations for all the groups in the corresponding key,
  resulting in possibly broken groups with no explicit actions or missing key fields.

  Fixed by skipping interpretations only for groups with explicit actions when parsing
  a keymap and setting relevant fields explicitly when serializing a keymap to a string. (#511.explicit-interp-per-group)

• registry: Fixed libxml2 global error handler not reset after parsing, which
  could trigger a crash if the corresponding rxkb_context has been freed.

  Contributed by Sebastian Keller. (#529.fix-xml-error-handler)

• xkb_keymap_new_from_names: Allow only one group per key in symbols sections.
  While the original issue was fixed in xkeyboard-config
  project, the previous handling in libxkbcommon of extra key groups was deemed unintuitive.

  Note: rules resolution may still produce more groups than the input layouts.
  This is currently true for some legacy rules in xkeyboard-config. (#262.only-one-group-per-key)

• xkb_keysym_to_utf8: Require only 5 bytes for the buffer, as UTF-8 encodes code
  points on up to 4 bytes + 1 byte for the NULL-terminating byte.
  Previous standard RFC 2279
  (1998) required up to 6 bytes per code point, but has been superseded by
  RFC 3629 (2003). (#418)

• rules: Fix handling of wild card * to match the behavior of libxkbfile.
  Previously * would match any value, even empty one. Now:

  • For model and options: always match.
  • For layout and variant: match any non-empty value.

  (#497)

• Fixed xkb_keymap_get_as_string truncating the 4 longest keysyms names, such as Greek_upsilonaccentdieresis.

• The following functions now allow to query also virtual modifiers, so they work
  with any modifiers (real and virtual):

  • xkb_state_mod_index_is_active
  • xkb_state_mod_indices_are_active
  • xkb_state_mod_name_is_active
  • xkb_state_mod_names_are_active
  • xkb_state_mod_index_is_consumed
  • xkb_state_mod_index_is_consumed2
  • xkb_state_mod_mask_remove_consumed

  Warning: they may overmatch in case there are overlappings virtual-to-real
  modifiers mappings.

Tools

New

• Added --test option to compile-keymap and compile-compose, to enable testing
  compilation without printing the resulting file.

• Added xkbcli dump-keymap-wayland and xkbcli dump-keymap-x11 debugging tools
  to dump a keymap from a Wayland compositor or a X server, similar to what
  xkbcomp -xkb $DISPLAY - does for X servers.

• xkbcli how-to-type: added new input formats and their corresponding documentation.

  Unicode code points can be passed in the following formats:

  • Literal character (requires UTF-8 character encoding of the terminal);
  • Decimal number;
  • Hexadecimal number: either 0xNNNN or U+NNNN.

  Keysyms can to be passed in the following formats:

  • Decimal number;
  • Hexadecimal number: 0xNNNN;
  • Name.

Fixes

• Fixed various tools truncating the 4 longest keysyms names, such as Greek_upsilonaccentdieresis.

Build system

Breaking changes

• Raised minimal meson version requirement to 0.58.

Fixes

• Gated the use of --undefined-version in the linker because it breaks on older
  GNU ld. (#481)

---

There are still a 2 MR initially planned for 1.8:

• Incorrect implementation of LatchToLock and ClearLock #549
• Experiment: state: support pure virtual modifiers #450
• state: add API for updating latched and locked mods & layout in server state #521

While I am eager to see both merged, a reasonable option is to cut a release now, even if we cut another one “soon”.

TODO:

• Decide if we wait for remaining MRs
• Fix typos in the changelog (i.e. in the fragments)
• Publish the changelog and delete fragments
• Comment
• Fix xkbcomp: Some lists are not allowed to be empty #565
• Fix Abort when augmenting key definition #564

---

Some previous release issues/MRs for reference:

• 1.7: Prepare for 1.7.0 #445
• 1.6: 1.6.0 release planning #354

@bluetech @whot

[wismill]

Hold on, #549 must be fixed before.

EDIT: false alarm.

[bluetech]

I will try this week to set up and document the website & release process so that all maintainers can do a release.

[bluetech]

I have moved the website over to github pages. It needs some time for DNS TTL to run out. After it moves over, the previous server should no longer see accesses (@fooishbar FYI).

I will write a RELEASING.md doc now explaining the steps needed for a release.

EDIT: See #560.

[fooishbar]

Yeah, good move. Thanks a lot for all your work on the release @bluetech and @wismill :)

[bluetech]

Rules: Added support for special layouts indexes:

Maybe we should mention the motivation for this change, namely, to prepare for support of more than 4 layouts.

While I am eager to see both merged, a reasonable option is to cut a release now, even if we cut another one “soon”.

Agree, it's not a problem to do many releases.

[wismill]

Maybe we should mention the motivation for this change, namely, to prepare for support of more than 4 layouts.

While the initial motivation was indeed supporting >4 layouts, the change would be also useful if we keep the 4 layouts limit. But I agree, it’s a kind of sneak preview.

While I am eager to see both merged, a reasonable option is to cut a release now, even if we cut another one “soon”.

Agree, it's not a problem to do many releases.

Indeed, so we can release single “risky”¹ changes rather than group them.

Footnotes

1. Generalized virtual mods, >5 layouts, etc. By “risky” I do not mean it is allowed to fail QA, only that it is a breaking change that may challenge hard-coded assumptions. ↩