The hash map served a single exact-match lookup that the trie already
answers, at the price of a second container, a second file loader with
its own drift, and a Str-to-UTF-8 conversion on every chain probe. The
files are small (kanji.dict is 7.5k lines), so the trie holds both.
trieopen validates keys and each space-separated value word against Str
and reports path:line; trielookup takes the Str every caller holds and
treats a nil trie as an unloaded map; trienew/trieput exist for tests.
The overflow guards on growth, Trie.root (always 0) and the per-language
init loop written twice are gone.
ipc.c: an AF_UNIX nonblocking connect completes at once or fails with
EAGAIN, so the EINPROGRESS/poll/SO_ERROR path and the fcntl juggling
were dead; the deadline plumbing checked a clock that cannot fail and
re-tested the deadline before every transfer although only waitfd
blocks; readfield's truncation and discard loop served the unit test,
since every caller owns char[Ipcfieldmax+1]; NULL-argument checks on
in-tree encoders are gone (peer validation stays). The primary key frame
is Ipckey, not 'legacy'; ipckeysym names the keysym mapping.
srv.c: the per-connection capability and caret already lived in the
persistent Keyreq; the mirror locals and 'negotiated' flag guarded a
protocol rule no client relied on. proccreate never fails in libthread.
gtk: focus-out no longer round-trips a reset before closing the socket
that releases the engine; the caret dedup compares the packed frame;
srvconnect's preedit no-ops and the insimple flag are gone.
The keysym-to-engine-key rule lived in ibus.c, xim/xim.c and gtk/main.c,
and the copies disagreed: GTK sent keypad digits as special keys while
IBus and XIM sent '1'..'9'. Modifiers were rebuilt bit by bit in two
places although Mmask already is the X core layout; only the virtual
Super bit (26) that GDK and IBus set needs folding. ipckey/ipcmod in
ipc.c serve the daemon, the module and the tests.
bench.sh copied ipcpath() and the IBus address-file discovery into shell,
although the benchmark client only speaks IPC; retrying the client itself
is the readiness test. It also forwarded arguments to a daemon that takes
exactly one. README claimed Enter confirms the selection, but Enter and Tab
commit the reading unless a candidate was moved to.
Only ibus.o needs the D-Bus and xkbcommon flags, so give them the same
per-object scoping as the popup and XIM objects. The docker targets used
-B, which recursive makes inherit: the daemon was linked three times and
the live tests built twice per docker-check. gtk/Makefile repeated the
module-directory check and cache refresh in install and uninstall.
bits_per_rgb_value is the colormap component precision, not the pixel
layout; the NVIDIA driver reports 11 for its ordinary x8r8g8b8 visuals,
which left the popup disabled there. Depth, class, masks, byte order and
the pixmap format already pin the layout that putimage writes.
Avoid reading inactive Str storage when deciding whether to publish a redraw. Keep the focused engine suite compact by retaining one case for each page, navigation, and empty-result invariant.
