確定アンドゥした位置にカーソルを戻す - #261
Merged
Merged
Conversation
Contributor
|
@delphinus #260 は既にマージしたので、こちらは rebase してください |
Taking a kakutei back now walks the cursor to it, so confirming something
else leaves the cursor in the middle of the line rather than where the undo
was asked from. Typing on is the case the previous commit is for, which makes
this the common way out of it: notice the mis-conversion, undo, pick, and then
have to walk back by hand to carry on.
Remember where the undo was asked from and put the cursor there once the
replacement has been written. Something of another length shifts the column by
the difference, so the cursor keeps its place in the text rather than its
column.
ddskk has this as skk-undo-kakutei-return-previous-point and leaves it off by
default. It is on here, and not configurable, because the situation only
arises from the previous commit: before it, the undo could not move the cursor
at all, so there was nothing to come back from. The length adjustment is the
one skk-kakutei-subr does.
The place reaches Vim by two routes, because the text reaches the buffer by
two:
- A kakutei out of a henkan is written by the keys the key handling returns,
so the place rides along at the end of the same feedkeys(). That needs
restoreLnum / restoreCol on the handle result -- the one part of this which
touches the return path of every key handling rather than the undo alone.
- A completion source has already written when it reports through
completeCallback, so there are no keys to ride along with and
skkeleton#restore_point() is called right away.
Three things have to hold for the cursor to be sent back.
The undo has to still own the spot. That is asked at the start of each key
handling, and before the pre-edit mismatch resets the state: a source which
writes the candidate it is only selecting makes the buffer disagree with the
pre-edit, and the reading it is about to complete is very much still pending.
Taking the reading all the way back with cancel does drop it.
Nothing else may have been started where the cursor now is. One key can
confirm and open a henkan after it -- an uppercase key does both -- and the
cursor belongs to the one it opened. For a kakutei out of a henkan that is
settled once the handling is over; for a completion the confirmation can
arrive later still, since CompleteDone fires after the mapping of the key
which ended the completion (measured) and an engine of its own may schedule
the report on top of that, so it is answered with whether skkeleton is done or
is still on the very reading being confirmed.
The cursor has to be on the line the undo happened on. A kakutei out of <CR>
inserts a newline unless eggLikeNewline is set, and a column of the line it
came from means nothing on the next one.
An undo which never moved the cursor records the column the kakutei ends at
anyway, so nothing visibly happens for it.
確認:
deno task fmt-check / lint 指摘なし
deno check main.ts 通る
DENOPS_TEST_DENOPS_PATH を設定して全テスト
前のコミット 118 passed / 0 failed
このコミット 126 passed / 0 failed (追加した 8 件ぶん)
doc/skkeleton-functions.jax 追加分は表示幅 78 桁以内
autoload/skkeleton.vim Vim 9.1 / Neovim の双方で読める
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
delphinus
force-pushed
the
feat/kakutei-undo-return-point
branch
from
September 22, 2026 05:48
607a57d to
329ead1
Compare
delphinus
marked this pull request as ready for review
September 22, 2026 05:49
Contributor
Author
|
@Shougo 直しました。ご確認ください。 |
Contributor
|
マージします |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
問題
#260 で確定アンドゥがカーソルを戻すようになったため、選び直した候補を確定した後、カーソルが行の途中に取り残されます。
続きを打った後に使うのが #260 の狙いですが、そのために再確定後の手順が増えています。
直し方
確定アンドゥを実行した場所を覚えておき、選び直したものが書かれた後にそこへ戻します。長さの違う候補を選んだ場合はその差だけずらすので、カーソルは桁ではなく文章中の位置を保ちます。
カーソルを動かさなかった確定アンドゥでは戻す先が確定した文字列の末尾と重なるので、見た目には何も起きません。
実装としては、▼ からの確定は戻す先を
HandleResultに載せてキー処理の戻り値と一緒に返し、補完での確定は既にバッファへ書かれた後なのでskkeleton#restore_point()を直接呼びます。前者がこの PR で唯一、確定アンドゥ以外の戻り値に触る部分です。ddskk との違い
ddskk にも
skk-undo-kakutei-return-previous-pointとして存在します。2007 年に後から足されたユーザオプションで、デフォルトはnilです。長さの差だけずらす計算もskk-kakuteiの中のものと同じにしました。今回は確定アンドゥ自体実装したばかりですから、オプションにせず常に有効にしています。カーソルが飛ぶようにした以上、その後始末までで一つの機能だと考えました。
動作確認
deno task fmt-check/lint指摘なし、deno check通ります。DENOPS_TEST_DENOPS_PATHを設定して全テストを実行し、続きを打った後でも確定アンドゥ出来るようにする #260 をマージした main で 118 passed / 0 failed → この PR で 126 passed / 0 failed。テスト(9 バイト) を試(3 バイト) に選び直し、カーソルが行末 (桁 10) に戻ること、素直に確定した場合の桁 4 ではないこと。<C-g>で読みごと取り消した後は、無関係な確定でカーソルが動かないこと。<CR>の確定が改行を入れたとき、前の行の桁へ戻らないこと。doc/skkeleton-functions.jaxに記述を足しています。🤖 Generated with Claude Code