Skip to content

Latest commit

 

History

History
886 lines (813 loc) · 58.1 KB

CHANGELOG.md

File metadata and controls

886 lines (813 loc) · 58.1 KB

1.36.7:

  • FIX: Reported on Windows 10, search cause 'filter' of undefined exception on search-model #1155.

1.36.6:

  • FIX: Revert #1150 to avoid unexpected workflow disruption. Sorry for my lack of scrutiny.

1.36.5:

  • FIX: Disable Esc for Normal mode during autocomplete #1150 by @charlesangus.

1.36.4:

  • FIX: a-f (a-function) in JS when using tree-sitter #1134 by @jonboiser.

1.36.3:

  • FIX: groupChangesWhenLeavingInsertMode configuration was not properly honored in previous version but now works. #1130
    • In previous version, even user change it to false it still group all changes user make into one change.

1.36.2:

  • New: Config param askOptInToEditorScrollPastEnd to avoid continuously asking to enable editor.scrollPastEnd config. #1062

1.36.1:

  • FIX: Fix sort operator. Thanks @bennypowers ! #1125

1.36.0: Big default changes and custom surround

  • Breaking, new Default: #1106
    • setCursorToStartOfChangeOnUndoRedo: now false for precise cursor placement on undo/redo
      • To say simply, this is just directly using atom's undo/redo marker for cursor placement.
      • To make this work well, careful marker snapshotting(transact, checkpoint) is crucial(#603).
      • I consider this change as part of default stayOnXXX = true.
    • replaceByDiffOnSurround: now true for better cursor placement and highlight on undo/redo
      • Old: Replaced whole text. e.g. Replacing word with (word)
      • New: Modify surrounding text only, intact inner text.
  • New: custom surround #1104, #1102(inspirational PR by @isundaylee)
    • New customSurroundPairs config is introduced, its value is JSON string with following key-value pair.
      • key: char to access
      • value: array [openText:string, closeText:string, addSpace:boolean]
        • addSpace is optional, it control padding space.
    • Custom surround can be used in all surround operation(surround, delete-surround, change-surround)
    • Example configuration and possible operations
      • Config
        "vim-mode-plus":
          customSurroundPairs: '''
            {
              "p": ["<?php", "?>", true],
              "%": ["<%", "%>", true],
              "=": ["<%=", "%>", true],
              "s": ["\\"", "\\""]
            }
          '''
      • y s c p: surround(y s) inner-word(c or i w) with p(PHP tag)
      • c s p s: change-surround(c s) from p(PHP tag) to s(double quotes)
      • d s =: delete-surround(d s) of =(ruby-IRB tag)

1.35.0: TreeSitter support

  • Support: Set minimum engines to ^1.31.0-beta1.
  • Fix: TextObject function and fold now work with TreeSitter parser have enabled.
    • Function TextObject #1096
    • Fold TextObject #1096

1.34.0:

  • Diff: here
  • Improve: ChangeCase family operators(~, g u and g U) now target non-ascii chars. #1085 by @zensoup
    • Ex: g U can upper-case αρρ to ΑΡΡ.

1.33.0:

  • Diff: here
  • New: BlockComment text-object. #1082
    • inner-block-comment(i *)
      • Select from first non-white-space-char to last non-white-space-char within block comment.
    • a-block-comment(a *)
      • Select whole block-comment.

1.32.1:

  • Diff: here
  • Improve: vim-mode-plus:clip-debug-info no longer report diff for default value with type:array config.

1.32.0: Big improvement for hardTab editor(j, k and visual-blockwise-mode)!!

  • Diff: here
  • Improve: Now BlockwiseSelection properly select block in hardTab using editor #1052.
    • Calculate accurate column by expanidng/de-expanding hardTab of selected row.
  • Improve: j, k motions now handle hardTab appropriately when moving up and down.
    • Calculate accurate column by expanidng/de-expanding hardTab of selected row.
  • Fix: Apply workaround to broken paragraph motion({, })
    • Which was broken from Atom v1.25.0 because of atom-core's bug. #1047.
  • Tweak: move-to-(next/previous)-fold-(start/end)-with-same-indent command.
    • Previously it used to use indent level of cursor's row.
    • Now it use indent level of closest fold cursor's row is contained.
    • New: Add following commands for consistency
      • move-to-previous-fold-end-with-same-indent, move-to-next-fold-end-with-same-indent

1.31.1:

  • Diff: here
  • Fix: Now properly disable readOnly state on package deactivation.
    • Condition
      • When user enabled autoDisableInputMethodWhenLeavingInsertMode
      • Then deactivate vim-mode-plus package.
    • editor's hiddenInput remain readOnly = true which make editor un-editable.

1.31.0:

  • Diff: here
  • New: Introduce new configuration to affect word finding in motion and text-object.
    • Purpose
      • When finding word in motions(w, e etc) and text-objet(i w etc), vmp use nonWordCharacters which vary from grammars.
      • Some grammar treat - as non-word, but some are not, this inconsistency is a bit confusing.
      • So user can avoid this inconsistencies by enabling this new setting.
    • useLanguageIndependentNonWordCharacters: Enable this to always use same nonWordCharacters regardless of grammar.
    • languageIndependentNonWordCharacters: This value is used only when useLanguageIndependentNonWordCharacters was enabled.
      • Default value is borrowed from DEFAULT_NON_WORD_CHARACTERS in text-editor.js.

1.30.1:

  • Diff: here
  • Improve: deMaximize maximized state also on atom.workspace.onDidAddPaneItem timing.

1.30.0:

  • Diff: here
  • New, Experiment: autoDisableInputMethodWhenLeavingInsertMode config to auto-disable IME when leaving insert-mode.
    • Default false.
    • Once this feature was brought by PR#151 then reverted because of unfixable BUG.
    • Now retrying to bring same feature in different approach
      • setting readOnly attribute to false than setting input.type = "password".
    • Credit: Borrowed whole idea from @frapples's vim-mode-plus-patch-switch-ime package. Thanks @frapples!!

1.29.0:

  • Diff: here
  • New: hideCommandsFromCommandPalette config which hide commands from command-palette when set to true #1033, #1034
    • Default false to to make it compatible with older version.
    • This feature is once implemented in #943 but reverted before releasing it. Because I find no big perf gain by this feature.
    • But this time I changed mind, reducing noise when searching palette is good idea.
  • New, Support: Add support helper command. #1037
    • To let user paste clipped info when they open issue.
    • vim-mode-plus:clip-debug-info, vim-mode-plus:clip-debug-info-with-package-info
    • These command write collected debug info to clipboard with the form of foldable(by detail tag) to not mess issue board.

1.28.1:

  • Diff: here
  • FIX: Add altgraph using version of keymap to to workaround atom-keymap issue @lydell #661, #1031
    • In many keyboard layouts, such as German, French and Swedish, some characters have to be typed using the AltGr key.
    • For example, i { keymap doesn't work for these keyboard, since to type {, they need to type altgraph {.
    • But atom-keymap doesn't treat altgraph as modifier, so it immediately terminate pendingstate when it see altgraph.
    • To workaround this issue, we defined i altgraph { keymap. other keymap is also defined in same manner.
    • This fix is temporal workaround should be removed after atom-keymap solve this altgraph issue.

1.28.0:

  • Diff: here
  • Internal: Code cleanup
  • Improve: ChangeOrder family operators now work for non-separated word
    • Previously, ChangeOrder family operators(e.g. reverse) cannot reverse characters in word.
    • But now it can reverse, rotate, sort... characters when target have not space or comma separated.
      • Old: Can change only order of separated text(space or comma or new line char).
        • reverse: a b c get c b a
        • reverse: a, b, c get c, b, a
        • reverse: abc get abc(do nothing)
      • New: Can change order of non separated text.
        • reverse: a b c get c b a
        • reverse: a, b, c get c, b, a
        • reverse: abc get cba(this is new)
      • Affect child of ChangeOrder operators bellow.
        • Reverse, ReverseInnerAnyPair
        • Rotate, RotateBackwards,
        • RotateArgumentsOfInnerPair, RotateArgumentsBackwardsOfInnerPair,
        • Sort, SortCaseInsensitively, SortByNumber,
  • New: Support activating replace-mode mode from insert-mode.
    • Previously it worked, but have some odd behavior, now officially supported.
    • New toggle-replace-mode misc command by which you can shift mode between insert and insert.replace. #1026
    • This command works only in insert-mode, use this to quickly switch to replace-mode then back to insert-mode.
    • No keymap by default, here is example keymap.
      'atom-text-editor.vim-mode-plus.insert-mode':
        'ctrl-cmd-r': 'vim-mode-plus:toggle-replace-mode'
  • Internal: Introduce operationStack.runNext to queue operations to run next after current operation finished.
  • Breaking, Improve: Invert rotation order for following operators(since it's counter-intuitive)
    • Rotate, RotateBackwards, RotateArgumentsOfInnerPair, RotateArgumentsBackwardsOfInnerPair

1.27.0:

  • Diff: here
  • New, Experimental: change-subword-occurrence command for direct keymap use.
    • One-key version of c O.
  • Apply standard linting/formatting. As a result fixed lots of minor variable scope related bugs.

1.26.0:

  • Diff: here
  • Fix: confirmThresholdOnOccurrenceOperation now properly applied to subword occurrence operation like g O, c O etc.
    • Previously confirmThresholdOnOccurrenceOperation check was just bypassed for subword occurrence.
  • New, Experimental: Replace text via diff for TransformString operators.
    • Currently this new way of text mutation is used by trim-string(g |) and surround family(disabled by default).
    • To enable it for surround family enable replaceByDiffOnSurround new configuration.
    • How different from normal replace?
      • Scenario: foo - surround -> (foo) - change surround -> {foo} - delete surround -> foo
      • Normal replace:
        • All text transformation is done by replacing old text with new text.
      • Replace by diff:
        • All text transformation is done inserting/deleting minimum amount of text.
        • surround: adding ( and ),
        • change-surround: replace ( to {, ) to }
        • delete-surround: remove { and }
    • Pros vs Cons
      • Pros: Cleaner undo/redo highlight, more accurate cursor staying by using marker to track original position.
      • Cons: When redoing/undoing delete-surround, change-surround happens at off-screen position, user doesn't see flash highlight
        • Because inner text of surround pair is not mutated(so not flashed), This is most significant disadvantage.
  • Internal: Simplify commands registration for vimState bound command in main.js.

1.25.1:

  • Diff: here
  • Improve: Avoid centering editor-in-editor on vim-mode-plus:maximize-pane #1014
    • No longer centering text-editor which is rendered within normal text-editor using block decoration
    • This editor-in-editor is created by pkg git-diff-details or inline-git-diff(my fork of git-diff-details)

1.25.0:

  • Diff: here

  • Fix: Hide bottom gutter on maximize-pane

  • New: Motion, TextObjet, Operator for work with diff buffer. #1009

    • Operator: yank-diff-hunk

      • Yank diff-hunk under cursor with auto-removing leading + or - char
    • Motion: move-to-next-diff-hunk and move-to-previous-diff-hunk

      • Move to next diff-hunk which starts with + or - char
    • TextObject: a-diff-hunk and inner-diff-hunk(No behavior differences in these two text-object)

      • Just for used by yank-diff-hunk operator
    • Intending to be used in diff output text buffer(e.g. output of git diff).

    • No keymaps are provided, set as you like in your keymap.cson, here is example.

      # "source diff" is provided by language-diff package
      'atom-text-editor.vim-mode-plus.normal-mode[data-grammar="source diff"]':
        'tab': "vim-mode-plus:move-to-next-diff-hunk"
        'shift-tab': "vim-mode-plus:move-to-previous-diff-hunk"
  • New: vim-mode-plus:resolve-git-conflict operator. #1011

    • Detail: #1011

    • Quickly replace git-conflict's hunk with ours or theirs based on cursor position when you execute this command.

    • No keymap provided. Keymap example is here

      'atom-text-editor.vim-mode-plus.normal-mode':
        'space g c': 'vim-mode-plus:resolve-git-conflict'

1.24.4:

  • Diff: here
  • Fix: j, k threw exception on folded row which is also soft-wrapped, but no longer. #1003

1.24.3:

  • Diff: here
  • Fix: Remove no longer necessary temporal imperfect workaround of Atom-v1.24.0-beta0 issue. #998
    • After Atom-v1.24.0-beta1, no need to workaround beta0 issue since it's fixed.
    • Detailed information is here atom/atom#16294.

1.24.2:

  • Diff: here
  • Fix: All code fold related features were broken in Atom-v1.24.0-beta0, but no longer. #1002
    • Since editor.tokenizedBuffer.getFoldableRanges(N) now seems require explicit indentLevel(N).
  • Fix: vim-mode-plus:reload and vim-mode-plus:reload-with-dependencies which are used for faster dev cycle. #1002

1.24.1:

  • Diff: here
  • Fix: quick fix for cursor become invisible in Atom-v1.24.0-beta0.

1.24.0:

  • Diff: here
  • New, Breaking: a-fold now select "conjoined" fold #996
    • Old: a-fold just select current fold only.
    • New: a-fold select all conjoined fold range.
      • select previous fold if previous fold end at startRow of current fold.
      • select next fold if next fold start at endRow of current fold.
    • So you can do
      • delete(d z) or yank(y z) whole if/else-if/else clause by just two keystrokes.
      • delete(d z) or yank(y z) whole try/catch/finally clause by just two keystrokes.
  • Fix: j and k no longer stuck at row when screenRow contains multiple fold. #994, #995
    • e.g. fold else if-fold, then if-fold, then try to cross row by j or k but couldn't.

1.23.0:

  • Diff: here
  • Doc: add youtube movie link on README.md
  • Performance: Now base.js is not loaded on startup. So we got extra performance boost. #993.
    • Now in my MacBook Pro Late 2016. it score under 30ms(27ms is fastest record) activation time.
  • Support: Now deprecate use of service.Base provided by provideVimModePlus() service, warn to use service.getClass instead.
    • new service.getClass: since I decided to not expose bare Base class as it was.
    • new service.registerCommandsFromSpec: for adding multiple commands which load body lazily
  • Internal:
    • Move command-table.json, file-table.json to under directly lib/json. #990
    • Keep command-table-pretty.json, file-table-pretty.json for human read. #990
  • Improve: Smooth scroll now no longer depends on jQuery, use window.requestAnimationFrame() instead #991
  • Improve: TransformStringBySelectList now use atom-select-list instead of space-pen-view's SelectListView
    • As a result, vmp no remove dependency to space-pen-view

1.22.1:

  • Diff: here
  • Fix: Allow numpad0 to be used when entering a number before a command #989 by @sunjay
    • Now: Can set number 1 0 by numpad1 numpad0.
    • Old: numpad0 incorrectly moved cursor to head of line by incorrect keymap.

1.22.0:

  • Diff: here
  • Improve: Now Function TextObject(i f, a f) can detect more function. #984
    • Previously vmp can detect only function which parameter-list and body forms single fold.
    • Now vmp can detect function if parameter-list and body form different fold.
  • Fix: No longer editor freeze when big count was set for MoveToRelativeLine targeted operation. #985
    • E.g. 1 0 0 0 0 0 0 0 0 0 d d, 1 0 0 0 0 0 0 0 0 0 y y

1.21.0:

  • Diff: here
  • Support: No longer warning when user enabled vim-mode.
    • In old days, I frequently got report who enabled both vim-mode and vim-mode-plus.
    • But now vim-mode is obviously unmaintained, and I think less chance to confuse user.
  • New, Breaking: Use change-case npm in TransformString operator.
    • As a result, added new operators, also non-listed exiting operator improve translate-ability between different case.
    • SwapCase: Same as existing ToggleCase but add to reflect original change-case's function name.
    • ParamCase: Same as existing DashCase but add to reflect original change-case's function name.
    • PathCase: New transform a_b_c to a/b/c, camelCase to camel/case.
    • HeaderCase: New, transform HeaderCase to Header-Case, header_case to Header-Case.
    • ConstantCase: New, transform ConstantCase to CONSTANT_CASE, constant-case to CONSTANT_CASE.
    • SentenceCase: New, transform SentenceCase to Sentence case, sentence_case to Sentence case.
    • UpperCaseFirst: New, transform upperCaseFirst to UpperCaseFirst, abc def to Abc Def.
    • LowerCaseFirst: New, transform LowerCaseFirst to lowerCaseFirst, ABC DEF to aBC dEF.
  • Breaking: Now transform-string-by-select-list just simply shows "Title Case"-ed operator class name.
    • No longer display different name like it didsplayed Underscore _ for SnakeCase operator. Now it just show Snake Case.
    • Which might confuse you if you've been familier with old names. Sorry, but I think it was unnecessary.
  • Performance: Delay loading underscore-plus in operator.
  • Internal: Move commandTable generation logic to developer.js

1.20.0:

  • Diff: here
  • Fix: No longer throw exception on keystroke o escape . when groupChangesWhenLeavingInsertMode have enabled #966
  • Performance: Reduce IOs at package activation further for faster startup #965
    • Separate file-table.json from command-table.json.
    • Optimize command-table data format to reduce file size.
    • Now in my MacBook Pro Late 2016. it score around 40ms(38ms-45ms) activation time(measured by package cop).
    • This is NOT BAD when considering vmp adds 300+ commands on activation.
    • 2017-11-06: 40ms (vmp-v1.20.0 + Atom-v1.23.0-beta1) Now!
    • 2017-04-17: 60-80ms (vmp-v0.90.0 + Atom-v1.16.0)
    • 2017-01-04: 150ms (vmp-v0.78.0 + Atom-v1.13.0-beta10)
  • Improve: Longer search-flash duration from 0.5s to 1.0s to more easy to spot found word.
  • Support, Doc: No longer warn that CoffeeScript based extension is no longer supported on pkg activation.
    • Instead it is now explained in README.md with english correction PR made by @filipewl
    • Why I remove now is I noticed that warning was made on newly installed case(bad), it's been a month from this warning was added.
  • Support: Update issue template so that issue opener will see TODOs without jumping to link.

1.19.0:

  • Diff: here
  • Improve: Scroll motions ctrl-f, ctrl-b... now works more correctly in editor having block decoration by @jdanbrown.
    • Previously amount of scroll was based on screen rows, which doesn't block-decoration take into account.
    • Now scroll pixel based, so accurate even if editor have block decoration inline image.
  • Improve: Smoother scroll in sequential scroll motion execution.
    • Previously when multiple scroll request was made in very short time-gap, it immediately finalize previous request.
    • This means user see non-smooth scrollTop change although user enabled smooth scroll feature.
    • Now previous request is just cancelled and calculate new scrollTop based on previously requested value.
    • So in UX-wise, user no longer see sudden scrollTop change, always see smoooooth scroll congrats!!

1.18.2:

  • Diff: here
  • Change mind:
    • Update all vmp-plugins which affect changes in v1.18.0.
    • Remove registerCommandFromSpec migration code.
    • Why? migration for passed spec was not perfect, just confusing.
    • So I decided to simply let user update to latest, thats works.

1.18.1:

  • Diff: here
  • Fix: user's custom command in init.js throw exception when calling registerCommand().
    • Regression introduced in v1.18.0 sorry!

1.18.0: BIG overhaul in different aspect

  • Diff: here
  • Performance: Small activation performance gain by Base.js refactoring and JSON form of command-table.
  • Fix: Fix broken word motion family from Atom v1.23.0-beta0. #946, #953.
    • w, b, e, g e and it's variant have been broken after v1.23.0-beta0 was out.
    • This is because vmp depends Atom's cursor's method to find word.
    • But this dependency also make vmp's word motion vulnerable to changes in atom-core.
    • So this time, did overhaul these word-motion to not use cursors's method
  • Internal:
    • Changes in how vmp commands are loaded.
      • Previously, each operation class need to calll register() in it's own file.
        • e.g. call MoveDown.register() in ./motion.js.
        • This mechanism was once useful, especially as executable-class-body when vmp was written in coffee-script.
      • Now all operations class are jus requre-ed by /base.js and manually registerd.
        • So less magic, more explicit, easy to understand.
    • Use JSON format for command-table for faster activation. #958
      • This involve changes in field name of command spec.
      • So also updated vmp plugins which use registerCommandFromSpec function provided as vmp service.
    • Introduce findInEditor utility and cleanup lot of boilerplate code by using this new utility.

1.17.0:

  • Diff: here
  • Improve: Hide some vmp commands from command-palette #943
    • Hide only super-basic small num of commands only whic is defined in main.js
    • I evaluated hide all vmp commands in #943 but reverted
    • I was expected hiding all vmp commands improve command-palette's responsiveness but it was not.
    • So I took benefit to invoke all vmp command from palette as of now.
  • Improve: c j, c k at first or last buffer row no longer enter insert-mode.
  • Fix: In v1.23.0-beta0, some TextObject(e.g fold, comment) did not work.
  • Support: set minimum engines to 1.22.0
  • Internal, Breaking: Remove ModeManager class and re-blend it to VimState.
    • In original vim-mode, mode handling was done in VimState.
    • I extracted mode handling as ModeManager.
    • But now ModeManager’s task now get very small, I’m OK to re-blend it again.
    • Add deprecation warning when calling old ModeManager's event API.
  • Internal: insert-mode's task done in ModeManager is now handled in ActivateInsertMode operation class.
    • Remove replace-mode-backspace command
      • This was mapped from backspace in insert.replace mode.
      • But now achieve same functionality by overriding core:backspace.
      • So, this intermediate command is no longer necessary.

1.16.0:

  • Diff: here
  • New: Motion to scroll to function and redraw at uppper middle.
    • Following two commands are defined(no keymap by default)
      • move-to-previous-function-and-redraw-cursor-line-at-upper-middle
      • move-to-next-function-and-redraw-cursor-line-at-upper-middle
    • I created this command to use in vmp-demo at vimConf2017.
  • Internal: Refacotoring.
  • Fix: Amount of scroll rows was not symmetric in ctrl-d and ctrl-u, but now fixed.
  • Improve: Use TextBuffer's new onDidChangeText event to flash for undo/redo. #941.

1.15.0:

  • Diff: here
  • Breaking: Rename confusing ScrollXXX(It easily be confused with scroll motions(ctrl-f etc..)
    • ScrollDown(ctrl-e) to MiniScrollDown
    • ScrollUp(ctrl-y) to MiniScrollUp
  • New: Smooth scroll option plus more for MiniScroll(ctrl-y, ctrl-e)
    • smoothScrollOnMiniScroll: default false
    • smoothScrollOnMiniScrollDuration: default 200
    • defaultScrollRowsOnMiniScroll: default 1
  • New: Smooth scroll option for redraw-cursor-line commands.
    • redraw-cursor-line is z begging command like z t, z u, z z, z b etc..
    • Default disabled, I need this when I do demo. With smooth scroll on z u, less chance to leave behind audiences.
    • smoothScrollOnRedrawCursorLine: default false
    • smoothScrollOnRedrawCursorLineDuration: default 300
  • New, Experimental: New scroll motion to scroll 1/4.
    • ScrollQuarterScreenDown: keymap g ctrl-d
    • ScrollQuarterScreenUp: keymap g ctrl-u
    • Smooth scroll options are NOT explicitly provided, it use ScrollHalf's config.
  • Internal: Rename non-straightforward naming
    • From SelectInVisualMode to VisualModeSelect
    • From isAsTargetExceptSelectInVisualMode to isTargetOfNormalOperator

1.14.1:

  • Diff: here
  • Fix: Remove spec for now removed feature.

1.14.0:

  • Diff: here
  • Fix: transform-string-by-select-list throw exception when executed in visual-mode.
    • Regression by operator execution model redesign from v1.13.0.
  • Internal:
    • Improve how transform-string-by-external-command is executed.
    • Remove OperationAbortedError which is vmp specific error used to abort() operation, but now no longer used.
  • New, Experimental: JoinTarget operator(wanted to name just Join but it's already taken by J)

1.13.0:

  • Diff: here
  • Improve: Better multi cursors support for toggle-persist-selection.
  • Improve: Refold temporarily opened fold by / and ? when confirmed #931
  • New: z X family to redraw cursor line at upper-middle. #932
    • Keymap and Command
      • z u: redraw-cursor-line-at-upper-middle
      • z space: redraw-cursor-line-at-upper-middle-and-move-to-first-character-of-line
    • Conflict: z u with pureVim's spellfile related command. But OK, vmp have no plan for this feat.
    • Here is summary table of keymap and where to draw
      | where        | no move | move to 1st char |
      |--------------|---------|------------------|
      | top          | z t     | z enter          |
      | upper-middle | z u     | z space          |
      | middle       | z z     | z .              |
      | bottom       | z b     | z -              |
      
  • Breaking: rename confusing ScrollCursorToTop commands. #932
    • z enter: ScrollCursorToTop to RedrawCursorLineAtTopAndMoveToFirstCharacterOfLine
    • z t: ScrollCursorToTopLeave to RedrawCursorLineAtTop
    • z .: ScrollCursorToMiddle to RedrawCursorLineAtMiddleAndMoveToFirstCharacterOfLine
    • z z: ScrollCursorToMiddleLeave to RedrawCursorLineAtMiddle
    • z -: ScrollCursorToBottom to RedrawCursorLineAtBottomAndMoveToFirstCharacterOfLine
    • z b: ScrollCursorToBottomLeave to RedrawCursorLineAtBottom
  • Internal: Remove Operator.prototype.requireTarget, now all operator have target.
    • Use target = "Empty" for old requireTarget = false equivalent.
  • Internal: Remove casual use of isComplete and renamed to isReady as prep for upcoming refactoring. #933
    • Now isReady(was isComplete) is used only in operationStack.
    • For input taking MiscCommand family command such as mark, insert-register now executed in async.
    • All operators which take extra input now executed in async(no longer user requireInput mechanism).
    • At this point, requireInput is used only by Motion, which I cannot make it simply transform to use async execution.
    • Since motion is used as Operator's target and Operator has not yet support target executed in async scenario.

1.12.1:

  • Diff: here
  • Improve: duplicate-with-comment-out-original flashes correct range(only changed range).

1.12.0: Internal redesign of replace and surround(no behavior diff).

  • Diff: here
  • New: Operator duplicate-with-comment-out-original(idea by @nwaywood) #929
    • Duplicate targeted lines(always works as linewise) and comment out original target(selected) text.
    • No keymap by default. But available from select-list(ctrl-s for macOS user).
    • This is maybe useful when you change code with keeping original code as reference.
      • e.g. If I keymap g shift-cmd-D to vim-mode-plus:duplicate-with-comment-out-original.
      • g shift-cmd-D p: Duplicate paragraph with original paragraph commentout.
      • g shift-cmd-D z: Duplicate fold with original fold commentout.
      • g shift-cmd-D f: Duplicate function with original function commentout.
  • Improve: Remove operator's supportEarlySelect flag used by surround and replace operator #926
    • This option's purpose is to select target immediately after target provided and before reading user's input.
    • Now achieve same UX with more simpler way.
      • Old: Select target before execute() and skip selectTarget by checking if it's already selected.
      • New: Just execute and await user's input(simple and no complex skip selectTarget scenario).
    • As result of this re-design, surround and replace operator now executed in async.
      • This should be no diff from UX perspective. Just diff in internal execution model(require spec code change).
  • Improve: Respect occurrence wise when updating register
    • When c o p update register's content register's type
      • Old: Save with register's type linewise, immediate paste(cmd-v) in insert-mode surprise user.
        • Since replaced text by c was characterwise, but pasted text add extra newline(\n).
      • New: Save with register's type characterwise, and immediate paste(cmd-v) in insert-mode works as expected.
  • Improve: Spec helper
    • New: ensureWait is like ensure, but it wait till finish operation for async execution
    • Breaking: Remove ensureByDispatch since it's non essential, just wrapper. I want manage minimum set of ensure family.
    • Breaking: Disallow keystroke helper(after confirmed that it's not used by vmp and vmp-plugins).
    • Improve: ensure now recognize 1st arg as keystroke(not like keystroke or ensureOption as old ensure).
      • This is more clearer and explicit, and allow 2nd ensureOption empty to just dispatch keystroke(as replacement of old keystroke helper).

1.11.6:

  • Diff: here
  • Fix: move-up-to-edge, move-down-to-edge motion did not work correctly in soft-wrapped editor.
    • Regression from v1.11.0.

1.11.5:

  • Diff: here
  • Fix: No r with ' or " input properly work on some international keyboard which require TWO keystroke to input these quotes.
    • Fix provided 1.11.4 was just fix issue partially, not perfect.

1.11.4:

  • Diff: here
  • Fix: No r with ' or " input properly work on some international keyboard which require TWO keystroke to input these quotes.

1.11.3:

  • Diff: here
  • Fix: No longer throw exception when change-surround-any-pair cannot find surround pair.

1.11.2:

  • Diff: here
  • Fix: No longer throw exception when demo-mode pkg was activated, regression from v1.11.0.
  • Fix remove-leading-white-spaces operator work properly again regression from v1.11.0.
  • Improve: #701, When wrapLeftRightMotion is enabled and l in vC now can select new-line(can stop at column 0 of next-line).
  • Internal: No longer expect Operator.protoype.setTarget return instance.
    • This eliminate unessential contract between caller/callee.

1.11.1:

  • Diff: here
  • Fix: z c no longer throw exception, this is regression from v1.11.0.
  • Fix: highlightSearch, fix minor not-highlighted issue where it should be.
    • Fix: Highlight newly opened editor is not highlighted when mainMoudle.globalState was not populated.
      • E.g. pane:split-right, pane:split-down on fuzzy-finder's select-list
    • Fix: Now highlight newly opened editor even when it's not initially activated.(e.g. open on next pane without activating it).

1.11.0:

  • Diff: here
  • Fix: toggle-highlight-search command now correctly clear/re-highlight when value changed #918
    • This is regression introduced in v1.10.0 sorry!
  • Improve: toggle-fold now work with multiple cursors.
  • Improve, Breaking: How text-object function detect function scopes
    • Improve: function scope detection for type-script.
    • Improve, Breaking: source.js and source.jsx.
      • Ignore arrow function. Use a-fold(e.g. y z) for arrow-function.
      • Now no longer detect instantiation (new Class) call as function text-object.
  • Breaking: Rename following align operator since it was confusing.
    • Old: align-start-of-occurrence, New: align-occurrence-by-pad-start
    • Old: align-end-of-occurrence, New: align-occurrence-by-pad-end
  • Internal: Convention change, return value of Base.prototype.initialize() is not used, so no need to return this.

1.10.0:

  • Diff: here
  • New: numbering-lines operator, which adds line number to each line.
    • No default keymap, accessible via transform-string-by-select-list(ctrl-s for macOS user) command.
  • Improve: z t, z enter works properly when last screen row was visible. @dcalhoun #915.
    • editor.scrollPastEnd need to be true to work these command properly, so show notification if not enabled.
  • Fix: z e did not scroll until next cursor move is happens, but now scroll immediately.
  • Improve, Performance: Lazy load further
  • Improve: Improve integration with demo-mode package by making hover and flash highlight fadeout more in-sync.
  • Internal: Remove intermediate class ScrollWithoutChangingCursorPosition used for misc-scroll commands.

1.9.1:

  • Diff: here
  • Fix: [CRITICAL] No longer throw exception by lack of semver dependency. Sorry!

1.9.0: Converted to JS, as a result, CoffeeScript based customization is no longer supported.

  • Diff: here
  • Maintenance: Convert from CoffeeScript to JavaScript for operator, motion, text-object codes.
    • Now all running code is written in JavaScript.
    • Still test-spec is written in CoffeScript.
    • Through rewriting to JS, introduced lots of refactoring(architectural simplification, minor bug fixes).
  • Breaking: CoffeeScript based custom-vmp-operation is no longer supported.
    • From this version, all operations are defined as ES6 class which is NOT extend-able by CoffeeScript.,
    • If you have custom-vmp-operation in your init.coffee, require rewrite to JS. See Wiki.
    • Also all vmp-plugin pkg I'm maintaining is rewritten this time, see #895 for detail.
  • Fix: Broken features broken from Atom-v1.22.0-beta0 now work again.
    • Fold related commands: a-fold, inner-fold, move-to-next-fold-start etc..
    • Comment text-object.
  • New: Operator AlignOccurrence, AlignStartOfOccurrence, AlignEndOfOccurrence. #904 #906
    • No default keymap(set it by yourself if necessary).
    • Available from transform-string-by-select-list(ctrl-s for macOS user) commands.
    • How align operator works.
      • I introduce this operator with great simplicity by intention.
      • It's always add space to start or end of occurrence to align occurrence.
      • Add only, not trim existing spaces, if you want to remove consecutive spaces, use compact-spaces operator(g space for macOS).
    • For general purpose aligning(such as align lines by = assignment), use pkg like aligner.
    • This operator's goodness is explicitness and manual control for pattern to use for align, which make it possible general purpose aligner tools is not good at.
  • New: blackholeRegisteredOperators to disable register update for selected operator commands. #901, #902
    • Old dontUpdateRegisterOnChangeOrSubstitute are deprecated. it's setting is auto-migrated on first startup.
    • Set list of operator commands to blackholeRegisteredOperators.
      • E.g. change, change-to-last-character-of-line, delete-right, delete-left.
    • change*, substitute*, delete* is special value available to specify ALL same family operators.

1.8.2:

  • Diff: here
  • Fix: TransformStringByExternalCommand operator now correctly shift to normal-mode after operation finished.
    • This operator is specifically used by vim-mode-plus-replace-with-execution pkg(was broken, but recover now).

1.8.1:

  • Diff: here
  • Maintenance: Add Base.initClass as alias of Base.extend for upcoming vmp changes.

1.8.0: Expose select operator as normal command

  • Diff: here
  • New: select operator just for select target.
    • select is super essential operator which have been used in every motion in visual-mode.
    • But this select operator was not available as command.
    • Now vmp expose this as select operator with slight modification.
      • Original Select operator used in visual-mode was renamed to SelectInVisualMode.
      • The diff between Select and SelectInVisualMode is
        • Select accept preset-occurrence and persistent-selection but SelectInVisualMode is not so that user can modify selection without being interfered by existing preset-occurrence and persistent-selection.
  • New Config: keymapSToSelect conditional keymap. When enabled, s behaves as select operator.
    • s p: select paragraph. Equivalent to v i p.
    • s i i: select inner-indentation Equivalent to v i i.
    • s o p: select occurrence in paragraph.
    • g o s p: select occurrence in paragraph(use preset-occurrence by g o).
    • s o p o escape: Place cursors to each start position of occurrence in paragraph.
    • s o p I: insert at start position of occurrence in paragraph.
    • s o p A: insert at end position of occurrence in paragraph.
  • Improve: [(move-up-to-edge) and ](move-down-to-edge) now motion stops at first and last row again.
    • Now stoppable as long as target column is exist at first row or last row.
    • This behavior is added at #314(v0.49.0) but removed at #481(v0.66.0).
    • Now re-introduced this feature with avoiding edge case reported in #481.
  • Improve: Confirm on occurrence operation #888, #894
    • Now ask confirmation before starting to create occurrence-markers in g o(preset-occurrence), or c o(using o modifier) operation.
    • Allows cancellation to avoid editor become unresponsive while creating tons of markers.
      • e.g. If you g o accidentally for single-space and editor have huge matches, no longer freeze editor if you cancel confirmation.
    • New: config confirmThresholdOnOccurrenceOperation(default 2000) control confirmation threshold.
  • Keymap: Shorthand keymap for inner-entire in operator-pending-mode for Linux and Windows.
    • Windows and Linux user can ctrl-a as shorthand of i e(inner-entire).
      • Usage example: y ctrl-a to yank all text in buffer.
    • For macOS user cmd-a is provided as shorthand of i e in older version(v0.88.0).
  • New: Operator command insert-at-head-of-occurrence, insert-at-head-of-subword-occurrence.
    • Previously only start and end version of this commands are provided.

1.7.0:

  • Diff: here
  • Fix: f escape f no longer throw exception when reuseFindForRepeatFind was enabled #883
  • New: insert-at-head-of-target operator, in the past I removed this operator, but I need this now #881.
  • Improve: Cancelling in the middle of operation no longer clear reset multiple-cursors #882, #885
    • Motion: Find family f, F, t, T
    • Operator: Replace r surround family, split-string family, join-by-input family
  • Improve: No longer hide cursor when focus is at mini-editor.
    • /, ?: was hidden in all mode in older version.
    • f, t: was hidden in operator-pending-mode in older version.
  • Improve: cleanup CSS in vim-mode-plus.less.
  • Improve: Respect original selection's reversed state on occurrence-operation operation.
  • Doc: Wrote wiki and update README to for cursor-line modifying syntax-theme issue #887.
    • See FAQ section of README: "Flash effect not appear on cursor-line..."
  • Internal: Add debug code to investigate known issue #875(for cursor jumped unexpectedly).

1.6.0: Occurrence respects operator-bound-wise. #879

  • Diff: here
  • Improve: occurrence-operation now aware of operator-bound-wise.
    • Behavior diff: Diff appears in occurrence-operation.
      • Old: Always worked as characterwise.
      • New: Works as linewise when
        • V operator-modifier is used.
        • linewise-bound-operator(e.g. g /, > in normal-mode, D, Y in visual-mode etc) is used.
    • What is occurrence-operation and operator-bound-wise ?
      • occurrence-operation: Operation with o modifier(e.g. c o f) or operation with preset-occurrence(e.g. g o c f).
      • operator-bound-wise: D, C, Y in visual-mode and V forced operation like d V p.
    • Example(See also animation GIF in PR #879)
      • Assume non consecutive console.log lines scattered in function and you want to bulk appply operation linewise-ly.
      • First place cursor at console of console.log. Then you can do variety of operation linewise-ly.
      • g o v i p D: Mark occurrence(g o), select-paragraph(v i p), then delete-line(D).
      • g / o p: toggle-line-comments(g /) for occurrence(o) in paragraph(p).
      • g U o V p or g U V o p: upper-case(g U) for occurrence(o) with force linewise(V) in paragraph(p).
    • How this works?
      • Most operator doesn't have wise in operator-level, in this case operation's wise is determined by target(motion or text-object).
      • When V operator-modifier or wise-bound-operator is used, operation's wise is forced to this bound wise.
      • When occurrence-operation is forced linewise, it total steps is here.
        1. Select target(1)
        2. select occurrences contains in (1)
        3. range > select linwisely and merge selections with immediately adjacent row.
    • Takeout
      • Some operators(e.g. C, Y, D, g /, >, < etc) have pre-bound to linewise.
        • When you use these operation with o or g o, it works as linewise.
      • You can force wise to linewise by V operator-modifier, in this case occurrence-operation also works linewise.
      • Also remember you can distinguish [c and C], [y and Y], [d and D] in visual-linewise mode.
        • Use lower letter(c, y, d) if you want characterwise behavior.
        • Use capital letter(c, y, d) if you want linewise behavior.

1.5.0: Reconcile visual-mode with outer-vmp command in better way. #878

  • Diff: here
  • Summary: This is ambitious and dangerous change.
    • If success, lots of potential issue would be fixed.
    • But dangerous since it use more frequently fired hook(editor.onDidChangeSelectionRange).
      • I'm not 100% sure if this change works properly for all other commands provided by atom-core and packages.
  • Fix: #874 cmd-d(find-and-replace:select-next) in normal-mode respect SelectNext#wordSelected state for find-and-replace.
    • Original cmd-d select word only(exclude partial matches) if original selection is also created via cmd-d.
    • But this useful feature did not work in vmp's normal-mode.
    • Because when cmd-d is executed in normal-mode, vmp modify selection internally, which breaks state find-and-replace keep internally.
    • Behavior diff with text atom atomic atom.
      • Old: cmd-d twice select 1st and 2nd atom. Here, atom of atomic is selected, bad!.
      • New: cmd-d twice select 1st and 3rd atom.
  • Fix: #872 cmd-f(find-and-replace:show), search word, confirm then escape to clear selection no longer reset cursor position to where cmd-f was started.
    • This happens when cmd-f started with non-empty selection.
  • Fix: #873 Now whenever outer-vmp command modify selection, vmp starts visual-mode accordingly.
    • How vmp handle temporal selection modification done in single-command?
    • When outer-vmp command select some range(1) and clear(2) in single-command.
    • Vmp start visual-mode at (1), then reset to normal-mode at (2).
    • This is NOT elegant solution, but there is no other better way.
    • We cannot determine selection is eventually cleared or not within editor.onDidChangeSelectionRange event.
    • Delaying, debouncing to minimize useless mode-shift is bad for UX, user see slight delay for cursor updated.

1.4.0:

  • Diff: here
  • New: Numbered register(0-9) and small delete register(-). #871
    • When are they updated?
      • 0 is for yank, 1-9 is for change and delete.
      • - is for small change and delete, what small means is "content is less-than-one-line".
    • Currently no command to display register's content
      • Package author can access register by vimState.register.get(REGISTER_NAME).
  • Fix: f, r did not work correctly when user modified .plain.text grammar. #869
    • This issue only happens when user set .plain.text grammar to have softWrap to true and softWrapHangingIndent to non-zero value.
  • New, Experimental: SequentialPaste for p, P and replace-with-register.
    • Unnamed register(") maintain history at maximum sequentialPasteMaxHistory.
    • When user execute p sequentially, it pop content from older history.
    • Intended to be used as lazy quick escape hatch from very recent unwanted register mutation.
    • New configuration to control this new feature.
      • Config: sequentialPaste(default false) when enabled, pop history on sequential paste by p, P, and replace-with-register.
      • Config: sequentialPasteMaxHistory(default 3). maintain history specified this value.
  • Improve: f family related
    • Improve: Restore original scrollTop when findAcrossLines enabled and cancelled after scroll.
    • Rename config: keymapSemicolonToConfirmFind to keymapSemicolonToConfirmOnFindInput(auto-migrate)
    • Improve: Empty confirm no longer move cursor when findCharsMax > 1.
    • New: Commands find-next-pre-confirmed and find-previous-pre-confirmed.
      • Scope: Available when f family waiting for input(atom-text-editor.vim-mode-plus-input.find).
      • Keymap: tab, and shift-tab is mapped by default.
      • Why?
        • Allow you to adjust landing target BEFORE confirm to skip un-aimed stop interactively.
        • In most case, using ; or , after f finished is OK.
        • But when f is used with operator, such as c t f " and " matched earlier spot than you aimed.
        • This situation is not recoverable/retry-able by . repeat, but manually avoid-able by new commands.
    • New, Experimental: Conditional keymap keymapSemicolonAndCommaToFindPreConfirmedOnFindInput(default false).
      • When enabled, ; and , is mapped to new find-next-pre-confirmed and find-previous-pre-confirmed.
    • Maintenance: Remove [EXPERIMENTAL] tag from description of findAcrossLines settings, since we found it useful.
    • Improve: Fix several highlight inconsistencies.
  • Fix: Incorrect cursor rendering when automaticallyEscapeInsertModeOnActivePaneItemChange is enabled. #855
    • When automaticallyEscapeInsertModeOnActivePaneItemChange is set and active-tab change and back.
    • Cursor rendered odd way.
    • This is started from Atom v1.19.0 with new editor-rendering change.
    • To workaround issue, now use more appropriate onDidStopChangingActivePaneItem hook.
  • New, Experimental: Conditional keymap keymapIAndAToInsertAtTargetWhenHasOccurrence(default false). #862
    • This is revival of old default keymap which is removed because it was too aggressive, confusing as default-keymap.
    • When enabled and edigtor has preset-occurrence marker, I and A behave as operator which take target.
    • e.g. I p, A p to insert at start or end of preset-occurrence in paragraph(p).
  • Improve: Now maximize-pane no longer automatically de-maximized on active pane item change(tab-change). #866
    • So user can switch active tab with keep maximized.
    • This was original behavioral design I intended. Noticed it's broken(not sure when, or initially broken), so fixed.
  • Improve: move-to-next-occurrence, move-to-previous-occurrence now visit in ordered by buffer position. #864
    • Following behavioral change is noticeable only when user create multiple preset-occurrence for different word.
      • old: tab, shift-tab visit preset-occurrence in created order.
      • new: tab, shift-tab visit preset-occurrence in ordered by buffer position.
  • Internal: Cleanup RegisterManager code to reduce my confusion.

1.3.3:

  • Diff: here
  • Improve: highlight-find-char now highlight unconfirmed-current-match differently( with thicker border ).
    • You now visually notified "no extra keytype is required to land this position" while typing.

1.3.2:

  • Diff: here
  • Improve: highlight-find-char now highlight all chars in next lines when findAcrossLines was set.
    • This gives important feedback for your keystroke when your eye is needling destination keyword while typing.

1.3.1:

  • Diff: here
  • Fix: Fix confusing description in setting, no behavioral diff.

1.3.0:

  • Diff: here
  • Breaking, New: findCharsMax options to find arbitrary length of chars by f.
    • Involves renaming, auto-value-conversion of existing configuration introduced in v1.1.0( yesterday ).
    • New, Rename: findCharsMax: default 1.
      • If find's input reaches this length, confirm without waiting explicit enter.
      • So default f a(move to a char) is behavior when findCharsMax is 1.
      • By setting bigger number(such as 100) with also enabling findConfirmByTimeout, you can expect FIXED timing of confirmation against your input.
        • FIXED-timing-gap for "f > chars-input > wait > land(always by timeout)" flow.
      • Old findByTwoChars = true is equals to findCharsMax = 2, and migrated on first time activation.
    • Rename: findByTwoCharsAutoConfirmTimeout to findConfirmByTimeout( generalize naming )

1.2.0:

  • Diff: here
  • New, Experimental: findAcrossLines: default false
    • When true, f searches over next lines. Affects f, F, t, T.
    • [Collecting feedback] twitter: @t9md or #851
      • Not sure if this is really necessary feature since / and ? is available for multi-line search.
  • Breaking: New default: flashOnMoveToOccurrence = true
    • When preset-occurrence( g o ) is exist on editor, you can move between occurrences by tab and shift-tab
    • When set to true, flash occurrence under cursor after move
    • This feature is NOT new, Just "flashing by default" is new change in this release.

1.1.1:

  • Diff: here
  • Fix: Fix confusing description in setting, no behavioral diff.

1.1.0: This release is all for better f by making it tunable

  • Diff: here
  • [CAUTION: added at release of v1.3.0]: Config option name changed.
    • I basically don't want update old CHANGELOG, but this is for reducing confusion.
    • Because of short release timing gap with braking config params change.
      • v1.1.0: 2017.9.1
      • v1.3.0: 2017.9.2
    • When you read this below, keep in mind, change in available config params.
      • Parameterized: findByTwoChars = true to findCharsMax = 2
      • Renamed: findByTwoCharsAutoConfirmTimeout to findConfirmByTimeout
  • New: [Summary] Now f is tunable. #852.
    • Inspired pure-vim's plugins: clever-f, vim-seek, vim-sneak.
    • Highlighting find-char. It help you to pre-determine consequence of repeat by ;, , and ..
    • Aiming to get both benefit of two-char-find(vim-seek, vim-sneak) and one-char-find( vim's default ).
      • Even after two-char-find was enabled, you can auto-confirm one-char input by specified timeout.
    • Can reuse f, F, t, T as repeat-find like clever-f.
    • Maybe reading test-spec for these feature is clearer than reading following explanation.
      • describe "vmp unique feature of `f` family", ->
        describe "ignoreCaseForFind", ->
        beforeEach ->
        settings.set("ignoreCaseForFind", true)
        it "ignore case to find", ->
        set textC: "| A ab a Ab a"
        ensure "f a", textC: " |A ab a Ab a"
        ensure ";", textC: " A |ab a Ab a"
        ensure ";", textC: " A ab |a Ab a"
        ensure ";", textC: " A ab a |Ab a"
        describe "useSmartcaseForFind", ->
        beforeEach ->
        settings.set("useSmartcaseForFind", true)
        it "ignore case when input is lower char", ->
        set textC: "| A ab a Ab a"
        ensure "f a", textC: " |A ab a Ab a"
        ensure ";", textC: " A |ab a Ab a"
        ensure ";", textC: " A ab |a Ab a"
        ensure ";", textC: " A ab a |Ab a"
        it "find case-sensitively when input is lager char", ->
        set textC: "| A ab a Ab a"
        ensure "f A", textC: " |A ab a Ab a"
        ensure "f A", textC: " A ab a |Ab a"
        ensure ",", textC: " |A ab a Ab a"
        ensure ";", textC: " A ab a |Ab a"
        describe "reuseFindForRepeatFind", ->
        beforeEach ->
        settings.set("reuseFindForRepeatFind", true)
        it "can reuse f and t as ;, F and T as ',' respectively", ->
        set textC: "| A ab a Ab a"
        ensure "f a", textC: " A |ab a Ab a"
        ensure "f", textC: " A ab |a Ab a"
        ensure "f", textC: " A ab a Ab |a"
        ensure "F", textC: " A ab |a Ab a"
        ensure "F", textC: " A |ab a Ab a"
        ensure "t", textC: " A ab | a Ab a"
        ensure "t", textC: " A ab a Ab | a"
        ensure "T", textC: " A ab a| Ab a"
        ensure "T", textC: " A a|b a Ab a"
        describe "findByTwoChars", ->
        beforeEach ->
        settings.set("findByTwoChars", true)
        describe "can find one or two char", ->
        it "can find by two char", ->
        set textC: "| a ab a cd a"
        ensure "f a b", textC: " a |ab a cd a"
        ensure "f c d", textC: " a ab a |cd a"
        it "can find by one-char by confirming explicitly", ->
        set textC: "| a ab a cd a"
        ensure "f a enter", textC: " |a ab a cd a"
        ensure "f c enter", textC: " a ab a |cd a"
        describe "autoConfirmTimeout", ->
        beforeEach -> settings.set("findByTwoCharsAutoConfirmTimeout", 500)
        it "auto-confirm single-char input on timeout", ->
        set textC: "| a ab a cd a"
        ensure "f a", textC: "| a ab a cd a"
        advanceClock(500)
        ensure textC: " |a ab a cd a"
        ensure "f c d", textC: " a ab a |cd a"
        ensure "f a", textC: " a ab a |cd a"
        advanceClock(500)
        ensure textC: " a ab a cd |a"
        ensure "F b", textC: " a ab a cd |a"
        advanceClock(500)
        ensure textC: " a a|b a cd a"
  • Config: [Detail] Following configuration option is available to tune f.
    • keymapSemicolonToConfirmFind: default false.
      • See explanation for findByTwoChars.
    • ignoreCaseForFind: default false
    • useSmartcaseForFind: default false
    • highlightFindChar: default true
      • Highlight find char, fadeout automatically( this auto-disappearing behavior/duration is not configurable ).
        • Fadeout in 2 second when used as motion.
        • Fadeout in 4 second when used as operator-target.
    • findByTwoChars: default false
      • When enabled, f accept TWO chars.
        • Pros. Greatly reduces possible matches, avoid being stopped at earlier spot than where you aimed.
        • Cons. Require explicit confirmation by enter for single char-input. You might mitigate frustration by.
          • Confirm by ;, easier to type and well blend to forwarding repeat-find( ; ).
            • Enable "keymap ; to confirm find motion"( keymapSemicolonToConfirmFind ) configuration.
            • e.g. f a ; to move to a( better than f a enter?). f a ; ; to move to 2nd a(well blended to default repeat-find(;)).
          • Enable auto confirm by timeout( See. findByTwoCharsAutoConfirmTimeout )
    • findByTwoCharsAutoConfirmTimeout: default 0.
      • "When findByTwoChars was enabled, automatically confirm single-char input on timeout( msec ).
      • 0 means no timeout.
    • reuseFindForRepeatFind: default false
      • When true you can repeat last-find by f and F(also t and T).
      • You still can use , and ;.
      • e.g. f a f move cursor to 2nd a.
    • My configuration( I'm still in-eval phase, don't take this as recommendation ).
      keymapSemicolonToConfirmFind: true
      findByTwoChars: true # [converted] to `findCharsMax = 2`
      findByTwoCharsAutoConfirmTimeout: 500 # [converted] to `findConfirmByTimeout = 500`
      reuseFindForRepeatFind: true
      useSmartcaseForFind: true

1.0.0: New default stayOn all true.

  • Diff: here
  • Version: Decided to bump major version.
  • Breaking: Default config change/Renamed config name.
    • Summary:
      • Now all stayOn prefixed configuration have new default false.
      • New default behavior is NOT compatible with pure-Vim.
        • Set all stayOn prefixed configuration to false to revert to previous behavior.
      • Some configuration parameter name is renamed to have stayOn prefix.
        • Automatically migrate existing config on activation of vmp.
      • What is stayOnXXX configuration?
        • Respect original cursor position as much as possible after operation( select, move, operate ).
        • It keep both cursor's row and column or column only( if vertical move was necessary ).
    • Config params renamed and changed default value
      • New default: incrementalSearch = true
      • New default: stayOnTransformString = true
      • New default: stayOnYank = true
      • New default: stayOnDelete = true
      • Renamed/New default: keepColumnOnSelectTextObject > stayOnSelectTextObject = true
      • Renamed/New default: moveToFirstCharacterOnVerticalMotion !> stayOnVerticalMotion = true
        • Renamed with meaning inverted: !moveToFirstCharacterOnVerticalMotion === stayOnVerticalMotion