[WIP] support transparent & ctermbg cursor + minimal highlight api #6566

teto · 2017-04-21T22:57:40Z

support for a transparent TUI cursor (set guibg=None to the cursor highlight).
ctermbg can also affect cursor color even whithout termtruecolors.
notify UIs about changed highlights
add api to play with highlights: nvim_hl_from_id/name
simplify handling of the 'Normal' group. :hi Normal should work even with -u NORC.
add to wiki a table describing terminal supported ? (for cursor color/transparency etc)
reset cursor for VTE (seems to be a VTE extension printf "\x1b[0 q")

As for setting cursor foreground color, VTE api allows it (termite has a cursor_foreground parameter for instance) but there is no OSC yet. https://bugzilla.gnome.org/show_bug.cgi?id=695011 proposes OSC20 for that but rxvt has a pre-existing use of OSC 20 (for background pixmap). Thomas Dickey (xterm maintainer) seems ok with a patch that would set cursor foreground color, so anyone feel free to propose such a patch.

brcolow · 2017-04-21T23:28:40Z

src/nvim/tui/tui.c

+
+    if (ui->rgb) {
+
+      RgbValue bg = aep->rgb_bg_color;


Is this more succinct using a ternary?

// if gui=reverse, swap bg and fg colors RgbValue bg = aep->rgb_ae_attr == HL_INVERSE ? aep->rgb_fg_color : aep->rgb_bg_color;

Just a random thought..might not be worth changing.

Makes a line 80 chars wide line so it ends up not being clearer

brcolow · 2017-04-21T23:33:28Z

src/nvim/tui/tui.c

    const char *val)
 {
-  if (!unibi_get_str(ut, str)) {
+  const char *temp = unibi_get_str(ut, str);


Why use temp here? Can't one just:

if (!unibi_get_str(ut, str)) { unibi_set_str(ut, str, val); return true; } return false;

Leftover of a debug session. Removed the temp variable. Kept returning the bool as it helps debug and should be optimized out by compiler anyway.

justinmk · 2017-04-22T16:36:58Z

src/nvim/tui/tui.c

+      RgbValue bg = aep->rgb_bg_color;
+
+      // if gui=reverse, swap bg and fg colors
+      if (aep->rgb_ae_attr == HL_INVERSE) {


should it be aep->rgb_ae_attr & HL_INVERSE ?

justinmk · 2017-04-22T20:11:08Z

In #6425 (comment) @zhou13 mentioned control sequence \x1b]112\x07 for "resetting to default cursor color (i.e., gui=reverse)".

teto · 2017-04-23T00:37:01Z

@zhou13 I can't find out what \x1b]112\x07 does ? In gvim gui=reverse swaps guifg and guibg. This PR changes the guibg of the cursor but not the guifg as it should (aka the text color underneath the cursor). Not sure how to properly do that on a terminal though.

I don't get why I have this error: terminal tests look ok for the rest :/

[  ERROR   ] 1 error, listed below:
[  ERROR   ] test/functional/terminal/tui_spec.lua @ 282: tui focus event handling can handle focus events in terminal mode
./test/functional/helpers.lua:252: 
retry() attempts: 3
./test/functional/ui/screen.lua:301: Row 1 did not match.
Expected:
  |*ready $                                           |
  |[Process exited 0]{1: }                               |
  |                                                  |
  |                                                  |
  |                                                  |
  |gained                                            |
  |{3:-- TERMINAL --}                                    |
Actual:
  |*nal                                               |
  |:terminal                                         |
  |{1: }                                                 |
  |{4:~                                                 }|
  |{4:~                                                 }|
  |gained                                            |
  |{3:-- TERMINAL --}                                    |

Thus I checked out terminals (I don't use the feature) and saw that in both master and this PR I get some (red) artifacts:

could that be linked ?

zhou13 · 2017-04-23T00:40:08Z

echo -ne '\e]12;#ff00ff\a'  # your cursor becomes purple.
echo -ne '\e]112\a'  # reset your cursor (gui=reverse)

tested in xfce4-terminal.

justinmk · 2017-04-23T00:56:12Z

Thus I checked out terminals (I don't use the feature) and saw that in both master and this PR I get some (red) artifacts:

What terminal? I didn't see that on iTerm.

Is the test failing consistently?

teto · 2017-04-23T01:17:07Z

Stupid me, it's just a plugin showing space at end of line. Thought it was linked to guicursor because I used to test with a red cursor.

The tests pass on master, but fail consistently for this PR (on travis too). Seems like feed_data(':terminal\n') is not run as it appears in output. Will debug this later, I am clueless right now :'(

jamessan · 2017-04-23T01:21:01Z

I can't find out what \x1b]112\x07 does ?

It's an Operating System Command (OSC). They're of the form OSC PS ; PT BEL.

Any terminal supporting these should respond to the PR. xterm does, if allowFontOps is enabled. Funnily, pangoterm doesn't yet.

The 10 colors (below) which may be set or queried using 1 0
through 1 9 are denoted dynamic colors, since the correspond-
ing control sequences were the first means for setting xterm's
colors dynamically, i.e., after it was started. They are not
the same as the ANSI colors. These controls may be disabled
using the allowColorOps resource. At least one parameter is
expected for Pt. Each successive parameter changes the next
color in the list. The value of Ps tells the starting point
in the list. The colors are specified by name or RGB specifi-
cation as per XParseColor.
...
Ps = 1 0 -> Change VT100 text foreground color to Pt.
Ps = 1 1 -> Change VT100 text background color to Pt.
Ps = 1 2 -> Change text cursor color to Pt.
Ps = 1 3 -> Change mouse foreground color to Pt.
Ps = 1 4 -> Change mouse background color to Pt.
...
The dynamic colors can also be reset to their default
(resource) values:
Ps = 1 1 0 -> Reset VT100 text foreground color.
Ps = 1 1 1 -> Reset VT100 text background color.
Ps = 1 1 2 -> Reset text cursor color.
Ps = 1 1 3 -> Reset mouse foreground color.
Ps = 1 1 4 -> Reset mouse background color.

teto

supporting the "gui=inverse" thing seems hard so I dropped it.
I haven't tested this much but cursor color seems to follow ctermbg with set notermguicolors.
It is possible to hide cursor with guibg=None.
guicursor is resent for each highlight_changed(). Less then optimal but cursor is updated properly.

The easiest way to deal with guicursor would be to send the full mode description on mode change (=stateless) but it was suggested that as this information doesn't change often, it can be stateful. Hence it is sent via mode_info_set. Sending guicursor colors in mode_change seems like a bad compromise between the 2 as cursor highlights shouldn't change much either.

UIs will need to retrieve highlights in all cases (because externalized components will need to match vim colors), so I believe letting the UIs subscribe to chosen highlight changes (such as the cursor's highlights) would be the best approach ? .
Meanwhile as a quickfix, I sent cursor colors along with guicursor info.
If this is blocking for 0.2 feel free to take over as I am not sure when I can complete this.

teto · 2017-04-27T20:08:00Z

src/nvim/api/private/helpers.c

+      } else {
+        attrs.background = aep->cterm_bg_color - 1;
+      }
+      ILOG("attr=%d  color to %d", attr_code, attrs.background);


todo remove

teto · 2017-04-27T20:09:58Z

src/nvim/syntax.c

+  }
+
+  return NULL;
+}


shouldn't this function along with cterm_color_name and color_numbers_X live in one of the terminal libraries ?

Not that I know of.

I do not know the scope of libtermkey or unibilium but this part should apply to all terminals so it might make sense to have it there.

teto · 2017-04-27T20:10:52Z

src/nvim/syntax_defs.h

 typedef struct attr_entry {
-  int16_t rgb_ae_attr, cterm_ae_attr;  // HL_BOLD, etc.
+  int16_t rgb_ae_attr, cterm_ae_attr;  ///< \ref HL_ATTRIBUTES
  RgbValue rgb_fg_color, rgb_bg_color, rgb_sp_color;
  int cterm_fg_color, cterm_bg_color;


if the hl_group retains the original cterm values, cterm_Xg_color could now be RgbValues too ?!

teto · 2017-04-27T20:11:27Z

src/nvim/tui/tui.c

@@ -91,13 +91,14 @@ typedef struct {
    int enable_mouse, disable_mouse;
    int enable_bracketed_paste, disable_bracketed_paste;
    int set_rgb_foreground, set_rgb_background;
-    int set_cursor_color;
+    int set_cursor_bg_color;


renamed for clarity as some terms (iTerm?) seem to have a set_cursor_fg_color

teto · 2017-04-27T20:11:50Z

src/nvim/tui/tui.c

@@ -482,6 +500,9 @@ static cursorentry_T decode_cursor_entry(Dictionary args)
      r.blinkoff = (int)value.data.integer;
    } else if (strequal(key, "hl_id")) {
      r.id = (int)value.data.integer;
+    } else if (strequal(key, "hl_dict")) {
+      r.attrs = decode_hl_entry(value.data.dictionary);
+      ILOG("decoded hl_dict with bg=%d", r.attrs.background);


teto · 2017-04-27T20:13:01Z

src/nvim/tui/tui.c

+    if (cursor_bg == -1) {
+      unibi_out(ui, unibi_cursor_invisible);
+    } else {
+      unibi_out(ui, unibi_cursor_normal);  // display if previously invisible


this call could be prevented with an additionnal check if we save the visibility status. worth it ?

It only happens on mode-change, so it's ok. Incidentally we do unibi_cursor_normal quite often, it's part of UI "flush".

In fact this call may not be needed at all, the next UI flush will show the cursor again.

teto · 2017-04-27T20:13:35Z

src/nvim/ui.c

-  api_free_array(style);
+  UI_CALL(mode_info_set, enabled, mode_style_array(ui->rgb));
+  // todo
+  // api_free_array(style);


with the new style I can't free the array.

justinmk · 2017-04-27T20:48:55Z

src/nvim/syntax.c

+  "Cyan", "LightCyan", "Red", "LightRed", "Magenta",
+  "LightMagenta", "Yellow", "LightYellow", "White", "NONE"
+};
+// terminals with less than 16 colors...


We don't care about them for 'guicursor' purposes. (Oh, I see you factored this out of do_highlight())

@mhinz asked to be able to change cursor via setting ctermbg. ctermbg=ColorName is straightforward but ctermbg=4 needed a function to convert the integer to the colorname.

justinmk · 2017-04-27T21:06:09Z

src/nvim/tui/tui.c

@@ -1060,7 +1093,7 @@ static void flush_buf(UI *ui, bool toggle_cursor)
  uv_buf_t buf;
  TUIData *data = ui->data;

-  if (toggle_cursor && !data->busy) {
+  if (toggle_cursor && !data->busy && cursor_bg != -1) {


Would be better to communicate this via the existing toggle_cursor. The only call that needs modification is the one in tui_flush.

justinmk · 2017-04-27T21:21:31Z

src/nvim/tui/tui.c

        ut, NULL, "\033]Pl%p1%06x\033\\");
  } else {
-    data->unibi_ext.set_cursor_color = (int)unibi_add_ext_str(
+    // new form rgb:X/X/X does not seem to work so well


outdated comment?

It's up to date but might be more relevant in the git log. rgb:X/X/X is supposed to work (see xparsecolor) but there are different forms such as rgb:XXX/XXX/XXX, rgb:XX/XX/XX and not all had an effect in termite. I would have to ask on vte tracker.

vte is supposed to convert the rgb:rrr/ggg/bbb (independent of the arity of each color) to the #rrrgggbbb form and then pass that on to pango_color_parse.

Older versions of vte used gdk_parse_color instead, but either one should handle the format fine.

If you're seeing different behavior, then I'd either guess it's an old version of vte or the terminal using vte is doing something wrong.

teto · 2017-04-30T00:22:02Z

Current code might be hard to follow as there is some dead code (I stopped implementing some functionality as it seemed overkill).
The idea was to have UIs register the highlight (=hl) groups (via nvim_ui_subscribe_to_highlights) for which they want to get notifications upon change. These subscriptions could be registered in ui->subscribed_highlights.
The patch tracks updated highlights in the global value (was easier for tests but it doesn't need to be global) changed_highlights. Then when highlight_changed() is called, ui_notify_changed_highlights sends the updated hl to the uis. In the end, I did not filter updates based on ui->subscribed_highlights (hence the dead code) as it seemed overkill, there are not so many hl updates after startup.
In the TUI code, tui_event checks if the updaded highlights can change the cursor appearance (see map_hlid_to_mode) in which case it refreshes the cursor.

A few interrogations:

having each UI check if the updated highlights should trigger a cursor update seems a waste. Core nvim should tell the UI that cursor might need to be redrawn, via a specialization of refresh/flush for cursor.
how for the UI to get highlights at startup ? current hl values could be sent right upon subcriptions nvim_ui_subscribe_to_highlights.
I am not sure the current code tracks all highlight changes. Sounds ok when changing on command line. A (better) API to deal with hl might help there.

teto · 2017-05-18T21:28:01Z

transparent cursor works in termite. cursor color can also now be set via ctermbg + notermguicolors. I've added tests for the highlight api (nvim_hl_*). Also nvim -u NORC -c ":hi Normal" now works as it should.
There are many things in the PR so I might polish a few things and split it into separate PRs.

bfredl · 2017-05-19T04:41:27Z

src/nvim/api/vim.c

+///
+/// @param highlights a list of highlights id (string or integer) to look for
+/// todo return array instead ? ArrayOf(HlAttrs)
+Object nvim_hl_get_list(Array highlights, Error *err)


if this is a batched version of the methods above then it is not needed (use call_atomic or lua)

bfredl · 2017-05-19T04:43:44Z

src/nvim/tui/tui.c

@@ -805,6 +841,16 @@ static void tui_set_icon(UI *ui, String icon)
 // to make a copy for the tui thread
 static void tui_event(UI *ui, char *name, Array args, bool *args_consumed)


this shouldn't be used. In fact ui_event impl should be just deleted from ui_bridge and tui. (the infrastructure checks for funptr being NULL, so this is safe). Instead, mark events as not REMOTE_ONLY to have them acessible from TUI.

justinmk · 2017-05-21T11:03:27Z

src/nvim/syntax.c

-    for (int i = 1; i <= highlight_ga.ga_len && !got_int; ++i)
-      /* TODO: only call when the group has attributes set */
+    for (int i = 0; i < highlight_ga.ga_len && !got_int; i++) {
+      // todo: only call when the group has attributes set


TODO(vim):

justinmk · 2017-05-21T11:05:07Z

src/nvim/syntax.c


 static bool did_syntax_onoff = false;
+static int invalid_group_id = -1;


should be either a constant named kFoo or a define named FOO.

I renamed it to kInvalidGroupId.

justinmk · 2017-05-21T11:08:05Z

src/nvim/syntax.c

-/*
- * Translate a group ID to the final group ID (following links).
- */
+bool hl_invalid_id(int hl_id)


FUNC_ATTR_PURE. can it also be static?

same for hl_is_valid().

I added the FUNC_ATTR_PURE. I renamed hl_is_valid to hl_valid_id. I think it makes sense to keep one function available (either hl_invalid_id or hl_valid_id) for other units for instance it is used in tui.c. I can inline or remove the 2nd one.

justinmk · 2017-05-21T11:17:38Z

src/nvim/api/ui_events.in.h

@@ -58,6 +58,8 @@ void set_title(String title)
    FUNC_API_SINCE(3);
 void set_icon(String icon)
    FUNC_API_SINCE(3);
+void highlights_changed(Array highlights_changed)


The naming precedent is "set", we don't have "changed" (though we do have "change"... oh dear). We already have highlight_set, which is a UI change, whereas this new event is a property change.

Kind of a mess. Naming this highlight_info_set at least is analogous to mode_info_set. We can use "info" as the pattern for "property changes".

Indeed "set" is better. The UI event stream tells the current state, not whether something changed or not

justinmk · 2017-05-21T11:27:20Z

1/ use a different value for a NONE color and for an unset guibg (right now, both use -1 making the distinction impossible between the two. I propose to use -2 for NONE and -1 for unset)
2/ use a cursor_on/cursor_off ui event as it already exists for the mouse (and these functions seem to exist as well in vim).

@teto No objection to 1/, but 2/ sounds like the UI is making decisions to work around TUI limits.

In general I suggest: give UIs all relevant info and let the UIs make the decisions. If TUI sees that guifg=bg it can decide to hide the cursor. It does that by inspecting highlight_info_set.

bfredl · 2017-05-21T17:12:36Z

If TUI sees that guifg=bg it can decide to hide the cursor.

I think it is better to encode explicitly that the cursor should be hidden, one would expect this to happen consistently in both TUI and GUI, for instance by using -2 like option 1.

teto · 2017-05-21T18:45:08Z

I went with option 1 as it was just a few lines of code. Setting guibg=NONE now translates to -2 in the datastructure instead of -1. The PR now seems to behave as intended. I'll clean up a bit, try to split it into 2 PRs as the changes to hl group index is not related and might be seen as potentially harmful.

justinmk · 2017-05-21T18:58:14Z

runtime/doc/options.txt

-			for the cursor
+			for the cursor. If an invalid group is set, the TUI won't change colors
+      contrary to vim. If the group has guibg=NONE then the cursor becomes
+      invisible.


did you mean UI instead of TUI? (We can't control what GUIs do, but we can add to the guidelines at :help dev-ui which should be considered the "official UI behavior").

Can elide "contrary to vim". (mention it at vim_diff.txt if you want to call out contrary-to-vim things)

justinmk · 2017-05-21T19:02:18Z

src/nvim/tui/tui.c

@@ -474,7 +474,7 @@ CursorShape tui_cursor_decode_shape(const char *shape_str)
 // if the cursor should be visible according to guicursor
 static bool cursor_visible(void)
 {
-  return cursor_bg != -1;
+  return cursor_bg != -2;


kColorFoo ?

justinmk · 2017-05-21T19:06:06Z

src/nvim/ui.c

@@ -142,6 +142,104 @@ void ui_event(char *name, Array args)
  }
 }

+bool attr2hlattr(int attr_code, bool use_rgb, HlAttrs *out)


sorry, I should have clarified. These functions should live in syntax.c (wasn't that where they originally were?)

jdebp · 2017-05-22T19:50:48Z

src/nvim/tui/tui.c

  }
+  return false;


This is rather pointless, since nothing uses the return value; and for what it's worth I have something in the works that I hope will do away with quite a number of calls to this function.

yevhen-m · 2017-05-31T07:51:48Z

Guys, so cool that you are working on this, thanks a lot!

teto · 2017-06-10T00:21:13Z

Trying to write some tests to check for certain sequences such as \x1b]0 q IIRC. In some PR @justinmk you wrote:

You can test terminal responses using a :terminal and hooking up a on_stdout handler. We have some tests doing that. This would at least verify that for a given $TERM we send/receive expected sequences.

Any precise test in mind ? I've been lookling at job_spec.lua and do sthg like:

    helpers.source([[
    function! s:OnEvent(id, data, event) dict
      let userdata = get(self, 'user')
      let data     = a:data
      call rpcnotify(g:channel, a:event, userdata, data)
    endfunction
    let g:job_opts = {
    \ 'on_stdout': function('s:OnEvent'),
    \ 'on_stderr': function('s:OnEvent'),
    \ 'on_exit': function('s:OnEvent'),
    \ 'pty': 1,
    \ 'user': 0
    \ }
    ]])

-- feedkeys('itoto') should enter insert mode so that I can generate guicursor for insert mode and check for the presence of matching CSI
      nvim('command', "let j = jobstart([ '".. helpers.nvim_prog.."', '-u', 'NONE', '-i', 'NONE', '--cmd', 'set noswapfile','--cmd', 'call feedkeys(\"itoto\")' ], g:job_opts)")
    end

    local msg = next_msg()
      print(msg[3][0])
      local line = msg[3][2]
    -- here I would like to check if the cursor change sequence is present, with some regex for instance:
    -- line.find("]3 q\007toto")

    -- eq({'notification', 'stdout', {0, {'abc', ''}}}, )

my trouble is that the on_stdout seems to be called many times, not sure against which line to check or how many times I should call next_msg (if incorrect, it leaves the test pending).

teto

I am a bit at a loss as how to use jobsend to send input to a background nvim.
I would like for instance to insert "toto" and then go back to normal mode.
So I send jobsend(j, "itoto\<c-c>:q!") but the on_stdout callback isn't called for "toto", just for <c-c>:q!.
then I try sthg like jobsend(j, "itoto\<c-r>\<c-l>:q!")
and it gives me 2 lines of output or is it the same line redrawn ? why the \13 ?

{
  "notification",
  "onout",
  {
    {
      "itoto^R\13",
      "itoto^L:q!"
    }
  }
}

This is still WIP but to see the behavior, one can run
LANG=C TEST_FILE=test/functional/terminal/tui_spec.lua make functionaltest 2>&1 >log && nvim log

teto · 2017-06-15T14:59:02Z

test/functional/terminal/tui_spec.lua

+    local pattern = shape_ver_noblink.."(.*)toto"
+    -- :append (line("."), "zaza")
+    -- :append (line(\".\"), \"zaza\")
+    gen('n-v-c-sm:block,i-ci-ve:ver25,r-cr-o:hor20', [[itoto\e:q!]], pattern)


bfredl · 2017-06-19T06:30:41Z

src/nvim/api/vim.c

@@ -704,12 +756,21 @@ void nvim_unsubscribe(uint64_t channel_id, String event)
  channel_unsubscribe(channel_id, e);
 }

+/// Translate to integer if \p name is an hex value (e.g. #XXXXXX),
+/// else look into \p color_name_table to translate a color name to  its


color_name_table sounds like an internal datastructure. Also the descriptions of nvim_get_color_map sounds like they use two different color tables.

bfredl · 2017-06-19T06:37:08Z

src/nvim/po/sv.po

@@ -5,7 +5,7 @@ msgid ""
 msgstr ""
 "Project-Id-Version: Vim 7.1\n"
 "Report-Msgid-Bugs-To: \n"
-"POT-Creation-Date: 2014-07-02 21:54+0200\n"


Normally we don't update po files in unrelated PRs?

mistake on my side.

chrisduerr · 2018-05-18T14:12:48Z

Is this still being worked on @teto? I'm interested in reset cursor for VTE (seems to be a VTE extension printf "\x1b[0 q").

If this PR is not active anymore, maybe it's possible to pick out some of the features and implement them separately? I'd be happy to help getting the VTE cursor reset working.

teto · 2018-05-18T14:31:21Z

@chrisduerr I promised some users to get the transparent cursor working but the lack of dynamic update when changing highlights was a pain so I wanted to solve it first in #8006. For many reasons, I can't work on neovim at the moment so please go ahead. I would
be very happy to see the reset behavior as a VTE-based terminal user. The code has changed much since the PR so you might be better off starting from scratch. Here is some place where the reset occurs https://github.com/neovim/neovim/pull/6566/files#diff-3ac88981a139ac0eee448a6bd19c7569R198

chrisduerr · 2018-05-18T14:46:51Z

Thanks for the quick response. I'm not sure if I can get to it this weekend, but I'll definitely look into it when I've got some time and this PR is still open. :)

teto · 2018-05-18T14:52:51Z

The PR is likely to stick for a few more months/years xD feel free to ask if you have any question.

teto · 2019-12-05T13:39:42Z

this is very old and almost everything has been implemented apart from the transparent TUI cursor. see #11519 for that

marvim added the WIP label Apr 21, 2017

brcolow reviewed Apr 21, 2017

View reviewed changes

justinmk reviewed Apr 22, 2017

View reviewed changes

teto mentioned this pull request Apr 22, 2017

UI: mode_info_set (cursor) enhancements #6428

Closed

teto force-pushed the tuicursor branch from 5e1614d to 2a65609 Compare April 23, 2017 00:34

teto force-pushed the tuicursor branch from ef280a5 to 35cfc86 Compare April 27, 2017 20:07

teto commented Apr 27, 2017

View reviewed changes

justinmk reviewed Apr 27, 2017

View reviewed changes

teto force-pushed the tuicursor branch from bbc8bb8 to 1a8818f Compare May 9, 2017 23:53

This was referenced May 13, 2017

Make the 'Normal' highlight a normal highlight #6738

Closed

synIDattr() return ctermbg even if &ctermguicolors == 1. #6633

Closed

teto force-pushed the tuicursor branch from 1a8818f to bcb95dc Compare May 15, 2017 23:30

teto closed this May 18, 2017

teto deleted the tuicursor branch May 18, 2017 21:18

jszakmeister removed the WIP label May 18, 2017

teto reopened this May 18, 2017

teto changed the title ~~[WIP] tui: support gui=inverse and transparent cursor~~ [WIP] support transparent & ctermbg cursor + minimal highlight api May 18, 2017

bfredl reviewed May 19, 2017

View reviewed changes

justinmk reviewed May 21, 2017

View reviewed changes

reviews

be6fb6e

justinmk reviewed May 21, 2017

View reviewed changes

teto mentioned this pull request May 22, 2017

TUI: cursor shape not correctly initialized and restored #4396

Open

jdebp reviewed May 22, 2017

View reviewed changes

frangio mentioned this pull request May 22, 2017

incsearch not working properly with nvim haya14busa/incsearch.vim#79

Open

Test starts looking good

5247929

teto commented Jun 15, 2017

View reviewed changes

marvim added the WIP label Jun 18, 2017

bfredl reviewed Jun 19, 2017

View reviewed changes

teto mentioned this pull request Jul 6, 2017

[RDY]:hi Normal works with -u NONE #6973

Merged

teto mentioned this pull request Oct 24, 2017

tui: restore cursor at exit #7393

Closed

teto mentioned this pull request Feb 13, 2018

[RFC] ui: notify changed highlights #8006

Closed

teto closed this Dec 5, 2019

		@@ -805,6 +841,16 @@ static void tui_set_icon(UI *ui, String icon)
		// to make a copy for the tui thread
		static void tui_event(UI ui, char name, Array args, bool *args_consumed)


		static bool did_syntax_onoff = false;
		static int invalid_group_id = -1;

Uh oh!

[WIP] support transparent & ctermbg cursor + minimal highlight api #6566

[WIP] support transparent & ctermbg cursor + minimal highlight api #6566

Uh oh!

Conversation

teto commented Apr 21, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

teto Apr 22, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

justinmk commented Apr 22, 2017

Uh oh!

teto commented Apr 23, 2017

Uh oh!

zhou13 commented Apr 23, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

justinmk commented Apr 23, 2017

Uh oh!

teto commented Apr 23, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jamessan commented Apr 23, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

teto left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

justinmk Apr 27, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

justinmk Apr 27, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

justinmk Apr 27, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

teto commented Apr 21, 2017 •

edited

Loading

teto Apr 22, 2017 •

edited

Loading

zhou13 commented Apr 23, 2017 •

edited

Loading

teto commented Apr 23, 2017 •

edited

Loading

jamessan commented Apr 23, 2017 •

edited

Loading

teto left a comment •

edited

Loading

justinmk Apr 27, 2017 •

edited

Loading

justinmk Apr 27, 2017 •

edited

Loading

justinmk Apr 27, 2017 •

edited

Loading

teto commented Apr 30, 2017 •

edited

Loading

teto commented May 18, 2017 •

edited

Loading

justinmk May 21, 2017 •

edited

Loading

justinmk May 21, 2017 •

edited

Loading

justinmk May 21, 2017 •

edited

Loading