Update docs

docs/src/features.md

@@ -174,7 +174,10 @@ Inlay hints are disabled by default and can be enabled with the `"show_inlay_hin
 
 Inline completions are typically provided by an AI code assistant.
 They can span multiple lines and are rendered directly in the source code as grayed out text ("ghost text").
-Currently inline completions are only requested when you manually trigger auto-completions (<kbd>Ctrl</kbd> + <kbd>Space</kbd>).
+
+!!! note
+    Currently inline completions are only requested when you manually trigger auto-completions (<kbd>Ctrl</kbd> + <kbd>Space</kbd>).
+    Inline completions are disabled if you have enabled `"mini_auto_complete"`.
 
 ## Server Commands
 

docs/src/keyboard_shortcuts.md

@@ -8,6 +8,7 @@ Refer to the [Customization section](customization.md#keyboard-shortcuts-key-bin
 | Feature | Shortcut | Command |
 | ------- | -------- | ------- |
 | Auto Complete | <kbd>ctrl</kbd> <kbd>space</kbd> (also on macOS) | `auto_complete`
+| Commit Inline Completion | <kbd>alt</kbd> <kbd>enter</kbd> | `lsp_commit_inline_completion`
 | Expand Selection | unbound | `lsp_expand_selection`
 | Find References | <kbd>shift</kbd> <kbd>f12</kbd> | `lsp_symbol_references`<br>Supports optional args: `{"include_declaration": true | false, "output_mode": "output_panel" | "quick_panel"}`.<br>Triggering from context menus while holding <kbd>ctrl</kbd> opens in "side by side" mode. Holding <kbd>shift</kbd> triggers opposite behavior relative to what `show_references_in_quick_panel` is set to.
 | Fold | unbound | `lsp_fold`<br>Supports optional args: `{"strict": true/false}` - to configure whether to fold only when the caret is contained within the folded region (`true`), or even when it is anywhere on the starting line (`false`).
@@ -28,13 +29,15 @@ Refer to the [Customization section](customization.md#keyboard-shortcuts-key-bin
 | Next Diagnostic | unbound | `lsp_next_diagnostic`
 | Previous Diagnostic | unbound | `lsp_prev_diagnostic`
 | Rename | unbound | `lsp_symbol_rename`
+| Request Inline Completions | unbound | `lsp_inline_completion`
 | Restart Server | unbound | `lsp_restart_server`
 | Run Code Action | unbound | `lsp_code_actions`
 | Run Code Lens | unbound | `lsp_code_lens`
 | Run Refactor Action | unbound | `lsp_code_actions`<br>With args: `{"only_kinds": ["refactor"]}`.
 | Run Source Action | unbound | `lsp_code_actions`<br>With args: `{"only_kinds": ["source"]}`.
 | Save All | unbound | `lsp_save_all`<br>Supports optional args `{"only_files": true | false}` - whether to ignore buffers which have no associated file on disk.
 | Show Call Hierarchy | unbound | `lsp_call_hierarchy`
+| Show next Inline Completion | unbound | `lsp_next_inline_completion`
 | Show Type Hierarchy | unbound | `lsp_type_hierarchy`
 | Signature Help | <kbd>ctrl</kbd> <kbd>alt</kbd> <kbd>space</kbd> | `lsp_signature_help_show`
 | Toggle Diagnostics Panel | <kbd>ctrl</kbd> <kbd>alt</kbd> <kbd>m</kbd> | `lsp_show_diagnostics_panel`

docs/src/language_servers.md

@@ -26,13 +26,37 @@ The following steps describe a local installation on a Windows PC with compatibl
 
 2. Download and extract a CUDA version of Tabby from the [GitHub releases page](https://github.com/TabbyML/tabby/releases) (click on "Assets"); e.g. `tabby_x86_64-windows-msvc-cuda122.zip`
 
-3. Install the `tabby-agent` language server via npm (requires NodeJS):
+3. On macOS and Linux it might be necessary to change the access permissions of `lama-server` and `tabby` to be executable:
+
+    ```sh
+    $ tabby_aarch64-apple-darwin chmod +x llama-server
+    $ tabby_aarch64-apple-darwin chmod +x tabby
+    ```
+
+    !!! note "On macOS you might get an error that “tabby” cannot be opened because it is from an unidentified developer."
+        After changing the permission to executable, right click on `tabby` and select "Open", that will get rid of the error.
+
+4. Download a completion model (see <https://tabby.tabbyml.com/docs/models/> for available model files and GPU requirements):
+
+    ```sh
+    tabby download --model StarCoder-1B
+    ```
+
+5. If necessary, edit the configuration file under `~/.tabby-client/agent/config.toml`, which is generated automatically on the first start of tabby-agent.
+   For example, to disable anonymous usage tracking add
+
+    ```toml
+    [anonymousUsageTracking]
+    disable = true
+    ```
+
+6. Install the `tabby-agent` language server via npm (requires NodeJS):
 
     ```sh
     npm install -g tabby-agent
     ```
 
-4. Open `Preferences > Package Settings > LSP > Settings` and add the `"tabby"` client configuration to the `"clients"`:
+7. Open `Preferences > Package Settings > LSP > Settings` and add the `"tabby"` client configuration to the `"clients"`:
 
     ```jsonc
     {
@@ -49,29 +73,15 @@ The following steps describe a local installation on a Windows PC with compatibl
     }
     ```
 
-5. Download a completion model (see <https://tabby.tabbyml.com/docs/models/> for available model files and GPU requirements):
-
-    ```sh
-    tabby download --model StarCoder-1B
-    ```
-
-6. If necessary, edit the configuration file under `~/.tabby-client/agent/config.toml`, which is generated automatically on the first start of tabby-agent.
-   For example, to disable anonymous usage tracking add
-
-    ```toml
-    [anonymousUsageTracking]
-    disable = true
-    ```
-
-7. Manually start the Tabby backend:
+8. Manually start the Tabby backend:
 
     ```sh
     tabby serve --model StarCoder-1B --no-webserver
     ```
 
     The language server communicates with this backend, i.e. it needs to be running in order for `tabby-agent` to work.
 
-8. Now you can open a file in Sublime Text and start coding.
+9. Now you can open a file in Sublime Text and start coding.
    Inline completions are requested when you manually trigger auto-complete via <kbd>Ctrl</kbd> + <kbd>Space</kbd>.