* lisp/progmodes/eglot.el (eglot-server-programs): Add
'vscode-markdown-language-server' for Markdown and
'dot-language-server' for GraphViz. (Bug#62844)
The DEFUN is hard to incorporate because it's made of two nodes rather
than one, and most tree-sitter functionalities assume a defun is one
node. I fixed the indent-defun and add-log functionality, but
beginning/end-of-defun and imenu still don't recognize DEFUN.
* lisp/progmodes/c-ts-mode.el (c-ts-mode-emacs-devel): New variable.
(c-ts-mode--defun-name): Support DEFUN.
(c-ts-mode--defun-valid-p): Support DEFUN.
(c-ts-mode--emacs-defun-p)
(c-ts-mode--emacs-defun-at-point): New functions.
(c-ts-mode-indent-defun): Use c-ts-mode--emacs-defun-at-point.
(c-ts-mode--emacs-current-defun-name): New function.
(c-ts-mode, c++-ts-mode): Optionally setup custom defun-name function.
This was supposed to be a cherry-pick from master, but
isn't, due to the unnecessary rush to fix master without
cleaning up the mess first and without separating changes
that can and cannot be backported.
* src/treesit.c (treesit_traverse_cleanup_cursor): Fix indentation
style.
(Ftreesit_search_subtree, Ftreesit_search_forward)
(Ftreesit_induce_sparse_tree): Fix specpdl unwinding. (Bug#62823)
Do not merge to master.
Earlier we switched to using cursors rather than nodes to traverse the
parse tree. Because cursors need cleanup, we have to catch signals
thrown by the predicate functions and free the cursor. Failing to do
this will result in leaking the cursor whenever the predicate function
signals in a search function. This change fixes the leak.
* src/treesit.c (treesit_traverse_cleanup_cursor): New function.
(Ftreesit_search_subtree)
(Ftreesit_search_forward)
(Ftreesit_induce_sparse_tree): Catch signals. (Bug#62823)
(cherry picked from commit a5eb9f6ad4)
* lisp/progmodes/ruby-mode.el (ruby-add-log-current-method):
Reduce the use of 'nreverse' (bug#62761).
* test/lisp/progmodes/ruby-mode-tests.el
(ruby-add-log-current-method-singleton-referencing-outer):
New test.
* doc/emacs/package.texi (Specifying Package Sources): List the
accepted keys in a new subsection of Fetching Package Sources.
* lisp/emacs-lisp/package-vc.el (package-vc-selected-packages):
Mention the `:doc` key. Add the `:doc` key to the Customize form,
mention the new Info node, correct "TexInfo" to "Texinfo", avoid
Git-specific terms for the description of `:branch`, mention guessing
`:vc-backend` based on the URL.
* lisp/vc/vc.el (vc-checkout): Try to use the vc-dir's backend
before looking for the file's one. The latter doesn't work when
the file doesn't exist on disk yet (bug#62674).
* doc/lispref/sequences.texi (Sequence Functions):
Remove inaccurate and over-specific claims about how `sort` works for
lists: there is no guarantee that it doesn't modify the `car` fields
of the input list (which is precisely what it does at this time).
(cherry picked from commit c753a95923)
* test/lisp/image/wallpaper-tests.el (wallpaper--find-setter)
(wallpaper--find-setter/call-predicate)
(wallpaper--find-setter/set-current-setter)
(wallpaper-set/calls-init-action)
(wallpaper-set/calls-wallpaper-set-function)
(wallpaper-set/runs-command, wallpaper-set/runs-command/detach):
Skip these tests if native functions are used for changing
wallpaper. These tests are irrelevant in that case.
* test/lisp/progmodes/eglot-tests.el
(eglot-test-rust-analyzer-watches-files): Bump timeout.
(eglot-test-json-basic): Check for yas-minor-mode before using it,
like other tests do (bug#61637).
We cannot call 'sh-set-shell' inside 'sh-base-mode', since various
settings of 'sh-mode', in particular the syntax table, is not yet
set. Likewise with various hooks: since 'sh-base-mode' is not a
mode any file should be visited with, it makes no sense to set up
stuff like 'hack-local-variables-hook' in 'sh-base-mode'; it
should be set in the descendant modes instead.
* lisp/progmodes/sh-script.el (sh-base-mode): Move the call to
'sh-set-shell' from here...
(sh-mode): ...to here...
(bash-ts-mode): ...with a copy here. (Bug#62748)
* lisp/progmodes/c-ts-mode.el (c-ts-base--before-indent): Try to
guess when the parse tree is incomplete, and provide a better node
to indent against (bug#62717).
(c-ts-base-mode): Set up advice for local treesit-indent-function.
The functionality under test in eglot.el is exactly the same, but use
the clangd server only, as that is used in more tests, and it is much
easier to check if it misbehaves or not.
Tests pass with clangd version 15.
* test/lisp/progmodes/eglot-tests.el (python): Don't require it.
(eglot--call-with-fixture): Simplify.
(eglot--wait-for-clangd): New helper.
(eglot-test-basic-completions)
(eglot-test-non-unique-completions, eglot-test-basic-xref)
(eglot-test-snippet-completions)
(eglot-test-snippet-completions-with-company)
(eglot-test-eldoc-after-completions, eglot-test-multiline-eldoc):
Use clangd, not pylsp.
(eglot-test-formatting): Renamed from
eglot-test-python-autopep-formatting.
(eglot-test-python-yapf-formatting): Remove.
(cherry picked from commit fc8230f336)
* lisp/mouse.el (context-menu-region): Make the entries be
symbols, not just integers, to prevent "C-h c" from interpreting
them as character keys. (Bug#62626)
* doc/lispref/frames.texi (Drag and Drop): Rephrase and rearrange
the documentation of XDS support. Add indexing. Document
'x-dnd-save-direct' and 'x-dnd-save-direct-immediately'. Original
patch from Po Lu <luangruo@yahoo.com>.
* lisp/x-dnd.el (x-dnd-types-alist, x-dnd-test-function)
(x-dnd-default-test-function, x-dnd-direct-save-function): Doc
fixes.
(x-dnd-save-direct, x-dnd-save-direct-immediately): Rename the
second argument to FILENAME. Doc fix.
* lisp/gnus/mm-view.el (mm-inline-image): Use "x" as text on which
to put the image 'display' property, so that indentation commands
don't destroy the image inadvertently. (Bug#62637)
* src/macfont.m (CG_SET_FILL_COLOR_WITH_FRAME_CURSOR): New macro.
(CG_SET_FILL_COLOR_WITH_FRAME_BACKGROUND): New macro.
(macfont_draw): When the cursor's color matches the face background,
set the fill color of the cursor to the face foreground.
* src/nsterm.m (ns_maybe_dumpglyphs_background): When dumping the
background of a glyph string, apply the logic mentioned
above. (Bug#62573)
* lisp/image.el (image-map): Make it inherit from
'image-slice-map' instead of repeating the bindings.
* doc/emacs/files.texi (Image Mode): Document the key bindings set
by 'insert-image'. Add indexing.
* doc/lispref/display.texi (Showing Images): Make the description
of user commands more concise. Add index entries and
cross-reference to the Emacs manual.
* etc/NEWS: Rearrange entries relevant to image commands.
* lisp/image.el (image-slice-map): New keymap, without some
bindings that make no sense with sliced images.
(insert-image, insert-sliced-image): Use it.
(insert-sliced-image): Make the 'keymap' property rear-nonsticky,
to prevent calling image commands when point is to the right of
the slice. (Bug#62679)
* lisp/image/image-crop.el (image-cut, image-crop): Doc fixes.
(image-crop): Don't try using stock MS-Widows convert.exe
program. Use 'image--get-image' to support sliced images.
* doc/misc/efaq.texi (Fullscreen mode on MS-Windows): Tell how to
avoid the initial screen changing size on display without using
the Registry. Suggested by David Hedlund <public@beloved.name>.
Related to https://github.com/joaotavora/eglot/discussions/1202, where
because of this very wide binding to inhibit-read-only to t, Eglot was
managing to write into read-only buffers from the response handlers
that ran from within the stack of the jsonrpc.el process filter.
This is of course illegal and dangerous, but Eglot wasn't made aware
because of the binding.
* lisp/jsonrpc.el (jsonrpc--process-filter): Don't bind
inhibit-read-only so early.
* lisp/international/quail.el (quail-minibuffer-message): Bind
deactivate-mark to allow setting active region in the minibuffer
when an input method is used. (Bug#62709)
