* lisp/treesit.el (treesit-forward-sexp): Check if point is strictly
inside a comment or string, only then use the default forward-sexp
function; otherwise use tree-sitter's forward-sexp routine.
Sorry for sneaking in a sizable commit so late. But I just found out
about this bug and it has to be fixed. Before this change, we weren't
properly cleaning up overlays that store local parsers. And in the case
of doxygen local parser in C files, the doxygen local parser overlay
sometimes bleeds out of comments and into other code, and interferes
with font-lock and indentation.
This commit adds a cleanup function that'll cleanup any overlays that
aren't being used. I tested with doxygen in C files and everything
works smoothly now, including tricky tests like removing the ending "*/"
of a doxygen comment and adding it back.
The idea is simple, at the end of each call to (treesit-update-ranges
BEG END), we remove any overlay within BEG and END that wasn't touched
by the range setting code.
* lisp/treesit.el (treesit--cleanup-local-range-overlays): New function.
(treesit--update-ranges-local): Remove code for cleaning up zero-length
overlays since we have the cleanup function now.
(treesit-update-ranges): Wrap the function body inside a let form, which
defines modified-tick; and add a call to
treesit--cleanup-local-range-overlays at the very end.
* src/xfaces.c (recompute_basic_faces): Revert the change which
caused recalculation of all the faces, as it made cursor motion
too slow. Reported by Juri Linkov <juri@linkov.net> (bug#72692).
This also affects the listing of `should' forms produced by hitting
the L key on a test button in an ERT buffer.
* lisp/emacs-lisp/ert.el (ert--pp-with-indentation-and-newline):
Indent the pretty-printed result to match the caller's current column
as a reference indentation.
* test/lisp/emacs-lisp/ert-tests.el
(ert--pp-with-indentation-and-newline): New test. (Bug#72561)
* lisp/progmodes/eglot.el (eglot--dumb-tryc): Check that the
expanded string matches every completion strictly (bug#72705).
And in the fallback case, check whether the table matches the
original prefix at all. Return nil otherwise.
* test/lisp/progmodes/eglot-tests.el
(eglot-test-stop-completion-on-nonprefix)
(eglot-test-try-completion-nomatch): Corresponding tests.
* etc/EGLOT-NEWS: New entry.
* test/lisp/progmodes/eglot-tests.el
(eglot-test-common-prefix-completion)
(eglot-test-try-completion-inside-symbol)
(eglot-test-rust-completion-exit-function): New tests.
(eglot--wait-for-rust-analyzer): New function.
* lisp/emacs-lisp/checkdoc.el (checkdoc-next-docstring): Use
'beginning-of-defun-raw' instead of 'beginning-of-defun', as the latter
always moves back to beginning of line and thus is not guaranteed to
advance point when 'open-paren-in-column-0-is-defun-start' is non-nil.
(Bug#72759)
* src/minibuf.c (minibuffer_unwind): Make sure that the buffer
referenced by the first element of the list of previous buffers
of the minibuffer window is live before assigning it to the
minibuffer window (Bug#72487).
* src/window.c (set_window_buffer): Assert that BUFFER is live.
This is a change specific to emacs-30. Don't merge to master.
* lisp/progmodes/eglot.el (Version): Mark it 1.17.30.
* etc/EGLOT-NEWS (1.17.30): Fill in section.
* src/sfnt.c (sfnt_read_cmap_format_2): Properly compute
subtable count, and append the empty table at position 0.
(sfnt_lookup_glyph_2): Update commentary.
* src/fns.c (Fsecure_hash):
* doc/lispref/text.texi (Checksum/Hash):
* lisp/subr.el (sha1): Fix documentation wrt to the number of
bytes 'secure-hash' and its variants return when BINARY is
non-nil. Reported by Pip Cet <pipcet@protonmail.com>.
