* test/lisp/comint-tests.el (comint-tests/test-password-function):
On MS-Windows, call 'w32-native-executable-find' instead of
'executable-find', to find the native version of the 'cat'
program.
(w32-native-executable-p, w32-native-executable-find): New
functions.
This refactor simplifies Eglot's source code.
* lisp/progmodes/eglot.el:
(Package-Requires): Require external-completion.
(external-completion): Require it.
(xref-backend-identifier-completion-table): Use external-completion-table.
(completion-category-overrides): No longer set it.
("Backend Completion"): Delete section.
To more closely emulate ido-mode, fido-mode and fido-vertical-mode
strongly lean to the 'flex completion style. But overriding
completion-category-defaults is not the correct way to do it, since
sometimes even the 'flex style isn't possible.
* lisp/icomplete.el (icomplete--fido-mode-setup): Don't set
completion-category-defaults here.
This table uses a new 'external' completion style which has been in
use in Eglot and other third-party extensions for a long time. It's
meant for libraries that interface with an external tool which
provides completions, such as a shell utility, an inferior process, an
http server.
The new file lisp/external-completion.el provides a helper function,
external-completion-table which is given an interfacing function to
the external tool and returns a "programmed completion" table that is
bound to the 'external' completion style. That table can then be used
with completing-read or any other facility expecting a completion
table.
When the 'external' is in use, the usual styles configured by the user
or other in 'completion-styles' are completely overriden. This
relatively minor inconvenience is the price to pay for responsive
completion where the full set of completion candidates doesn't need to
be transfered into Emacs's address space.
* lisp/external-completion.el: New file.
* etc/NEWS: (Lisp Changes in Emacs 29.1): Mention
external-completion-table.
When Eglot receives diagnostics for a file not yet visited in Emacs,
it stores them in flymake-list-only-diagnostics, which feed M-x
flymake-show-project-diagnostics. If the file is eventually visited
in a buffer and Eglot starts receibing diagnostics for it, the
flymake-list-only-diagnostics database has to be updated accordingly,
since the previous diagnostics are now stale.
* lisp/progmodes/eglot.el (eglot-handle-notification): Reset
flymake-list-only-diagnostics
Besides the problem mentioned by Juri, which is fixed by removing
-cp tree-sitter-lang.in "tree-sitter-${lang}/src"
-cp emacs-module.h "tree-sitter-${lang}/src"
(we removed those files in an earlier commit, because they are not
used anymore.)
Now it also more parameterized and builds typescript and tsx
separately.
* admin/notes/tree-sitter/build-module/build.sh (topdir)
(repo, sourcedir, grammardir): New variables.
(Build): Make it parametric.
(Copy out): Use absolute path.
* admin/notes/tree-sitter/build-module/batch.sh (languages): Add tsx.
* lisp/progmodes/c-ts-mode.el (c-ts-mode--keywords): Remove some keywords.
(c-ts-mode--type-keywords): New variables.
(c-ts-mode--font-lock-settings): New rule.
* lisp/emacs-lisp/subr-x.el (add-display-text-property): Fix the
case where OBJECT is not nil. (Bug#59857)
* test/lisp/emacs-lisp/subr-x-tests.el
(subr-x-test-add-display-text-property): Add test for this case.
* test/infra/gitlab-ci.yml (.filenotify-gio-template): Do not run
scheduled.
(build-native-comp-speed1, build-native-comp-speed2): Comment out.
* test/lisp/international/ucs-normalize-tests.el
(ucs-normalize-part1): Skip also on EMBA.
* src/lread.c (read_bool_vector, skip_lazy_string):
Replace `|` with `||` to explicitly introduce sequence points since
the variables, `length` and `nskip`, are mutated more than once.
The `|` was just a weak attempt at micro-optimisation in any case;
sorry about that.
Before this change Ftreesit_query_capture doesn't convert character
position to byte position for BEG and END parameters. I observed
fontification issue in css files but couldn't figure out why, now I
know :-)
I decide to keep treesit--font-lock-query-expand-range, since it might
provide a escape hatch for problems we discover in the future, and it
should be very cheap so no downside of keeping it.
* lisp/textmodes/css-mode.el (css-ts-mode): Stop setting
treesit--font-lock-query-expand-range.
* lisp/treesit.el (treesit--font-lock-query-expand-range): Update
docstring.
* src/treesit.c (Ftreesit_query_capture): Convert BEG and END to byte
position. Also added parentheses wround "beg_byte - visible_beg" in
the call to ts_query_cursor_set_byte_range (i.e., style change).
- Highlight property and field definitions in class declaration.
- Highlight property-keys in object initializaters.
- Highlight variable-names in function/method return-statements.
* lisp/progmodes/typescript-ts-mode.el
(typescript-ts-mode--font-lock-settings): Add fontification rules.
* lisp/progmodes/json-ts-mode.el (json-ts-mode--font-lock-settings):
Add back in removed pair feature. Also alphabetize features.
(json-ts-mode): Use the new feature.
* doc/lispref/display.texi (Display Margins): More accurate
explanation of how to show text in the display margin without
concealing buffer text. (Bug#59841)
This also fixes fontification problem with c-ts-mode--fontify-defun.
Now treesit-fontify-with-override clips the fontification region for
the user, so no need for (max start node-start) shenanigans anymore.
More importantly it doesn't fontify unless the region between
node-start and node-end intersects with the region between start and
end, which fixes the problem with c-ts-mode--fontify-defun.
* lisp/treesit.el (treesit-fontify-with-override): Add optional
parameter BOUND-START and BOUND-END. Wrap the function body in a
when-form.
* lisp/progmodes/c-ts-mode.el (c-ts-mode--fontify-declarator)
(c-ts-mode--fontify-variable)
(c-ts-mode--fontify-defun)
(c-ts-fontify-error)
* lisp/progmodes/js.el (js--fontify-template-string)
* lisp/progmodes/python.el (python--treesit-fontify-string): Use the
new signature.
Previously, if 'server-stop-automatically' was configured for
'kill-terminal' or 'delete-frame', killing a client via
'save-buffers-kill-terminal' wouldn't prompt about the saving files in
the client's buffer list (as it does when not using those settings).
This change ensures that those settings only apply when killing the
last client, as described in the manual (bug#51993).
* lisp/server.el (server-save-buffers-kill-terminal): Handle
'server-stop-automatically' behavior in this function, rather than
calling 'server-stop-automatically--handle-delete-frame'.
* lisp/tab-bar.el (tab-bar-separator)
(tab-bar-format-align-right, tab-bar-auto-width):
* lisp/tab-line.el (tab-line-format-template): Replace the variable
'window-system' with the function call '(window-system)'.
* lisp/emacs-lisp/lisp-mode.el (lisp-cl-font-lock-keywords-2)
(lisp-el-font-lock-keywords-2): Restore use of type face instead of
builtin face for &symbol keywords. This fixes what appears to be
a copy paste error that changed the face for common lisp and emacs
lisp &symbol style keywords that was introduced in commit
a498e5f83 by restoring the type face to font-lock-type-face as
consistent with the comments.
* nsterm.m ([EmacsView resetCursorRects:]): Be defensive when
accessing FRAME_OUTPUT_DATA. [resetCursorRects:] can be called
from the event loop after the frame is deleted. When this
happens, emacsframe is NULL. This means there is an underlying
leak of the EmacsView object! (Bug#59794)
Do not merge to master.
Copyright-paperwork-exempt: yes
After this change, ucs-normalize-tests are still very slow but
somewhat less disastrously so (from 100 to 20 min on this machine).
* test/lisp/international/ucs-normalize-tests.el
(ucs-normalize-tests--normalization-equal-p)
(ucs-normalize-tests--normalization-chareq-p)
(ucs-normalize-tests--rule1-holds-p)
(ucs-normalize-tests--rule2-holds-p)
(ucs-normalize-tests--part1-rule2):
Run only over the Unicode code space.
Hoist `with-current-buffer` to reduce overhead.
* test/lisp/emacs-lisp/comp-tests.el
(with-test-native-compile-prune-cache)
(test-native-compile-prune-cache)
(test-native-compile-prune-cache/delete-only-eln)
(test-native-compile-prune-cache/dont-delete-in-parent-of-cache):
Check that the last directory in `native-comp-eln-load-path' is
not affected by `native-compile-prune-cache'.
* lisp/startup.el (command-line): Interpret non-absolute file
names in '--init-directory' relative to the directory from
which Emacs is started. (Bug#59795)
* lisp/emacs-lisp/easy-mmode.el (define-minor-mode)
(easy-mmode-define-keymap): Don't declare obsolete, since we are
still using it in gud.el. (Bug#59769) (Bug#59605)
Do not merge to master.
This reverts commit 8bb5c1bfec.
That commit lost too many useful features in the GUD menus
and caused several bugs, the last of them bug#59769.
Do not merge to master.
