Do not merge to master.
Before working on bugfixes to Eglot in its non-ELPA form delivered
with Emacs 29, it's important to distinguish between this emacs-29
version and the one in master, which will also get the bugfixes (via
routine gitmerge.el) but with a bumped version number.
* lisp/progmodes/eglot.el (Version): Emphasize Emacs 29, but hint this
is 1.12 + something.
(Package-Requires): Remove. Everything required is already in Emacs
29.
Running "M-x htmlfontify-buffer" in one buffer, the exported HTML contains
lines like:
body, pre { text-decoration: none; font-family: Migu 2M; font-stretch:
normal; font-weight: 500; font-style: normal; color: #ffffff;
background: #000000; font-size: 15pt; }
Standards-compliant web browsers should ignore this font-family.
MDN Web Docs says:
https://developer.mozilla.org/en-US/docs/Web/CSS/font-family
Valid family names
Font family names must either be given quoted as strings, or unquoted as
a sequence of one or more identifiers. This means that punctuation
characters and digits at the start of each token must be escaped in
unquoted font family names.
It is a good practice to quote font family names that contain white
space, digits, or punctuation characters other than hyphens.
An unquoted font-family is valid as long as it doesn't start with a digit,
but MDN Web Docs also says:
The following example is technically valid but is not recommended:
font-family: Gill Sans Extrabold, sans-serif;
So it makes sense to quote all font-family.
* lisp/htmlfontify.el (hfy-family): Quote 'font-family'. (Bug#62054)
Many language server configuration options are of the JSON array
datatype, for example argument lists for executables, but there wasn't
any example of that in the Eglot manual.
* doc/misc/eglot.texi (User-specific configuration)
(User-specific configuration): Tweaks.
(JSONRPC objects in Elisp): Mention JSON arrays. Tweak example.
Copyright-paperwork-exempt: Yes
* lisp/emacs-lisp/package-vc.el (require): Do not load `inline' during
compilation.
(package-vc--query-spec): Remove function.
(package-vc--clone): Do not call 'package-vc--query-spec'.
* lisp/emacs-lisp/package-vc.el (package-vc--main-file)
(package-vc--unpack-1): Query 'pkg-spec' instead of
'package-desc-extras'.
(package-vc--unpack): Do not update 'package-desc-extras'.
This simplification is possible due to the change in 2718bbb3bc, since
we now ensure that package specifications are not lost. They are
either provided by a package archive or stored in
'package-vc-selected-packages'.
* doc/emacs/package.texi (Fetching Package Sources): Do not promote
the usage of 'package-vc-selected-packages' to install packages.
* lisp/emacs-lisp/package-vc.el (package-vc-selected-packages): Remove
custom setter and change docstring according to these changes.
(package-vc--desc->spec): Consult 'package-vc-selected-packages' for
package specifications.
(package-vc--unpack): Add unknown package specifications to
'package-vc-selected-packages'
There is a possible use for a global setting
eglot-workspace-configuration, which is to (ab)use it to set
user-specific configuration when the server doesn't permit other
methods. Rearrange the "Advanced server configuration" section and
describe that use in the manual.
* doc/misc/eglot.texi (Project-specific configuration): Correct
mistake about global usefulness of eglot-workspace-configuration.
(Advanced server configuration): Swap order of sections.
(User-specific configuration): Mention possibility of globally
setting eglot-workspace-configuration.
The fix contained a spurious check for this-command that shouldn't
have made it in..
* lisp/progmodes/eglot.el (eglot-ensure): Don't check this-command.
'eglot-ensure', typically used in the major-mode-hook, use
'post-command-hook' to schedule an automated, non-interactive
connection attempt to a server. The goal is to connect when the
buffer is ready, i.e. after the user command that found the file.
However, if there are dir-local or buffer-local variables to confirm,
finding the file will cause a minibuffer prompt to appear.
In that case, 'eglot-ensure's addition to the global post-command-hook
runs before it was intended too and a connection is started
prematurely.
In turn, this means that a call to 'hack-dir-local-variables' -- which
is part of the connection process -- which also needs a minibuffer
prompt, collides with the previous one. This generates an error and
confuses the user, who doesn't know if the directory-local variables
have been applied or not.
This commit fixes the clash by having 'eglot-ensure' set
'post-command-hook' buffer-locally. This causes the automated
connection to take place, as intended, after the user's original
file-finding command has ended.
However, the problem reported in bug#62065 is not completely fixed.
If the user answers "no" to the first "confirm local variables"
"prompt, she will be prompted again in the second one. A subsequent
commit will address this separate problem.
* lisp/progmodes/eglot.el (eglot-ensure): Use buffer-local post-command-hook.
* lisp/novice.el (disabled-command-function): Improve wording, and
make sure the prompt doesn't exceed one screen line, so that the
mini-window is not resized (which gets in the way of fitting the
window to the size of the help text), and the help text is fully
visible when Emacs prompts for input. (Bug#62146)
When Emacs is built, temacs is run in batch mode, so if we enlarge
'gc-cons-percentage' in that case, the enlarged value will be
"remembered" by the dumped Emacs, and will show confusing
information in "C-h v", claiming that the original value was 1.0.
Keeping the value at 0.1 during dumping avoids that.
* src/emacs.c (main): Increase 'gc-cons-percentage' in batch mode
only if we are not initializing (a.k.a. "dumping") Emacs.
* lisp/files-x.el (connection-local-set-profiles)
(connection-local-set-profile-variables):
Use `custom-set-variables'. It serves the purpose to not keep
unsaved user options, and it is less invasive than
`customize-save-variable'.
* lisp/files-x.el (connection-local-set-profiles)
(connection-local-set-profile-variables):
Use `customize-save-variable' instead of `customize-set-variable'.
(Bug#62106)
* lisp/arc-mode.el (archive-zip-summarize): On MS-Windows,
override archive-file-name-coding-system for files compressed
on MS-Windows, and use the local console encoding instead.
* lisp/progmodes/ruby-ts-mode.el (ruby-ts--s-p-query):
Don't match ':' in symbol node text. Or '_', I suppose.
(ruby-ts--syntax-propertize): Make sure to only put the '_' syntax
on punctuation syntax characters, and not on the whole symbol (to
e.g. have symbols like :foo? include text recognized as word).
* test/lisp/progmodes/ruby-ts-mode-tests.el
(ruby-ts-syntax-propertize-symbol): New test.
* lisp/progmodes/python.el
(python-info-dedenter-opening-block-positions): Make the check
stricter. Require that block starts only at indentation.
* test/lisp/progmodes/python-tests.el
(python-indent-after-bare-match): Another test (bug#62031).
This isn't a typical autoload: the progn block is plced in the
autoloads file, but the eglot.el file itself isn't loaded as a result
when the function inside the progn block is called.
* lisp/progmodes/eglot.el (eglot--debbugs-or-github-bug-uri):
autoload, but in progn.
* doc/misc/eglot.texi (Top): Add section "Advanced server configuration"
(Setting Up LSP Servers): Rework.
(Advanced server configuration): New chapter.
A recent commit in tree-sitter-java (issue #141) removed support for
the text_block node type from the grammar. To support the old released
version and the new one, a helper function was added on which the
string syntax highlighting now depends. (bug#61913)
* lisp/progmodes/java-ts-mode.el
(java-ts-mode--string-highlight-helper): New helper function that
return the right query string
(java-ts-mode--font-lock-settings): Use it
Copyright-paperwork-exempt: yes
* lisp/progmodes/python.el (python-nav-end-of-statement)
(python-nav-end-of-block): Revert workaround introduced in Bug#56271
as the bug is fixedin Bug#58780.
* lisp/progmodes/python.el
(python-info-dedenter-opening-block-positions): Check that the
supposed block start is not a method call (bug#62031).
* test/lisp/progmodes/python-tests.el (python-indent-after-re-match):
New test.
Co-authored-by: Lele Gaifax <lele@metapensiero.it>
