No reason not to. Tested with gfm-view-mode and markdown-ts-mode
* lisp/progmodes/eglot.el (eglot-client-capabilities): Annouce
markdown support for completion docs.
The fringe indicator uses a custom lightning-bolt bitmap, an alternative
to the margin indicator on GUI frames. It is non-interactive, however.
* lisp/progmodes/eglot.el (eglot--fringe-action): New fringe bitmap.
(eglot-code-action-indications): Add 'left-fringe' to default value
and to docstring. Update incompatibility note.
(eglot-code-action-suggestion): Handle 'left-fringe' indication.
Eglot previously needed gfm-view-mode from markdown-mode.el to render
Markdown from LSP servers. It now prefers markdown-ts-view-mode when
available.
* lisp/progmodes/eglot.el (eglot--accepted-formats): Recognize
markdown-ts-view-mode as a Markdown renderer.
(eglot--format-markup): Rework with cl-labels; prefer
markdown-ts-view-mode over gfm-view-mode.
* doc/misc/eglot.texi (Eglot Features): Don't mention
markdown-mode directly.
* etc/EGLOT-NEWS: Mention change
Github-reference: https://github.com/joaotavora/eglot/discussions/1588
When the server negotiates positionEncoding utf-8 or utf-16,
ParameterInformation.label vector offsets are byte/code-unit counts
into the signature label, not character counts. Using them raw caused
wrong highlights and crashes on Unicode-rich signatures.
* lisp/progmodes/eglot.el (eglot--sig-info): Mostly rewrite.
(eglot-move-to-utf-8-linepos-function): Tweak docstring.
(eglot-move-to-utf-8-linepos, eglot-move-to-utf-16-linepos): Return
position moved to.
See discussion at
https://yhetil.org/emacs/86y0hy9e6g.fsf@gnu.org/
* lisp/emacs-lisp/elisp-scope.el (elisp-scope-analyze-form):
Add example in docstring.
* lisp/progmodes/elisp-mode.el (elisp-fontify-semantically):
Clarify use of 'cursor-sensor-mode'.
This ensures we only compute the 'help-echo' string once per
symbol in a certain position.
* lisp/progmodes/elisp-mode.el
(elisp--annotate-symbol-with-help-echo): Add caching for
when the symbol role's :help property is a function.
* lisp/progmodes/eglot.el (eglot-code-action-indicator): No lighbulb, no
fancy lightning bolt, just use zigzags which seem to display well on
typical fonts and typically have a width of 1.
* lisp/progmodes/project.el (project-prompter)
(project-prompt-project-dir, project-prompt-project-name):
Delete ALLOW-EMPTY parameter. Default to the current project if
there is one.
* lisp/vc/vc.el (project-root): Declare.
(vc--prompt-other-working-tree): Replace ALLOW-EMPTY parameter
with new ALLOW-CURRENT parameter.
(vc-working-tree-switch-project): Allow selecting the current
working tree, for symmetry with project-switch-project.
* etc/NEWS: Update.
e72afa9dbf replaced nil, which is invalid as a face definition,
with ((t (:inherit default))), but that does not actually result in
an "empty" face, which doesn't set any attributes. Instead it doubles
down on all attributes set for the default face. Replace that with
((t)), which accomplishes what appears to have been the intention.
* lisp/net/shr.el (shr-h4, shr-h5, shr-h6):
* lisp/nxml/nxml-mode.el (nxml-text):
* lisp/progmodes/make-mode.el (makefile-shell): Use empty face
definition ((t)), instead of ((t (:inherit default))).
* lisp/progmodes/compile.el (compilation-search-extra-path): New
option (bug#80279).
(compilation-start): Use it.
(compilation-search-path):
* doc/emacs/building.texi:
* etc/NEWS: Document it, and also compilation-search-path.
* lisp/progmodes/eglot.el (eglot--server-buffer-name): New function.
(eglot--connect): Rename events and stderr buffers after serverInfo
arrives to include the server's self-reported name.
(eglot-show-workspace-configuration, eglot-describe-connection)
(eglot--define-hierarchy-command): Use eglot--server-buffer-name.
(eglot--propose-changes-as-diff): Add SERVER argument; use
eglot--server-buffer-name.
(eglot--apply-workspace-edit): Add SERVER argument; pass server to
eglot--propose-changes-as-diff.
(eglot-handle-request workspace/applyEdit, eglot-execute)
(eglot-rename): Pass server to eglot--apply-workspace-edit.
When the user renames a buffer's file via set-visited-file-name,
notify the server with didClose (old URI) and didOpen (new URI).
The TextDocumentIdentifier cache still holds the old URI when
after-set-visited-file-name-hook fires, so didClose uses it
correctly; didOpen then clears and rebuilds it from the new name.
* lisp/progmodes/eglot.el (eglot--managed-mode): Hook and unhook
eglot--after-set-visited-file-name-hook.
(eglot--after-set-visited-file-name-hook): New function.
Use box-drawing characters for the tree guides in hierarchy
buffers when the display supports them.
* lisp/progmodes/eglot.el (eglot--hierarchy-2): Rework.
Now that 'eglot-uri-to-path' tries to return
"project-root-familiar" paths, unresolving the absolute path
name, it's important to update 'eglot--find-buffer-visiting' to
consider the two alternative ways to refer to a file. IOW, this
was already a bug, but the latest changes to 'eglot-uri-to-path'
made it more evident.
* lisp/progmodes/eglot.el (eglot--find-buffer-visiting): Also
consider buffer-file-name.
Commit 6942d508f0 made js-mode-regexp-syntax-bug-25529 fail,
because it changes the moment and place in which
`syntax-propertize` gets called, uncovering a bug where
`js-jsx--detect-after-change` narrows the buffer, then calls
`re-search-forward` which calls `syntax-propertize`, so we end
up doing `syntax-propertize` while narrowed which can fail in
all kinds of ways.
* lisp/progmodes/js.el (js-jsx--detect-and-enable): Change optional arg
to BEG and END.
(js-jsx--detect-after-change): Use them to avoid narrowing.
* lisp/emacs-lisp/syntax.el (syntax-propertize): Emit a message
when called narrowed and the narrowing starts later than the
nearest already propertized position.
Also add more bindings to M-x eglot-list-connections
* lisp/progmodes/eglot.el (eglot-list-connections-mode-map): Add
bindings for e, w, RET.
(eglot-events-buffer-of-listed-connection)
(eglot-show-workspace-configuration-of-listed-connection)
(eglot-describe-listed-connection): New commands.
(eglot-describe-connection): New command.
* doc/misc/eglot.texi: Document new keys and eglot-describe-connection.
* etc/EGLOT-NEWS: Announce.
* lisp/progmodes/eglot.el (eglot-list-connections-mode-map): New
defvar.
* lisp/progmodes/eglot.el (eglot--list-connections-entries): New
function, extracted from eglot-list-connections.
* lisp/progmodes/eglot.el (eglot--list-connections-cmd): New macro.
* lisp/progmodes/eglot.el (eglot-shutdown-listed-connection)
(eglot-reconnect-listed-connection): New commands.
* lisp/progmodes/eglot.el (eglot-list-connections): Use
eglot--list-connections-entries.
* doc/misc/eglot.texi: Document eglot-list-connections and new keys.
* etc/EGLOT-NEWS: Announce.
Some LSP servers prefer to canonicalize all document URIs to
their "true name", chasing all fs symbolic links. If the user
is visiting a project under a symlink, say ~/myprojects/foo
which points to /data/true/projects/foo. When visiting a
~/myprojects/foo/f.c and trying to jump to a nearby file f.h via
LSP 'xref-find-definitions', the path should probably be
~/myproject/foo/f.h, not /data/true/projects/foo/f.h.
Even though Emacs can recognize that the two files are the same,
'buffer-file-name' is not. This confuses tools like C-x C-f,
ibuffer, and ultimately the users themselves who probably
oblivious to the true place of the project. After all they
started Eglot under a syminked file.
* lisp/progmodes/eglot.el (eglot-lsp-server): Add trueroot slot.
(eglot-uri-to-path): Rework to consider trueroot and project root.
* lisp/progmodes/compile.el
(compilation-error-regexp-alist-alist): Make leading '|'
optional when there are leading spaces (bug#80681).
* etc/compilation.txt (GNU style): New error with two leading
spaces.
* test/lisp/progmodes/compile-tests.el
(compile-test-error-regexps): Increment expected
compilation-num-errors-found.
(compile-tests--test-regexps-data): New error test case for
'gnu' with two leading spaces. Change expected match for
'cucumber' to match again 'gnu'.
Copyright-paperwork-exempt: yes
When jit-lock is not in use, `font-lock-flush` is too costly to
justify using it just to maybe update the highlighting of a few macro calls.
* lisp/progmodes/elisp-mode.el (elisp--font-lock-flush-elisp-buffers):
Skip buffers that use jit-lock, like the `*ielm*-comint-indirect` buffer.
* lisp/progmodes/elisp-mode.el (elisp--local-variables): Restore
point before calling `elisp--safe-macroexpand-all` in case this
triggers a temporary redisplay.
