* test/src/comp-resources/comp-test-funcs.el (comp-test-62537-1-f)
(comp-test-62537-2-f): New functions.
* lisp/emacs-lisp/comp.el (comp-jump-table-optimizable): Make it
stricter add a comment.
Also fix one problem evident in test file decls-10.cc.
* lisp/progmodes/cc-engine.el (c-brace-stack-at): Bind
c-record-type-identifiers to nil to prevent called functions recording
identifiers spuriously.
(c-forward-<>-arglist-recur): Revert the ill-advised optimization from autumn
2022 which attempted to avoid re-marking c-type text properties inside angle
bracket arglists.
(c-forward-decl-or-cast-1): Accept the semicolon at the end of "t8 * id;" as
sufficient evidence to fontify as a declaration (not a multiplication).
* lisp/progmodes/cc-fonts.el (c-font-lock-c++-using): No longer fontify the
last component of foo::bar with c-reference-face-name.
* lisp/files.el (remote-file-name-inhibit-locks): Fix docstring.
* lisp/userlock.el (userlock--handle-unlock-error):
Display warning only when `create-lockfiles' is non-nil. (Bug#62614)
* lisp/net/tramp.el (tramp-handle-unlock-file): Raise a warning
only when `create-lockfiles' is non-nil or
`remote-file-name-inhibit-locks' is nil.
Because of outstanding bug#62576, it's way to easy for users to get
confused by the M-x package-install process for Eglot.
Whenever an Eglot release depends on a new version of a GNU ELPA :core
package which is _also_ provided built-in, M-x package-install will
install that dependency, but will not always load it on top of the
built-in one.
The solution is to not use "require" for now. This may potentially
lead to double "loads", but that should in principle be idempotent.
* lisp/progmodes/eglot.el (project, eldoc, seq, flymake, xref, jsonrpc)
(external-completion): Load, don't require.
When answering the :textDocument/completion request, LSP servers
provide a :isIncomplete flag in the response, which allows Eglot to
know if "further typing should result in recomputing [the completions]
list.
If :isIncomplete is false (i.e. the full set was returned), Eglot
caches the response in a global variable eglot--capf-cache that
persists for the duration of the "completion session", taken to be the
interval between two calls to completion-in-region-mode.
If the cache has been set, and Eglot detects that "further typing" has
happened, it is safe to use the cache instead of making a request to
the server.
Thus eglot--capf-cache-flush, added to completion-in-region-mode-hook,
is used to flush this cache. Since the popular Company completion
package doesn't use completion-in-region-mode, eglot--capf-cache-flush
is also added to its company-after-completion-hook.
* lisp/progmodes/eglot.el (eglot--managed-mode): Set
'completion-in-region-mode-hook and company-after-completion-hook.
(eglot--capf-cache): New variable.
(eglot--capf-cache-flush): New function.
(eglot-completion-at-point): Rework.
* etc/EGLOT-NEWS: Update.
Only echo the "active signature", send all the other signatures for
the *eldoc* buffer.
* lisp/progmodes/eglot.el (eglot--sig-info): Rework protocol.
(eglot-signature-eldoc-function): Rework.
The vast majority of Eglot sync requests to the server need to inform
the server of any pending changes to the buffer beforehand, so that
the server has up-to-date information to do its job.
But doing so at the expense of ugly advice of jsonrpc-request is ill
advised. Introduce eglot--request helper instead and use that.
Use this opportunity to conduct a review of Eglot sync requests and
come to the conclusion that all need to send any changes beforehand.
Nevertheless, an IMMEDIATE kwarg to eglot--request was added to
bypassing this.
* lisp/progmodes/eglot.el
(eglot--request): New helper.
(eglot-shutdown)
(eglot-execute-command)
(eglot-workspace-configuration)
(eglot--signal-textDocument/willSave)
(eglot--workspace-symbols)
(eglot--lsp-xrefs-for-method)
(xref-backend-apropos)
(eglot-format)
(eglot-completion-at-point)
(eglot-imenu)
(eglot-rename)
(eglot-code-actions): Use eglot--request.
* lisp/eshell/eshell.el (eshell-read-command): New function...
(eshell-command): ... use it. Additionally, require the COMMAND
argument, and rename ARG to TO-CURRENT-BUFFER.
This regressed due to the patch for bug#53715, which changed how
Eshell pipelines return the processes in the pipeline (bug#62556).
* lisp/eshell/esh-cmd.el (eshell-parse-command): When creating
background commands, wrap the process(es) in a cons cell whose CAR is
':eshell-background'. This lets us use fewer heuristics...
(eshell-eval-command): ... here. Additionally, keep the result and
the incomplete delimiter separate.
* lisp/eshell/eshell.el (eshell-command): Check ':eshell-background'
and use a more-robust method for setting the output target.
* test/lisp/eshell/eshell-tests.el (eshell-test/eshell-command/simple)
(eshell-test/eshell-command/pipeline)
(eshell-test/eshell-command/background)
(eshell-test/eshell-command/background-pipeline): New tests.
* lisp/simple.el (inhibit-auto-fill): New var.
(internal-auto-fill): Obey it.
(newline): Use it instead of binding `auto-fill-function`, so
as to avoid messing up the state if something wants to enable/disable
`auto-fill-mode` during the course of the let binding (bug#62419).
Normally, Eshell only uses synchronous processes on MS-DOS, so this is
hard to test. To get around this, let the tests explicitly request
synchronous processes.
* lisp/eshell/esh-proc.el (eshell-supports-asynchronous-processes):
New variable...
(eshell-gather-process-output): ... use it, and remove some incorrect
code updating Eshell's internal markers (the async code path doesn't
do this, so neither should the sync path).
* lisp/eshell/esh-cmd.el (eshell-execute-pipeline): Use
'eshell-supports-asynchronous-processes'.
* test/lisp/eshell/esh-proc-tests.el
(esh-proc-test/emacs-command): New function.
(esh-proc-test/emacs-echo, esh-proc-test/emacs-upcase): New variables.
(esh-proc-test/synchronous-proc/simple/interactive)
(esh-proc-test/synchronous-proc/simple/command-result)
(esh-proc-test/synchronous-proc/pipeline/interactive)
(esh-proc-test/synchronous-proc/pipeline/command-result): New tests.
* lisp/eshell/esh-cmd.el (eshell-rewrite-for-command): Make
'for-items' an uninterned symbol.
(eshell-as-subcommand): Correct docstring.
(eshell-do-command-to-value): Mark obsolete.
(eshell-command-to-value): Move binding of 'value' outside of the
macro's result, and remove call to 'eshell-do-command-to-value'.
* test/lisp/eshell/esh-cmd-tests.el
(esh-cmd-test/subcommand-shadow-value)
(esh-cmd-test/for-loop-for-items-shadow): New tests.
(esh-cmd-test/for-name-loop, esh-cmd-test/for-name-shadow-loop):
Rename to...
(esh-cmd-test/for-loop-name, esh-cmd-test/for-loop-name-shadow):
... these.
* lisp/gnus/gnus.el: New variable gnus-prev-cwc.
* lisp/gnus/gnus-art.el (gnus-article-edit-mode):
New local variable gnus-prev-cwc.
(gnus-article-edit-article): Store original
gnus-current-window-configuration as gnus-prev-cwc.
(gnus-article-edit-done, gnus-article-edit-exit): Restore
gnus-current-window-configuration from gnus-prev-cwc.
* lisp/gnus/gnus-eform.el (gnus-edit-form-mode):
New local variable gnus-prev-cwc.
(gnus-edit-form): Store original
gnus-current-window-configuration as gnus-prev-cwc.
(gnus-edit-form-done, gnus-edit-form-exit): Restore
gnus-current-window-configuration from gnus-prev-cwc.
* lisp/gnus/gnus-sum.el (gnus-summary-limit-to-marks)
(gnus-summary-limit-exclude-marks): Use completing-read to complete on
marks in the current summary buffer.
* lisp/gnus/gnus-sum.el (gnus-current-move-article): New variable to
track article while moving.
(gnus-summary-move-article): Set gnus-current-move-article when
moving/copying/crossposting or respooling.
* lisp/shell.el (shell-get-old-input-include-continuation-lines): New
defcustom (default nil).
(shell-get-old-input): New function. Like
'comint-get-old-input-default' but include all continuation lines if
'shell-get-old-input-include-continuation-lines' is non-nil.
(shell-mode): Install shell-get-old-input. (Bug#61069)
* etc/NEWS: Advertise the new defcustom.
Because of a drive-letter casing mismatch between 'buffer-file-name'
and the return value of 'url-generic-parse-url', the hash-table test
'equal' in 'eglot-current-server' failed.
This failed to recognize that the file xref landed us on really is
managed by the language server that facilitated that jump.
The function w32-long-file-name seems to convert "C:/Users/" to
"c:/Users" consistently and so is a good addition to
eglot--uri-to-path.
* lisp/progmodes/eglot.el (eglot--uri-to-path): Use
w32-long-file-name.
* lisp/progmodes/typescript-ts-mode.el:
(typescript-ts-mode-jsx-tag-face)
(typescript-ts-mode-jsx-attribute-face): New faces.
(typescript-ts-mode--font-lock-settings): Use new faces.
(tsx-ts-mode): Mention the new faces in the docstring.
* test/src/comp-tests.el (61917-1): New test.
* src/comp.c (syms_of_comp): New variable.
* lisp/loadup.el: Store primitive arities before dumping.
* lisp/emacs-lisp/comp.el (comp--func-arity): New function.
(comp-emit-set-call-subr): Make use of `comp--func-arity'.
* lisp/net/tramp-gvfs.el (tramp-gvfs-enabled): Do not check for
:system bus.
(tramp-gvfs-mounttypes): New defconst.
(tramp-gvfs-maybe-open-connection): Use it.
(top): Call zeroconf only when :system bus is available.
`unwind-protect` without unwind forms is not just pointless but often
indicates a mistake where the intended unwind part is misplaced, as in
(unwind-protect (progn PROT-FORMS UNWIND-FORMS)) ; oops
or
(unwind-protect PROT-FORM) UNWIND-FORMS ; also oops
or entirely forgotten for that matter. Warning about this makes
sense, and the warning can always be silenced by removing the
`unwind-protect` altogether if it shouldn't be there in the first
place.
* lisp/emacs-lisp/macroexp.el (macroexp--expand-all): Implement
warning.
* etc/NEWS: Announce.
* test/lisp/emacs-lisp/bytecomp-tests.el
(bytecomp-test--with-suppressed-warnings): Add test case.
