This is more in line with how we generally handle user options for
commands to call. Later, we can get rid of the 'eval' call.
* lisp/textmodes/tex-mode.el (tex-view): Warn if tex-dvi-view-command is
set to an sexp instead of a string; say that this use is obsolete.
(tex-dvi-view-command): Change the default value to a string. Update
docstring to reflect the above obsoletion.
* lisp/treesit.el (treesit-font-lock-level): Allow the
value to be an alist mapping major modes to font lock levels.
(treesit--font-lock-level-setter): Rewrite.
(treesit--compute-font-lock-level): New function.
(treesit-font-lock-recompute-features): Use new function.
* doc/lispref/modes.texi (Parser-based Font Lock): Minor fix.
* lisp/progmodes/lua-ts-mode.el (lua-ts-send-file): Remove
restriction on interactive use. Unlike related send-* commands
it can be useful to send a file to the Lua interpreter outside of
a 'lua-ts-mode' buffer. (Bug#74705)
* lisp/progmodes/flymake-cc.el (flymake-cc-command): Use
customization type 'function' instead of 'symbol' to allow
other callable objects in addition to named functions, and
update the doc string accordingly.
(flymake-cc): Use 'functionp' to check if 'flymake-cc-command'
is callable. (Bug#74378)
* lisp/net/tramp.el (tramp-format-spec): New defun.
(tramp-dissect-file-name, tramp-compute-multi-hops)
(tramp-expand-args):
* lisp/net/tramp-cmds.el (tramp-default-rename-file): Use it.
When tree-sitter grammar make breaking changes, major modes
breaks completely: no highlighting is shown. This new function
will run in treesit-major-mode-setup, validate each
font-lock features, and disable the ones that are not compatible
with the new grammar, so that the rest of the features still
work. It also displays a warning explaining the situation,
which looks like this:
Warning (treesit-font-lock-rules-mismatch): Emacs cannot
compile every font-lock rules because a mismatch between the
grammar and the rules. This is most likely due to a
mismatch between the font-lock rules defined by the major
mode and the tree-sitter grammar.
This error can be fixed by either downgrading the
grammar (tree-sitter-c) on your system, or upgrading the
major mode package. The following are the temporarily
disabled features:
- `preprocessor' for c.
* lisp/treesit.el (treesit-validate-font-lock-rules): New function.
(treesit-major-mode-setup): Validate font-lock rules.
Now treesit-explore-mode will prompt the user to select a parser
to explore, rather than a language. It'll also include the
local parsers at point for selection.
* lisp/treesit.el (treesit--explorer-language): Change to
treesit--explorer-parser.
(treesit--explorer--nodes-to-highlight):
(treesit--explorer-refresh): Change to use parser.
(treesit--explorer-generate-parser-alist): New function.
(treesit-explorer-switch-parser): New command.
(treesit-explore-mode): Use switch-parser to setup.
* doc/lispref/parsing.texi (Language Grammar): Mention
treesit-explorer-switch-parser.
* lisp/gnus/message.el (message-goto-to, message-goto-from,
message-goto-subject, message-goto-cc, message-goto-bcc,
message-goto-fcc, message-goto-reply-to,
message-goto-newsgroups, message-goto-distribution,
message-goto-followup-to, message-goto-mail-followup-to,
message-goto-keywords, message-goto-summary): Mention that these
will insert an empty header if the header is not found.
(message-position-on-field): Explain that this inserts an empty
header if the header is missing, and that insertion is done
after the headers mentioned in AFTERS.
(cherry picked from commit 9f266e2d7c)
* doc/lispref/keymaps.texi (Creating Keymaps): Add :continue-only
to :repeat part of defvar-keymap.
* lisp/keymap.el (defvar-keymap): Add support for new :repeat
keyword :continue-only.
* lisp/repeat.el (repeat-post-hook): The property 'repeat-continue-only'
is handled as a list of repeat-maps.
* test/lisp/repeat-tests.el (repeat-tests-repeat-map): Use new
:repeat keyword :continue-only.
* lisp/gnus/message.el (message-goto-to, message-goto-from,
message-goto-subject, message-goto-cc, message-goto-bcc,
message-goto-fcc, message-goto-reply-to,
message-goto-newsgroups, message-goto-distribution,
message-goto-followup-to, message-goto-mail-followup-to,
message-goto-keywords, message-goto-summary): Mention that these
will insert an empty header if the header is not found.
(message-position-on-field): Explain that this inserts an empty
header if the header is missing, and that insertion is done
after the headers mentioned in AFTERS.
* lisp/gnus/message.el (message-send-mail): Pass a fixed string
instead of regexp to 'message-remove-header' (Bug#74658).
Copyright-paperwork-exempt: yes
Specifically, standalone-parent and prev-sibling. The c-ts-mode
custom version skips labels and proproc directives.
* lisp/progmodes/c-ts-mode.el:
(c-ts-mode--standalone-parent-skip-preproc): Rename to
c-ts-mode--standalone-parent, and make it skip labels too.
(c-ts-mode--preproc-indent-rules): Rename standalone-parent and
prev-sibling.
(c-ts-mode--label-indent-rules): Use custom standalone-parent
and prev-sibling.
Remove functions that became unused due to the previous change.
* lisp/progmodes/c-ts-mode.el:
(c-ts-mode--standalone-grandparent):
(c-ts-mode--else-heuristic):
(c-ts-mode--first-sibling):
(c-ts-mode--parent-is-not-top-compound):
(c-ts-mode--indent-styles):
(c-ts-mode--top-level-label-matcher): Remove functions.
(c-ts-mode--simple-indent-rules): Move two label rules into
c-ts-mode--label-indent-rules.
(c-ts-mode--label-indent-rules): Add rules
Rework the indent rules in c-ts-mode using
c-ts-common-baseline-indent-rule. Also reworked internal
functions for setting indent styles to make it simpler and (more
importantly) more readable. Now there's a single function,
c-ts-mode--simple-indent-rules, that returns a value that can go
straight to treesit-simple-indent-rules.
* lisp/progmodes/c-ts-mode.el:
(c-ts-mode--indent-style-setter): Simplify.
(c-ts-mode--get-indent-style): Remove function.
(c-ts-mode--prompt-for-style): Use a hard-coded list of styles.
This is a tiny downgrade from the computed lists but is
acceptable.
(c-ts-mode-set-style): Use c-ts-mode--simple-indent-rules.
(c-ts-mode--standalone-grandparent): Docstring change.
(c-ts-mode--for-loop-indent-rule): New function.
(c-ts-mode--preproc-indent-rules): New variable. Copied from
old indent rules and added some new rule.
(c-ts-mode--macro-heuristic-rules):
(c-ts-mode--simple-indent-rules):
(c-ts-mode--parenthesized-expression-indent-rule):
(c-ts-mode--label-indent-rules): New functions.
(c-ts-mode):
(c++-ts-mode): Use c-ts-mode--simple-indent-rules.
I found a really good baseline indent rule that handles a wide
range of situations very well. Now major modes can just start
with this rule and add exceptions on top.
This is worth mentioning in the manual, but that'll be a large
change, and doesn't have to be included in this commit.
* lisp/progmodes/c-ts-common.el:
(c-ts-common-list-indent-style): New variable.
(c-ts-common--standalone-parent):
(c-ts-common--prev-standalone-sibling):
(c-ts-common-parent-ignore-preproc):
(c-ts-common-baseline-indent-rule): New function.
* lisp/treesit.el (treesit--indent-prev-line-node): New function.
(treesit-simple-indent-presets): Add new preset.
* lisp/treesit.el (treesit-simple-indent-rules): Allow the rule
to be a single function. Also replace cl-loop with dolist plus
catch-throw.
(treesit--indent-rules-optimize): Handle the case when a rule is
a function.
* doc/lispref/modes.texi (Parser-based Indentation): Update
manuel.
Properly indent the body of compound expressions, even when then
compound expression is not at the beginning of line and the
parent is not an if/for/while/etc., and matches the behavior of
c-mode.
This fixes a problem that is common with macros and in testing
frameworks. For example, you expect this to indent:
TEST_CASE(1) {
assert (...);
}
If the compound statement is the function body itself, don't
apply this new rule and instead guide by the parent and first
sibling.
I'm sure there are subtle interactions that aren't handled
properly by checking for "function_definition" rather than
something more general, but it does fix the test case and the
check can be improved as more cases are found.
* lisp/progmodes/c-ts-mode.el:
(c-ts-mode--parent-is-not-top-compound): New function.
(c-ts-mode--indent-styles): Use it.
* test/lisp/progmodes/c-ts-mode-resources/indent.erts: New
compound statement test.
* etc/ERC-NEWS: Add entry explaining changes to entry point 'erc-tls'
and library functions `erc-normalize-port' and `erc-compute-port'.
* lisp/erc/erc.el (erc-normalize-port): Map "ircu" to 6665 instead of
6667, and add related IANA service mappings. Return 0 for unknown
nonempty strings.
(erc-open): Pass `erc-session-port' through `erc-string-to-port' before
handing it to `erc-server-connect'. This prevents the network machinery
from ever seeing a numeric string, like "6667".
(erc-select-read-args): Since `erc-compute-port' no longer relies on
`erc-normalize-port', ensure its input is a number. Use
`erc-port-equal' instead of `eql'.
(erc-tls): Respect a configured non-nil `erc-port' option when the user
does not provide a :port keyword argument from Lisp code.
(erc-determine-parameters): Use `erc-compute-port' for initializing
`erc-session-port'.
(erc-compute-port): Don't pass the result through `erc-normalize-port',
which can convert it to an unintuitive form.
(erc--url-default-connect-function): Use `erc-compute-port' instead of
`erc-normalize-port'.
(erc-handle-irc-url): Use `erc-port-equal' for comparison.
* test/lisp/erc/erc-scenarios-auth-source.el
(erc-scenarios-common--auth-source): Allow tests to convey the automatic
port number to `erc-open' via `erc-port'.
(erc-scenarios-base-auth-source-server--dialed): Use `erc-port' option
instead of passing a :port parameter to entry-point command.
* test/lisp/erc/erc-tests.el (erc-normalize-port): New test.
(Bug#74516)
* lisp/erc/erc.el (erc--auth-source-determine-params-defaults): Allow
arbitrary strings for `ers-session-port'. Previously, if a port/service
was any string other than "irc", the auth-source query would fail for a
seemingly unknown reason. Restricting the value to "irc" is unnecessary
since "irc" is already added to the list of ports, and
`make-network-process' already consults /etc/services for well-known
service names, like "ircs-u", etc. This change allows a user to (setopt
erc-port "1234"), intentionally or accidentally, while still being able
to use .authinfo for password management. (Bug#74516)
Copyright-paperwork-exempt: yes
