* src/sfnt.c (sfnt_table_names): Introduce name of post table.
(sfnt_read_post_table): New function.
(main): New tests.
* src/sfnt.h (struct sfnt_post_table): New struct.
* src/sfntfont.c (sfnt_enum_font_1): Read post table, and set
spacing from its is_fixed_pitch value.
(sfntfont_list_1): Compare spacing between both fonts if
supplied in the font spec.
(sfntfont_open): Set FONT_FILE_INDEX as well as
FONT_FULLNAME_INDEX.
Adjust the type operations to account for the fact that types can have
several parents.
* lisp/emacs-lisp/comp-cstr.el (comp--cl-class-hierarchy):
Use `cl--class-allparents`. Add FIXME.
(comp--direct-supertype): Declare obsolete.
(comp--direct-supertypes): New function.
(comp--normalize-typeset0): Rewrite to use `comp--direct-supertypes`;
adjust to account for the DAG structure; use `cl-set-difference`.
(comp--direct-subtypes): Rewrite.
(comp--intersection): New function.
(comp-supertypes): Rewrite and change return type.
(comp-subtype-p): Simplify.
(comp-union-typesets): Use `comp-supertypes` instead of iterating over
`comp-cstr-ctxt-typeof-types`.
* lisp/emacs-lisp/comp.el (comp--native-compile): Don't catch
errors if we're debugging.
* test/lisp/emacs-lisp/comp-cstr-tests.el: Adjust tests.
* lisp/emacs-lisp/cl-macs.el (cl-defstruct): Fix mishap when we
evaluate (cl-defstruct cl-structure-object ..) during the compilation
of `cl-preloaded.el`.
* lisp/emacs-lisp/cl-preloaded.el: Add corresponding assertion.
Now file-name-at-point-functions, which runs
ffap-guess-file-name-at-point, will pick up on a file name at point if
that file name is a relative file name which exists when looked up
from the root of the project.
For example, in test/lisp/progmodes/eglot-tests.el there is the string
lisp/progmodes/eglot.el; if you put point on that and C-x C-f,
lisp/progmodes/eglot.el under the root of the Emacs repo will now be
part of future history.
* lisp/ffap.el (ffap-alist): Add entry for ffap-in-project.
(ffap-in-project): Add. (bug#66668)
Nested repetitions such as (A*)* potentially take exponential time but
can usually be rewritten in a faster and more readable way without
much trouble. These were all found by Relint.
* lisp/obsolete/terminal.el (te-parse-program-and-args):
* lisp/org/org.el (org-make-tags-matcher):
Apply the transform (A+B*)+ -> A(A|B)*
* lisp/textmodes/fill.el (adaptive-fill-regexp):
Apply the transform A*(B+A*)* -> (A|B)*
* lisp/progmodes/idlw-shell.el (idlwave-shell-filter):
Find the last newline or CR in a more direct way.
* lisp/progmodes/vhdl-mode.el (vhdl-port-copy, vhdl-subprog-copy):
Trim trailing whitespace from a string in a more direct way.
All-whitespace strings are left unchanged as before.
* lisp/progmodes/gdb-mi.el (gdb-python-guile-commands-regexp):
Remove, integrate into...
(gdb-control-commands-regexp): ...this. Translate into rx.
Remove useless submatches which broke earlier attempts at using other
submatches. Rewrite tail expression to avoid superlinear nested
repetition, eliminating a Relint complaint.
* lisp/progmodes/gdb-mi.el (gdb-send): Simplify use of the regexp
above, and use the correct submatch this time. Remove unnecessary and
incorrect second regexp matching.
Passing arbitrary arguments to functions through
"emacsclient --eval" sometimes requires complicated escaping
to avoid them being parsed as Lisp (as seen in
emacsclient-mail.desktop before this change).
The new variable 'server-eval-args-left' allows access to the
arguments before they are parsed as Lisp. By removing
arguments from the variable before they're parsed, a snippet
of Lisp can consume arguments, as in emacsclient-mail.desktop.
org-protocol might be able to use this as well, which might allow it
to drop its current advice on server-visit-files.
* etc/emacsclient-mail.desktop: Use 'server-eval-args-left'.
* lisp/server.el (server-eval-args-left): New variable.
(server-process-filter, server-execute): Make '-eval' arguments
available through 'server-eval-args-left'.
* lisp/startup.el (argv): Mention 'server-eval-args-left' in
docstring.
* etc/NEWS: Announce 'server-eval-args-left'.
* doc/emacs/misc.texi (emacsclient Options): Document
'server-eval-args-left'. (Bug#65902)
In some programming languages and styles, a symbol (or every
symbol in a sequence of symbols) might be capitalized, but the
individual words making up the symbol should never be capitalized.
For example, in OCaml, type names Look_like_this and variable names
look_like_this, but it is basically never correct for something to
Look_Like_This. And one might have "aa_bb cc_dd ee_ff" or "Aa_bb
Cc_dd Ee_ff", but never "Aa_Bb Cc_Dd Ee_Ff".
To support this, the new variable 'case-symbols-as-words' causes
symbol constituents to be treated as part of words only for case
operations.
* src/casefiddle.c (case_ch_is_word): New function.
(case_character_impl, case_character): Use 'case_ch_is_word'.
(syms_of_casefiddle): Define 'case-symbols-as-words'.
* src/search.c (Freplace_match): Use 'case-symbols-as-words'
when calculating case pattern.
* test/src/casefiddle-tests.el (casefiddle-tests--check-syms)
(casefiddle-case-symbols-as-words): Test 'case-symbols-as-words'.
* etc/NEWS: Announce 'case-symbols-as-words'.
* doc/lispref/strings.texi (Case Conversion): Document
'case-symbols-as-words'.
(Bug#66614)
Previously, the header line would stay around even when after
disabling 'which-function-mode', although it may be empty. Now
the 'which-function-mode' element is properly removed from
'header-line-format', so the header line will disappear if
there's nothing else in 'header-line-format'.
Also, previously, when we ran (which-function-mode), we would
enable
'which-function-mode' for all buffers even if they didn't support
imenu. We didn't run the normal logic in 'which-func-ff-hook' to
disable 'which-func-mode' if imenu wasn't present. Now we do run
that logic, by just calling 'which-func-ff-hook'. This is
especially important when the header line is enabled, because
otherwise there's a very noticeable header line added to every
buffer, including e.g. *Help* and *Buffer List*.
Also, we now check that 'header-line-format' is a list before trying
to add to it; this makes us work properly when enabling and
disabling 'which-function-mode' for modes which set
'header-line-format' to a string or symbol, such as eww.
* lisp/progmodes/which-func.el (which-func-try-to-enable): Re-add
'which-func-format' to the header line.
(which-func--header-line-remove): New function.
(which-func--disable): Call 'which-func--header-line-remove'.
(which-function-mode): Call 'which-func-ff-hook' and
'which-func--header-line-remove'. (bug#66283)
* test/lisp/progmodes/which-func-tests.el: New test.
* java/org/gnu/emacs/EmacsNative.java (close): New declaration.
* java/org/gnu/emacs/EmacsSdk11Clipboard.java
(getClipboardData): Catch SecurityException and guarantee file
descriptors are closed even if exceptions arise.
* src/android.c (dup): Export another function.
Benchmarking shows seq-filter to be ~30% faster on this machine.
* lisp/language/hanja-util.el (hanja-filter): Make into obsolete alias
for 'seq-filter'. Update single caller.
em-smart was forcibly re-displaying the screen upwards of 500 times
per screen of output. This caused the Eshell to feel quite slow when
the module was in use. By using fewer hooks and never explicitly
calling 'redisplay' (which was unnecessary) the performance issues go
away (bug#57367).
lisp/eshell/em-smart.el:
(em-smart-unload-hook, eshell-smart-unload-hook): Remove
'eshell-smart-scroll' instead of the now deleted
'eshell-refresh-windows'.
(eshell-smart-displayed, eshell-currently-handling-window)
(eshell-refresh-windows): Delete.
(eshell-smart-scroll-window): Rename to 'eshell-smart-scroll-windows'
and add a bunch of logic originally from 'eshell-refresh-windows'.
(eshell-smart-initialize): Don't add a hook onto
'window-scroll-functions'. Replace 'eshell-refresh-windows' with
'eshell-smart-scroll-windows'.
(eshell-smart-display-setup): Don't refresh windows.
(eshell-smart-redisplay): Rename to 'eshell-smart-scroll'. Delete
'eobp' case.
Benchmarking shows seq-filter to be ~30% faster on this machine.
* lisp/net/rcirc.el (rcirc-condition-filter): Make into an
obsolete alias for 'seq-filter'. Update single caller.
5f60913208 Fix State button for customize-icon (Bug#66635)
27c71979ff ; Another Texinfo fix
889a550ca0 ; Fix Texinfo warnings
893c344b4e Fix the use of adaptive-fill-regexp in treesit indent preset
1098c114b7 Fix treesit-install-language-grammar (bug#66673)
491ee428c0 Fix treesit-explore-mode (bug#66431)
ee043a2703 tsx-ts-mode--font-lock-compatibility-bb1f97b: Re-fix the ...
* lisp/dnd.el (dnd-protocol-alist): Redefine file name handlers
to match solely the local or remote URIs they understand.
(dnd-handle-multiple-urls): Prevent calling the same handler
multiple times for a single URI.
* lisp/gnus/mml.el (mml-dnd-protocol-alist): Apply an anologous
adjustment here. Delete now redundant redefinition of
dnd-open-file.
(mml-dnd-attach-file): Inquire whether to apply the default
disposition and such only once even if more than one file is
dropped.
* test/lisp/dnd-tests.el (dnd-tests-receive-multiple-urls)
(dnd-tests-default-file-name-handlers): New tests.
* src/window.h:
* src/window.c (window_from_coordinates): In the no toolkit
build, add support for the menu bar window. Add a new
'menu_bar_p' argument so the function's signature has changed.
All callers changed.
* src/androidterm.c (handle_one_android_event):
* src/haikuterm.c (haiku_read_socket):
* src/keyboard.c (make_lispy_position):
* src/nsterm.m ([EmacsView mouseDown:]):
* src/pgtkterm.c (button_event):
* src/w32term.c (w32_read_socket):
* src/xdisp.c (note_mouse_highlight):
* src/xterm.c (handle_one_xevent): Set menu_bar_p to true.
* java/org/gnu/emacs/EmacsWindow.java (EmacsWindow)
<dndXPosition, dndYPosition>: New fields initialized to -1.
(onDragEvent): Remember the position of the previous event to
avoid sending duplicates.
* src/androidvfs.c (EMACS_PATH_MAX): New define.
(android_saf_tree_rename, android_saf_tree_opendir)
(android_name_file, android_fstatat, android_faccessat)
(android_fchmodat, android_readlinkat): Use EMACS_PATH_MAX where
SAF file names might be encountered.
The new handling of aliases in `provided-mode-derived-p`
introduced in Emacs-28.1 caused a regression where
(provided-mode-derived-p MODE MODE) returns nil if MODE is an alias.
Rework the loop so we consider an alias as a kind of parent.
* lisp/subr.el (provided-mode-derived-p): Step over aliases separately.
* test/lisp/subr-tests.el (subr-tests--derived-mode-1)
(subr-tests--derived-mode-2): Move out of `provided-mode-derived-p`
and give them properly namespaced names.
(provided-mode-derived-p): Add more tests for aliases.
* lisp/erc/erc-fill.el (erc-fill--spaced-commands): Remove unused
internal variable originally intended for ERC 5.6.
(erc-fill): Check for `erc-msg' being `msg', which carries the same
meaning as `erc-cmd' being `PRIVMSG' or `NOTICE', except that inserted
outgoing messages now no longer normally have an `erc-cmd' property.
(erc-fill-wrap-mode, erc-fill-wrap-disable): Kill
`erc-fill--wrap-last-msg'.
(erc-fill--wrap-max-lull): Convert from buffer-local to normal
variable.
(erc-fill--wrap-continued-message-p): Rework slightly to guard against
resetting the "last speaker" marker when the `erc-ephemeral' text
property is present. This tells insert- and send-related hook members
to pretend the current message doesn't exist when performing stateful
operations. That is, modules should expect the message being inserted
to possibly disappear or be replaced. Also, look for `erc-msg' being
`msg' instead of `erc-cmd' being `PRIVMSG', and fix bug involving only
checking `erc-ctcp' in the current message.
* lisp/erc/erc-stamp.el (erc-add-timestamp): Don't insert timestamps
when the `erc-ephemeral' text property is present.
* lisp/erc/erc.el (erc--msg-props): Add doc string explaining the
purpose of this variable and the various text properties most commonly
present in its value.
(erc--msg-prop-overrides): Add doc string.
(erc-send-action): Don't set `erc-cmd' prop on outgoing CTCP ACTIONs.
(erc-display-message): Reverse overrides to prefer items toward the
front of the alist.
(erc-process-ctcp-query): Include existing overrides from environs.
(erc-send-current-line): Include existing overrides from environs.
(erc-display-msg): Fix doc string and reverse overrides.
* test/lisp/erc/erc-fill-tests.el (erc-fill-tests--insert-privmsg):
Remove stray comment.
(erc-fill-tests--save-p): Set value from environment variable.
(erc-fill-tests--compare): Limit writing snapshots to one test at a
time.
(erc-fill-wrap--merge-action): Fix expected output for non-action
messages that follow action messages. These were previously merged
but escaped detection.
* test/lisp/erc/resources/fill/snapshots/merge-wrap-01.eld: Update.
(Bug#60936)
* lisp/erc/erc-stamp.el (erc-stamp--propertize-left-date-stamp): Run
`erc-stamp--insert-date-hook' separately here instead of via
`erc-insert-modify-hook'.
(erc-stamp--insert-date-stamp-as-phony-message): Don't include value
of `erc-stamp--insert-date-hook' in let-bound `erc-insert-modify-hook'
because its members can run twice if buffer-local. Remove
`erc-send-modify-hook' because it only runs via `erc-display-msg'.
Shadow "pre" and "done" hooks because they don't expect to run in a
narrowed buffer. Call getter for `erc-stamp--current-time'.
(erc-stamp--lr-date-on-pre-modify,
erc-insert-timestamp-left-and-right): Use function form of
`erc-stamp--current-time' for determining current time stamp.
* lisp/erc/erc.el (erc--get-inserted-msg-bounds): Fix off-by-one like
thinko.
(erc--traverse-inserted): Create temporary marker when END is a buffer
position so that insertions and deletions are accounted for in
the terminating condition.
(erc--delete-inserted-message): New function.
* test/lisp/erc/erc-tests.el (erc--delete-inserted-message): New test.
(erc--update-modules/unknown): Improve readability slightly.
* test/lisp/erc/resources/erc-d/erc-d-t.el (erc-d-t-make-expecter):
Indicate assertion flavor in error message. (Bug#60936)
* etc/ERC-NEWS: Mention that date stamps no longer optionally affect
the mode line. Also mention but discourage new variable
'erc-stamp-prepend-date-stamps-p'.
* lisp/erc/erc-stamp.el (erc-stamp-prepend-date-stamps-p): New
variable, an escape hatch to allow date stamps to once again be
prepended to messages.
(erc-insert-timestamp-left-and-right): Don't insert stamps as
independent messages when legacy support flag
`erc-stamp-prepend-date-stamps-p' is non-nil.
* lisp/erc/erc-track.el (erc-track--skipped-msgs): New internal
variable.
(erc-track-modified-channels): In previous versions, a date stamp
attached to a message for an IRC command in `erc-track-exclude-types'
would have no effect on the mode line. That they were able to
otherwise was probably a bug. Regardless, this distinction was lost
for the worse after date stamps became independent messages with
c68dc7786f "Manage some text props for ERC insertion-hook members".
To sidestep this regression, the `track' module will ignore date
stamps completely from now on. Thanks to Corwin Brust for spotting
this.
* test/lisp/erc/erc-scenarios-stamp.el
(erc-scenarios-stamp--left/display-margin-mode): Remove redundant
binding.
(erc-scenarios-stamp--legacy-date-stamps): New test. (Bug#60936)
* lisp/erc/erc-goodies.el (erc--scrolltobottom-on-post-command):
Redo obsolete doc string.
(erc--scrolltobottom-at-prompt-minibuffer-active,
erc--scrolltobottom-on-win-conf-change): Rename former to latter to
better reflect actual role. Remove conditional guard so it always
runs.
(erc--scrolltobottom-setup): Set `scroll-step' locally when a user
hasn't customized `scroll-conservatively'. Update
`window-configuration-change-hook' member name. (Bug#64855)
5f60913208 Fix State button for customize-icon (Bug#66635)
27c71979ff ; Another Texinfo fix
889a550ca0 ; Fix Texinfo warnings
893c344b4e Fix the use of adaptive-fill-regexp in treesit indent preset
1098c114b7 Fix treesit-install-language-grammar (bug#66673)
491ee428c0 Fix treesit-explore-mode (bug#66431)
ee043a2703 tsx-ts-mode--font-lock-compatibility-bb1f97b: Re-fix the ...
