* lisp/gnus/message.el (message-insert-screenshot): Force scaling
to 1, since the screenshot image will already be suitable for
displaying directly (it's the same resolution).
* doc/misc/message.texi (MIME): Document it.
* lisp/gnus/message.el (message-screenshot-command): New variable.
(message-mode-map): New keystroke and menu item. Also add
mml-attach-file to the menu.
(message-insert-screenshot): New command.
* lisp/gnus/mml.el (mml-parse-1): Allow having
content-transfer-encoding already in the part, so that we can have
inline base64-encoded binaries in the Message buffers.
* lisp/progmodes/sh-script.el (sh-font-lock-keywords-var): Fontize
$(...) slightly better (bug#42417). Instead of just fontizing the
first word in the expression, fontize until the closing
parenthesis. This doesn't work well if you have nested $(...)
expressions.
* lisp/erc/erc.el (erc--unignore-user): Separate into own function
(bug#40137).
(erc-cmd-IGNORE): Ask if the user wants a timeout.
(erc--read-time-period): New function.
* src/image.c (image_set_transform [HAVE_XRENDER]): Use different filter
when scaling up vs scaling down.
* src/nsimage.m (ns_image_set_smoothing):
([EmacsImage setSmoothing:]): New functions.
* src/nsterm.h: Add definitions.
* src/nsterm.m (ns_dumpglyphs_image): Disable smoothing if requested.
GCC has removed the -fcheck-pointer bounds option, and the Linux
kernel has also removed support for Intel MPX, so there’s no point
to keeping this debugging option within Emacs.
* src/bytecode.c (BYTE_CODE_THREADED):
* src/lisp.h (DEFINE_LISP_SYMBOL, XSYMBOL, make_lisp_symbol):
Assume __CHKP__ is not defined.
* src/ptr-bounds.h: Remove. All uses of ptr_bounds_clip,
ptr_bounds_copy, ptr_bounds_init, ptr_bounds_set removed.
* test/lisp/gnus/mml-sec-tests.el
(mml-first-secure-en-decrypt-sign-1): mml-secure-en-decrypt-sign-1
fail sometimes, on some machines, unless it's the first test. I'm
guessing there's a race condition somewhere in the test, but put
it first now to avoid build reports.
* lisp/gnus/mml-sec.el
(mml-secure-allow-signing-with-unknown-recipient): New variable
(bug#18393) (but this should probably be fixed in a different way).
(mml-secure-epg-sign): Use it.
(mml-secure-check-user-id): Protect against recipients that aren't
email addresses, like "No recipient".
These tests are all written by Gemini Lasswell <gazally@runbox.com>.
* test/lisp/emacs-lisp/cconv-tests.el
(top-level): Add two commented out tests which the byte-compiler
can't handle.
(cconv-tests-lambda-:documentation)
(cconv-tests-pcase-lambda-:documentation)
(cconv-tests-defun-:documentation)
(cconv-tests-cl-defun-:documentation)
(cconv-tests-function-:documentation)
(cconv-tests-cl-defgeneric-literal-:documentation)
(cconv-tests-defsubst-:documentation)
(cconv-tests-cl-defsubst-:documentation): New tests.
(cconv-tests-cl-iter-defun-:documentation)
(cconv-tests-iter-defun-:documentation)
(cconv-tests-iter-lambda-:documentation)
(cconv-tests-cl-function-:documentation)
(cconv-tests-cl-defgeneric-:documentation): New failing tests.
* test/lisp/net/network-stream-tests.el
(network-test--resolve-system-name): New function.
(echo-server-with-dns): Skip test if (system-name) doesn't look
like it's going to resolve (bug#42535).
* test/lisp/vc/vc-tests.el (vc-test--svn-enabled): Macos machines
may have a dummy svn program that helpfully just outputs "There's
no svn program here", so also test for the svnadmin program
(bug#42536).
* doc/misc/tramp.texi (Predefined connection information):
Add "direct-async-process".
(Remote processes): New subsection "Improving performance of
asynchronous remote processes".
* lisp/net/tramp-adb.el (tramp-methods) <adb>: Add `tramp-login-program'
and `tramp-login-args'.
(tramp-adb-handle-make-process): Use `tramp-handle-make-process'.
(tramp-adb-maybe-open-connection): Add "set +o vi +o emacs" command.
* lisp/net/tramp-sh.el (tramp-sh-handle-make-process):
Use `tramp-handle-make-process'.
(tramp-sh-file-name-handler-p, tramp-multi-hop-p): New defuns.
(tramp-compute-multi-hops): Use `tramp-multi-hop-p'.
* lisp/net/tramp.el (tramp-dissect-file-name, tramp-dissect-hop-name):
Use `tramp-multi-hop-p'.
(tramp-handle-insert-file-contents, tramp-local-host-p):
Use `tramp-sh-file-name-handler-p'.
(tramp-handle-make-process): New defun.
* test/README: Add another example how to use SELECTOR.
* test/lisp/net/tramp-tests.el (tramp-test03-file-name-method-rules):
Adapt test.
(tramp--test-sh-p): Use `tramp-sh-file-name-handler-p'.
* src/xdisp.c (with_echo_area_buffer): Pass void * instead of
ptrdiff_t, since the values are typically pointers and this ports
better to (mostly-theoretical) hosts where ptrdiff_t is narrower
than intptr_t. All uses changed.
* configure.ac: Use AC_CHECK_FUNCS_ONCE for __lsan_ignore_object.
* src/buffer.c, src/data.c, src/emacs-module.c, src/regex-emacs.c:
* src/search.c: Use __lsan_ignore_object unconditionally, and don’t
include sanitizer/lsan_interface.h.
* src/lisp.h (__lsan_ignore_object): Provide a dummy in the
typical case where leak sanitization is not available.
* src/alloc.c (mark_maybe_object):
Use simpler way to avoid -fsanitize=undefined false alarms,
by converting the word tag to intptr_t first.
Omit now-unnecessary runtime overflow check.
(mark_memory): Work even if UINTPTR_MAX <= INT_MAX (!).
When Edebug backtracks, it nevertheless generates definitions for the
non-matching branches, see Bug#41988 and Bug#42701. This should be
fixed eventually (probably by deferring the definition until a branch
is known to match), but for now add a band-aid to avoid these
duplicate symbols, at least for anonymous forms.
* lisp/emacs-lisp/edebug.el (edebug-make-enter-wrapper): Regenerate
anonymous names.
* test/lisp/emacs-lisp/edebug-tests.el
(edebug-tests-duplicate-symbol-backtrack): New regression test.
The following commits were skipped:
d767418b76 Backport: Make checking for liveness of global values more...
8c94ca94dc Backport: Fix subtle bug when checking liveness of module ...
8ecca2f09f Backport: Fix memory leak for global module objects (Bug#4...
* lisp/so-long.el (so-long-detected-long-line-p): Ensure that
`comment-start-skip' and `comment-end-skip' are both set if
`comment-use-syntax' is nil, as `comment-forward' requires them
to be bound in this scenario.
As this `require' is not at the top-level (it is only conditionally
evaluated, when loading the library over the top of an earlier
version), we need `eval-and-compile' to ensure that both macros and
functions from advice.el are accounted for.
Reverting certain changes from commits b0f20651e3 and d1a791f8ed.
Please refer to the comments on spelling at the end of the library.
M-x ispell-buffer should find no misspellings in the documentation.
See also test/lisp/so-long-tests/spelling-tests.el
(The current spelling will persist while so-long.el is maintained in
its own Savannah repository, to avoid unnecessary conflicts between
the two versions. If in the future it is maintained solely in the
Emacs repository, changing the spelling would become an option.)
Note that "mitigations" (plural) is intentional -- this library
identifies a collection of different performance mitigations,
multiple of which will typically be in effect together.
* lisp/image-file.el (image-file-name-regexp): Use it to make
`n'/`p' in image mode work (bug#39994).
* lisp/image/image-converter.el
(image-converter-file-name-extensions): New variable to keep track
of all suffixes.
(image-convert-p): Update.
(image-converter--find-converter): Set.
* lisp/image-mode.el (image-toggle-display-image): Make it
possible to view images (via external formatters, like webp) from
zip files (and other archive modes) (bug#39994).
* lisp/image-mode.el (image-mode): Even when
`image-user-external-converter' is on, we may get
`unknown-image-type' (bug#39994). Adjust the error message in
that case.
* src/process.c (wait_reading_process_output): Before the select,
check every interesting gnutls stream for available data in the
buffer. If some of them hit, and either there is no wait_proc or
the wait_proc is one of the gnutls streams with new data, set the
select timeout to 0 after the select, and merge the gnutls buffer
status into the select returns (bug#40665). This fixes a problem
where TLS connections would sometimes hang.
