GitLab

This commit fixes Bug#31492.
* lisp/replace.el (replace-match-maybe-edit): Preserve match data.

* test/lisp/replace-tests.el (query-replace-undo-bug31492): Add test.

e3f00f56 (origin/emacs-26) Clarify when to use advice-add vs add-function
c0f3e6b5 Minor formatting tweak in the Emacs manual
fadf6535 Update comments in Intro to Emacs Lisp
52ccb24a ; * doc/os.texi (Batch Mode): use = after --eval
d65430f6 * etc/NEWS: Mention change in `edebug-prin1-to-string'.
ef4aec14 ; * doc/lispref/hash.texi (Defining Hash): Fix typos.
2981952b Another attempt to fix sql.el
b1b96d7c Update binding of 'M-.' in Intro to Emacs Lisp
b239a096 Fix a typo in last change in sql.el
0d8bae5c Fix capitalization of mail headers
845fe038 Fix buffer names in sql.el
a3885f56 Minor fixes in the Emacs manual
542f8302 Fix a typo in rmail.texi
eb0bc6f7 * etc/PROBLEMS: Document Bug#31305.

Conflicts:
	doc/emacs/sending.texi
	etc/NEWS

The following commit was skipped:

152a5569 (gnus-blocked-images): Clarify privacy implications

73bc6f86 Fix a typo in describing input methods
6b4bafed ; Add a comment to emacs-lisp-intro.texi

* lisp/emacs-lisp/nadvice.el (add-function): Note in docstring that
advice-add should be used for named functions.

* test/lisp/electric-tests.el (electric-pair-define-test-form):
Improve automatically generated docstring.

* src/alloc.c (PNTR_ADD): Put attributes after ‘static’ and before
returned type.  Problem reported by Eli Zaretskii in:
https://lists.gnu.org/r/emacs-devel/2018-05/msg00559.html

Problem reported by Philipp Stephani (Bug#31556).
* src/lread.c (read_vector): When load_force_doc_strings, check
for byte code vectors that are invalid because they are too short.
* test/src/lread-tests.el (lread-invalid-bytecodes): New test.

* doc/emacs/custom.texi (Directory Variables): Don't have 2
footnotes back to back, it looks bad in HTML.  (Bug#31555)

* doc/lispintro/emacs-lisp-intro.texi: Explain in the comments
what are roman page numbers in Preface about.  (Bug#31541)

* src/xwidget.c (webkit_javascript_finished_cb)
(Fxwidget_webkit_execute_script): Use intptr_t to avoid warnings
in the (unlikely) event that ptrdiff_t and void * differ in width.
(save_script_callback): Simplify by using xlispdstrdup and
larger_vector.

* lisp/net/secrets.el (secrets-search-item-paths): New function.
(secrets-search-items): Use it.
(secrets-create-item): Adapt docstring.

* test/lisp/net/secrets-tests.el (secrets-test04-search): Extend test.

* src/xwidget.h (struct xwidget): Add script_callbacks.
* src/xwidget.c (save_script_callback): New function.
(Fxwidget_webkit_execute_script): Use it.  Encode script
before passing to execution engine.  Always use a callback.
(webkit_javascript_finished_cb): Deallocate script.
(kill_buffer_xwidgets): Deallocate remaining scripts.
(Fxwidget_webkit_zoom): Doc fix.
(Fxwidget_resize): Doc fix.

* test/src/process-tests.el (process-tests--mixable): New function.
(make-process/mix-stderr): Don’t assume stdout is merged before
stderr.  POSIX does not require this, and the assumption failed to
hold on my Fedora 28 platform.  See Bug#31214.

This incorporates:
2018-05-21 crypto: omit stream ops Emacs doesn’t need
2018-05-13 truncate: Fix compilation error on Android
2018-05-13 imaxdiv: Fix compilation error on Android
2018-05-13 Support selective inclusion of recent mingw.org headers
2018-05-13 Add cross-compilation guesses for Linux systems sans glibc
2018-05-13 stdioext: Fix compilation errors with newer Android headers
2018-05-07 af_alg: Pacify --enable-gcc-warnings
2018-05-06 af_alg: Fix bug with streams that are not at position 0
2018-05-06 Followup to 'af_alg: New module'
2018-05-05 crypto/{md5,sha1,sha256,sha512}: simplify
2018-05-05 af_alg: New module
2018-05-05 af_alg: Improve function signature
2018-04-28 md5sum: Use AF_ALG when available
2018-04-28 sha512sum: Use AF_ALG when available
2018-04-28 sha256sum: Use AF_ALG when available
2018-04-28 sha1sum: Use AF_ALG when available
2018-05-05 all: Replace more http URLs by https URLs
2018-05-03 maint: port more modules to GCC 8
2018-05-03 Simplify code; drop support for Borland C++ on Windows
* admin/merge-gnulib (GNULIB_MODULES): Use crypto/md5-buffer
rather than crypto/md5, since Emacs doesn’t use the stream
operations that in recent Gnulib pull in other stuff Emacs doesn’t
need.  Similarly for crypto/sha1-buffer, crypto/sha256-buffer,
crypto/sha512-buffer.
* build-aux/config.guess, build-aux/config.sub, lib/dosname.h:
* lib/dup2.c, lib/errno.in.h, lib/euidaccess.c, lib/fcntl.c:
* lib/fcntl.in.h, lib/fpending.c, lib/fsync.c, lib/getdtablesize.c:
* lib/getopt.c, lib/gettimeofday.c, lib/inttypes.in.h, lib/md5.c:
* lib/md5.h, lib/open.c, lib/pipe2.c, lib/putenv.c, lib/sha1.c:
* lib/sha1.h, lib/sha256.c, lib/sha256.h, lib/sha512.c:
* lib/sha512.h, lib/stat-time.h, lib/stdio-impl.h, lib/stdio.in.h:
* lib/stdlib.in.h, lib/sys_stat.in.h, lib/sys_types.in.h:
* lib/timespec.h, lib/unistd.in.h, lib/utimens.c, m4/c-strtod.m4:
* m4/gnulib-common.m4, m4/inttypes.m4, m4/lstat.m4, m4/nocrash.m4:
* m4/pselect.m4, m4/readlink.m4, m4/stdio_h.m4, m4/symlink.m4:
* m4/unistd_h.m4, m4/utimens.m4:
Copy from Gnulib.
* lib/gnulib.mk.in, m4/gnulib-comp.m4: Regenerate.

The Flymake test suite fails if the "rubocop" program is installed,
because the ruby-flymake-rubocop backend is selected automatically by
ruby-flymake-auto.  The test was designed for ruby-flymake-simple,
tough, and fails.

* test/lisp/progmodes/flymake-tests.el (ruby-backend): Ensure
this test runs exclusively with the ruby-flymake-simple backend.

* doc/misc/tramp.texi (Frequently Asked Questions):
Mention `tramp-ignored-file-name-regexp'.  Improve index.

; * etc/NEWS: Mention `tramp-ignored-file-name-regexp'.

* lisp/net/tramp.el (tramp-ignored-file-name-regexp): New defcustom.
(tramp-tramp-file-p): Use it.  Check also for `tramp-mode'.
(tramp-file-name-handler): Don't check for `tramp-mode'.  (Bug#31489)

* test/lisp/net/tramp-tests.el (tramp-test01-file-name-syntax):
Extend test.

* lisp/progmodes/sql.el (sql-product-interactive): Fix calculation
of the SQL buffer name.  (Bug#31446)

* doc/lispintro/emacs-lisp-intro.texi (On Reading this Text)
(Finding More, Buffer Related Review, Buffer Exercises)
(Find a File, Conclusion): 'M-.' is now bound to
'xref-find-definitions'.  (Bug#31542)

* lisp/progmodes/sql.el (sql-product-interactive): Use 'null',
not 'zerop'.  (Bug#31446)

* doc/misc/message.texi:
* doc/misc/efaq.texi:
* doc/emacs/sending.texi:
* doc/emacs/rmail.texi: Revert the CC => Cc etc. conversions.

Also fix similar problems with Bcc, Fcc, In-Reply-To,
and similar email headers.  See thread starting at:
https://lists.gnu.org/r/emacs-devel/2018-05/msg00463.html

String delimiters, including escaped new lines, of correctly terminated
strings are left in font-lock-string-face.  All others get
font-lock-warning-face.  The latter get syntax-table text properties on the
opening string delim and the "terminating EOL".

Correct two miscellaneous bugs: the handling of text properties on Java Mode's
generic delimiters; the handling of c-just-done-before-change.

* lisp/progmodes/cc-defs.el (c-point): New position 'eoll "end of logical line".
(c-characterp): New macro.

* lisp/progmodes/cc-fonts.el (c-font-lock-invalid-string): Removed.
(c-basic-matchers-before): Use a simple matcher in place of the form around
c-font-lock-invalid-string.

* lisp/progmodes/cc-langs.el (c-get-state-before-change-functions): Add
c-before-change-check-unbalanced-strings to the value for all modes except AWK
Mode.  Also add c-before-change-check-<>-operators to Java Mode, correcting an
error in that mode's handling of generic delimiters.
(c-before-font-lock-functions): Add c-after-change-re-mark-unbalanced-strings
to the value for all modes except AWK Mode.
(c-single-quotes-quote-strings, c-string-delims): New lang variables for
future enhancements.
(c-string-innards-re-alist): New lang variable.

* lisp/progmodes/cc-mode.el (c-just-done-before-change): Do not set this
variable when a change is the alteration of text properties.
(c-basic-common-init): Set parse-sexp-lookup-properties (and the XEmacs
equivalent) also for Pike Mode.
(c-neutralize-CPP-line): No longer neutralize unbalanced quotes here.
(c-unescaped-nls-in-string-p, c-multiline-string-start-is-being-detached)
(c-pps-to-string-delim, c-before-change-check-unbalanced-strings)
(c-after-change-re-mark-unbalanced-strings): New functions.
(c-after-change): Fix a bug with the handling of c-just-done-before-change.

* doc/emacs/sending.texi (Mail Format, Mail Headers)
(Mail Aliases, Header Editing): Use consistent capitalization of
CC, BCC, and FCC.  (Bug#31532)

* lisp/progmodes/sql.el (sql-product-interactive): Fix the way the
buffer name is determined by prefix arg.  (Bug#31446)

* doc/emacs/misc.texi (Gnus Group Buffer): Fix a typo.  (Bug#31534)
(Single Shell): Rephrase an unclear sentence.  (Bug#31535)
(Remote Host): Use @Command where appropriate.  (Bug#31536)

* doc/emacs/rmail.texi (Rmail Summary Edit): Add markup to 'nil'.
(Bug#31533)

This allows to encode HELLO in UTF-8, thus supporting the entire
repertory of Unicode, while still keeping the charset info where
that is important.  Suggested by Michael Welsh Duggan <mwd@md5i.com>.
* lisp/textmodes/enriched.el (enriched-translations): Add
translations for 'charset'.
(enriched-decode-charset, enriched-handle-charset-prop): New
functions.
* lisp/facemenu.el (facemenu-special-menu): Add sub-menu for
'charset' property.
(facemenu-set-charset): New function.
(facemenu-remove-special): Remove the 'charset' property as well.

* etc/NEWS: Announce the new feature of Enriched mode.
* etc/HELLO: Recode in UTF-8 and place under Enriched mode.

* doc/emacs/text.texi (Enriched Properties): Mention the support
for 'charset'.

In GCC 8, gcc -fsanitize=undefined flags the undefined behavior
that Emacs relies on in its XPNTR and XSYMBOL low-level functions.
Disable undefined sanitization in these functions.  Although this
disabling doesn’t suffice if DEFINE_KEY_OPS_AS_MACROS is true, it
works for -fsanitize=undefined -DINLINING=0, which is good enough.
* src/alloc.c (macro_PNTR_ADD): New macro.
(PNTR_ADD): New function and macro.
The function disables -fsanitize=undefined.
(macro_XPNTR): Use it.
* src/conf_post.h (ATTRIBUTE_NO_SANITIZE_UNDEFINED): New macro.
* src/lisp.h (XSYMBOL): Disable -fsanitize=undefined.

* lisp/gnus/gnus-art.el (gnus-blocked-images): Clarify the privacy
implication of altering the value of this variable.

(cherry picked from commit f4d9fd3d)

* lisp/gnus/gnus-art.el (gnus-blocked-images): Clarify the privacy
implication of altering the value of this variable.