Paul Eggert [Fri, 19 May 2017 07:11:48 +0000 (00:11 -0700)]
Attempt to work around macOS vfork bug
Problem reported by YAMAMOTO Mitsuharu in:
http://lists.gnu.org/archive/html/emacs-devel/2017-05/msg00342.html
This is related to the fix for Bug#26397.
* src/callproc.c (call_process_cleanup, call_process) [!MSDOS]:
Report internal error if wait_for_termination fails.
* src/sysdep.c (get_child_status): Return -1 if waitpid is
buggy, instead of aborting.
(wait_for_termination): Return bool success value.
All callers changed.
Paul Eggert [Fri, 19 May 2017 03:40:42 +0000 (20:40 -0700)]
Fix DARWIN_OS_CASE_SENSITIVE_FIXME==2 false alarm
* src/fileio.c (file_name_case_insensitive_p):
Don’t compile the (DARWIN_OS_CASE_SENSITIVE_FIXME == 2)
code unless DARWIN_OS_CASE_SENSITIVE_FIXME is 2.
Problem reported by Philipp Stephani in:
http://lists.gnu.org/archive/html/emacs-devel/2017-05/msg00495.html
Paul Eggert [Fri, 19 May 2017 03:27:11 +0000 (20:27 -0700)]
Port --enable-gcc-warnings to clang 3.9.1
* configure.ac (WERROR_CFLAGS): Omit -Wdouble-promotion if clang.
Problem reported by Philipp Stephani in:
http://lists.gnu.org/archive/html/emacs-devel/2017-05/msg00495.html
Paul Eggert [Fri, 19 May 2017 03:02:42 +0000 (20:02 -0700)]
Clean up compiler warning in emacs-module.c
* src/emacs-module.c (MODULE_SETJMP_1): Use the local var
instead of leaving it unused, to pacify picky compilers.
(module_reset_handlerlist): Now takes a dummy pointer to a struct
handler *, instead of a dummy pointer to an int. All uses changed.
Eli Zaretskii [Thu, 18 May 2017 20:18:29 +0000 (23:18 +0300)]
Support hscrolling only the current line
* src/xdisp.c (hscrolling_current_line_p): New function.
(init_iterator): If auto-hscrolling just the current line, don't
increment the iterator's first_visible_x and last_visible_x
variables.
(hscroll_window_tree): Recompute window's hscroll when moving
vertically to another screen line.
(redisplay_window): If we are hscrolling only the current line,
disable the optimizations that rely on the current matrix being
up-to-date.
(display_line): Accept an additional argument CURSOR_VPOS, the
vertical position of the current screen line which might need
hscrolling; all callers changed. Compute first_visible_x and
last_visible_x specially when auto-hscrolling current line, by
repeating the calculation that is done in init_iterator in other
modes.
(syms_of_xdisp) <auto-hscroll-mode>: No longer boolean, it can now
accept a 3rd value 'current-line, to turn on the mode where
only the current line is hscrolled.
* etc/NEWS: Mention the new auto-hscroll-mode value.
Eli Zaretskii [Thu, 18 May 2017 19:54:35 +0000 (22:54 +0300)]
Fix last change in line-move-finish
* lisp/simple.el (line-move-finish): Fix last change. This corrects a
regression in C-n and C-p when lines are truncated, introduced by the
change in 2017-05-10.
Tino Calancha [Thu, 18 May 2017 05:47:06 +0000 (14:47 +0900)]
Use the expression angle units while simplifying it
Don't use the angle mode, use the angle units included
in the expression instead (Bug#23889).
* lisp/calc/calc-alg.el (calc-input-angle-units): New defun.
(math-simplify): Use it.
* lisp/calc/calc-forms.el (math-to-hms, math-from-hms):
Don't use calc-angle-mode if math-simplifying-units is non-nil.
* lisp/calc/calc-math.el (calcFunc-nroot, math-from-radians)
(math-to-radians-2, math-from-radians-2): Don't convert angle
to radians if math-simplifying-units is non-nil.
* test/lisp/calc/calc-tests.el (test-calc-23889): Add test.
Paul Eggert [Wed, 17 May 2017 20:35:52 +0000 (13:35 -0700)]
Avoid undefined behavior in struct sockaddr
Problem noted by Philipp Stephani in:
http://lists.gnu.org/archive/html/emacs-devel/2017-05/msg00391.html
* src/conf_post.h (ATTRIBUTE_MAY_ALIAS, DECLARE_POINTER_ALIAS):
New macros.
* src/process.c (conv_sockaddr_to_lisp, conv_lisp_to_sockaddr)
(connect_network_socket, network_interface_info)
(server_accept_connection): Use it when aliasing non-char objects.
Stefan Monnier [Wed, 17 May 2017 19:44:36 +0000 (15:44 -0400)]
* lisp/vc/smerge-mode.el (smerge-refine-regions): Work in multi-bufs
Rename from smerge-refine-subst. Allow the `beg's to be markers.
Add autoload cookie.
(smerge--refine-forward): Rename from smerge-refine-forward.
(smerge--refine-chopup-region): Rename from smerge-refine-chopup-region.
Assume that its `beg` arg is a marker.
(smerge--refine-highlight-change): Rename from
smerge-refine-highlight-change. Remove `buf` arg.
(smerge-refine-subst): Redefine as an obsolete alias.
Paul Eggert [Wed, 17 May 2017 17:58:11 +0000 (10:58 -0700)]
Work around AddressSanitizer bug with vfork
Problem reported by Jim Meyering in:
http://lists.gnu.org/archive/html/emacs-devel/2017-05/msg00246.html
* src/conf_post.h (vfork) [ADDRESS_SANITIZER]: Define to fork.
Unfortunately with the AddressSanitizer in Fedora 25 x86-64, the
vforked child messes up the parent’s shadow memory. This is too
bad, as we’d rather have AddressSanitizer catch memory-access bugs
related to vfork.
Paul Eggert [Wed, 17 May 2017 17:27:31 +0000 (10:27 -0700)]
Catch IPv4/IPv6 issues at compile time
* src/process.c (connect_network_socket): Use verify,
not eassert, so that any problems are caught at compile-time.
Avoid dodgy cast by using a local var of the correct type.
Eli Zaretskii [Wed, 17 May 2017 15:12:58 +0000 (18:12 +0300)]
Remove redundant code in connect_network_socket
* src/process.c (connect_network_socket) [HAVE_GETSOCKNAME]:
Remove redundant type-casting and variables. Don't call
'getsockname' to find the port for AF_LOCAL sockets.
[AF_INET6]: Add an assertion to verify that the ports in the IPv4
and IPv6 structures are at the same offset and have the same size.
Paul Eggert [Tue, 16 May 2017 21:29:18 +0000 (14:29 -0700)]
Simplify procname code to avoid GCC bug
* src/process.c (server_accept_connection): Simplify and avoid
multiple calls and struct literals in the last case of a switch.
The old code ran afoul of GCC bug 80659, which caused an internal
compiler error. Problem reported by Jim Meyering in:
http://lists.gnu.org/archive/html/emacs-devel/2017-05/msg00182.html
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80659
Paul Eggert [Tue, 16 May 2017 17:24:19 +0000 (10:24 -0700)]
Merge with gnulib, pacifying GCC 7
This incorporates:
2017-05-16 manywarnings: update for GCC 7
2017-05-15 sys_select: Avoid "was expanded before it was required"
* configure.ac (nw): Suppress GCC 7’s new -Wduplicated-branches and
-Wformat-overflow=2 options, due to too many false alarms.
* doc/misc/texinfo.tex, lib/strftime.c, m4/manywarnings.m4:
Copy from gnulib.
* m4/gnulib-comp.m4: Regenerate.
* src/coding.c (decode_coding_iso_2022):
Fix bug uncovered by -Wimplicit-fallthrough.
* src/conf_post.h (FALLTHROUGH): New macro.
Use it to mark all switch cases that fall through.
* src/editfns.c (styled_format): Use !, not ~, on bool.
* src/gtkutil.c (xg_check_special_colors):
When using sprintf, don’t trust Gtk to output colors in [0, 1] range.
(xg_update_scrollbar_pos): Avoid use of possibly-uninitialized bool;
this bug was actually caught by Clang.
* src/search.c (boyer_moore):
Tell GCC that CHAR_BASE, if nonzero, must be a non-ASCII character.
* src/xterm.c (x_draw_glyphless_glyph_string_foreground):
Tell GCC that glyph->u.glyphless.ch must be a character.
Michael Albinus [Tue, 16 May 2017 13:55:06 +0000 (15:55 +0200)]
Make autoloading Tramp more robust
* lisp/net/tramp.el (tramp-file-name-for-operation):
Use `default-directory' where appropriate.
(tramp-file-name-handler): Do not autoload.
(tramp-autoload-file-name-handler): Reintroduce function.
(tramp-register-autoload-file-name-handlers): Use it.
Glenn Morris [Tue, 16 May 2017 01:01:30 +0000 (21:01 -0400)]
Stop some epg tests failing on rhel7 with gpg 2.0.22 (bug#23619)
* test/lisp/epg-tests.el (with-epg-tests):
Also set GNUPGHOME in the environment of child processes.
This avoids problems if gpg does not pass --homedir to spawned agent.
; Ref eg https://gpgtools.tenderapp.com/discussions/problems/29537-gpg2-doesnt-start-agent-properly-if-using-homedir
Eli Zaretskii [Mon, 15 May 2017 19:45:57 +0000 (22:45 +0300)]
Remove unneeded stuff from nt/inc/sys/time.h
* nt/inc/sys/time.h (_TIMEVAL_DEFINED, struct timevat, timerisset)
(timercmp, timerclear): Don't define. Instead, include the system
header sys/time.h, and add only the interval timers stuff. This
avoids compiler warnings about 'gettimeofday's prototype, and also
avoids redefinition of macros from system headers.
Paul Eggert [Mon, 15 May 2017 01:44:36 +0000 (18:44 -0700)]
Fix address violation found by AddressSanitizer
* src/process.c (connect_network_socket):
Use struct sockaddr_storage, not struct sockaddr_in, to store info
about a socket address. Problem reported by Philipp Stephani in:
http://lists.gnu.org/archive/html/emacs-devel/2017-05/msg00314.html
This fix is based on a patch by Philipp in:
http://lists.gnu.org/archive/html/emacs-devel/2017-05/msg00357.html
Gemini Lasswell [Sat, 13 May 2017 18:35:49 +0000 (11:35 -0700)]
Make edebug-step-in work on generic methods (Bug#22294)
* lisp/emacs-lisp/edebug.el (edebug-match-cl-generic-method-args):
New function to implement the edebug-form-spec property of
the symbol cl-generic-method-args.
(edebug-instrument-function): If the function is a generic
function, find and instrument all of its methods. Return a list
instead of a single symbol.
(edebug-instrument-callee): Now returns a list. Update docstring.
(edebug-step-in): Handle the list returned by edebug-instrument-callee.
* lisp/emacs-lisp/cl-generic.el (cl-defmethod): Use name and
cl-generic-method-args in its Edebug spec.
* lisp/emacs-lisp/eieio-compat.el (defmethod): Use name and
cl-generic-method-args in its Edebug spec.
* lisp/subr.el (method-files): New function.
* test/lisp/subr-tests.el (subr-tests--method-files--finds-methods)
(subr-tests--method-files--nonexistent-methods): New tests.
Eli Zaretskii [Sun, 14 May 2017 16:02:50 +0000 (19:02 +0300)]
Remove gettimeofday from w32 sources
* lib-src/ntlib.c (gettimeofday):
* nt/inc/sys/time.h (gettimeofday, struct timezone): Remove unused
function 'gettimeofday' and all of its supporting code.
Paul Eggert [Sun, 14 May 2017 08:29:05 +0000 (01:29 -0700)]
Merge from gnulib
This incorporates:
2017-05-13 largefile: Simplify
2017-05-13 largefile: Improve and document
2017-05-13 truncate: New module
2017-05-13 windows-stat-timespec: New module
2017-05-13 windows-stat-override: New module
2017-05-11 getopt-posix: port to mingw
2017-05-11 gettimeofday: Increase precision on mingw
2017-05-10 time: Fix missing initialization of HAVE_TIMEZONE_T
2017-05-10 Implement a way to opt out from MSVC support
2017-05-09 tzset: Expand comment about TZ problem on native Windows
* build-aux/config.guess, lib/dup2.c, lib/fcntl.c, lib/fsync.c:
* lib/getdtablesize.c, lib/getopt.c, lib/gettimeofday.c:
* lib/mktime.c, lib/stat-time.h, lib/sys_stat.in.h, lib/unistd.in.h:
* lib/utimens.c, m4/gettimeofday.m4, m4/largefile.m4:
* m4/sys_stat_h.m4, m4/sys_time_h.m4, m4/time_h.m4, m4/time_rz.m4:
* m4/unistd_h.m4: Copy from gnulib.
* lib/gnulib.mk.in, m4/gnulib-comp.m4: Regenerate.
Philipp [Sat, 6 May 2017 20:23:03 +0000 (22:23 +0200)]
Make `old-style-backquotes' variable internal
* src/lread.c (load_warn_old_style_backquotes, Fload, read1)
(syms_of_lread): Rename `old-style-backquotes' to
`lread--old-style-backquotes', and clarify that it's for internal
use only.
* lisp/emacs-lisp/bytecomp.el (byte-compile-from-buffer): Rename
variable.
* test/src/lread-tests.el (lread-tests--old-style-backquotes): Add
unit test.
* emacs-lisp/bytecomp-tests.el
(bytecomp-tests--old-style-backquotes): Add unit test.
Alan Mackenzie [Fri, 12 May 2017 21:11:49 +0000 (21:11 +0000)]
Fontify C++ for loop variable as variable, even when followed by parentheses
In the following: "for (auto *Friend : Class->friends()) {", "Friend" was
getting fontified as a function, due to insufficient checking of the tokens
between it and "()".
* lisp/progmodes/cc-langs.el (c-:-op-cont-tokens, c-:-op-cont-regexp): New
lang-consts/vars.
* lisp/progmodes/cc-engine.el (c-forward-declarator): After finding a putative
declarator's identifier, check for a ":" token inside a for's parentheses, and
abort the search for "(" if this is found.
Michael Albinus [Fri, 12 May 2017 16:36:41 +0000 (18:36 +0200)]
Make Tramp backward compatible
* lisp/net/tramp-cmds.el (tramp-change-syntax):
Set tramp-autoload cookie.
* lisp/net/tramp-compat.el: Run `tramp-change-syntax' at
startup, if necessary.
* lisp/net/tramp.el (tramp-syntax): Use `tramp-compat-user-error'.
(tramp-register-autoload-file-name-handlers): Do not mark
`operations' for `tramp-file-name-handler'.
(tramp-register-file-name-handlers): Remove also
`tramp-autoload-file-name-handler' for backward compatibility.
(tramp-register-foreign-file-name-handler): Use `delete-dups'.
* test/lisp/emacs-lisp/lisp-tests.el (elisp-tests-with-temp-buffer):
Don't refer to the =!NAME= as "markers" since they produce variables
with just plain positions, not marker objects. Explicitly specify
that CONTENTS is evaluated at compile time. Don't re-evaluate
CONTENTS at runtime. Fix debug specification. Suppress warnings due
to BODY not using =!NAME= variables.
(elisp-test-point-position-regex): Rename from
`elisp-test-point-marker-regex'.
(mark-defun-test-buffer): Wrap in `eval-and-compile'.
Marcin Borkowski [Fri, 31 Mar 2017 11:06:06 +0000 (13:06 +0200)]
Fix Bug#21072 and rework `mark-defun'
* test/lisp/progmodes/elisp-mode-tests.el (mark-defun-test-buffer):
New variable
(mark-defun-no-arg-region-inactive)
(mark-defun-no-arg-region-active)
(mark-defun-arg-region-active)
(mark-defun-pos-arg-region-inactive)
(mark-defun-neg-arg-region-inactive, mark-defun-bob): Add tests for
the new `mark-defun'.
* lisp/emacs-lisp/lisp.el (beginning-of-defun--in-emptyish-line-p):
New function.
(beginning-of-defun-comments): New function.
(mark-defun): Fix bug#21072, also rewrite large parts of `mark-defun'
to accept a numerical prefix argument.
Andrew Robbins [Fri, 12 May 2017 08:19:46 +0000 (11:19 +0300)]
Extend DNS lookup commands to allow specifying the name server
* lisp/net/net-utils.el (ffap-string-at-point): Removed due to
'net-utils-machine-at-point' obviating this autoloaded
function (Bug#25426).
(dig-program-options): New customization variable.
(nslookup-host, dns-lookup-host, run-dig): Can now specify
optional name server argument interactively (by prefix arg) and
non-interactively.
* etc/NEWS: Mention the extension of DNS lookup commands.
Eli Zaretskii [Wed, 10 May 2017 17:06:15 +0000 (20:06 +0300)]
Ensure cursor's foreground color is in sync with 'default' face
* src/w32term.c (x_set_cursor_gc): Don't reuse cursor GC if its
foreground color is different from the background of the glyph
string's face. (Bug#26851)
Eli Zaretskii [Wed, 10 May 2017 16:58:06 +0000 (19:58 +0300)]
Fix vertical cursor motion when columns are of unequal size
* lisp/simple.el (line-move-finish): In line-move-visual mode, use
vertical-motion to move to the goal column, as the goal column
should in that case be interpreted in units of frame's canonical
character width. (Bug#26852)
Tino Calancha [Wed, 10 May 2017 02:53:20 +0000 (11:53 +0900)]
Tweak a recent test
This test fails in my local machine because the data files
are compiled, and the test doesn't expect that.
* test/src/lread-tests.el (lread-test-bug26837): Match a suffix
ending with '.elc' when the data files are compiled.
Glenn Morris [Wed, 10 May 2017 01:31:46 +0000 (21:31 -0400)]
Put license information in each generated uni-*.el
* admin/unidata/unidata-gen.el (unidata-gen-file):
Get Copyright line from copyright.html.
Put information in file header, not separate README.
(unidata-gen-charprop): Mention the source location.
* lisp/international/README: Remove file.
Fix lisp-indent-region and indent-sexp (Bug#26619)
The new lisp-indent-region introduced in 2017-04-22 "Add new
`lisp-indent-region' that doesn't reparse the code." is broken because
it doesn't save the calculated indent amounts for already seen sexp
depths. Fix this by unifying the indent-sexp and lisp-indent-region
code. Furthermore, only preserve position 2 of the running parse
when the depth doesn't change.
* lisp/emacs-lisp/lisp-mode.el (lisp-ppss): Use an OLDSTATE that
corresponds with the start point when calling parse-partial-sexp.
(lisp-indent-state): New struct.
(lisp-indent-calc-next): New function, extracted from indent-sexp.
(indent-sexp, lisp-indent-region): Use it.
(lisp-indent-line): Take indentation, instead of parse state.
* test/lisp/emacs-lisp/lisp-mode-tests.el
(lisp-mode-tests--correctly-indented-sexp): New constant.
(lisp-indent-region, lisp-indent-region-defun-with-docstring):
(lisp-indent-region-open-paren, lisp-indent-region-in-sexp): New
tests.
Dmitry Gutov [Wed, 10 May 2017 00:34:16 +0000 (03:34 +0300)]
Simplify url-encode-url and add a test
* lisp/url/url-util.el (url-encode-url): Simplify.
url-generic-parse-url copes with multibyte strings just fine
(https://debbugs.gnu.org/cgi/bugreport.cgi?bug=24117#185).
* test/lisp/url/url-parse-tests.el
(url-generic-parse-url/multibyte-host-and-path): New test.
Glenn Morris [Tue, 9 May 2017 23:44:09 +0000 (19:44 -0400)]
Put re-loaded file back at start of load-history (bug#26837)
* src/lread.c (readevalloop): Fix the "whole buffer" check to
operate in the correct buffer.
(Feval_buffer): Move point back to the start after checking
for lexical binding.
* test/src/lread-tests.el (lread-test-bug26837): New test.
* test/data/somelib.el, test/data/somelib2.el: New test data files.
Marcin Borkowski [Tue, 14 Feb 2017 10:30:36 +0000 (11:30 +0100)]
Add elisp-tests-with-temp-buffer, a new testing macro
* test/lisp/emacs-lisp/lisp-tests.el
(elisp-test-point-marker-regex) New variable.
(elisp-tests-with-temp-buffer): New macro to help test functions
moving the point and/or mark.
Noam Postavsky [Tue, 9 May 2017 02:17:53 +0000 (22:17 -0400)]
Revert "Output number of characters added to file (Bug#354)"
The extra message text turned out to be quite annoying in practice,
and is generally more trouble than it's worth. Also revert several
related changes.
Partially revert "Handle `write-region' messages in Tramp properly"
Revert "New var write-region-verbose, default nil"
Revert "* src/fileio.c (write_region): Don't say "1 characters". (Bug#26796)"
Revert "Minor tuneup of write-region change"
Revert "Adjust write-region so file name is at the beginning again"
Revert "Fix handling of non-integer START param to write-region"
Revert "Output number of characters added to file (Bug#354)"
Ken Brown [Mon, 8 May 2017 19:11:23 +0000 (15:11 -0400)]
Skip a test from filenotify-tests.el on Cygwin
* test/lisp/filenotify-tests.el (file-notify-test02-rm-watch):
Skip the last part of the test on Cygwin; it fails due to timing
issues.
(file-notify--test-read-event): Remove `sit-for' that was added
for Cygwin.
Paul Eggert [Mon, 8 May 2017 17:46:21 +0000 (10:46 -0700)]
Merge from gnulib
This incorporates:
2017-05-08 intprops: don’t depend on ‘verify’
2017-05-07 utimens: on native Windows, improve resolution if fd < 0
2017-05-07 utimens: Improve error code on native Windows
* lib/intprops.h, lib/utimens.c: Copy from gnulib.
Wilson Snyder [Mon, 8 May 2017 17:44:47 +0000 (13:44 -0400)]
Fix various verilog-mode.el issues.
* lisp/progmodes/verilog-mode.el (verilog-read-decls): Fix SystemVerilog
2012 import breaking AUTOINST. Reported by Johannes Schaefer.
(verilog-auto-wire-type, verilog-insert-definition): Fix AUTOWIRE using
logic in top-level non-SystemVerilog module, bug1142. Reported by Marcin K.
(verilog-define-abbrev-table) (verilog-mode-abbrev-table): Don't expand
abbrev inside comment/strings, bug1102. Reported by Slava Yuzhaninov.
(verilog-auto): Fix AUTORESET widths pulling from AUTOREGINPUT,
msg2143. Reported by Galen Seitz.
(verilog-modify-compile-command): Fix expansion of __FLAGS__ when
compile-command is globally set, bug1119. Reported by Galen Seitz.
Fix glyph string generation for multi-font compositions (Bug#26742)
* src/xdisp.c (glyph_string_containing_background_width): New function.
(draw_glyphs): Use it to get correct background width.
(compute_overhangs_and_x): Don't change x in the middle of composite
characters.
Alan Mackenzie [Sun, 7 May 2017 09:27:00 +0000 (09:27 +0000)]
CC Mode internal cache: Handle a cache pos being inside a two-char construct.
Cache c-state-semi-nonlit-pos-cache was failing when a cache position was,
e.g., between the two characters of an opening comment "/*", and additionally
there were an odd number of quote marks (apostrophes) in the comment. This
happened in .../src/xdisp.c in the Emacs master branch around 2017-05-02 at
buffer position 615001.
* lisp/progmodes/cc-defs.el (c-emacs-features): Repurpose symbol
pps-extended-state to mean that there are at least 11 elements in the parser
state.
* lisp/progmodes/cc-engine.el (c-cache-to-parse-ps-state)
(c-parse-ps-state-to-cache): Rewrite these to use enhanced cache element list
types which indicate potentially being inside two-char constructs.
(c-parse-ps-state-below): Rewrite to use the new versions of the above two
functions.
Glenn Morris [Sun, 7 May 2017 01:20:21 +0000 (18:20 -0700)]
Evaluate mh-require when compiling
* lisp/mh-e/mh-alias.el, lisp/mh-e/mh-folder.el:
* lisp/mh-e/mh-gnus.el, lisp/mh-e/mh-search.el:
Evaluate mh-require when compiling, as require is automatically.
* lisp/mh-e/mh-gnus.el: No longer disable byte-compilation.
Paul Eggert [Sun, 7 May 2017 01:00:23 +0000 (18:00 -0700)]
New var write-region-verbose, default nil
By popular demand, write-region char counts are now off by default
(Bug#26796).
* src/fileio.c (write-region-verbose): New Lisp var.
(write_region): Output char count only if the var is non-nil.
* doc/emacs/files.texi (Misc File Ops), etc/NEWS: Document this.
Glenn Morris [Sun, 7 May 2017 00:58:20 +0000 (17:58 -0700)]
Write autoloads file atomically
* lisp/emacs-lisp/autoload.el (autoload--save-buffer):
New function, to save buffer atomically.
(autoload-save-buffers, update-directory-autoloads):
Use autoload--save-buffer.
* lisp/Makefile.in ($(lisp)/loaddefs.el):
No longer write to a temp file by hand.
Paul Eggert [Sat, 6 May 2017 22:29:16 +0000 (15:29 -0700)]
Port .gdbinit to GDB 7.11.1 + Python 2.7.12
* src/.gdbinit (Lisp_Object_Printer.to_string):
Explicitly convert integer val to 'int', so that
older GDBs do not complain about the conversion.
* src/lisp.h (Lisp_Object) [CHECK_LISP_OBJECT_TYPE]:
Give the struct a tag, so that older GDB pretty-printers have a
tag to hang their hat on.
Paul Eggert [Sat, 6 May 2017 21:24:12 +0000 (14:24 -0700)]
Pretty-print const Lisp_Objects in .gdbinit
* src/.gdbinit (Emacs_Pretty_Printers.__call__):
Compare unqualified type to Lisp_Object, to do the right thing
when the expression has type ‘Lisp_Object const’.
Problem reported by Eli Zaretskii in:
http://lists.gnu.org/archive/html/emacs-devel/2017-05/msg00138.html
Paul Eggert [Sat, 6 May 2017 21:21:19 +0000 (14:21 -0700)]
Pacify GCC setjmp/longjmp warning
* src/eval.c (internal_lisp_condition_case): Do not modify local
var VAR, to pacify GCC’s setjmp/longjmp warning which in some
cases mistakenly diagnoses VAR possibly being modified between a
setjmp and a longjmp.
Philipp Stephani [Sat, 22 Apr 2017 16:04:29 +0000 (18:04 +0200)]
Introduce new misc type for module function
This resolves a couple of FIXMEs in emacs-module.c.
* src/lisp.h (MODULE_FUNCTIONP, XMODULE_FUNCTION): New functions.
* src/alloc.c (make_module_function): New function.
(mark_object): GC support.
* src/data.c (Ftype_of, syms_of_data): Handle module function type.
* src/print.c (print_object): Print support for new type.
* src/emacs-module.c (module_make_function, Finternal_module_call):
Use new module function type, remove FIXMEs.
(module_format_fun_env): Adapt and give it external linkage.
* test/src/emacs-module-tests.el (module-function-object): Add unit
test.
Eli Zaretskii [Sat, 6 May 2017 08:22:44 +0000 (11:22 +0300)]
Turn on GC_CHECK_MARKED_OBJECTS by default under ENABLE_CHECKING
* src/alloc.c (GC_CHECK_MARKED_OBJECTS): Define to 1 by default of
ENABLE_CHECKING is defined.
(mark_object): Test for GC_CHECK_MARKED_OBJECTS being non-zero,
instead of being defined.