... as per the compiler's target.
The value provided by MSYS2 `uname' is not correct when MSYS2 runs on
one architecture (x86_64) and the user is building for another (i686).
* configure.ac [mingw]: Set $canonical default value as per the
compiler's target.
* configure.ac (comma_version, comma_space_version) [mingw32]:
New output variables.
(nt/emacs.rc, nt/emacsclient.rc) [mingw32]: New output files.
* make-dist: Update nt/ for *.rc -> *.rc.in changes.
* nt/emacs.rc.in, nt/emacsclient.rc.in: Rename from nt/emacs.rc, emacsclient.rc.
Let configure generate the real files, and set the version numbers.
* lib-src/Makefile.in (emacsclient.res): Update deps for nt/emacsclient.rc
now being in the build directory, not the source directory.
* nt/Makefile.in (distclean): Delete *.rc.
(emacs.res): Update deps for nt/emacsclient.rc now being in the
build directory, not the source directory.
* admin/admin.el (set-version): No more need to update nt/*.rc.
* admin/authors.el (authors-renamed-files-alist): Add .rc.in files.
2014-11-14 extern-inline: update commentary about GCC bugs
2014-11-06 unistd: port to iOS
2014-11-04 update from texinfo
* doc/misc/texinfo.tex, lib/unistd.in.h, m4/extern-inline.m4:
Update from gnulib.
On platforms this old, building with _FORTIFY_SOURCE equal to 2
results in duplicate definitions of standard library functions.
Problem reported by Nelson H. F. Beebe.
* configure.ac (_FORTIFY_SOURCE): Sort after GNULIB_PORTCHECK.
By default, do not enable this unless GNULIB_PORTCHECK is defined.
This better matches the original intent, which as I recall was to
enable these extra checks only with --enable-gcc-warnings.
All bzr revision IDS, and all CVS revision IDs for which a commit
could be identified, were changed to time-date!committer version
stamps. All .cvsignore files in the history became .gitignore files.
Fixes-bug annotations from bzr were copied into the corresponding
commit comments.
(The first .cvsignore commit was 1999-09-30T14:07:54Z!fx@gnu.org>.
The last CVS commit was <2009-12-27T08:11:12Z!cyd@stupidchicken.com>)
Committer/author email addresses are generally correct for the
transition day, not necessarily when the commit was originally made.
All bzr revision IDS, and all CVS revision IDs for which a commit
could be identified, were changed to time-date!committer version
stamps. All .cvsignore files in the history became .gitignore files.
Fixes-bug annotations from bzr were copied into the corresponding
commit comments.
(The first .cvsignore commit was 1999-09-30T14:07:54Z!fx@gnu.org>.
The last CVS commit was <2009-12-27T08:11:12Z!cyd@stupidchicken.com>)
Committer/author email addresses are generally correct for the
transition day, not necessarily when the commit was originally made.
* configure.ac (doc/man/emacs.1): Generate it.
* Makefile.in (top_bootclean): Remove doc/man/emacs.1.
* make-dist: Do not distribute doc/man/emacs.1.
* admin/admin.el (set-version): No need to update doc/man/emacs.1.
* doc/man/emacs.1.in: Rename from emacs.1.
* .bzrignore: Add doc/man/emacs.1.
* Makefile.in (epaths-force-w32): Simplify, assuming that the shell is
bash (as is the case in MSYS).
(msys_w32prefix_subst, msys_sed_sh_escape): Remove (no longer used).
* build-aux/msys-to-w32: simplify the initial over-engineered
interface, and the implementation.
* Makefile.in (epaths-force-w32): Update for the above.
admin: Changes in several documents.
autogen.sh: Neutralize language specific to a repository type.
doc/misc/efaq-w32.texi: Neutralized language specific to a repository type.
doc/misc/gnus-coding.txt: Neutralized language specific to a repository type.
lisp/Makefile.in: Change some production names so they're neutral
about the repository type.
* Makefile.in (AWK, srcdir_doc_info_dir_inputs, info_dir_inputs):
New macros.
(clean): Remove info-dir.*.
(info_dir_deps): Depend on make-info-dir too.
Fix bug with wildcards that weren't expanded.
(${srcdir}/info/dir): Make sure info directory exists.
Don't call pwd; just redirect make-info-dir's stdout to temp file.
* build-aux/make-info-dir: Send output to stdout.
Take input file names from args.
Fix a "FIXME inefficient looping" by using awk's associative arrays.
Basically, this rewrites the script from scratch, and speeds up
'make info/dir' from 2.6s to 0.07s on my platform.
Do all putenv calls before Emacs creates any threads.
Use a safer way to modify the TZ environment variable in the
presence of multiple threads. For further thread-safety,
prefer localtime_r and gmtime_r to localtime and gmtime,
and prefer struct tm's tm_gmtoff (if available) to calling
both localtime_r and gmtime_r.
* configure.ac (LOCALTIME_CACHE): Remove.
We needn't worry about SunOS 4 any more; Sun dropped support in 2003.
All uses of LOCALTIME_CACHE removed. This simplifies the fix.
(tzalloc): Add check for this function.
* admin/merge-gnulib (GNULIB_MODULES): Add time_r, since Emacs now
calls localtime_r and gmtime_r directly.
* src/dbusbind.c (Fdbus__init_bus): Move xputenv call from here ...
(init_dbusbind): ... to this new function.
* src/emacs.c (main) [HAVE_DBUS]: Call it before creating threads.
* src/xterm.c (x_term_init): Move xputenv call from here ...
(init_xterm): ... to this new function.
* src/emacs.c (main) [USE_GTK]: Call it before creating threads.
* src/editfns.c (HAVE_TM_GMTOFF): Default to false.
(dump_tz_string): New constant.
(init_editfns): Use it. This centralizes the dump_tz stuff.
Call set_time_zone_rule here, so that its xputenv is done
before Emacs goes multithreaded.
(mktime_z) [!HAVE_TZALLOC]: New function, which is typically
thread-safe enough for Emacs.
(format_time_string, Fdecode_time, Fcurrent_time_string)
(Fcurrent_time_zone):
Prefer localtime_r and gmtime_r, which are more thread-safe, to
localtime and gmtime. Remove now-unnecessary calls to block_input.
(tm_gmtoff): New static function.
(Fdecode_time, Fcurrent_time_zone): Use it.
(Fencode_time): Use mktime_z, for better thread-safety.
(set_time_zone_rule): Now static. Rewrite to be mostly thread-safe,
i.e., not quite thread-safe but good enough for Emacs typical usage.
Do not reclaim storage that is in the environment; let it leak.
Always call tzset, since localtime_r does not.
* src/emacs.c (dump_tz, Fdump_emacs) [HAVE_TZSET]: Remove dump_tz stuff.
This is now done in init_editfns.
* src/systime.h (mktime_z, timezone_t, tzalloc, tzfree) [!HAVE_TZALLOC]:
New macros and declarations, for platforms lacking tzalloc & friends.
Fixes: debbugs:8705
README: Bump version to 25.0.50.
configure.ac (AC_INIT): Bump version to 25.0.50.
nt/makefile.w32-in (VERSION): Bump version to 25.0.50.
nt/emacsclient.rc: Bump version to 25.0.50.
nt/emacs.rc: Bump version to 25.0.50.
nt/config.nt (VERSION): Bump version to 25.0.50.
msdos/sed2v2.inp: Bump version to 25.0.50.
etc/refcards/ru-refcard.tex: Bump version to 25.0.50.
etc/refcards/emacsver.tex: Bump version to 25.0.50.
src/msdos.c (internal_terminal_init): Bump version to 25.
doc/man/emacs.1: Bump version to 25.0.50.
doc/emacs/emacsver.texi (EMACSVER): Bump to 20.0.50.
