* src/gmalloc.c (_fraghead) [!HYBRID_MALLOC]:
Make it static in this case, too. This avoids having both
‘extern struct list _fraghead[];’ and
‘static struct list _fraghead[BLOCKLOG];’, which
GCC 9 complains about.
* doc/lispref/nonascii.texi (Explicit Encoding): Document what
happens when DESTINATION of decoding is a unibyte buffer.
* src/coding.c (Fdecode_coding_region)
(Fdecode_coding_string): Document what happens if DESTINATION
is a unibyte buffer.
The GCC + valgrind fix caused the m68k build to fail (Bug#35711).
Simplify string allocation a bit to make similar problems less
likely in the future.
* src/alloc.c (sdata, SDATA_NBYTES, SDATA_DATA) [GC_CHECK_STRING_BYTES]:
Use the same implementation as with !GC_CHECK_STRING_BYTES,
as the special case is no longer needed.
(SDATA_ALIGN): New constant.
(SDATA_SIZE): Remove this macro, replacing with ...
(sdata_size): ... this new function. All uses changed.
Properly account for sizes and alignments even in the m68k case,
and even if GC_CHECK_STRING_BYTES is not defined.
Don't merge to master, this has already been fixed there by 2019-01-15
"Fix unlikely races with GnuTLS, datagrams".
* src/gnutls.c (emacs_gnutls_read): Similar to emacs_gnutls_write,
when gnutls_record_recv returns GNUTLS_E_AGAIN set errno to EGAIN.
This fixes bug #35647. State that when indent-tabs-mode is non-nil, spaces
are inserted before the tab rather than the tab being replaced by spaces.
* doc/lispref/text.texi (columns)
* src/indent.c (move-to-column): Make the above documentation amendment.
* configure.ac: Use EMACS_CHECK_MODULES fors LCMS2 rather than
AC_SEARCH_LIBS.
* src/Makefile.in: Get LCMS2_LIBS and LCMS2_CFLAGS from configure,
instead of just LIBLCMS2.
(cherry picked from commit cb3863370c)
* src/msdos.c (the_only_tty_output): Define.
* src/msdos.h (the_only_tty_output): Declare.
* src/frame.c (make_terminal_frame) [MSDOS]:
* src/dispnew.c (init_display) [MSDOS]: Set up
f->output_data.tty pointer using the_only_tty_output, before
dereferencing the pointer. This prevents crashes with DPMI
servers that provide NULL pointer protection.
* src/frame.c (free_monitors) [USE_GTK]: Define in the GTK case as
well.
* src/xfns.c (x-display-monitor-attributes-list) [USE_GTK]: Plug
memory leak. Use dupstring over xstrdup as gdk_monitor_get_model may
return NULL.
* src/minibuf.c (Fread_command): Document the return value
when DEFAULT-VALUE is nil and the user enters nothing.
* doc/lispref/minibuf.texi (High-Level Completion): Document
the printed representation of a symbol whose name is empty.
(Bug#3522)
* doc/lispref/windows.texi (Window Start and End):
* src/window.c (Fset_window_start): Document that reliable
setting of a window start position requires to adjust point to
be visible. (Bug#34038)
* src/window.c (Fset_window_margins, Fset_window_fringes)
(Fset_window_scroll_bars): In doc-strings tell that a window
must be large enough to accommodate fringes, sroll bars and
margins of the desired size.
* doc/lispref/display.texi (Fringe Size/Pos, Scroll Bars)
(Display Margins): Tell that windows must be large enough to
accommodate fringes, sroll bars and margins of the desired
size.
* src/keyboard.c (read_char, read_event_from_main_queue):
Ensure the global value of getcjmp is restored when the stack
is unwound by the likes of 'throw', by calling
record_unwind_protect_ptr instead of restoring the value
manually. (Bug#34394)
(restore_getcjmp): Argument is now 'void *', to match the
signature of record_unwind_protect_ptr.
(cherry picked from commit 10527fca66)
This fixes a regression from Emacs 25.3 (Bug#34134).
* src/process.c (server_accept_connection):
Set host correctly, fixing a bug introduced in
2017-09-16T21:29:18Z!eggert@cs.ucla.edu
when working around a GCC bug.
* src/nsterm.m ([EmacsView viewWillDraw]): Cancel drawing if the frame
has been garbaged.
* src/xdisp.c (expose_window_tree, expose_frame): Remove NS only
exceptions.
* src/w32uniscribe.c (uniscribe_otf_capability): Set
inhibit-quit around calls to otf_features, because the latter
cons Lisp data structures while we are in a critical section.
* src/xdisp.c (ALLOCATE_HDC) [HAVE_NTGUI]: Set inhibit-quit.
(RELEASE_HDC) [HAVE_NTGUI]: Restore inhibit-quit.
(OPTIONAL_HDC, DECLARE_HDC): Remove macros, their job is now
done by ALLOCATE_HDC and by a single #ifdef.
(draw_glyphs): Adapt to the above changes in macros.
(Bug#34059)
Suggested by Eli Zaretskii (Bug#20890#31).
* src/font.h (font_data_structures_may_be_ill_formed): New function.
* src/ftfont.c (ftfont_close):
* src/ftcrfont.c (ftcrfont_close): Use it.
(cherry picked from commit d02fd482fb)
* doc/lispref/processes.texi (Accepting Output):
* src/process.c (Faccept_process_output):
Document that (accept-process-output P) can return non-nil
even after P has exited, and that it can return nil even if P
is still running (Bug#33839).
* src/process.c (Fprocess_send_region, Fprocess_send_string):
Document that process input longer than the process input
buffer may be split into bunches. Remove an outdated
reference to a 500 character split boundary.
* doc/lispref/processes.texi (Asynchronous Processes): Remove
mention of "stray character injections" in PTY processes. See
also the comment about ICANON in src/sysdep.c, function
child_setup_tty.
* src/keyboard.c (read_char): Events put into `unread-command-events'
with the form (t . EVENT) should always have the t stripped when read
out.
* test/src/keyboard-tests.el: New tests for `unread-command-events'.
(cherry picked from commit 1f3f4b1296)
Backport from master.
I ran into this when building Emacs 26 with GCC 8 on Fedora 29 x86.
* lwlib/lwlib-Xaw.h (xaw_update_one_value, xaw_popup_menu):
* lwlib/lwlib-Xlw.h (xlw_update_one_value, xlw_pop_instance):
* lwlib/lwlib.h (lw_allow_resizing, lw_set_main_areas) [!USE_MOTIF]:
No longer const.
* src/emacs-module.c: Ignore -Wcast-function-type.
