070e82b ; * src/window.c (window_scroll): Improve commentary.
60457d7 Improve documentation of the window tree
ea1a014 Fix window scrolling on TTY frames when there's no mode line
df7ed10 Fix decoding XML files encoded in ISO-8859
7851ae8 (tag: emacs-26.1.90) ; ChangeLog.3 update
1958808 * etc/AUTHORS: Update.
7252507 Fix description of some window hooks
88762b4 Run 'window--adjust-process-windows' when frame size changes ...
d6542ea Avoid errors in zone.el when there's overlay at EOB
* doc/lispref/windows.texi (Windows and Frames): More accurate
wording regarding the relation of a mini-window to its frame's
window tree.
* src/window.h (struct window): Improve commentary to some
fields.
* src/window.c (window_internal_height): Remove tests for
next, prev, and parent pointers, as they are unrelated to
whether a window has a mode line. (Bug#33363)
POSIX says sh ~/foo should act like $HOME/foo even if $HOME is
relative, so be consistent with that (Bug#33255).
* admin/merge-gnulib (GNULIB_MODULES): Add dosname.
* src/buffer.c (init_buffer): Use emacs_wd to get
initial working directory with slash appended if needed.
(default-directory): Say it must be absolute.
* src/emacs.c (emacs_wd): New global variable.
(init_cmdargs): Dir arg is now char const *.
(main): Set emacs_wd.
* src/emacs.c (main) [NS_IMPL_COCOA]:
* src/fileio.c (Fexpand_file_name):
Use get_homedir instead of egetenv ("HOME").
* src/fileio.c: Include dosname.h, for IS_ABSOLUTE_FILE_NAME.
(splice_dir_file, get_homedir): New functions.
* src/xrdb.c (gethomedir): Remove. All callers changed
to use get_homedir and splice_dir_file.
* test/src/fileio-tests.el (fileio-tests--relative-HOME): New test.
Problem reported by Thomas Fitzsimmons (Bug#33174).
Do not merge to master, as we have a better fix there.
* src/Makefile.in (emacs$(EXEEXT)):
(bootstrap-emacs$(EXEEXT)):
Unset EMACS_HEAP_EXEC before invoking temacs.
Problem reported by Thomas Fitzsimmons (Bug#33174).
* src/emacs.c (main): Adjust to sysdep.c changes.
* src/sysdep.c (exec_personality): New static var.
(disable_address_randomization): Remove, replacing with ...
(maybe_disable_address_randomization): ... this new function.
Do not set or use an environment variable; use a command-line
argument instead, and set the new static var. Migrate the emacs.c
personality-change code to here, where it belongs.
(emacs_exec_file): Simplify by using new static var.
I ran into this problem on Ubuntu 18.04.1 LTS.
* src/gnutls.c (HAVE_GNUTLS_CIPHER_GET_IV_SIZE)
(HAVE_GNUTLS_CIPHER_GET_TAG_SIZE, HAVE_GNUTLS_DIGEST_LIST):
New macros. This uses the same style as the other
macros that depend on GnuTLS version, as opposed to trying
to do things a bit more cleverly.
(gnutls_cipher_get_iv_size, gnutls_cipher_get_tag_size)
(gnutls_digest_list, gnutls_digest_get_name):
Define these macros only if they will be used.
* src/process.c (dissociate_controlling_tty): New function.
(create_process): Use it to dissociate controlling tty if setsid
fails, which happens on Darwin after a vfork (Bug#33154).
Do this on all platforms, not just on Darwin, as a similar
problem is plausible elsewhere.
* src/callproc.c (call_process): Use the new function here, too,
for consistency and to avoid duplicate code.
* src/editfns.c (Fgroup_name): Fix the doc string. Move
closer to the "group" functions.
* src/w32.c (getgrgid): Return NULL if GID is not the group ID
of the user of this Emacs session
* test/src/editfns-tests.el (test-group-name): Rename from
'group-name'. Add tests for non-Posix hosts. Test error when
the argument to group-name is invalid.
* etc/NEWS: Fix wording of last added entry.
bd1d617 Avoid race in rcirc process filter (bug#33145)
88ef31a Avoid file-name errors when viewing PDF from Gnus
c939042 Avoid crashes with remapped default face in Org mode
97660fa Doc fix for checkdoc-continue
96f055b Fix a typo in autoload.el
* src/timefns.c (emacs_mktime_z): Remove; no longer needed.
(time_error): New function, replacing invalid_time.
All callers changed.
(decode_float_time, decode_ticks_hz, decode_time_components):
Return an error number instead of merely a boolean.
All callers changed.
(decode_lisp_time): Signal an error based on the error number,
instead of merely returning a boolean to the caller.
All callers changed.
(format_time_string, Fdecode_time, Fencode_time)
(Fcurrent_time_string):
Do not assume that a failure of a system time function
must be due to time overflow.
(Fencode_time): Don't report an error merely because mktime
returned ((time_t) -1), as that may be a valid time_t value.
Use a simpler error check. See:
https://www.sourceware.org/ml/libc-alpha/2018-11/msg00062.html
* src/fringe.c (draw_window_fringes): Switch to window's
buffer to get the local value of face-remapping-alist, if
necessary. (Bug#33244)
* src/xfaces.c (syms_of_xfaces) <Qface_remapping_alist>: New
DEFSYM.
* src/xfaces.c (face_at_buffer_position): Look up BASE_FACE_ID
anew if it is not in the frame's face cache. This avoids
crashes when Org mode sets up for a new major mode in embedded
code fragment, and the default face is remapped. (Bug#33222)
Backport from master.
* src/bytecode.c (exec_byte_code): Save VECTOR into stack slot
so that it survives GC. The stack slot was otherwise unused,
so this doesn’t cost us memory, only a store insn.
Although the previously-applied fix worked for its platform,
it doesn’t suffice in general.
* src/bytecode.c (exec_byte_code): Save VECTOR into stack slot
so that it survives GC. The stack slot was otherwise unused,
so this doesn’t cost us memory, only a store insn.
* src/eval.c (Ffuncall): Do not make FUN volatile, reverting
2018-10-14T19:12:04Z!gazally@runbox.com. Adding ‘volatile’
does not suffice, since storage for a volatile local can be
reclaimed after its last access (e.g., by tail recursion
elimination), which would make VECTOR invisible to GC.
* src/eval.c (Ffuncall): Make local variable 'fun' volatile.
* test/src/eval-tests.el
(eval-tests-byte-code-being-evaluated-is-protected-from-gc):
Add regression test for Bug#33014.
(eval-tests-33014-var): New variable.
(eval-tests-33014-func, eval-tests-33014-redefine): New functions.
* src/nsterm.m (ns_clip_to_rect):
(ns_reset_clipping): Remove gsaved variable and associated code.
(ns_flush_display): Remove function.
(ns_copy_bits): use translateRectsNeedingDisplayInRect:by: to copy any
pending drawing actions along with the image.
([EmacsView windowWillResize:toSize:]): Remove unneeded call.
([EmacsView drawRect:]): Remove redundant call to ns_clear_frame_area,
and optimize the exposed rectangles.
(ns_draw_window_cursor): Remove unneeded disabling of screen updates.
Thus when entering edebug, the focus will be moved to the frame with the
source being debugged, and when its finished, the focus will move back.
Commands edebug-visit-eval-list (`E') and edebug-bounce-point (`p') have been
likewise amended.
* src/keyboard.c (readable_events): Handle FOCUS_OUT_EVENT as an invisible
event. This prevents input-pending-p returning t when one of these events
arrives, and thus obviates an instant termination of sit-for when there's no
"real" event waiting.
* lisp/emacs-lisp/edebug.el (edebug-focus-frame): New function.
(edebug-default-enter, edebug--display-1, edbug-where, edebug-bounce-point)
(edebug-visit-eval-list): Call edebug-focus-frame to move focus into the
window newly selected by edebug-pop-to-buffer.
Problem reported by Joseph Mingrone in:
https://lists.gnu.org/r/emacs-devel/2018-10/msg00238.html
While we’re at it, apply a similar fix to struct Lisp_Subr; this
removes the need for GCALIGNED_STRUCT_MEMBER and thus can shrink
struct Lisp_Subr a bit.
* configure.ac (HAVE_STRUCT_ATTRIBUTE_ALIGNED): Bring back this macro.
Although used only for performance (not to actually align
structures), we might as well take advantage of it.
* src/lisp.h (GCALIGNED_STRUCT_MEMBER): Remove; all uses removed.
(union Aligned_Lisp_Subr): New type, like struct Lisp_Subr but aligned.
* src/lisp.h (XSUBR, DEFUN):
* src/lread.c (defsubr): Use it. All callers changed.
* src/thread.c (union aligned_thread_state): New type.
(main_thread): Now of this type, so it’s aligned.
All uses changed.
* src/xmenu.c (syms_of_xmenu) [USE_GTK || USE_X_TOOLKIT]:
Adjust to union Aligned_Lisp_Subr change.
* src/thread.c (unmark_main_thread): New function.
* src/lisp.h (unmark_main_thread): Prototype it.
* src/alloc.c (garbage_collect_1): Call it after sweeping.
(Bug#33073)
* test/src/thread-tests.el (threads-test-bug33073): New test.
73babba (origin/emacs-26) Clarify documentation of fractional vertica...
b20c51d * lisp/isearch.el (isearch-cmds): Recall absent isearch--stat...
700acbd doc/lispref/edebug.texi (Specification List) Remove obstrusiv...
1902450 Fix wording in module API documentation
e724a8f Fix redisplay of glyphless characters
8fc892d Update --without-toolkit-scroll-bars doc
80e0bfa Call GTK functions only on GTK scrollbars
91c4c46 Update the description of startup in ELisp manual
18b42c6 Use the 'line-number' face for line-number fields past EOB
a6ab8db Ensure NS frame is redrawn correctly after scroll
Conflicts:
lisp/isearch.el
* doc/lispref/windows.texi (vertical scrolling): Clarify the meaning of
vertical scrolling by referring to tall screen lines, images, and the display
action. Clarify an ambiguous English tense.
* src/window.c (window-vscroll, set-window-vscroll): Amend doc strings to
refer to display.
Problem reported by Paul Pogonyshev (Bug#33034).
* src/lisp.h (union specbinding): New member unwind.eval_depth.
* src/eval.c (record_unwind_protect, set_unwind_protect): Set it.
(do_one_unbind): Use it.
* src/conf_post.h (bool_bf): Use 'unsigned int' in the MinGW
builds. Suggested by Tom Tromey <tom@tromey.com>. (Bug#33017)
* src/dispnew.c (scrolling_window): Update commentary
regarding xwidget builds.
* src/json.c (struct json_buffer_and_size): New member
inserted_bytes.
(json_insert): Instead of creating a string and inserting it
into the current buffer, copy the unibyte text into the gap.
(struct json_insert_data): New member inserted_bytes.
(json_insert_callback): Update commentary. Pass the
inserted_bytes value to json_insert and on its return copy the
updated value back into DATA.
(Fjson_insert): Decode the unibyte text inserted into the gap.
Call before-change-functions and after-change-functions only
once, before and after processing the insertion of the entire
JSON representation.
* test/src/json-tests.el (json-insert/throw): Adapt to the
modified implementation of json-insert: it no longer calls the
modification hooks once for each inserted chunk of JSON
representation.
* src/xdisp.c (get_phys_cursor_geometry): Treat rows at and
beyond ZV specially. Don't let the cursor exceed the
vertical dimensions of the row.
(maybe_produce_line_number): Use the 'line-number' face
instead of 'default' for blank fields beyond ZV. Don't update
the IT metrics when displaying blank line-number fields beyond
ZV. (Bug#32337)
