GitLab

* lisp/mail/rmail.el (rmail-file-coding-system): It's only ever used
in a boolean sense, so just make it a boolean, and fix the doc.
(rmail-show-mime-function, rmail-mime-feature)
(rmail-require-mime-maybe): Doc fixes.
(rmail-show-message-1): Check rmail-show-mime-function is non-nil.

* lisp/mail/rmailmm.el (rmail-show-mime): Doc fix.

* lisp/mail/rmailmm.el (rmail-mime-entity, rmail-mime-entity-segment)
(rmail-mime-shown-mode, rmail-mime-hidden-mode, rmail-mime-raw-mode)
(rmail-mime-toggle-hidden, rmail-mime-insert-tagline)
(rmail-mime-insert-header, rmail-mime-handle, rmail-mime-parse)
(rmail-mime, rmail-show-mime): Doc fixes.

* lisp/term/ns-win.el (mode-line-frame-identification): Leave it alone.

* lisp/mail/rmailout.el (rmail-output-to-rmail-buffer):
Handle empty buffers.

 src/xdisp.c (display_line): Move the call to
 highlight_trailing_whitespace before the call to
 compute_line_metrics, since the latter needs to see the final
 faces of all the glyphs to compute ROW's hash value.  Fixes
 assertion violations in row_equal_p.

src/w32proc.c (reader_thread): Don't check pending input if cp->fd < 0,
just return.

If the server hangs up while we're talking to it, just `message' the error instead of throwing an error.

* lisp/mail/rmailsum.el (rmail-summary): Remove movement to beginning
of message - not necessary, and causes problems.

I believe the call to rmail-summary-beginning-of-message was only
thought to be necessary due to a couple of bugs in rmail-new-summary
that are hopefully fixed now (`mesg' not being set,
rmail-summary-buffer's buffer locality not being considered).

Fixes: debbugs:9831

Nothing was setting the local variable `mesg'.
Eg when looking at message 10, pressing h to get a summary would
send you to message 1.

* lisp/progmodes/executable.el
(executable-make-buffer-file-executable-if-script-p):
Handle file-modes returning nil.

* lisp/mail/rmailsum.el (rmail-summary, rmail-new-summary)
(rmail-new-summary-1): Allow empty summaries. 
(rmail-new-summary): Remember that rmail-summary-buffer is buffer-local.

Fixes: debbugs:9964

 src/makefile.w32-in (HEAPSIZE): New variable, allows to build temacs
 with user-defined heap size.  Bump the default size of the temacs
 heap to 27MB, to avoid memory warning when running temacs.
 ($(TEMACS)): Use HEAPSIZE instead of a hardcoded value.

 src/dispnew.c (verify_row_hash) [XASSERTS]: New function.
 (adjust_glyph_matrix, row_equal_p): Use it in xassert to verify
 that the hash value of glyph rows is correct.

 src/dispnew.c (scrolling_window): Fix incorrect indices in accessing
 current_matrix and desired_matrix.  (Bug#9990)

* window.h (window): Remove splits slot.
* window.c (Fwindow_splits, Fset_window_splits): Remove.
(Fdelete_other_windows_internal, make_parent_window)
(make_window, Fsplit_window_internal, Fdelete_window_internal)
(Fset_window_configuration, save_window_save): Don't deal with
split status of windows.
(saved_window): Remove splits slot.
(Vwindow_splits): Rewrite doc-string.
* window.el (window-resize, delete-window): Use window-splits
variable instead of function.
(window-state-get-1, window-state-put-2, window-state-put):
Don't deal with windows' splits status.
* windows.texi (Splitting Windows, Deleting Windows): Remove
references to splits status of windows.

* lisp/apropos.el (apropos-do-all, apropos-library, apropos-value)
(apropos-documentation): Doc fixes.

* nsfns.m (unwind_create_frame):
* w32fns.c (unwind_create_frame): Use Fmemq to check if frame is in
Vframe_list.

Fixes: debbugs:9999

* lisp/electric.el (electric-indent-post-self-insert-function): Make it
possible for a char to only indent in some circumstances.
(electric-indent-mode): Simplify.

* window.el (window-size-ignore, window-size-fixed-1)
(window-in-direction-2): Prefix with "window--".
(window-tree-1): Rename to window--subtree, fix doc-string.

* window.el (windows-with-parameter): Remove unused function.
(windows-at-side): Rename to window-at-side-list.
(window-check, window-atom-check, window-atom-check-1)
(window-side-check): Prefix with "window--".