I've seen segfaults where `infile` is nil when we get to
readbyte_from_file, presumably because Fload set it to NULL (via
close_infile_unwind) just before returning to its caller which was
probably itself within another read/load and for some reason
readevalloop didn't get to re-set `infile` like it used to do at every
iteration. I was not able to really track down the bug, but the way
`infile` was set/reset seemed fragile and managing it like a standard
dynamically-scoped var seems both safer (and more efficient since we
don't need readevalloop to constantly re-set it).
(readchar): Assert that `infile` is set if using a function the depends on it.
(readbyte_from_file): Assert that `infile` is set.
(close_infile_unwind): Reset `infile` to its previous value rather than
to NULL.
(Fload): Remember the previous value of `infile` before chaning it.
(readevalloop): Don't set `infile` any more.
* doc/lispref/positions.texi (List Motion): Remove xref, the text it
references was removed in 2018-02-15 "Document
open-paren-in-column-0-is-defun-start being of less importance".
(message-send-mail-function) <function>: Remove `local-library` tests
for libs distributed with Emacs.
(message-use-send-mail-function): New function.
(message-default-send-mail-function): Default to it, and remove cases
already handled by it.
(message--default-send-mail-function): New function.
(message-send-mail-function) <variable>: Use it as new default.
(message-sendmail-f-is-evil): Obey mail-specify-envelope-from if available.
(message-check, message-with-reply-buffer): Use `declare`.
(message-smtpmail-send-it): smtpmail accepts mail-header-separator,
so simplify and declare obsolete.
(message-send-mail-with-mailclient): Declare obsolete.
(message-check-news-body-syntax): Don't presume that the checksum is
a fixnum.
* test/lisp/progmodes/flymake-tests.el
(flymake-tests--wait-for-backends): Guard against
flymake-no-changes-timeout being nil (in personal configurations)
to help when running tests interactively.
* doc/lispref/display.texi (Button Buffer Commands)
(Button Buffer Commands): Document this.
* lisp/button.el (backward-button, forward-button): Accept a
NO-ERROR parameter.
(button-activate): Make it possible to have specific data in the
callback action.
* doc/emacs/programs.texi (Left Margin Paren): Add index for "\( in
strings".
* doc/lispref/positions.texi (List Motion): Add index, and cross
reference.
This reverts commit 848712b481.
There could be dictionaries available (set via command-line options for the
speller, for instance) that would not be returned.
* lisp/descr-text.el (describe-property-list): Don't special-case
for symbols that have widget properties here (bug#22957). It's
not documented that this function should do that, and looking at
the code, it doesn't seem like this function is actually used for
doing that, either. This makes describing some text properties
that are on the form `(string ...)' work.
* lisp/profiler.el (profiler--xref-backend): New function (bug#23455).
(profiler-report-mode): Use it to use the elisp xref handler when
hitting `M-.' in profiler buffers.
(vhdl--pushnew): New macro. Use it instead of `pushnew`.
(emacs-major-version): Don't bother checking (featurep 'xemacs)
since that is always (< emacs-major-version 25) as well.
Use it to improve the docstring of the accessor functions.
* doc/misc/cl.texi: Rename menu entry to "CL-Lib".
(Structures): Add ':documentation' and mention ':type' as well,
which we don't completely ignore any more.
* lisp/gnus/mm-decode.el (mm-display-external): Don't delete other
buffers when displaying parts (bug#22861) because that should be
up to the user, probably.
