This is shorter, simplifies use inside rx expressions, and removes
need for eval-when-compile elsewhere (for later exploitation).
* lisp/emacs-lisp/lisp-mode.el (lisp-mode-symbol): New rx-define.
(lisp-mode-symbol-regexp): Redefine using lisp-mode-symbol.
(lisp-imenu-generic-expression, lisp--el-match-keyword)
(lisp-fdefs, lisp-string-in-doc-position-p):
* lisp/emacs-lisp/checkdoc.el (checkdoc--error-bad-format-p):
* lisp/emacs-lisp/shorthands.el (shorthands-font-lock-shorthands):
Use lisp-mode-symbol instead of lisp-mode-symbol-regexp.
* lisp/emacs-lisp/subr-x.el (string-limit): Return more correct
results in the CODING-SYSTEM case for coding systems with BOM and
charset designations (bug#48324). Also amend the algorithm to
return complete glyphs, not just complete code points.
Try and reduce the pain caused by font-lock and syntax-propertize's
wholeline-based operation in buffers made up of a few very long lines
(bug#45898).
* lisp/emacs-lisp/syntax.el (syntax-wholeline-max): New var.
(syntax--lbp): New function.
(syntax-propertize-wholelines): Use it.
* lisp/jit-lock.el (jit-lock--antiblink-post-command): Use `syntax--lbp`.
* lisp/font-lock.el (font-lock-extend-region-wholelines): Rewrite,
using `syntax-propertize-wholelines`.
* lisp/emacs-lisp/gv.el:
* lisp/emacs-lisp/macroexp.el (internal-macroexpand-for-load): Since
eager macro-expansion errors are no longer warnings, don't say so.
* lisp/emacs-lisp/rx.el: `>=` and `=` are much more likely functions
than RX constructs and the indentation machinery currently has
no way to tell them apart.
Suggested by Michael Herdeegen.
* doc/lispref/loading.texi (How Programs Do Loading): Update
documentation.
* lisp/emacs-lisp/macroexp.el:
(internal-macroexpand-for-load): We've been warning about eager
macro expansion for many years, so finally change that into an
error (bug#18154).
That change (introduced to circumvent an error now that `seq.el` is
preloaded) caused all dispatchers to be left uncompiled, which slows down
method dispatch very significantly. Fix the problem in the old way,
i.e. by adding an explicit call to `cl--generic-prefill-dispatchers`.
* lisp/emacs-lisp/cl-generic.el (cl--generic-compiler): Revert last change.
Add (cl--generic-prefill-dispatchers 1 integer) instead to handle
the new dispatchers needed for `seq.el`.
(cl--generic-prefill-generalizer-sample): New function.
(cl--generic-get-dispatcher): Use it to signal an error giving precise
instructions for what to do if we're about the load the byte-compiler
during the preload.
(cl--generic-oclosure-generalizer): Rename from
`cl-generic--oclosure-generalizer` for consistency with all
other generalizers.
* lisp/emacs-lisp/byte-opt.el (byte-optimize-form-code-walker):
Turn functions into nil when compiled for-effect since they have no
side-effects on their own. This may enable further improvements such
as the elimination of variable bindings.
`unwind-protect` forms can be treated as plain function call at this
point. In particular, their unwind function argument should be
not optimised for effect since it's a function.
Nested calls to `syntax-ppss` and `syntax-propertize` can easily
happen unexpectedly via ondemand propertizing or `forward-sexp`.
Refine the handling of nested calls so we detect them more reliably
(e.g. also within `syntax-propertize-extend-region-functions`)
and so that the `syntax-ppss` cache is automatically flushed in case
it might have been filled with data that's become obsolete since.
(syntax-propertize--inhibit-flush): Delete var.
(syntax-propertize--in-process-p): New function to replace it.
(syntax-ppss-flush-cache): Use it.
(syntax-ppss--updated-cache): New var.
(syntax-propertize): Make `syntax-propertize--done` binding apply to
`syntax-propertize-extend-region-functions` as well, as intended (fixes
bug#46713). Use `syntax-ppss--updated-cache` to flush
syntax-ppss cache at the end when needed.
Don't bind `syntax-propertize--inhibit-flush` any more.
(syntax-ppss): Set `syntax-ppss--updated-cache` when applicable.
* lisp/term/haiku-win.el (require):
* lisp/progmodes/elisp-mode.el (require):
* lisp/isearch.el (require): Require subr-x at compile time, since
these use defsubsts/macros from there.
* lisp/emacs-lisp/subr-x.el (string-empty-p): Move from here...
* lisp/simple.el (string-empty-p): ... to here. This is to help
with a build problem where files.el is using the defsubst, but
requiring subr-x.el at compile time leads to load errors.
This is a cosmetic change only; there is no change in behaviour.
* lisp/emacs-lisp/bytecomp.el:
* src/bytecode.c (BYTE_CODES, exec_byte_code):
Update and/or remove incorrect, outdated or useless comments.
Clarify. Reorder where appropriate. Rename Bsave_current_buffer to
Bsave_current_buffer_OBSOLETE and Bsave_current_buffer_1 to
Bsave_current_buffer, reflecting the state since 1996.
* doc/lispref/commands.texi (Reading One Event): Document it.
* lisp/emacs-lisp/rmc.el (read-multiple-choice): Allow using
long-form answers instead of single character ones.
(read-multiple-choice--long-answers): New function.
(read-multiple-choice--short-answers): Refactored out from the
main function.
* lisp/sqlite-mode.el (require):
* lisp/net/eudc.el (require):
* lisp/arc-mode.el (require): Require subr-x, since these files
are using macros from there.
* lisp/emacs-lisp/subr-x.el (with-memoization): Move from here...
* lisp/subr.el (with-memoization): ... to here, as it's used from
the preloaded cl-generic.el file.
* lisp/emacs-lisp/cl-generic.el (cl--generic-compiler): Don't use
the autoloaded `byte-compile' function during bootstrap.
(cl--generic-get-dispatcher): Don't require subr-x, either.
cl-generic has been preloaded since 2015, and most usages of it (in
preloaded files) work fine. In particular, using `cl-defgeneric' is
unproblematic. However, `cl-defmethod' would end up pulling in the
byte compiler (at load time), which would make it impossible to use
`cl-defmethod' in pre-loaded files, and this change fixes that (but
possibly not in the most self-evidently correct way).
* lisp/emacs-lisp/bindat.el (bindat--pack-strz): For fixed-length strz
fields, explicitly write a null terminator after the packed string if
there is room (bug#56048).
* doc/lispref/processes.texi (Bindat Types): Update documentation.
* test/lisp/emacs-lisp/bindat-tests.el (bindat-test--str-strz-prealloc):
Update tests.
Motivation/rationale:
* Improve code readability. Now `bindat--pack-strz` is used for all
`strz` packing, not just variable-length `strz` packing.
* Make it easier to change the behavior of fixed-length `strz`
packing without also affecting the behavior of `str` packing. (A
future commit will modify `strz` to write a null terminator if
there is room.)
This fixes Bug#54648.
* lisp/emacs-lisp/loaddefs-gen.el (loaddefs-generate--make-autoload):
Add `iter-defun' and `cl-iter-defun' to the list of "complex cases"
that are macro-expanded and recursed on.
* lisp/emacs-lisp/bytecomp.el (byte-compile-file): Document behavior
if 'no-byte-compile' is set.
* test/lisp/emacs-lisp/bytecomp-tests.el
(byte-compile-file/no-byte-compile): New unit test.
* test/lisp/emacs-lisp/bytecomp-resources/no-byte-compile.el: New test
file.
