This improves performance of base64-decode-region by about 7.5% on
my platform, and gets rid of some macros.
* src/fns.c (IS_ASCII, IS_BASE64, IS_BASE64_IGNORABLE)
(READ_QUADRUPLET_BYTE): Remove.
(base64_value_to_char, base64_char_to_value):
Now an array of two arrays. All uses changed.
(base64url_value_to_char, base64url_char_to_value):
Remove. All uses changed to the other array.
(base64_char_to_value): Entries are now of type signed char, not
short, since we can assume C99. Use C99 initializers; this is
clearer and caters to the (theoretical) possibility of systems
that do not use ASCII or do not have 8-bit bytes. Allow any index
in the range 0..UCHAR_MAX instead of limiting it to 0..127, so
that uses need not check for in-range indexes. Also record
padding chars. All uses changed.
(base64_decode_1): Always store number of chars in *NCHARS_RETURN,
for simplicity. All callers changed. Speed up the byte-fetching.
Implement the RFC4648 variant of base64 encoding used by URLs.
* doc/lispref/text.texi (base64url-encode-region,
base64url-encode-string): Document new functions.
(base64-decode-region, base64-decode-string): Document new optional
parameter 'base64url' used to use url variant when decoding data.
* src/fns.c (base64url-encode-region, base64url-encode-region): New
functions to manage url variant.
(base64-decode-region, base64-decode-string): Add optional
parameter to indicate use of url-variant.
(base64_encode_region_1, base64_encode_string_1): Internal functions
with extracted code from 'base64_encode_region' and
'base64_encode_string' and optional parameters to manage padding and
url variant.
(base64-encode-region, base64-encode-string) : Use
base64_encode_region_1 and base64_encode_string_1.
(base64-encode-1): Add parameters to manage padding and url variant.
(base64-decode-1): Add parameter to manage url variant.
* test/src/fns-tests.el (fns-tests--with-region): New helper macro to
test region variant of base64 encode / decode functions.
(fns-tests--string-repeat): Helper function used in base64 tests.
(fns-tests-base64-encode-region, fns-tests-base64-encode-string):
Tests for standard base64 function.
(fns-test-base64url-encode-region,
fns-test-base64url-encode-string): Tests for url variant.
(fns-tests-base64-decode-string): Tests for decoding part.
* configure.ac (HAVE_X_WINDOWS): Add xfont.o to FONT_OBJ if HAVE_CAIRO.
* doc/lispref/frames.texi (Font and Color Parameters): Mention X core font
driver with Cairo drawing.
* src/font.c (syms_of_font) [HAVE_X_WINDOWS && USE_CAIRO]: Call syms_of_xfont.
* src/xfns.c (x_create_tip_frame) [USE_CAIRO]: Register xfont_driver.
* src/xterm.c (x_cr_gc_clip) [USE_CAIRO]: New function extracted from
x_begin_cr_clip.
(x_begin_cr_clip) [USE_CAIRO]: Use it.
(xlib_surface_key, saved_drawable_key) [USE_CAIRO]: New variables.
(x_cr_destroy_xlib_surface, x_try_cr_xlib_drawable)
(x_end_cr_xlib_drawable) [USE_CAIRO]: New functions.
(x_draw_composite_glyph_string_foreground)
(x_draw_glyph_string_foreground) [USE_CAIRO]: Get Xlib surface when drawing
text with X core fonts into bitmap surfaces. Add fallback code for drawing
into outline surfaces.
* src/dispextern.h (struct image) [USE_CAIRO]:
* src/image.c (free_image, image_clear_image_1)
(image_set_crop, image_set_size, image_set_rotation)
(image_create_x_image_and_pixmap) [USE_CAIRO]: #ifdef out HAVE_XRENDER part.
* src/image.c (cr_create_surface_from_pix_containers) [USE_CAIRO]: Rename
from cr_create_surface_from_pix_containers. Change arguments to pair of
Emacs_Pix_Container:s. Move block_input and unblock_input to caller.
(cr_put_image_to_cr_data) [USE_CAIRO]: New function.
(prepare_image_for_display) [USE_CAIRO]: Use it.
(image_set_transform) [USE_CAIRO]: Create dummy solid color pattern equipped
with transformation matrix and set it to img->cr_data.
* src/xterm.c (fringe_bmp) [USE_CAIRO]: Change type to cairo_pattern_t **.
(x_cr_define_fringe_bitmap, x_cr_destroy_fringe_bitmap) [USE_CAIRO]: Create
or destroy cairo pattern.
(x_cr_draw_image) [USE_CAIRO]: Remove arguments image_width and image_height.
Change type of image to cairo pattern. All callers changed.
* src/gtkutil.c (xg_get_image_for_pixmap) [USE_CAIRO]: Get cairo surface
from img->cr_data, which is of cairo pattern now.
* src/dispextern.h (struct face):
* src/xfaces.c (hash_string_case_insensitive, lface_hash)
(cache_face, lookup_face): Use uintptr_t for face hashes
instead of discarding the upper pointer bits on 64-bit machines.
9254885 (origin/emacs-26) Resurrect display-line-number-mode in clien...
aecbbd5 * src/fns.c (Fmapconcat): Doc fix. (Bug#35710)
8e5fc38 Fix typo
ee21b40 * lisp/term/w32-win.el ([noname]): Bind to 'ignore'. (Bug#36...
f68b33f Fix styling of Unicode codepoints in manuals
ff7ec6f Fix a few uses of quotes in user manual
b67042b More minor copyedits in the Emacs manual
9734b5c Fix minor issues in the Emacs manual
c153250 Try to improve text on atomic windows in Elisp manual
fb314ba Don't recommend insert-before-markers in process filters
* src/image.c (image_set_crop) [HAVE_NATIVE_TRANSFORMS]: Don't call
compute_image_size, as it is already called from image_set_size.
(image_set_size, image_set_crop) [HAVE_NATIVE_TRANSFORMS]: Use harmless
matrix transformation code also for USE_CAIRO and HAVE_NTGUI cases,
though image_set_transform is not yet implemented on them.
The copy_file_range syscall (introduced in Linux kernel
version 4.5) can copy files more efficiently via server-side
copy etc.
* admin/merge-gnulib (GNULIB_MODULES): Add copy-file-range.
* lib/copy-file-range.c, m4/copy-file-range.m4:
New files, copied from Gnulib.
* lib/gnulib.mk.in, m4/gnulib-comp.m4: Regenerate.
* src/fileio.c (Fcopy_file): Try copy_file_range first,
falling back on read+write only if copy_file_range failed or
if the input is empty and so could be a /proc file.
Also, do some refactoring to simplify code.
* src/dispextern.h (INIT_MATRIX, COPY_MATRIX, MULT_MATRICES): Remove.
* src/image.c (matrix3x3): New type, replacing all uses of 3x3 double.
(matrix3x3_copy, matrix3x3_mult): New functions, replacing
COPY_MATRIX, MULT_MATRICES. Replace INIT_MATRIX by C initializers.
(image_set_rotation): Use Fmod to avoid undefined behavior on
double-to-int conversion and to reduce bignum rotations correctly.
(image_set_crop): Finish up previous correction, by not re-setting
width and height if compute_image_size has set them.
Prefer shifting right by 1 to dividing by 2 if either will do.
* lisp/image.el (image--get-imagemagick-and-warn): Only fallback to
ImageMagick if native transforms aren't available.
* src/dispextern.h (INIT_MATRIX, COPY_MATRIX, MULT_MATRICES): New
macros for matrix manipulation.
(HAVE_NATIVE_SCALING, HAVE_NATIVE_TRANSFORMS): Rename and change all
relevant locations.
* src/image.c (x_set_image_rotation):
(x_set_transform): New functions.
(x_set_image_size): Use transform matrix for resizing under X and NS.
(x_set_image_crop): New function.
(lookup_image): Use the new transform functions.
(Fimage_scaling_p, Fimage_transforms_p): Rename and update all
callers.
* src/nsimage.m (ns_load_image): Remove rotation code.
(ns_image_set_transform): New function.
([EmacsImage dealloc]): Release the saved transform.
([EmacsImage rotate:]): Remove unneeded method.
([EmacsImage setTransform:]): New method.
* src/nsterm.h (EmacsImage): Add transform property and update method
definitions.
* src/nsterm.m (ns_dumpglyphs_image): Use the transform to draw the
image correctly.
* src/xterm.c (x_composite_image): Use PictOpSrc as we don't care
about alpha values here.
* doc/lispref/display.texi (Image Descriptors): Add :rotation.
(ImageMagick Images): Remove :rotation.
* doc/lispintro/emacs-lisp-intro.texi (Digression into C):
Adjust to match current C code.
* lisp/emacs-lisp/ert.el (ert--force-message-log-buffer-truncation):
Simplify.
* src/.gdbinit (Lisp_Object_Printer.to_string): Return
a string that says "make_fixnum", not "make_number".
Without this fix, Emacs can get into a tight loop reporting
a range error when calculating timestamps.
* doc/lispref/numbers.texi (Integer Basics):
* src/alloc.c (syms_of_alloc): Document this.
* src/bignum.c (make_bignum_bits): Always allow bignums
of at least twice the width of (u)intmax_t.
* src/hbfont.c (hbfont_shape): Use ptrdiff_t, not int,
for indexes that in theory might overflow int.
Add UNINIT to pacify GCC 9 and omit an unnecessary initialization.
134edc1 Warn about wrong number of args for subrs (Bug#35767)
5f01af6 Use plain symbols for eieio type descriptors (Bug#29220)
4b24b01 Pacify GCC 9 -Wredundant-decls
* src/w32fns.c (Fx_create_frame, w32_create_tip_frame):
Register the Uniscribe font backend only if HarfBuzz is not
available, or if explicitly requested via frame parameters
or resources.
* src/hbfont.c: New file, with code moved from w32uniscribe.c
and renamed/modified as appropriate.
* src/w32uniscribe.c: Move to hbfont.c DEF_DLL_FN and macro
definitions for HarfBuzz functions used in hbfont.c
(load_harfbuzz_funcs): Move loading of HarfBuzz functions used
by hbfont.c to hbfont.c:hbfont_init_w32_funcs, and call that
function from here.
(syms_of_w32uniscribe_for_pdumper): Fill the 'shape' and
'combining_capability' members with hbfont.c function names.
* src/w32common.h (hbfont_init_w32_funcs) [HAVE_HARFBUZZ]: Add
prototype.
* src/font.h (hbfont_shape, hbfont_combining_capability)
[HAVE_HARFBUZZ]: Add prototypes.
* src/Makefile.in (SOME_MACHINE_OBJECTS): Add hbfont.o.
* configure.ac (FONT_OBJ): Add hbfont.o if HAVE_HARFBUZZ.
* src/w32reg.c (w32_get_string_resource): The argument V_RDB
is a 'char **', not a 'char *'. This fixes -xrm handling on
MS-Windows, broken by conversion of x_get_string_resource to
terminal-specific hook.
(cherry picked from commit 833097cbc4)
* src/w32font.c (w32font_draw): Convert the glyph_string's
char2b array to 16-bit WCHAR data that ExtTextOutW needs.
(cherry picked from commit 38564f8a66)
* src/w32uniscribe.c [HAVE_HARFBUZZ]: Include math.h and
hb.h.
(bswap_32): Define for GCC 4.3.0 and later; else include
<byteswap.h> from Gnulib.
(struct uniscribe_font_info): Extend for HarfBuzz; 'cache' is
now a 'void *' (all users changed).
[HAVE_HARFBUZZ]: Define typedefs for HarfBuzz functions to be
loaded dynamically from the HarfBuzz DLL. Define macros to
call those functions via function pointers.
(uniscribe_open) [HAVE_HARFBUZZ]: Use the HarfBuzz font driver
if the type of the font entity is 'harfbuzz'.
(uniscribe_close) [HAVE_HARFBUZZ]: For fonts using the
HarfBuzz backend, call hb_font_destroy to free memory used for
the cached hb_font data.
(uniscribe_shape): Fix assignment of character codepoints to
glyphs from a single cluster.
(w32hb_list, w32hb_match, free_cb, w32hb_get_font_table)
(w32hb_get_font, w32hb_encode_char, w32hb_begin_font)
(w32uni_combining, w32uni_general, w32uni_mirroring)
(get_hb_unicode_funcs, w32hb_shape)
(w32hb_combining_capability, load_harfbuzz_funcs)
[HAVE_HARFBUZZ]: New functions.
(syms_of_w32uniscribe_for_pdumper) [HAVE_HARFBUZZ]: Load the
HarfBuzz DLL and register the HarfBuzz backend with its
functions.
* src/w32font.c (syms_of_w32font) <Qharfbuzz>: New DEFSYM.
* src/w32fns.c (Fx_create_frame, w32_create_tip_frame)
[HAVE_HARFBUZZ]: Register the harfbuzz font backend.
* src/lisp.h (get_unicode_property): Declare prototype.
* src/font.h (harfbuzz_font_driver) [HAVE_NTGUI]: Declare.
* src/chartab.c (get_unicode_property): New function, body
taken from get-unicode-property-internal.
(Fget_unicode_property_internal): Call get_unicode_property
after validating input.
* doc/lispref/frames.texi (Font and Color Parameters):
* doc/emacs/msdos.texi (Windows Fonts): Document support for
HarfBuzz text shaping on MS-Windows.
* configure.ac (HAVE_HARFBUZZ): Move out of the X-specific
part, and consider HarfBuzz also for HAVE_W32 systems.
Require HarfBuzz v1.2.3 for w32.
* 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.
