- add WRAP_AUTO and WRAP_TERM (for terminal emulator)
WRAP_AUTO wraps lines if the window is more than 3/4 of the screen width
if window is narrower, lines are truncated.
WRAP_TERM does not display the \ wrapping mark to simulate terminal behavior
- add default_wrap member in ModeDef structure
- delegate to view pane for man, ssh and compile commands
- fix invalid widths in unicode_width.h
- avoid setting s->offset to invalid values in hex.c
- update s->unihex_mode on the fly if larger code-points are displayed
- display code-points with at least 4 hex digits
- compute default s->dump_width more accurately
- qe_isaccent() recognises all combining code points (to be refined)
- add computer generated width table unicode_width.h
- display difference in hex in compare-windows to distinguish
space and nbsp, normal and combined form for accented letters.
- display multiple combining code points in what-cursor-position (C-x =)
- attach combining code points to preceeding space in display_char_bidir
- handle accents specificaaly in flush_fragment
- handle combining code-points in screen backstore via indirection table
this transient table can be inspected via describe-screen
- add QECharset.encode_table as a default for CharsetDecodeState.table
- make CharsetDecodeState.table const
- use static initializers for utf8 tables
- add al32utf8 charset name (used in Oracle for true utf-8 encoding)
- use direct pointers in unicode_glyph_range_index
- initialize unicode_glyph_range_index at run time
- rename unicode_glyph_tty_width as unicode_tty_glyph_width
- changed unihex display to space out single width characters
- remove `eb_nextc_style()`
- `eb_nextc()` does not modify `b->cur_style`
- add `eb_get_style()` to retrieve style for given offset in full-style buffers
- add `eb_next()` and `eb_prev()` to skip chars
- pass `offset` to `eb_get_line()` and `generic_get_colorized_line()`...
- use `eb_skip_chars()` in `do_delete_char()` and `text_write_char()`
- replaced `eb_get_strline()` with `eb_fgets()` with similar semantics
except return value is number of bytes before the newline or end of line
- no longer pass end offset to `isearch_colorize_matches()`
- just pass number of code points to `combine_static_colorized_line()`
- no longer call `isearch_colorize_matches` from `generic_get_colorized_line`
- rename method dispatcher `text_display` as `display_line`
rename corresponding mode based handlers as `xxx_display_name`
- rename method dispatcher `text_backward_offset` as `backward_offset`
- add default text handlers for `move_bof` and `move_eof` in
hex, binary and unihex modes
- handle utf-8 in window prompt (ex: default buffer name for C-x b)
- add various TODOs
- add mode specific `delete_bytes()` method
- reorder `write_char()` and `mouse_goto()` mode methods
- pass `keep` argument to `do_kill()`
- pass `argval` to `do_kill_line()`
- add `do_kill_beginning_of_line()` command
- allow kill commands in read-only buffers, copies contents to kill-ring and
moves point.
- change `do_kill_line()` semantics when invoked with numeric argument
- add `cur_prompt` and `last_char` in ShellState to track start of
shell input line
- C-a moves to start of shell input, exits interactive mode if there already
- force interactive mode on shell-end-of-line and shell-end-of-buffer
- simplify shell_write_char(), remove hard coded dispatcher
- add `shell_delete_bytes()` to delete contents in shell input line
- change shell-kill-word and shell-kill-line to not rely on mode specific
cursor movement commands because of asynchronous behavior. This is also
a problem for macros
- make shell-yank insert data via shell input.
- pause for 1ms after each CR sent to the process
- get rid of ModeSavedData structures
- save window data upon buffer change and mode setting
- make generic_mode_close() and generic_mode_init() local
- simplify edit_set_mode() and switch_to_buffer()
- reset w.insert, w.hex_mode, w.hex_nibble, w.unihex_mode, w.wrap
to mode specific default values upon mode change.
- split do_set_next_mode() command into qe_set_next_mode() lower level
function with an extra status parameter report new mode.
- added probe function for binary mode so it gets selected when cycling
modes with next-mode (bound to M-m)
- renamed w->disp_width as dump_width and make it a window variable
- compute default dump-width more accurately for dired-mode
- dump-width can be controlled with C-left and C-right in
hex, unihex and binary modes
* always regsister callbacks on s->top and s->top_offset
* handle saved_data generically, remove window callback
this will be replaced with view history mecanism.
* support 3 types of end of line: Unix, Dos and old style Mac
* detect eol type automatically in detect_charset
* set buffer eol_type together with charset
* transparently convert eol sequence to \n upon reading buffers
characters in eb_nextc().
* transparently convert \n to eol sequence in eb_encode_uchar()
* handle eol types in convert-file-buffer-coding-system: either preserve
current value or force it via charset suffix -unix, -doc and -mac
* change read_charset() to accept eol_type suffix
* add eol_type and eol_char to CharsetDecodeState
* change charset methods to take CharsetDecodeState instead of charset pointers
* display eol_type in mode line if not unix
* mode initializers should call text-mode-init to register ccallback
for s->offset and s->top_offset, and other generic stuff.
* this is error prone, should simplify this with base_mode property and
keep unmapped windows lingering to get rid of ModeSavedData handlers
* determine hex dump width from max char value in buffer
* display byte offset in margin instead of char offset
* accept 0x1f in text-mode
* no longer output octal char value if not ASCII
* filter control characters and non BMP1 code points
* new mode specific modeline
* do not space out single width glyphs (not very readable)
* eb_get_char_offset: add validation on offset argument
* eb_get_char_offset: align offset argument on character boundary
- default to overtype mode
- fixed "cursor not found" bug at end of file
- simplified hex dump code
- display both glyph offset and byte offset in unihex mode
qe_register_binding takes ModeDef* and returns status
cosmetic changes in QEDisplay API
added int dpy_init(QEditScreen *s, QEDisplay *dpy, int w, int h);
copies *dpy (or dummy_dpy if NULL) to QEditScreen and calls method
simplified dpy_init methods and calls
inlined 5 more methods
moved dummy_dpy to display.c and simplified
simplified to_hex
simplified qe_colors array handling
changed API for GetColorizedLineFunc: offset passed by address,
updated to start of next line
added check_readonly(EditState *s)
added tentative code for unregistring bindings, cycling modes and charsets
experiment alternate method for region highlighting:
added EditState.region_style and EditState.curline_style
added QEmacsState.hilite_region
added DisplayState.style
added clear_color to reset style bits in unicode array
display_char_bidir takes style from DisplayState, code point is bare.
fixed problem in text_display where line_num was not computed in shell mode
handle region coloring in text_display (tentative)
display line numbers and hex dump offsets as QE_STYLE_COMMENT
display hex dump as QE_STYLE_FUNCTION
added eb_find_new -> find a buffer by name, create one from flags if not found.
qe_register_cmd_table takes ModeDef* instead of mode name
do_global_set_key -> do_set_key with local/global flag
added local-set-key
reuse *bufed* buffer if one exists
bind increase/decrease-width commands from command definition
cosmetics, comments, remarks
renamed kill buffers from *yank* to *kill-%d*
save QErrorContext in parse_args, load_all_modules:
improves readability in *messages* buffer
fixed qe_init to fall back from x11 display to terminal session upon failure
renamed canonize to canonicalize
added crude support for ~user -> /home/user
added win32 specific stuff in basename
added C-/ as an alias for C-_
some cosmetics, moved code
create files with mode 0644 in raw_save_buffer
untabified, removed redundant spaces
modified code generators to the same effect
fixed bug in cfb32_fill_rectangle
use c99 syntax for new style struct initializers instead of gcc specific
simplified tolower unsigned char hacks
fixed various sparse warnings
utf8_encode now returns the number of characters copied,
simplified code accordingly
made local functions static
tagged unused function arguments
fixed shadowing issues
added support for GCC attributes for unused arguments and format strings
use snprintf() instead of sprintf()
moved scattered declarations to qe.h
removed duplicate declarations in header files
eb_write() payload argument is now const
fixed crash bug: reset trace_buffer upon deletion
improved trace_buffer readability: trace all pseudo tty streams separately
added missing structure initializers
removed buffered string handling from cutils.h
added new buffered string API for statically allocated buffers (should unify)
small improvements in X11 initialization
improved consistency in tty.c (function names...)
added STDIN and STDOUT streams in QEditScreen to stop using stdin and stdout
use unlocked stream I/O if available (faster)
changed and documented terminal setup and exit sequences
improved charset handling
mode_line functions now return the character count (like snprintf)
