summaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2021-04-14fix "warning array subscript has type 'char'"Denys Vlasenko
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-14Makefile.flags: add a test for -lrt availabilityDenys Vlasenko
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-14echo: fix !ENABLE_FEATURE_FANCY_ECHO buildDenys Vlasenko
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-14fix gcc-11.0 warningsDenys Vlasenko
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-14*: --help text tweaksDenys Vlasenko
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-14touch: shorten --help textDenys Vlasenko
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-14touch: code shrinkDenys Vlasenko
function old new delta touch_main 424 421 -3 Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-14touch: fix -amDenys Vlasenko
function old new delta touch_main 414 424 +10 Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-13dd: support iflag=count_bytesRafał Miłecki
It allows passing amount of bytes in the count= function old new delta packed_usage 33599 33617 +18 static.iflag_words 29 41 +12 dd_main 1601 1607 +6 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 3/0 up/down: 36/0) Total: 36 bytes Signed-off-by: Rafał Miłecki <rafal@milecki.pl> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-13vi: allow writing to another file if this one is readonlyAlison Winters
Version 2. Same change but rebased after Ron's improvements. Fixes bug where if you open a read only file, you can't save it as a different filename. function old new delta colon 3160 3162 +2 Signed-off-by: Alison Winters <alisonatwork@outlook.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-13build system: avoid build failure during bisectionRon Yorston
Commit 4bdc914ff (build system: fix compiler warnings) added a test on the return value of fgets() in split-include.c. During bisection it's possible to go back to a state where a configuration value didn't exist. This results in an empty include file corresponding to the missing feature. If a subsequent bisection returns to a state where the feature exists split-include treats the empty file as an error and the build fails. Add a call to ferror() to distinguish between fgets() failing due to an error and due to there being no data to read. Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-13hush: beautify ^D handling to match ash / bashDenys Vlasenko
function old new delta fgetc_interactive 227 244 +17 Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-13watchdog: make open-write-close-open functionality a config knobRasmus Villemoes
The behaviour introduced by commit 31c765081dc4 ("watchdog: stop watchdog first on startup") causes warnings in the kernel log when the nowayout feature is enabled: [ 16.212184] watchdog: watchdog0: nowayout prevents watchdog being stopped! [ 16.212196] watchdog: watchdog0: watchdog did not stop! The latter may also appear by itself in case the watchdog is of the type that cannot be stopped once started (e.g. the common always-running gpio_wdt kind). These warnings can be somewhat ominous and distracting, so allow configuring whether to use this open-write-close-open sequence rather than just open. Also saves a bit of .text when disabled: function old new delta shutdown_on_signal 31 58 +27 watchdog_main 339 306 -33 shutdown_watchdog 34 - -34 ------------------------------------------------------------------------------ (add/remove: 0/1 grow/shrink: 1/1 up/down: 27/-67) Total: -40 bytes Make it default n: - It's a workaround for one specific type of watchdog (and that seems to be a defect in the kernel driver) - Even when not enabled in busybox config, it can easily be implemented outside busybox - Code size - Commit 31c765081dc4 should be considered a regression for all the boards that now end up with KERN_CRIT warnings in dmesg. - The author of that commit said "This use case is evidently rare, so if it is indeed causing problems for other people I'd OK then I understand whatever needs to be done." in the v1 thread. Cc: Matt Spinler <mspinler@linux.vnet.ibm.com> Cc: deweloper@wp.pl Cc: tito <farmatito@tiscali.it> Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-13touch: fix previous commitDenys Vlasenko
function old new delta touch_main 423 414 -9 Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-13touch: make FEATURE_TOUCH_NODEREF unconditionalDenys Vlasenko
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-13touch: code shrinkDenys Vlasenko
function old new delta touch_main 450 423 -27 Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-13touch: code shrinkDenys Vlasenko
function old new delta .rodata 103215 103218 +3 touch_main 460 450 -10 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 1/1 up/down: 3/-10) Total: -7 bytes Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-13touch: prevent usage of -r and -t at onceXabier Oneca
coreutils forbids this combination. Signed-off-by: Xabier Oneca <xoneca@gmail.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-13touch: switch to using utimensat() and futimens()Denys Vlasenko
This patch changes the functions used to update timestamps in touch. Before, utimes() and lutimes() were used, which had certain disadvantages. They are unable to handle nanosecond timestamps, and implementations of certain features like -a and -m require running stat() in a loop. Almost all implementations of utimes() and lutimes() are wrappers for utimensat(), this is the case for glibc, ulibc and musl libc. function old new delta __futimens_time64 - 24 +24 __lutimes_time64 80 - -80 touch_main 539 456 -83 ------------------------------------------------------------------------------ (add/remove: 2/2 grow/shrink: 0/1 up/down: 24/-163) Total: -139 bytes Signed-off-by: urmum-69 <urmum69@snopyta.org> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-12touch: unbreak -hDenys Vlasenko
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-12touch: remove unneeded GETOPT32 definesXabier Oneca
Long options handling (getopt32 vs getopt32long) is done in libbb.h, no need to care here of the same logic. This cleans the code a bit. Also, --no-create was grouped as a SUSv3 option, where as the short -c was not. Even if it is part of SUS, leave it out as was the short option. v2: Fix for disabled ENABLE_LONG_OPTS. getopt32long does not like IF_FEATURE_xxx() style conditionals... :/ Signed-off-by: Xabier Oneca <xoneca@gmail.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-12touch: fix atime set from reference fileXabier Oneca
When using a file's times as reference, use both atime and mtime for the files to be modified. Signed-off-by: Xabier Oneca <xoneca@gmail.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-12touch: add SUSv3 options -a and -mXabier Oneca
Add missing -a and -m options to be fully SUSv3 compliant. function old new delta touch_main 415 510 +95 packed_usage 33824 33865 +41 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 2/0 up/down: 136/0) Total: 136 bytes v2: Ignore -a/-m if not ENABLE_FEATURE_TOUCH_SUSV3. Signed-off-by: Xabier Oneca <xoneca@gmail.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-11lineedit: fix tab completion with equal signNatanael Copa
Fix tab completion for the path when equal sign (=) is used. For example: dd if=/dev/ze<tab> Signed-off-by: Natanael Copa <ncopa@alpinelinux.org> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-11vi: improvements to reporting of changesRon Yorston
Traditional vi is mostly silent about the results of yank, delete, change, undo or substitution commands. Vim reports some details about undo and substitution. BusyBox vi is positively verbose in comparison. Make some improvements to BusyBox vi: - Add vim-like reporting of changes caused by substitutions, of the form '64 substitutions on 53 lines'. This replaces a fairly useless report of the result of the last change made. - Ensure that the report about put operations correctly reflects the newly introduced repetition count. - Commit 25d2592640 tried to limit status updates for delete and yank operations by detecting whether the register had changed. This didn't always work because the previously allocated memory could be reused for the new register contents. Fix this by delaying freeing the old register until after the new one has been allocated. - Add a configuration option to control verbose status reporting. This is on by default. Turning it off make BusyBox vi as taciturn as traditional vi and saves 435 bytes. function old new delta colon 3212 3292 +80 yank_status - 74 +74 static.text_yank 99 86 -13 string_insert 130 76 -54 do_cmd 4842 4776 -66 ------------------------------------------------------------------------------ (add/remove: 1/0 grow/shrink: 1/3 up/down: 154/-133) Total: 21 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-11vi: issue a warning on failure to find a characterRon Yorston
When a search for a character within a line fails issue a warning. function old new delta ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 0/0 up/down: 0/0) Total: 0 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-11vi: make the substitute command more like viRon Yorston
Make the ':s/find/replace/g' command behave more like vi: - the final delimiter is optional if no flag is specified; - the cursor is moved to the first visible character of the last line where a substitution was made; - a warning is displayed if no substitution was made. function old new delta colon 3156 3212 +56 .rodata 105133 105142 +9 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 2/0 up/down: 65/0) Total: 65 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-11vi: make put commands more like viRon Yorston
Make the put commands 'p' and 'P' behave more like vi: - allow a repetition count to be specified; - when the text being inserted doesn't include a newline the cursor should be positioned at the end of the inserted text. function old new delta do_cmd 4765 4842 +77 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 1/0 up/down: 77/0) Total: 77 bytes v2: Don't break build when FEATURE_VI_UNDO is disabled. Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-11vi: allow the '.' command to have a repetition countRon Yorston
The '.' command repeats the last text change. When it has a repetition count replay the change the number of times requested. Update the stored count if it changes. For example, 3dw deletes 3 words . deletes another 3 words 2. deletes 2 words and changes the stored count . deletes 2 words function old new delta do_cmd 4746 4781 +35 .rodata 105133 105138 +5 edit_file 887 849 -38 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 2/1 up/down: 40/-38) Total: 2 bytes v2: Change implementation to include repetition count in string. Otherwise repeating 'r' doesn't work properly. Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-11vi: allow 'r' command to be aborted, repeatedRon Yorston
Make the 'r' command behave more like vi: - abort the command if ESC is entered after the 'r'; - allow a repeat count to be entered before the 'r'; - if the repeat count exceeds the space available on the line don't change any characters and issue an alert. function old new delta do_cmd 4679 4746 +67 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 1/0 up/down: 67/0) Total: 67 bytes v2: Don't break build when FEATURE_VI_UNDO is disabled. Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-11vi: improvements to undoRon Yorston
The left shift operator ('<') didn't support undo at all; right shift ('>') required changes to be undone separately for each line. Allow both types of shift to be undone as a single operation. Also, neither traditional vi nor vim yank the lines being shifted by the '<' and '>' commands, so remove that call to yank_delete(); When a repetition count was specified for the '~', 'x', 'X' or 's' commands the changes had to be undone one character at a time. Allow undo as a single operation (though the delete and change parts of the 's' command still have to be undone separately). function old new delta undo_push_insert 37 40 +3 do_cmd 4695 4663 -32 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 1/1 up/down: 3/-32) Total: -29 bytes v2: Don't break build when FEATURE_VI_UNDO is disabled. Don't reset 'undo_del' too early in '~' handling code. Code shrink '~'. Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-11vi: more fixes to range selection by wordRon Yorston
An example in my vi book presents different ways to fix the spelling of the last word in a line: ... anyweigh. With the cursor on the 'e' the command 'cway' should do the job. Since commit 776b56d77, though, 'cw' incorrectly includes the full stop in the range if we're on the last line of the file. (Prior to commit 776b56d77 BusyBox vi got 'cw' right in this case but 'cW' wrong: it *didn't* delete the full stop.) Reinstate some of the bloat removed by the earlier commit to fix this. Also, commit 7b4c2276a (vi: fix word operations across line boundaries) incorrectly ignores whitespace after a single character word. Adjust the condition to avoid this. function old new delta find_range 707 737 +30 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 1/0 up/down: 30/0) Total: 30 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-11vi: improvements to range selectionRon Yorston
Rewrite find_range(), pushing quite a bit of code from do_cmd() down into it. - The commands 'y', 'd', 'c', '<' and '>' can be given twice to specify a whole-line range. BusyBox vi actually accepted any second character from that group, e.g. 'dc' or '<y', with the latter being accepted even if yank was disabled. Require the two characters to match. - '<' and '>' commands followed by ESC incorrectly issued an alert. - Allow search commands and a marker (specified as "y'a", for example) to define a range for those operators that support it. function old new delta find_range 518 707 +189 .rodata 105119 105133 +14 get_motion_char 68 - -68 do_cmd 4860 4695 -165 ------------------------------------------------------------------------------ (add/remove: 0/1 grow/shrink: 2/1 up/down: 203/-233) Total: -30 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-11vi: code shrink search commandsRon Yorston
Changes to search commands ('/', '?', 'n' and 'N'): - Rewrite to be smaller and (possibly) clearer. - Issue a warning when a repeat search is requested without a previous search having been made. Vim and BusyBox vi support a repetition count for searches though the original vi doesn't. If the count exceeds the number of occurrences of the search string the search may loop through the file multiple times. function old new delta .rodata 105135 105119 -16 do_cmd 4898 4860 -38 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 0/2 up/down: 0/-54) Total: -54 bytes Signed-off-by; Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-04-08vi: changes to option handlingRon Yorston
Since commit 70ee23399 (vi: code shrink) the ':set' command is unable to process multiple options on a line. Fix this by temporarily null-terminating each option. Change the default setting for all options to off to match vim. Actually, 'flash' isn't an option in vim, only traditional vi, where it's on by default. In vim the corresponding option is 'visualbell' which defaults to off. POSIX doesn't have either of these. Allow the abbreviation 'ts' for the 'tabstop' option. Issue an error message if: - an option is not implemented - an option that takes a value has no '=' or has a 'no' prefix - a boolean option has a '=' function old new delta colon 2944 3003 +59 .rodata 103171 103189 +18 vi_main 274 270 -4 setops 73 - -73 ------------------------------------------------------------------------------ (add/remove: 0/1 grow/shrink: 2/1 up/down: 77/-77) Total: 0 bytes v2: Try harder to detect invalid options. Thanks to Peter D for pointing this out. Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-03-30vi: deal with invalid movements in shift commandsRon Yorston
Since commit 25d259264 (vi: make buffer handling more vi-like) find_range() can return early when an invalid movement is specified. The call to find_range() in the code that handles shift commands ('<' and '>') doesn't check for this condition. Previously this only resulted in the current line being shifted but it can now result in a segfault. Check for an invalid movement and notify the user without taking any further action. function old new delta do_cmd 4890 4898 +8 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 1/0 up/down: 8/0) Total: 8 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-03-29vi: code shrinkRon Yorston
- In the '+' and '-' commands the call to dot_skip_over_ws() is only needed for the final line processed so it can be moved out of the while loop. - Marking sync_cursor() NOINLINE doesn't seem to offer the same advantages it did in 2009 (commit adf922ec2). function old new delta refresh 694 974 +280 do_cmd 4900 4887 -13 sync_cursor 336 - -336 ------------------------------------------------------------------------------ (add/remove: 0/1 grow/shrink: 1/1 up/down: 280/-349) Total: -69 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-03-29vi: improve operations involving paragraph movementRon Yorston
Movement by paragraph doesn't always involve whole lines. If the cursor is positioned in the middle of a line deleting to either end of the paragraph will result in one partial line and zero or more full lines. Adjust the end of ranges delimited by paragraph movement to more closely match what vi does. function old new delta find_range 467 518 +51 at_eof - 49 +49 ------------------------------------------------------------------------------ (add/remove: 1/0 grow/shrink: 1/0 up/down: 100/0) Total: 100 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-03-29vi: improve motion by paragraphRon Yorston
When moving by paragraph ('{' and '}'): - Treat multiple empty lines as a single paragraph separator. - When no paragraph separator is found move to the start or end of the file depending on the direction of motion. function old new delta do_cmd 4821 4900 +79 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 1/0 up/down: 79/0) Total: 79 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-03-29vi: allow repetition count for paragraph motionRon Yorston
The paragraph motion commands '{' and '}' should accept a count. function old new delta do_cmd 5054 5071 +17 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 1/0 up/down: 17/0) Total: 17 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-03-29vi: allow motion count for change/delete/yank/shiftRon Yorston
The motion that determines the range of a change, delete, yank or shift operation can have its own count. Thus the commands '5dd' and 'd5d' are equivalent: both delete 5 lines. When the command itself also has a count the two values are multiplied. Thus the command '2d3w' deletes 6 words and '2D3G' deletes from the current line to line 6. (When dealing with structured data it might make sense to think in units of 3 words so '2d3w' is deleting 2 such units. It doesn't seem quite as sensible to express 'delete from current line to line 6' as '2D3G' but vi permits it.) function old new delta get_motion_char - 68 +68 ------------------------------------------------------------------------------ (add/remove: 1/0 grow/shrink: 0/0 up/down: 68/0) Total: 68 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-03-29vi: support more commands for range selectionRon Yorston
Add 'F', 'T' and '|' as commands that can be used to specify a range for change/delete/yank operations. function old new delta .rodata 105129 105135 +6 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 1/0 up/down: 6/0) Total: 6 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-03-29vi: improvements to character search within lineRon Yorston
- Use a common routine to handle all commands to search for a character in a line. - When searching for the nth occurrence of a character don't move the cursor if fewer than n occurrences are present. - Add support for the 'T' command, search backwards for character after next occurrence of given character. function old new delta do_cmd 4861 4805 -56 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 0/1 up/down: 0/-56) Total: -56 bytes v2: Add parentheses to avoid searches continuing past end of line. Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-03-29vi: remember cursor column during vertical motionRon Yorston
When the 'j'/'k' commands or up/down arrow keys are used to move the cursor vertically 'vi' remembers the original cursor column and positions the cursor there if possible. Also, if the '$' command has been used to position the cursor at the end of a line vertical movements keep the cursor at the end of the line. Make BusyBox 'vi' do the same. function old new delta refresh 674 694 +20 do_cmd 4853 4861 +8 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 2/0 up/down: 28/0) Total: 28 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-03-29vi: 'G'/'M' commands move to first visible characterRon Yorston
The 'G' command with no target (meaning 'go to last line') should position the cursor on the first visible character of the line, as it already does in other cases. The 'M' command should position the cursor on the first visible character (as 'H' and 'L' already do). function old new delta do_cmd 4842 4853 +11 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 1/0 up/down: 11/0) Total: 11 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-03-29vi: don't overwrite existing fileRon Yorston
If the name of the file being written doesn't match the current filename and the output file already exists vi should issue a warning and not overwrite the file. Because the test only compares the file names it's somewhat over- protective. If the current file name is 'my_text' and the user tries to save to './my_text' they'll be prevented from doing so. function old new delta colon 3092 3151 +59 .rodata 105118 105146 +28 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 2/0 up/down: 87/0) Total: 87 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-03-29vi: make buffer handling more vi-likeRon Yorston
Vi places text affected by change/delete/yank operations into a buffer. The contents of such buffers can be restored with the put commands, 'p' or 'P'. These behave differently depending on whether the buffer contains whole lines or partial lines. For whole lines the text is copied into the file on the line before (P) or after (p) the current line. For partial lines the text is copied before or after the current cursor position. Whether an operation results in whole or partial lines depends on the command used. BusyBox vi treats any buffer with a newline as though it contained whole lines. This is incorrect. Deleting multiple words across a line boundary results in a buffer with a newline but not having whole lines. Rework how buffers are handled to behave more like vi. function old new delta static.text_yank 79 99 +20 colon 3092 3097 +5 edit_file 885 887 +2 yank_delete 127 112 -15 .rodata 105139 105101 -38 find_range 514 467 -47 do_cmd 5088 4842 -246 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 3/4 up/down: 27/-346) Total: -319 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-03-29vi: code shrinkRon Yorston
I was puzzled by code in find_range() which handles forward word movement. It included a test to see if we're at the start of a word. Since these are forward word movements surely we'd expect to be at the start of a word? In fact, the test was intended to fix a problem with changes to the last word in a file, as discussed in the thread starting here: http://lists.busybox.net/pipermail/busybox/2004-January/044552.html The code can be simplified by testing directly for end of file instead of indirectly for not being at the start of a word. Since trailing whitespace is now handled in do_cmd() the code to back up off a newline is no longer required. function old new delta find_range 619 514 -105 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 0/1 up/down: 0/-105) Total: -105 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-03-29vi: fix word operations across line boundariesRon Yorston
Commit 4b49422a0 (vi: fix changes to word at end of line. Closes 11796) fixed a problem where an operation on a word at the end of a line followed by a line starting with whitespace incorrectly joined the lines. However it also broke the case where operating on multiple words across a line boundary *should* join the lines. Fix this by detecting when trailing whitepace in a word operation includes a newline. Whitespace beyond the newline is excluded from consideration. function old new delta do_cmd 5083 5088 +5 ------------------------------------------------------------------------------ (add/remove: 0/0 grow/shrink: 1/0 up/down: 5/0) Total: 5 bytes Signed-off-by: Ron Yorston <rmy@pobox.com> Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2021-03-26ntpd: code shrink (force not-inlining, stop returning structs)Denys Vlasenko
function old new delta d_to_sfp - 133 +133 lfp_to_d - 84 +84 sfp_to_d - 78 +78 d_to_lfp 141 137 -4 .rodata 103182 103174 -8 recv_and_process_peer_pkt 2380 2173 -207 recv_and_process_client_pkt 706 493 -213 ------------------------------------------------------------------------------ (add/remove: 3/0 grow/shrink: 0/4 up/down: 295/-432) Total: -137 bytes Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>