KO Myung-Hun
89d461dd83
exec: search commands in the current directory first on OS/2
...
On OS/2, commands(executables) are searched in the current directory
before in $PATH.
modified: exec.c
2016-12-19 15:50:58 +09:00
KO Myung-Hun
20dbf6b45b
Read in text mode in a needed place only
...
Converting CR+LF to LF in blocking_read() which is a underlying read
function, may affect to the functions which do not perform line-based
operation.
modified: funcs.c
modified: main.c
modified: misc.c
modified: os2.c
modified: shf.c
2016-12-19 08:27:06 +09:00
KO Myung-Hun
18ff277047
Write in binary mode through pipes
...
On kLIBC, a child inherits a translation mode of stdio/stdout/stderr
of a parent. So if setting stdio handles of a parent to text mode,
a child reads CR+LF as LF and write LF to CR+LF. Especially, LF to
CR+LF conversion causes many troubles in UNIX programs.
Instead, if a child inherts a child inherit a current translation mode,
where stdin is text mode, stdout and stderr are binary mode, a child
would write LF as is.
This is the way for OS/2 to survive in UNIX jungle.
For example, 'make check' of libiconv fails due to mismatch of CR+LF
and LF.
2016-12-15 17:37:04 +09:00
KO Myung-Hun
2b9eed78a0
move os2_init() from main() to main_init()
2016-12-14 11:05:50 +09:00
KO Myung-Hun
224fd5aebe
Merge remote-tracking branch 'mksh/master'
...
Conflicts:
misc.c
2016-11-12 15:41:31 +09:00
KO Myung-Hun
cc49df380e
fix mksh_sdirsep() and simplify mksh_vdirsep() on OS/2
...
mksh_sdirsep() returned a position of a more far directory separator
if '/' and '\' were mixed.
And mksh_sdirsep() returned a position of ':' even if a directory
separator follows. For example, in case of 'x:/', the position of
'/' should be returned not ':'.
mksh_vdirsep() were simplified with mksh_sdirsep().
2016-11-12 15:23:24 +09:00
KO Myung-Hun
a2e965cad3
Merge remote-tracking branch 'mksh/master'
...
Conflicts:
Makefile
edit.c
eval.c
exec.c
main.c
misc.c
sh.h
var.c
2016-11-12 13:22:58 +09:00
tg
7d387fd54e
fix embarassing pasto and OS/2 UNC logic; thanks komh!
2016-11-12 03:54:48 +00:00
tg
6d5496a81c
make both cats
2016-11-12 00:20:37 +00:00
tg
0ba220d2c4
gcc 6.2 warnings (stupid uninitialised that wasn't, plus FALLTHROUGH fixes)
2016-11-11 23:48:30 +00:00
tg
7b4bee7e58
collective R54 release preparation multi-merger:
...
install both lksh and mksh manpages from Build.sh (Martijn Dekker)
spelling fixes (Larry Hynes)
manpage improvements (Martijn Dekker)
initial port to Harvey-OS’ APEX (Ronald G. Minnich, Elbing Miss, Álvaro Jurado)
more from komh’s OS/2 port (KO Myung-Hun)
2016-11-11 23:31:39 +00:00
tg
3aac3c30b5
return 128+SIGALRM if read with timeout timeouts, inspired by GNU bash
2016-11-11 22:17:09 +00:00
tg
945ef46af1
document print builtin changes
2016-11-11 21:37:44 +00:00
tg
6267ea36a4
fix print -cn
2016-11-11 21:37:35 +00:00
tg
6c04e4a665
make print -c honour line separator
2016-11-11 21:13:25 +00:00
tg
26689a7c81
just 'print -c' outputs not even a newline; plug a memleak while here
2016-11-11 20:53:15 +00:00
tg
2f63478bc9
restructure so that the trailing separator is output separately from
...
the inner line separators
2016-11-11 20:22:52 +00:00
tg
5ba6be5837
put print_columns options into a helper struct, eases passing things around
2016-11-11 20:14:18 +00:00
tg
e16b0b9d5b
first implementation of “print -c” for columnising, currently hardcoded newlines
2016-11-11 19:59:39 +00:00
tg
6a9dae2144
implement print -N: set output word and line separator to NUL
2016-11-11 19:18:40 +00:00
tg
8c2eda2f0f
implement print -l: change output word separator to newline
2016-11-11 19:12:52 +00:00
tg
d4b81df050
goto keeps us from evaluating the loop condition twice in immediate succession
...
(and shaves off an indentation level)
2016-11-11 19:09:44 +00:00
tg
1c07e75fc0
implement: empty arg in print -A is input word separator
2016-11-11 19:02:26 +00:00
tg
4ed09871a9
refactor c_print into a loop
2016-11-11 18:50:09 +00:00
tg
4e855b1be2
pack local bools into a struct, for tighter stack packing
2016-11-11 18:44:32 +00:00
tg
5ee8ba10b3
fix lazy evaluation with side effects; spotted by ormaaj via IRC
2016-11-07 16:58:48 +00:00
tg
7fef6b129e
apply patch from Brian Callahan <bcallah@devio.us> for better pcc support
2016-11-07 16:55:51 +00:00
tg
038cfe7476
shave five spaces off .rodata with no functional difference
2016-10-26 22:55:51 +00:00
tg
ac405dd6b7
avoid even the chance at UB, it’s too risky with “postmodern” compilers
2016-10-22 23:56:50 +00:00
tg
56bdf24e54
fix error propagation in TAND/TOR constructs, noted by Martijn Dekker
...
still a bit unsure about the whole XERROK and *xerrok stuff, but…
it seems to work
2016-10-02 22:21:47 +00:00
tg
9c602791b2
today, Andreas Buschka learnt that the Oxford comma is an american thing
...
(except for exactly the University of Oxford or to disambiguate)
2016-09-29 09:04:56 +00:00
tg
d18f9133c8
fix English (thanks to Andreas Buschka); TIL:
...
• to start ⇒ a start
• to begin ⇒ a beginning
2016-09-01 12:59:12 +00:00
tg
a838564cec
Unicode 9.0.0 (code and data part; FixedMisc font tbd)
2016-09-01 12:55:21 +00:00
KO Myung-Hun
46ed22a2bb
Preserve coding styles of upstream
...
modified: exec.c
2016-08-26 08:42:29 +09:00
tg
e5e3fb3fc7
outstanding bumps
2016-08-25 16:21:34 +00:00
tg
61969809e4
read(2) and write(2) don’t EINTR for fast input (LP#1616692)
2016-08-25 16:21:14 +00:00
tg
fa94f0339e
experiment with GCC 5’s new -malign-data=abi
2016-08-24 20:50:11 +00:00
tg
bb796ff785
simplify manpage and code; streamline manpage formatting and Satzzeichen
2016-08-24 20:40:00 +00:00
tg
35bd16ddfa
this bug has long been fixed
2016-08-24 19:36:26 +00:00
tg
91b5d4b3ad
remove all mentions of “the null string”, it’s an empty string (in most
...
cases), or an unset parameter (in some cases)
2016-08-24 19:35:26 +00:00
KO Myung-Hun
ef42532640
Merge tag 'mksh-R53a' into HEAD
2016-08-13 15:11:41 +09:00
KO Myung-Hun
d4be6d4358
Merge tag 'mksh-R53' into HEAD
...
Conflicts:
exec.c
2016-08-13 15:05:49 +09:00
root
61a50ea6ba
This commit was manufactured by cvs2svn to create tag 'mksh-R53a'.
...
Sprout from master 2016-08-12 16:48:05 UTC tg <tg@mirbsd.org> 'fix refactoro, spotted by Natureshadow; add (working) testsuite for all operators'
Delete:
Makefile
2016-08-12 16:48:06 +00:00
tg
879c6a0911
fix refactoro, spotted by Natureshadow; add (working) testsuite for all operators
2016-08-12 16:48:05 +00:00
root
fe672422e7
This commit was manufactured by cvs2svn to create tag 'mksh-R53'.
...
Sprout from master 2016-08-10 18:20:18 UTC tg <tg@mirbsd.org> 'bump, required by XTaran'
Delete:
Makefile
2016-08-10 18:20:19 +00:00
tg
2db6d0a08c
bump, required by XTaran
2016-08-10 18:20:18 +00:00
tg
bb7226277b
fixes
2016-08-10 18:20:05 +00:00
tg
b58ff810a5
mention bases can (naturally…) only go up to 36, requested by izabera
2016-08-10 18:15:18 +00:00
tg
f26cf0562b
fixup a bad OpenBSD reaction on a bug afl showed:
...
set source to NULL only if the memory backing source is actually reclaimed;
fixes segfault due to NULL(+24) pointer dereference reported by Score_Under
(simplified testcase added; thanks!)
2016-08-04 20:51:35 +00:00
tg
e52a2bb23f
assorted code cleanup, while here anyway
2016-08-04 20:32:14 +00:00