PR 56983 Izumi Tsutsui: fix confusing message in test(1) with -DSMALL

s/existance/existence/ in comment.

Allow SMALL (and TINY) builds of test (for SMALL/TINY builds of sh)
which support only the defined modes of operation of test, to allow
the version of sh on small install media be kept as small as po

Allow SMALL (and TINY) builds of test (for SMALL/TINY builds of sh)
which support only the defined modes of operation of test, to allow
the version of sh on small install media be kept as small as possible.

show more ...

PR standards/34646

Make test(1) always use the POSIX "number of args" evaluation rules
when they apply.

Only fall back to the old expression evaluation when there are more
than 4 args, or when the

PR standards/34646

Make test(1) always use the POSIX "number of args" evaluation rules
when they apply.

Only fall back to the old expression evaluation when there are more
than 4 args, or when the args given cannot work as a test expression
using the POSIX rules. That is when the result is unspecified.

Also fix old bug where a string of whitespace is considered to be a
valid number (at least one digit is needed amongst it somewhere...)

XXX pullup -8

show more ...

Drop main() prototype.

test(1): Compare timestamp in nsec scale in -nt/-ot.

Add __printflike attribution to use vprintf and friends with an argument
as format string.

Use __dead

bin/39446: Aleksey Cheusov: always produce errors on empty numbers.

make getn return long long to support large numbers on all archs.

PR/38755: murray armfield: /bin posix programs missing setprogname and/or
setlocale

convert __attribute__s to applicable cdefs.h macros

PR/36534: Greg A. Woods: test(1) doesn't do -r and -w quite right
Use stat(2) instead of access(2) per POSIX.

Make sure we don't run over the end of the argument array while scanning
for input. This can happen if we have a unary not without an argument. When
we scan for the argument, we are already at the NU

Make sure we don't run over the end of the argument array while scanning
for input. This can happen if we have a unary not without an argument. When
we scan for the argument, we are already at the NULL element of the argument
array. Then when we scan ahead for a -a or -o, we end up testing the next
element after the NULL.

show more ...

PR/36092: B K: algorithmic inefficiency in bin/test/test.c:t_lex
test(1) scans for "operators" linearly in an array using strcmp() to
find a match. Since the list of "operators" is fixed, split them

PR/36092: B K: algorithmic inefficiency in bin/test/test.c:t_lex
test(1) scans for "operators" linearly in an array using strcmp() to
find a match. Since the list of "operators" is fixed, split them
into one and two character ones, and ones that start with a `-' and
ones they don't. This way we can optimize the compare function to
just check for one or two characters. Sort and use bsearch(3). We
could have used a single sorted array and bsearch(3), to save some
complexity, but I decided to be a bit fancier.

show more ...

Revert previous (==) - not worth the trouble.

Allow '==' as synonym to '=', just like our ksh and bash do.
This also affects the sh(1) builtin test command.

don't call get/set progname when we are a shell built-in.

use getprogname() instead of argv[0], because some programs that execv, pass
the full path in argv[0] instead of just the basename. In reality, those
programs should be fixed (hello gdb for example).

KNF, use NULL instead of (type *)0.

__STDC__ is always defined on NetBSD.

Some KNF fixes by Petri Koistinen and me.

ANSIfy and slight KNF. Mostly #13594 by Petri Koistinen.

- Enable test to be a shell builtin
- KNF
- Fix bug where test without arguments could access invalid memory

Operator precedence consistency; from "Andrew_L. Moore" <alm@SlewSys.Org>
in PR bin/7015.